FOREST ENTOMOLOGY Lethal and Sublethal Effects of Imidacloprid on (: Adelgidae) and Two Introduced Predator Species

1,2 1 1 3 BRIAN M. EISENBACK, SCOTT M. SALOM, LOKE T. KOK, AND ANTHONY F. LAGALANTE

J. Econ. Entomol. 103(4): 1222Ð1234 (2010); DOI: 10.1603/EC09270 ABSTRACT Eastern hemlock, Tsuga canadensis (L.) Carrie`re, branchlets were systemically treated with imidacloprid and bioassayed with hemlock woolly adelgid, tsugae (Annand) (Hemiptera: Adelgidae), and predators. After 10 d, two hemlock woolly adelgid predators, nigrinus Fender (Coleoptera: Derodontidae) and Sasajiscymnus tsugae Sasaji & McClure (Coleoptera: Coc- cinellidae), were allowed to feed on remaining hemlock woolly adelgid for 20 d on branches systemically treated with 1, 10, or 100 ppm imidacloprid. Every 5 d, mortality, mobility (measured as ßip time), number of hemlock woolly adelgid consumed, and degree of intoxication of each individual beetle were recorded. Liquid chromatography-mass spectrometry was used to quantify imidacloprid and some of its major metabolites in hemlock wood tissues and in the predator beetles postmortem. Probit analysis of hemlock woolly adelgid mortality and imidacloprid concentrations recovered from

branch wood tissues determined the 30 d LC50 to be 242 ppb. A topical application of imidacloprid to the ventral abdomen of individual beetles resulted ina6dLD50 value of 1.8 and 0.71 ng imidacloprid per beetle for L. nigrinus and S. tsugae, respectively. In no-choice tests, L. nigrinus mortality was signiÞcantly higher on hemlock branchlets treated with 100 ppm imidacloprid than on controls, but S. tsugae mortality was not. S. tsugae consumed the same number of adelgids on treated branchlets as on controls, but L. nigrinus consumed fewer adelgids from the 100 ppm branchlets than on controls. In choice tests, beetle mortality and ßip times were generally not signiÞcantly different from controls. At times, both beetle species displayed intoxication symptoms after feeding on adelgids from treated branchlets and imidacloprid was recovered from both beetle species postmortem. These results suggest that systemic imidacloprid displayed both lethal and sublethal effects on these two nontarget predators of the hemlock woolly adelgid.

KEY WORDS Tsuga canadensis, Laricobius nigrinus, Sasajiscymnus tsugae, nontarget, prey-mediated effects

Hemlock woolly adelgid, Adelges tsugae (Annand) avian species (Reay 1999, Yamasaki et al. 1999, Snyder (Hemiptera: Adelgidae), is an exotic invasive pest of et al. 2002, Tingley et al. 2002, Ross et al. 2003). two hemlock (Tsuga) species in the eastern United Hemlock woolly adelgid feeds on ßuids and nutri- States. Carolina hemlock, Tsuga caroliniana En- ents from parenchyma cells at the base of hemlock gelmann, is a relatively rare species conÞned to moun- needles (McClure 1987, 1991; Shields et al. 1995; tain ridges and isolated stands in the Appalachian Young et al. 1995). Infestation causes desiccation, bud highlands of western North Carolina, Tennessee, and mortality, needle loss, and a reduction of new growth Virginia. Eastern hemlock, Tsuga canadensis (L.) Car- (McClure 1987, McClure et al. 2001) and can lead to rie`re, is common along streams and cool, mountain dieback and tree death in 4Ð10 yr (McClure et al. slopes throughout eastern forests; its range extends 2001). Hemlock woolly adelgid has an anholocyclic from Maine to the southern extent of the Appalachian life cycle with two parthenogenic generations per year chain in northern Georgia and Alabama (Godman and on hemlocks. The asexual sistentes generation is Ϸ1 Lancaster 1990). Eastern hemlocks are an important mm in length and secretes a woolly white ßocculence economic (Rhea 1995, Holmes et al. 2005) and eco- or ovisac. Sistentes Þrst-instar crawlers aestivate on logical species; they provide habitats and nutrients hemlock stems throughout the summer. Breaking aes- that are beneÞcial for many aquatic, terrestrial, and tivation in the fall, they begin feeding and develop through four instars during the winter. Adults mature and lay eggs in March and April, which hatch as either 1 Department of Entomology, Virginia Polytechnic Institute and progredientes or sexuparae. Sexuparae production is State University, Blacksburg, VA 24061. initiated by high densities of sistentes that cause poor 2 Corresponding author: Department of Biology, Bryan College, host nutritional quality (McClure 1991). Sexuparae Dayton, TN 37321 (e-mail: [email protected]). 3 Department of Chemistry, Villanova University, Villanova, PA are winged and sexual; they do not feed on hemlocks 19085. but disperse to search for suitable (Picea) spe-

0022-0493/10/1222Ð1234$04.00/0 ᭧ 2010 Entomological Society of America August 2010 EISENBACK ET AL.: EFFECTS OF IMIDACLOPRID ON HEMLOCK WOOLLY ADELGID 1223 cies to complete an holocyclic life cycle (Havill and 1999). For example, in cotton, Gossypium hirsutum L., Foottit 2007). The progredientes generation is similar metabolites protect the plant from whiteßies even in body morphology to the sistentes. After hatching, after parent compound concentrations declined Þrst instar crawlers disperse, settle, and begin to feed (Nauen et al. 1999). at the base of needles. The progredientes develop Effective control of hemlock woolly adelgid can be rapidly through four instars, mature, and lay eggs. Eggs achieved when imidacloprid is applied by foliar sprays hatch into sistentes crawlers which migrate to new or through systemic trunk or soil injections (McClure growth, settle, and aestivate through the summer. et al. 2001, Doccola et al. 2003, Webb et al. 2003, Laricobius nigrinus Fender (Coleoptera: Derodon- Cowles et al. 2006). Trunk and soil injections are the tidae) and Sasajiscymnus (Pseudoscymnus) tsugae primary control techniques for hemlock woolly adel- Sasaji & McClure (Coleoptera: Coccinellidae) are two gid in large-scale injection programs in national forests predators that have been reared and released as bio- and state parks. The Great Smoky Mountains National logical control agents of hemlock woolly adelgid Park in northeastern Tennessee is an example of a (Cheah and McClure 2002, Cheah et al. 2005, Lamb et current initiative to limit hemlock decline from hem- al. 2006). L. nigrinus is native to the PaciÞc Northwest lock woolly adelgid infestation. In the park, 466,843 and is a specialist predator of hemlock woolly adelgid. S. tsugae and 7,857 L. nigrinus) have been released at It is univoltine, host speciÞc and closely synchronized 32 and 191 sites, respectively, and Ͼ140,000 hemlocks with hemlock woolly adelgidÕs life cycle (Zilahi- have been treated via soil or trunk injections of imi- Balogh et al. 2002, 2003, Mausel et al. 2010). Adults dacloprid as of 2010 (K. Johnson, personal communi- emerge from the soil and begin feeding on sistentes in cation). the fall after the adelgids break aestivation. L. nigrinus In settings where chemical and biological controls feeds throughout the winter. The adults are cold hardy are jointly applied, it is important to determine the and active at temperatures above 0ЊC (Mausel 2007). compatibility of the two control options. High-value In early spring, adults begin laying eggs; larvae develop hemlock stands are often the Þrst candidates for chem- through four instars and feed primarily on hemlock ical control, and these same stands are desired as sites woolly adelgid eggs. In late spring, larvae drop from for biological control predator releases. Often preda- the hemlock branches into the soil where they pupate tor releases are in proximity to chemically treated and aestivate through the summer. L. nigrinus can sites, occupying the same valleys or mountain ridges. survive, oviposit, and reduce hemlock woolly adelgid As predators are released into the environment, it is populations in Þeld cages during winter and spring in important to analyze factors that could affect their southwestern Virginia (Lamb et al. 2005, 2006; Flowers establishment (Mausel 2007). Subsequent relocation et al. 2006). Three generations of L. nigrinus were and dispersal of predators from release sites could recovered from release sites in eastern forests (Lamb possibly result in predators coming into contact with et al. 2006, Mausel et al. 2010). hemlock woolly adelgid on hemlocks that have been S. tsugae is a multivoltine predator that is native to chemically treated. Japan where it is found in association with hemlock Using imidacloprid to control hemlock woolly ad- woolly adelgid (Cheah and McClure 1995). It can feed elgid could impact predators either through direct and develop to maturity on all stages of hemlock exposure or indirectly by changing hemlock woolly woolly adelgid, and its life cycle suggests a seasonal adelgid availability, quality, or distribution. Systemic synchrony with hemlock woolly adelgid progredien- insecticides do not directly expose predators to the tes in the Þeld (Cheah and McClure 1998, 2000). S. degree that foliar applications do (Mizell and Scony- tsugae is less cold tolerant than L. nigrinus and devel- ers 1992); however, nontarget beneÞcials can come ops most rapidly at temperatures from 20 to 25ЊC into contact with systemic toxins by feeding on plant (Cheah and McClure 2000). More than 3 million S. parts, contaminated prey, or through environmental tsugae and Ͼ100,000 L. nigrinus have been released in factors such as contaminated soil (Groot and Dicke eastern forests as of spring 2009. 2002). Predator consumption of sucking pests feeding Imidacloprid is a neonicotinoid insecticide and is an on plants treated systemically with imidacloprid can agonist of nicotinic acetylcholine receptors cause predator mortality (Grafton-Cardwell and Gu (nAChRs) (Elbert et al. 1991). Imidacloprid binds 2003, Cole and Horne 2006, Walker et al. 2007, Pa- with postsynaptic nAChRs, activating them and initi- pachristos and Milonas 2008), and natural enemies ating nervous stimulation along the synapse, which who supplement their diet with plant products can be leads to neurotoxicity poisoning symptoms and death exposed to imidacloprid through contaminated nec- (Lind et al. 1999, Silcox 2002, Tomizawa and Casida tar, pollen, or leaf tissues (Smith and Krischik 1999, 2003). Toxicity symptoms in include loss of Stapel et al. 2000, Moser and Obrycki 2009). Predators coordination, excessive grooming, twitching, tremors, exposed to systemic imidacloprid through feeding on and paralysis (Kunkel et al. 2001, Decourtye et al. contaminated prey could be subjected to lethal and 2004). Imidacloprid is effective against a wide range of sublethal effects that affect their biology, survival, insects and when applied systemically, it is especially and population increase parameters (Papachristos and successful protecting plants from sucking insects (Mi- Milonas 2008). For example, a sublethal effect such as zell and Sconyers 1992, Lind et al. 1999). Some plant impaired mobility (Kunkel et al. 2001), changes in species biotransform imidacloprid into metabolites dispersal (Conway et al. 2003), reduced foraging that have insecticidal properties (Nauen et al. 1998, (Schmuck 1999, Yang et al. 2008), stimulated fecun- 1224 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 103, no. 4 dity (Wang et al. 2005) or reduced fecundity and and analysis was carried out on the progredientes longevity (Papachristos and Milonas 2008) could re- generation. sult from exposure to sublethal imidacloprid residues. Approximately 30 cm of new growth from each Imidacloprid could potentially be passed on to pred- branchlet was subsampled 10, 20, and 30 d after treat- ators as they feed on contaminated adelgids because ment. Adelgids were observed under a microscope both L. nigrinus and S. tsugae rely on hemlock woolly and noted as alive or dead at each sampling period. adelgid as a primary food source. Thus, it is important Those that moved when disturbed or exuded fresh to determine whether systemic imidacloprid in hem- aqueous hemolymph when punctured were recorded lock tissue has any impact on nontarget predators as alive. Adelgids that did not move or respond when feeding on hemlock woolly adelgid. In the hemlocks, disturbed or were dry and exuded thick black hemo- imidacloprid may have indirect effects on predators lymph when punctured were recorded as dead. After when it is used to manage hemlock woolly adelgid. examination, branchlets were placed into Ziploc freezer bags and frozen until extraction. Imidacloprid and Metabolite Recovery in Hemlock Tissues. To extract imidacloprid, branchlets were re- Materials and Methods moved from the freezer and dried in an oven overnight General Methods of Treating Beetles and Hem- at 40ЊC or until needles could easily be removed. lock Branches with Imidacloprid. Eastern hemlock Branches were then ground using a Wiley Mill branches infested with hemlock woolly adelgid were (Arthur H. Thomas Co., Philadelphia, PA), which me- collected from Montgomery, Smyth, and Wythe coun- chanically pulverized the tissue until it was able to pass ties in southwest Virginia during April and May 2007. through a 1- by 1-mm screen. To extract imidacloprid, Branches were cut from 20- to 80-yr-old hemlocks, and 150 mg of the dried, ground wood tissue was added to each tree had substantial new growth and healthy a 1.5-ml microcentrifuge tube. Each sample was com- green needles. Distal 60Ð150-cm sections were cut bined with 1.5 ml of 5% H2SO4 and shaken with a from the lower and mid-crown of each tree and re- tabletop vortexer for 5 s. Samples were placed on a turned to the laboratory in Blacksburg. Ends were cut tabletop shaker at room temperature and shaken over- again and submerged in water for storage. Distal 20-cm night. The next day, samples were centrifuged at branchlets with hemlock woolly adelgid populations 13,000 ϫ g for 13 min. Supernatant was removed with Ͼ1.5 hemlock woolly adelgid per cm and healthy new a 1-ml syringe and Þltered through a 0.45-␮m polyvi- growth were cut from each branch and suspended on nylidene dißuoride Þlter before being added to 1.8-ml foam in 8- by 3-cm vials containing 20 ml of 0, 1, 10, or autosampler vials to be analyzed by liquid chroma- 100 ppm imidacloprid in water. Floral foam acted as a tography-tandem mass spectrometry (LC/MS/MS). barrier in the mouth of the vial; the cut end of each Samples were then packed in ice and shipped over- hemlock branchlet main stem was pushed through the night to Villanova University (Villanova, PA) for LC/ ßoral foam into the vial. Serial dilutions were created MS/MS analysis. The LC/MS/MS procedure is similar by diluting a 100 ppm stock solution of technical grade to Scho¨ning and Schmuck (2003) and quantiÞes imi- imidacloprid (99.2% purity; Bayer, Research Triangle dacloprid, and the oleÞn, urea, 6-chloronicotinic acid Park, NC) with distilled water. Branchlets were stored (6-CNA), des-nitro, des-nitro-oleÞn, and 5-hydroxy in a growth chamber (model I-36 VL, Percival Scien- and dihydroxy metabolites. A Shimadzu Promi- tiÞc, Perry, IA) maintained at 18:16ЊC (day:night) with nence high-performance liquid chromatography a photoperiod of 14:10 (L:D) h. Two open-topped system (Shimadzu, Colombia, MD) was used under 50-ml plastic containers were Þlled with tap water and Analyst software control (Applied Biosystems, Fra- placed in the chamber to slow desiccation of the mingham, MA). Separation of imidacloprid and me- branchlets. All vials and containers were reÞlled with tabolites was achieved using a Dionex Acclaim 120 C18 water as needed over the course of the experiment. column (150 by 2.1 mm, 3-␮m particle, 120 Å) with a After 10 d, the branchlets were removed and the top 4- by 2-mm C18 guard cartridge (Phenomenex, Sunny- portions that had no contact with the imidacloprid vale, CA). This involved using an aqueous (solvent A, solution were clipped and placed in 8- by 3-cm vials water with 0.1% formic acid) and organic (solvent B, with 20 ml of fresh water for hemlock woolly adelgid methanol 0.1% formic acid) mobile phase at a total and predator studies. ßow of 0.3 ml minϪ1. A gradient programmed elution The experiment was a complete block design with of 20Ð35% B over 12 min followed by a 2 min column 12 blocks (time), and approximately six 20-cm ßush (90% B) was carried out. A multiple reaction branchlets per treatment concentration in each block. monitoring (MRM) ion transition for imidacloprid For each block, all branches were from the same tree. and each metabolite was optimized on an Applied The experiment was started on 17 April and repeated Biosystems 3200 QTRAP using an electrospray ion- approximately every 6Ð10 d. ization (ESI) source. The precursor/fragment ion m/z Imidacloprid Impacts on Hemlock Woolly Adelgid. transitions for imidacloprid (256.1/209.1), oleÞn The Þrst four blocks of the hemlock woolly adelgid (254.1/171.1), urea (212.1/128.2), 6-chloronicotinic experiment covered the overwintering sistentes gen- acid (158.1/122.1), des-nitro (209.1/126.2), des-nitro- eration and the subsequent eight blocks (beginning 14 oleÞn (209.1/126.2), 5-hydroxy (272.1/191.2) and di- May) covered the progredientes generation. Data hydroxy (288.1/207.2) permitted selective analysis for were pooled by hemlock woolly adelgid generation, each compound. Optimized source parameters for the August 2010 EISENBACK ET AL.: EFFECTS OF IMIDACLOPRID ON HEMLOCK WOOLLY ADELGID 1225

Applied Biosystems ESI source were as follows: cur- intoxication. Imidacloprid intoxication symptoms in- tain gas (CUR) ϭ 35 psi, CAD gas ϭ medium, ESI clude twitching, spasms, or paralysis. Beetles were nebulizer gas (GAS1) ϭ 60 psi, auxiliary gas (GAS2) ϭ rated at each observation period on an ordinal scale 60 psi, ESI probe temperature ϭ 550ЊC, ion spray according to the degree of intoxication symptoms ob- voltage (IS) ϭ 5,500 V. The collision cell exit potential served: 0, alive, healthy; 1, mobile but twitching; 2, (CXP) was maintained at 4 V and the Q0 entrance mobile but with severe tremors; 3, immobile with potential (EP) was maintained at 10 V for all compounds. severe tremors; 4, paralyzed, unable to right them- The optimized declustering potential (DP) and collision selves; and 5, dead. energy (CE) produced the following limits of detection A randomized complete block design with three in picograms on-column for each compound: imidaclo- treatment levels and 10 beetles per level was used. For prid (8.8), oleÞn (22.9), urea (4.6), 6-chloronicotinic each species there were three blocks (time), resulting acid (11.0), des-nitro (0.65), des-nitro-oleÞn (0.22), in 30 beetles of each species per treatment level. L. 5-hydroxy (11.5), and dihydroxy (15.2). nigrinus was stored at 12:10ЊC (day:night) and a pho- Beetles. All L. nigrinus beetles originated from a toperiod of 14:10 (L:D) h. S. tsugae was stored at colony maintained at Virginia Tech. Adults emerging 20:18ЊC (day:night) and 14:10 (L:D) h. L. nigrinus and from aestivation in fall 2006 were fed hemlock woolly S. tsugae trials were initiated on 16 and 23 May 2007, adelgid collected in the Þeld throughout winter 2006Ð respectively. 2007. A subset was set aside for these experiments in Impact on Predators: Choice Experiment. A control March 2007. S. tsugae adults originated from a colony and treatment (1, 10, or 100 ppm) branch were each maintained by Clemson University (Clemson, SC). placed in a 10-cm petri dish with a 9-cm Þlter paper. Adults (1Ð2 wk old) were shipped to Blacksburg in Controls received two untreated branchlets. March 2007, and the colony was reared on hemlock Branchlets were inserted into water-soaked ßoral woolly adelgid collected from the Þeld in spring 2007. foam blocks wrapped in ParaÞlm. The ParaÞlm of the To match normal environmental rearing conditions, L. ßoral foam block of each treated branchlet was nigrinus was stored at 12:10ЊC (day:night) and a pho- marked with a permanent marker to distinguish it from toperiod of 14:10 (L:D) h, and S. tsugae was stored at the control. One beetle was added to each petri dish. 20:18ЊC and a photoperiod of 14:10 (L:D) h. Half the replications received a male beetle and the Impact on Predators: No-Choice Experiment. A other half received a female beetle. As gender of L. subset of branchlets not used for the hemlock woolly nigrinus beetles cannot be determined by external adelgid experiments was used in a predator no-choice physical characteristics, adults were sexed by observ- experiment. A branchlet with at least 100 ovisacs was ing oviposition after 72 h as in Lamb et al. (2005). L. taken from each treatment level (0, 1, or 100 ppm nigrinus beetles referred to as “male” are nonovipos- imidacloprid) and was inserted into a wetted ßoral iting individuals. S. tsugae adults were observed under foam block wrapped in paraÞlm and placed into a a microscope with the two sexes identiÞed according 10-cm petri dish. A 9-cm piece of Þlter paper was to distinguishing morphological characteristics de- placed at the bottom of each petri dish and moistened scribed by Cheah and McClure (1998). every5dtoprevent desiccation. Every 5 d, beetle mortality, intoxication, ßip time To measure loss of coordination or mobility due to (also measured at day 1), and beetle location (on imidacloprid, a test was initiated based on Bai et al. control versus treated branchlet) were recorded. Bee- (2006). Before placement on a treated branchlet, each tles were rated on the intoxication scale described beetle was placed in the center of a petri dish with a above. The number of hemlock woolly adelgid eaten 9-cm Þlter paper and was ßipped onto its dorsum by on each branchlet was based on counts of disrupted hand using a small paintbrush. The length of time ovisacs. The experiment was a randomized complete taken for each beetle to right itself was timed by hand block design, with each block (time) consisting of six with a stopwatch. Time was measured to the nearest replications (beetle) per treatment level. There were 0.01 s, with a maximum time of 60 s. The amount of Þve blocks, resulting in a total of 30 beetles of each time taken by each beetle to right itself was recorded species tested for each treatment level. For all blocks, as “ßip time.” Flip time has been used to quantify observations were made every 5 d for a total of four beetle health and mobility (Bai et al. 2006; Smith and observations over 20 d. Each species was stored ac- Krischik 1999; Lundgren and Wiedenmann 2002, cording to temperature and light conditions men- 2005). One beetle was added to each petri dish con- tioned previously. L. nigrinus trials were started on 15 taining a hemlock branchlet from 0, 1, or 100 ppm April 2007 and S. tsugae trials on 25 May 2007. treatment group. Each beetleÕs ßip time was recorded Predator Topical Application. Technical grade imi- at each observation event every 5 d. Observations dacloprid was mixed with acetone to create a serial were terminated 20 d after beetles were placed on dilution resulting in concentrations of 0 (acetone branchlets (30 d after treatment). only), 0.01, 0.1, 1, 10, and 100 ppm. A mechanized The number of hemlock woolly adelgid consumed microapplicator applied 0.5 ␮l of each treatment rate was recorded at each observation period by examining to the ventral abdomen of individual beetles, resulting individual ovisacs by using a microscope. Evidence of in 0, 0.05, 0.5, 5, or 50 ng of imidacloprid applied per predator feeding was quantiÞed indirectly by count- beetle. After treatment, beetles were placed on an ing disrupted ovisacs. Also recorded at each observa- untreated 5-cm hemlock twig infested with abundant tion time were beetle mortality and evidence of hemlock woolly adelgid in a 5-cm petri dish. Before 1226 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 103, no. 4 topical application, each beetleÕs ßip time was re- Table 1. Percentage of mortality (mean ؎ SE) of hemlock corded and were subsequently examined every 24 h woolly adelgid sampled at 10, 20, and 30 d after branchlets were for 6 d. Also at each observation, individuals were placed in imidacloprid solutions rated on the level of intoxication and ßip time. Indi- Dose Days after treatment viduals were recorded as dead when no response was (ppm) 10 20 30 elicited after probing with a paintbrush. Each species was stored according to temperature and light condi- 045Ϯ 4a 25 Ϯ 5a 29 Ϯ 11a 153Ϯ 5ab 40 Ϯ 8a 70 Ϯ 16b tions mentioned previously. Approximately 10 L. ni- 10 55 Ϯ 4ab 70 Ϯ 5b 84 Ϯ 6b grinus or S. tsugae were tested at each treatment level 100 65 Ϯ 3b 78 Ϯ 6b 96 Ϯ 2b in three independent random complete blocks result- ing in 145 and 102 individuals for each species, re- Within each column, means followed by different letters are sig- ϭ spectively, on 3 and 11 June 2007. One S. tsugae block niÞcantly different (P 0.05; TukeyÕs HSD test). was excluded from analysis due to 95% mortality in 24 h, probably due to an imidacloprid-contaminated root. One-way ANOVA compared beetle ßip time paintbrush. means before placement in the various experiments. Imidacloprid and Metabolite Recovery in Beetles. Intoxication ratings were compared among treatments At the conclusion of each experiment, beetles were and tested for signiÞcance using the KruskalÐWallis stored in 1.5-ml microcentrifuge tubes and frozen at test. Differences among treatments were separated by Ϫ60ЊC. To have sufÞcient mass for the analysis, the pair-wise comparison of treatment ranks using a Wil- beetles were pooled by treatment group before ex- coxon signed rank test. Intoxication ratings were from traction. To extract imidacloprid, beetles were frozen the ordinal intoxication scale but did not include dead with liquid nitrogen and ground into a powder. Ace- beetles in the analysis. tonitrile (0.5Ð1.5 ml) was added to each sample and shaken overnight at room temperature. The LC/ Results MS/MS procedure was similar to that of the tissue. Statistical Analysis. Data were analyzed using SPSS Imidacloprid Impacts on Hemlock Woolly Adelgid. for Macintosh OS X version 11.0.4. Control mortality Imidacloprid affected hemlock woolly adelgid mor- was corrected using AbbottÕs formula (Abbott 1925). tality in a dosage-dependant manner over time (Table All mortality data were arcsine square-root trans- 1). SigniÞcant differences among dosages were found formed before analysis; reported means are untrans- for all three sample periods; 10 d (F ϭ 3.887; df ϭ 3, formed. For all tests, P Յ 0.05 was used to separate 156; P ϭ 0.01), 20 d (F ϭ 15.145; df ϭ 3, 79; P ϭ 0.001), means. and 30 d (F ϭ 14.628; df ϭ 3, 40; P ϭ 0.001) after Hemlock woolly adelgid mortality during each time treatment. Hemlock woolly adelgid mortality was rel- period (10, 20, and 30 d after treatment) was analyzed atively high in controls (mean, 45%) during the Þrst using one-way analysis of variance (ANOVA), with 10 d of the experiment, probably because of the move means separated by TukeyÕs honestly signiÞcant dif- from the outdoors to an environmental chamber. Mor- ference (HSD) test. Probit analysis was conducted for tality of controls leveled off after 20 d. After 30 d, hemlock woolly adelgid mortality after 30 d based on hemlock woolly adelgid mortality ranged from 29 to imidacloprid concentrations recovered by LC/ 96%. Probit analysis of hemlock woolly adelgid mor- MS/MS and for beetle mortality 3 and 6 d after topical tality after 30 d versus detected imidacloprid concen- application. Linear regression was performed compar- tration means determined the LC50 of imidacloprid ing imidacloprid concentration on a logarithmic scale and its 95% conÞdence limit (CL) to be 242 and 105Ð versus hemlock woolly adelgid mortality. Imidaclo- 411 ppb, respectively. After correcting for control prid and metabolite concentrations determined by mortality, hemlock woolly adelgid mortality means LC/MS/MS were corrected by subtracting any values 30 d after treatment had a strong dependent relation- in the control branchlets (Ͻ20 ppb) from treated ship with the average concentration of imidacloprid in branchlets. each treatment group determined by LC/MS/MS For the no-choice test, the proportion of beetle (Fig. 1). mortality was averaged within replications for each Imidacloprid and Metabolite Recovery in Hemlock treatment observation period combination and ana- Tissues. Imidacloprid and metabolite concentrations lyzed using repeated measures ANOVA. Means were in wood tissues generally increased with treatment separated using TukeyÕs HSD test. Repeated measures rate and time after treatment (Table 2). OleÞn, des- ANOVA tested for ßip time differences over time. nitro, urea, and 6-CNA were the metabolites detected For the choice test, a two-way ANOVA compared in the largest proportions. Imidacloprid made up 64Ð mortality by beetle sex and treatment. The total num- 98% of the total metabolite concentrations detected. ber of adelgids eaten by each beetle during the course OleÞn was the most prevalent metabolite, consisting of the experiment was analyzed using one-way of 2Ð27% of the total concentrations. Other metabo- ANOVA, with means separated by TukeyÕs HSD test. lites combined did not exceed 2% of the total concen- Repeated measures ANOVA tested for ßip time dif- tration in most samples. ferences over time. Impacts on Predators: No-Choice Experiment. Dif- Flip time data were transformed to achieve homo- ferences were observed in predator feeding, ßip time, geneity of variance using the inverse of the square mortality, and pooled intoxication symptoms among August 2010 EISENBACK ET AL.: EFFECTS OF IMIDACLOPRID ON HEMLOCK WOOLLY ADELGID 1227

Fig. 1. Hemlock woolly adelgid percentage of mortality (means Ϯ SE) 30 d after treatment and imidacloprid concen- trations (mean Ϯ SE) recovered from hemlock branchlets placed in 1, 10, and 100 ppm imidacloprid solutions. Mortality adjusted for control mortality using AbbottÕs formula. treatments in the no-choice test (Table 3). Predator than beetles from control branchlets (F ϭ 4.526; df ϭ mortality increased with hemlock woolly adelgid mor- 2, 85; P ϭ 0.014). Flip time means increased for S. tality and imidacloprid concentrations recovered from tsugae as the imidacloprid concentration increased, hemlock wood tissues (Fig. 2A). L. nigrinus mortality however, differences were not signiÞcant (F ϭ 0.653; was similar to hemlock woolly adelgid mortality and df ϭ 2, 87; P ϭ 0.523). imidacloprid recovery in the no-choice test (Fig. 3A), Treatment had a signiÞcant effect on mortality for but S. tsugae mortality was generally much lower than L. nigrinus (F ϭ 2.952; df ϭ 2, 6; P ϭ 0.048) but not for L. nigrinus. S. tsugae (F ϭ 1.50; df ϭ 2, 6; P ϭ 0.296). After cor- No differences in consumption of hemlock woolly recting for control mortality, L. nigrinus mortality av- adelgid by S. tsugae were observed between treated eraged Ͼ40%, whereas S. tsugae mortality was Ͻ7% in and untreated branchlets (F ϭ 0.229; df ϭ 2, 87; P ϭ the 100 ppm treatment group. Mortality generally in- 0.796), but L. nigrinus consumed signiÞcantly less creased with time and dose. hemlock woolly adelgid from treated branchlets than The percentage of beetles observed intoxicated or from untreated branchlets (F ϭ 6.153; df ϭ 2, 87; P ϭ moribund at least once during the course of the ex- 0.003). periment ranged from 0 to 20% and from 0 to 13% for Before placement on branchlets, there was no dif- L. nigrinus and S. tsugae, respectively. For L. nigrinus, ference in ßip times between individuals for L. nigri- 25% of the beetles observed intoxicated in the 1 ppm nus (F ϭ 0.40; df ϭ 2, 87; P ϭ 0.667) or S. tsugae (F ϭ treatment group, and 50% from the 100 ppm treatment 0.65; df ϭ 2, 87; P ϭ 0.523). Flip times for L. nigrinus group exhibiting intoxication eventually died. For S. from the 100 ppm branchlets were signiÞcantly longer tsugae, none of the intoxicated beetles in the one ppm

Table 2. Concentrations (ppb; mean ؎ SE) of imidacloprid and four of its most prevalent metabolites recovered from eastern hemlock wood tissuea

Dose DATa Imidacloprid OleÞn Des-Nitro Urea 6-CNA (ppm) 11068Ϯ 21 (64)b 30 Ϯ 9.1 (27) 2 Ϯ 0.4 (2) 0 (0) 0 Ϯ 0 (0) 20 216 Ϯ 206 (87) 21 Ϯ 7.9 (8) 3 Ϯ 2 (1) 0 (0) 1 Ϯ 1 (1) 30 200 Ϯ 97 (90) 16 Ϯ 8.9 (7) 2 Ϯ 1 (1) 2 Ϯ 1 (1) 2 Ϯ 2 (1) 10 10 89 Ϯ 56 (78) 17 Ϯ 4 (15) 3 Ϯ 1 (2) 0 (0) 1 Ϯ 0.4 (1) 20 470 Ϯ 247 (93) 23 Ϯ 11 (5) 4 Ϯ 3 (1) 6 Ϯ 4 (1) 3 Ϯ 1 (1) 30 1,254 Ϯ 638 (94) 35 Ϯ 18 (3) 9 Ϯ 6 (1) 19 Ϯ 12 (1) 7 Ϯ 3 (1) 100 10 4,488 Ϯ 2,497 (97) 93 Ϯ 31 (2) 38 Ϯ 17 (1) 13 Ϯ 11 (0) 8 Ϯ 4 (0) 20 7,799 Ϯ 4,454 (96) 172 Ϯ 86 (2) 56 Ϯ 35 (1) 88 Ϯ 77 (1) 29 Ϯ 16 (0) 30 15,282 Ϯ 4,002 (93) 442 Ϯ 155 (3) 110 Ϯ 53 (1) 294 Ϯ 187 (2) 84 Ϯ 24 (1)

a Days after treatment. b The percentage of total concentration each compound comprises for each doseÐtime combination is given in parentheses. The des-nitro- oleÞn, 5-hydroxy, and dihydroxy metabolites are not included because they generally constituted Ͻ2% of the total concentration. 1228 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 103, no. 4

Table 3. Impacts on two predators resulting from feeding on hemlock woolly adelgid on eastern hemlock branchlets treated with imidacloprid in no-choice and choice experiments

Dose HWA disturbed Mortalityc Intoxicatedd Exp Speciesa Flip timeb (s) (ppm) per dayb (%) (%) No choice Ln 0 2.2 Ϯ 0.14 a 19.9 Ϯ 2.7a 0a 0a 1 1.6 Ϯ 0.17b 28.4 Ϯ 3.6ab 15ab 13b 100 1.5 Ϯ 0.15b 34.4 Ϯ 3.4b 41b 20b St 0 3.6 Ϯ 0.26A 11.9 Ϯ 1.6A 0A 0A 1 3.4 Ϯ 0.18A 13.4 Ϯ 2.1A 3A 7A 100 3.4 Ϯ 0.28A 18.7 Ϯ 2.6A 7A 13A Choice Ln 0 16.4 Ϯ 3.6a 0a 0a 1 24.4 Ϯ 3.4a 11a 27b 10 23.6 Ϯ 3.8a 11a 23b 100 See Table 4 28.8 Ϯ 3.6a 26a 23b St 0 12.5 Ϯ 1.3A 0A 7A 1 13.9 Ϯ 1.4A 0A 0A 10 19.5 Ϯ 2.4AB 3A 13B 100 23.2 Ϯ 2.6A 3A 10B

a Ln, L. nigrinus; St, S. tsugae. b Data are mean Ϯ SE. For each experiment and species, means within each column followed by the same letter are not signiÞcantly different (P ϭ 0.05; TukeyÕs HSD test). c Displayed means are corrected for control mortality. For each species, means within each column followed by the same letter are not signiÞcantly different (P ϭ 0.05; TukeyÕs HSD test). d Percentage of beetles observed intoxicated (pooled symptoms) at least once during course of the experiment. Differences among treatment ranks were separated with a pairwise comparison using the Wilcoxon signed rank test (P ϭ 0.05); ranks not shown. group died, but as in L. nigrinus 50% of beetles ob- ments for L. nigrinus (␹2 ϭ 9.79, df ϭ 2, P ϭ 0.007), but served intoxicated in the 100 ppm group died. Intox- differences were not signiÞcant for S. tsugae (␹2 ϭ ication ranks were signiÞcantly different among treat- 4.44, df ϭ 2, P ϭ 0.109).

Fig. 2. No-choice (A) and choice (B) tests comparing hemlock woolly adelgid and beetle mortality to imidacloprid concentrations recovered from hemlock wood tissue 30 d after treatment. Mortality adjusted for control mortality using AbbottÕs formula. HWA, hemlock woolly adelgid; Ln, L. nigrinus; St, S. tsugae. August 2010 EISENBACK ET AL.: EFFECTS OF IMIDACLOPRID ON HEMLOCK WOOLLY ADELGID 1229

Fig. 3. No-choice (A) and choice (B) test results comparing beetle mortality and imidacloprid concentrations recovered from each species of beetle. Ln, L. nigrinus; St, S. tsugae. Mortality adjusted for control mortality using AbbottÕs formula.

Impacts on Predators: Choice Experiment. Differ- branchlets, there was a signiÞcant difference between ences in predator biology were evident in choice tests control and 100 ppm for L. nigrinus (P ϭ 0.006), but (Table 3). L. nigrinus mortality increased with hem- no signiÞcant differences were found for S. tsugae lock woolly adelgid mortality and imidacloprid con- (Table 4). The difference in the number of hemlock centrations recovered from hemlock wood tissue (Fig. woolly adelgid consumed on untreated and treated 2B). S. tsugae mortality in the choice test was not branchlets increased with dose for both species. For signiÞcant and did not show a relationship with both species, there were no signiÞcant differences hemlock woolly adelgid mortality or imidacloprid between sexes: L. nigrinus (F ϭ 3.698; df ϭ 1, 111; P ϭ concentrations (Fig. 3B). 0.057), S. tsugae (F ϭ 2.163; df ϭ 1, 112; P ϭ 0.144), or In paired t-tests comparing the number of hem- treatment levels: L. nigrinus (F ϭ 0.851; df ϭ 3, 115; P ϭ lock woolly adelgid eaten on control and treated 0.469), S. tsugae (F ϭ 0.217; df ϭ 3, 116; P ϭ 0.885).

Table 4. Number (mean ؎ SE) of disrupted hemlock woolly adelgid ovisacs on control and treated eastern hemlock branchlets in paired choice tests

Dose Speciesa Control Treated Difference Ϯ SE t-statistic df P value (ppm) Ln 0 7.3 Ϯ 0.95 6.2 Ϯ 0.632 1.1 Ϯ 0.66 1.71 29 0.098 1 8.8 Ϯ 1.1 7.3 Ϯ 0.75 1.5 Ϯ 1.1 Ϫ1.44 29 0.162 10 8.1 Ϯ 0.97 6.4 Ϯ 0.70 1.7 Ϯ 0.98 Ϫ1.77 29 0.087 100 7.8 Ϯ 0.84 5.3 Ϯ 0.59 2.5 Ϯ 0.84 Ϫ2.96 28 0.006 St 0 14.6 Ϯ 0.54 14.6 Ϯ 0.802 0.0 Ϯ 0.69 0.00 28 1.00 1 14.8 Ϯ 0.75 14.6 Ϯ 0.92 0.21 Ϯ 1.1 Ϫ0.19 28 0.853 10 14.6 Ϯ 0.75 13.9 Ϯ 0.62 0.59 Ϯ 0.99 Ϫ0.59 28 0.562 100 14.6 Ϯ 0.67 13.4 Ϯ 0.69 1.2 Ϯ 0.71 Ϫ1.74 29 0.093

a Ln, L. nigrinus; St, S. tsugae. b For the 0 dose group, “treated” branchlets are untreated. 1230 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 103, no. 4

Flip times before beetle placement in the choice Table 5. Effects of topical application of imidacloprid on L. arena were not signiÞcantly different for L. nigrinus nigrinus and S. tsugae biology 6 d after treatment (P ϭ 0.161) or S. tsugae (P ϭ 0.563), nor between sexes Dose (ng per Flip timeb Mortalityc Intoxicatedd ϭ ϭ ϭ Speciesa in L. nigrinus (F 2.674; df 1, 119; P 0.105) or S. beetle) (s) (%) (%) tsugae (F ϭ 0.083; df ϭ 1, 112; P ϭ 0.774), and data for Ln 0 15.9 Ϯ 3.7a 0a 0a the two sexes of each species were pooled. Flip times 0.005 30.5 Ϯ 4.3ab 19.4ab 70b increased over time and treatment level but were not 0.05 32.8 Ϯ 3.5ab 31.9abc 60b signiÞcantly different for L. nigrinus (F ϭ 1.238; df ϭ 0.5 43.6 Ϯ 3.5bc 46.3bc 83d 3, 106; P ϭ 0.30). S. tsugae ßip times in the 100 ppm 5 59.4 Ϯ 3.5c 82.1bc 77c 50 60.0c 100c 63bcd treatment group were signiÞcantly longer than con- St 0 15.1 Ϯ 2.8A 0A 0A trols (F ϭ 3.634; df ϭ 3, 113; P ϭ 0.015). 0.005 32.4 Ϯ 2.8B 5.3AB 55B A two-way ANOVA of mortality showed no differ- 0.05 37.8 Ϯ 2.8B 5.3AB 85D ences among treatments for L. nigrinus (F ϭ 1; df ϭ 3, 0.5 59.8 Ϯ 2.9C 31.6B 95E ϭ 5 60.0C 100C 70C 32; P 0.406) but females exhibited higher mortality 50 60.0C 100C 70C than males (F ϭ 0.937; df ϭ 1, 32; P ϭ 0.019). After analyzing mortality within each sex, treatment effects a Ln, L. nigrinus; St, S. tsugae. were not signiÞcant for females (F ϭ 1.588; df ϭ 1, 32; b Data are mean Ϯ SE. For each species, means within each column P ϭ 0.231) or males (F ϭ 0.940; df ϭ 3, 16; P ϭ 0.445). followed by the same letter are not signiÞcantly different (P ϭ 0.05; TukeyÕs HSD test). For S. tsugae, a two-way ANOVA revealed no differ- c Displayed means are corrected for control mortality. For each ences in mortality by treatment (F ϭ 0.667; df ϭ 3, 32; species, means within each column followed by the same letter are not P ϭ 0.579) or sex (F ϭ 2; df ϭ 1, 32; P ϭ 0.167). After signiÞcantly different (P ϭ 0.05; TukeyÕs HSD test). correcting for control mortality, L. nigrinus mortality d Percentage of beetles observed intoxicated (pooled symptoms) at least once during course of the experiment. Differences among treat- was as high as 26% in the 100 ppm treatment group, but ment ranks were separated with a pairwise comparison using the S. tsugae mortality was Յ3% (Table 3). Wilcoxon signed rank test (P ϭ 0.05); ranks not shown. More than 20% of L. nigrinus beetles showed signs of intoxication from feeding on treated branchlets and species from the highest treatment dose in the topical 25, 57, and 43% of them died from the 1, 10, and 100 application experiment; however, no metabolites ppm treatment groups, respectively. Signs of intoxi- were recovered from beetle cadavers from the choice cation for S. tsugae were less prevalent, ranging from or no-choice experiments. Beetle mortality corre- 7 to 13%. From these intoxicated beetles, 25 and 33% sponded with recovered imidacloprid concentrations eventually died from the 10 and 100 ppm treatment from the choice and no-choice experiment (Fig. 3) groups, respectively. Intoxication ranks were signiÞ- and the topical application experiment (Fig. 4). L. cantly different among treatments for both L. nigrinus nigrinus mortality was highest in the 100 ppm treat- (␹2 ϭ 8.71, df ϭ 3, P ϭ 0.033) and S. tsugae (␹2 ϭ 11.23, ment group in the no-choice test; the highest concen- df ϭ 3, P ϭ 0.011). tration of imidacloprid was also found in beetles from

Predator Topical Application. The6dLD50 and the 100 ppm treatment group in the choice test. In LD90 values and their 95% conÞdence intervals for L. general, higher concentrations of imidacloprid were nigrinus were 1.8 (1.7Ð133.7) and 5.8 (3.2Ð47.8) ng per recovered from L. nigrinus than S. tsugae. beetle, respectively. For S. tsugae, the6dLD50 and LD values and their 95% conÞdence intervals were 90 Discussion 0.71 (0.5Ð1.7) and 1.3 (0.88Ð3.6), respectively. Both beetle species required longer ßip times after imida- Our results show that in the laboratory, very low cloprid treatment than controls, with higher concen- concentrations (Ͻ242 ppb) of imidacloprid in hem- trations resulting in longer times than lower concen- lock wood tissue killed 50% of hemlock woolly adelgid trations (Table 5). Flip times were signiÞcantly within 30 d of systemic treatment. Our 30 d LC50 of different among treatments for L. nigrinus (F ϭ 18.51; imidacloprid and its 95% CL was 242 and 105Ð411 ppb, df ϭ 5,139; P ϭ 0.001) and for S. tsugae (F ϭ 37.29; df ϭ whereas Cowles et al. (2006) conducting a similar 5, 96; P ϭ 0.001). Mortality increased with dose. L. experiment with hemlock woolly adelgid sistentes es- nigrinus and S. tsugae mortality were signiÞcantly dif- timated the 20 d LC50 and its 95% conÞdence interval ferent among treatments (F ϭ 10.50; df ϭ 5, 9; P ϭ (CI) to be 300 and 150Ð600 ppb, respectively. Their 0.002) and (F ϭ 7.22; df ϭ 5, 9; P ϭ 0.039), respectively. results with sistentes and our study on progredientes Both beetle species displayed tremors and paralysis suggest that these two adelgid generations display after treatment, with increasing intensity of poisoning similar susceptibility to imidacloprid. Both studies in- symptoms and mortality over time and with increasing vestigated short-term imidacloprid impacts (Ͻ30 d), dose concentration. The ratings from the intoxication but in the Þeld hemlock woolly adelgid will be exposed scale were signiÞcantly different for both L. nigrinus to imidacloprid for times much longer than 30 d and ␹2 ϭ ϭ ϭ ␹2 ϭ ( 304.50; df 5; P 0.001) and S. tsugae ( over several adelgid generations. Field LC50 values 326.90; df ϭ 5; P ϭ 0.001). therefore are likely to be lower than 300 ppb, as shown

Imidacloprid and Metabolite Recovery in Beetles. by Cowles et al. (2006) who estimated the Þeld LC50 Imidacloprid was the main toxicant recovered from for hemlock woolly adelgid to be Ϸ120 ppb. both beetle species postmortem. The oleÞn and Based on analysis by LC/MS/MS, imidacloprid was 6-CNA metabolites were recovered from both beetle effectively taken up and partially metabolized by cut August 2010 EISENBACK ET AL.: EFFECTS OF IMIDACLOPRID ON HEMLOCK WOOLLY ADELGID 1231

Fig. 4. Topical applications results comparing beetle mortality with imidacloprid concentrations recovered by LC/ MS/MS from beetles 6 d after topical application of imidacloprid to the ventral abdomen. Ln, L. nigrinus; St, S. tsugae. Mortality adjusted for control mortality using AbbottÕs formula. hemlock branchlets within 10 d after treatment in the display any intoxication symptoms, possibly indicating laboratory. The primary metabolite recovered from natural mortality or mortality from reduced prey avail- hemlock wood tissues was oleÞn, a compound that, in ability and quality rather than direct insecticidal ef- some cases, has displayed even higher biological efÞ- fects. The recovery of imidacloprid from beetles post- cacy against insects than its parent compound imida- mortem by LC/MS/MS suggests that imidacloprid was cloprid. For example, oleÞn was 10 times as biologi- passed on to the predators through feeding on con- cally efÞcacious as imidacloprid against whiteßies taminated adelgids. If the imidacloprid concentrations Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae) that were found in these lab-treated branchlets are (Nauen et al. 1999). In the honey bee, Apis mellifera similar to concentrations in hemlocks after Þeld treat- L. (Hymenoptera: Apidae), oleÞn was a product of ments of imidacloprid, it may be possible for systemic biotransformation of oral doses of imidacloprid and imidacloprid to be prey-mediated to higher trophic was associated with the onset of mortality within hours levels and directly affect nontarget hemlock woolly after ingestion, whereas imidacloprid was the cause of adelgid predators. initial neurotoxicity symptoms (Suchail et al. 2003, In the choice test, both predator species consumed 2004). OleÞn could display similar increased efÞcacy more adelgids from untreated than treated branchlets against hemlock woolly adelgid and its predators, al- which may be a consequence of increased adelgid though this relationship has not been studied. If so, mortality and decreased adelgid quality on treated eastern hemlock biotransformation of imidacloprid branchlets. L. nigrinus was more susceptible to feeding into the oleÞn metabolite could provide additional on adelgids from treated branchlets compared with S. protection from hemlock woolly adelgid infestation. tsugae because its behavior was more intimately linked Predators feeding on adelgids surviving on treated to hemlock woolly adelgid. For example, egg laying is trees could potentially be directly affected by insec- prevalently within hemlock woolly adelgid ovisacs ticide toxicity from contact with the adelgids them- whereas S. tsugae egg laying tends to be on the bark. selves or indirectly from a general reduction in local- Predators could prefer to lay eggs on untreated ized prey quality, density and abundance. Some branchlets than treated branchlets to use healthier, beetles displayed intoxication symptoms after feeding denser, higher quality adelgid populations. Differ- on adelgids from treated branchlets, suggesting that ences in predator behavior may play a role in suscep- imidacloprid may be passed from an adelgid to a pred- tibility; L. nigrinus consumed whole adelgid adults, ator under certain conditions. Intoxication response whereas S. tsugae was more often observed feeding on was variable; some beetles showed intoxication symp- eggs or partially consuming adelgids. Adelgid eggs toms within5doffeeding on hemlock woolly adelgid might not have imidacloprid within them and could be from treated branchlets, others displayed symptoms a safer food source for both predators and their larvae, later, and some never displayed symptoms. Not all although further experiments are required to test this beetles that displayed intoxication symptoms died, hypothesis. suggesting that consumption of low concentrations of Topical applications demonstrated the intolerance imidacloprid can be sublethal as some beetles may be of both beetle species to direct exposure to imidaclo- able to metabolize the insecticide to nontoxic metab- prid. Both predator species displayed intoxication olites within a few days. Many beetles that died did not symptoms at the lowest dose 24 h after treatment. At 1232 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 103, no. 4 doses of Ͻ0.5 ng/beetle some individuals were able to like L. nigrinus that prey on hemlock woolly adelgid survive for the 6 d experiment, although they often during winter and early spring when few other food exhibited heavy tremors and paralysis. In some cases, sources are available. Whether or not indirect imida- mortality was not acute; delayed mortality may have cloprid exposure to predators through feeding on ad- resulted from starvation or dehydration because bee- elgids on treated trees is possible, predator preference tles were paralyzed and unable to move or feed. The for healthier food stock could drive them away from

LD50 values for L. nigrinus and S. tsugae are similar to treated stands toward denser, healthier adelgid pop- the LD50 value for honey bees, reported to be 46 ng ulations, although Þeld tests are required to test this per bee 96 h after oral dosing (Suchail et al. 2001), and hypothesis. abnormal honey bee foraging behavior was caused by In the laboratory, hemlock woolly adelgid and both oral ingestion of 1.82Ð4.33 ng imidacloprid per bee biological control agents are susceptible to low (ppb (Yang et al. 2008). Imidacloprid can have sublethal range) concentrations of imidacloprid. Prey-mediated effects in the range of nanograms per insect. Differ- effects were possible in a laboratory situation, as pred- ences in beetle size and weight could play a role in ators displayed intoxication and mortality from feed- susceptibility; L. nigrinus weighs Ϸ750 ␮g each and S. ing on adelgids surviving on treated hemlock tsugae weighs 390 ␮g each (unpublished data). The branchlets. Impacts varied among beetle species, imi-

LD50 value for L. nigrinus was more than twice the dacloprid concentration, and the availability of un- value for S. tsugae, which may reßect that they weigh treated hemlock woolly adelgid. Field tests are needed approximately twice as much. If LD50 values are cal- to investigate what range of imidacloprid concentra- culated by body weight, the LD50 values are 2.4 and 1.8 tions are likely in hemlock branches and whether or ␮g/g for L. nigrinus and S. tsugae, respectively, re- not predators could be exposed to those systemic ßecting that topical applications of imidacloprid are concentrations by consuming surviving adelgids. Even more acutely toxic to S. tsugae compared with L. ni- if prey-mediated exposure is unlikely under Þeld con- grinus. Another factor that could have affected sus- ditions, using imidacloprid to control hemlock woolly ceptibility is that the topical test was carried out in adelgid populations will impact predator establish- June when L. nigrinus adults had been alive and feed- ment and survivorship by reducing hemlock woolly ing for Ͼ9 mo and were at the end of their annual life adelgid availability as a food source. In high-value cycle. S. tsugae adults were Ͻ3 mo old and compared hemlock stands, biological and chemical control could with L. nigrinus they were younger and more robust. be advantageous when used together; chemicals could Thus, beetle age and health could be factors impacting eradicate hemlock woolly adelgid populations on a subsequent survivorship and it would be beneÞcial if proportion of hemlocks in a stand and predators could L. nigrinus assays were conducted during early winter subsequently be released on the remaining trees, al- when they were younger and more robust. though predator dispersal to treated trees with sur- In the Þeld, very low concentrations of imidacloprid viving adelgids could pose a hazard. Chemical control are capable of controlling hemlock woolly adelgid. using imidacloprid remains a short-term solution for Imidacloprid concentrations in Þeld-treated hemlock reducing hemlock woolly adelgid impacts, and bio- wood and needle tissue can be Ͻ250 ppb (Cowles et logical control of hemlock woolly adelgid remains a al. 2006) or generally Ͻ1 ppm (B.M.E., unpublished long-term goal. Chemical and biological control of data). In these experiments imidacloprid concentra- hemlock woolly adelgid are both important tools tions in the branchlets from the 100 ppm treatment needed to protect hemlocks in eastern forests, and this group ranged from 4.5 to 15 ppm, which are concen- study and future projects will help further deÞne the trations that are much higher than would be expected relationships among insecticide, tree, pest, and pred- from treating hemlock trees in the Þeld. The one ppm ator in the hemlock ecosystem. treatment group exhibited imidacloprid concentra- tions in the branchlets ranging from 60 to 200 ppb which are concentrations that are more likely to be Acknowledgments found in Þeld-treated hemlocks. In this treatment We thank Donald Mullins, Jeffrey Bloomquist, Frank group predator mortality and Þtness impacts in the Cook, Theresa Dellinger, Carrie Jubb, Ashley Lamb, Rob choice and no-choice tests were generally less pro- Flowers, David Mausel, LayLa Burgess, and Nick Reynolds. nounced than the 10 and 100 ppm treatments. These Bayer supplied technical grade imidacloprid. Funding pro- results suggest that the concentrations found in Þeld- vided by USDA Forest Service, Forest Health Protection, treated hemlocks may not directly impact predators to Southern Region 8, grants 04-DG-11083150-020 (Virginia a quantiÞable degree and any negative effects of imi- Tech) and 04-CA-11244225-397 (Villanova University). dacloprid on hemlock woolly adelgid predators would probably result from reduced prey quality and density. References Cited Acute insecticide effects can be so efÞcacious against pest populations that natural enemies starve, disperse Abbott, W. S. 1925. A method of computing the effective- ness of an insecticide. J. Econ. Entomol. 18: 265Ð267. or Þnd new food sources (Johnson and Tabashnik Bai, Y. Y., M. X. Jiang, J. A. Cheng, and D. Wang. 2006. 1999). Any antifeedant or survivorship impacts that Effects of Cry1Ab toxin on Propylea japonica (Thunberg) imidacloprid has on hemlock woolly adelgid will result (Coleoptera: Coccinellidae) through its prey, Nilapar- in lower quality and abundance of food for predators. vata lugens Stål (Homoptera: ), feeding on This effect would be greater on host-speciÞc predators transgenic Bt rice. Environ. Entomol. 35: 1130Ð1136. August 2010 EISENBACK ET AL.: EFFECTS OF IMIDACLOPRID ON HEMLOCK WOOLLY ADELGID 1233

Cheah, C.A.S.-J., and M. S. McClure. 1995. Exotic natural Havill, N. P., and R. G. Foottit. 2007. Biology and evolution enemies of Adelges tsugae and their prospect for biolog- of Adelgidae. Annu. Rev. Entomol. 52: 325Ð349. ical control, pp. 103Ð112. In S. M. Salom, T. C. Tigner, and Holmes, T. P., E. A. Murphy, and D. D. Royle. 2005. The R. C. Reardon [eds.], Proceedings of the First Hemlock economic impacts of hemlock woolly adlegid on residen- Woolly Adelgid Review. U.S. Dep. Agric. For. Serv. Publ. tial landscape values: Sparta, New Jersey case study, pp. FHTET 96-10, Charlottesville, VA. 15Ð24. In B. Onken and R. Reardon [eds.], Third Sym- Cheah, C.A.S.-J., and M. S. McClure. 1998. Life history and posium on Hemlock Woolly Adelgid in the Eastern development of Pseudoscymnus tsugae (Coleoptera: Coc- United States. U.S. Dep. Agric. For. Serv. Publ. FHTET- cinellidae), a new predator of the hemlock woolly adelgid 2005-01, Asheville, NC. (Homoptera: Adelgidae). Environ. Entomol. 27: 1531Ð Johnson, M. W., and B. E. Tabashnik. 1999. Handbook of 1536. biological control. Academic, San Diego, CA. Cheah, C.A.S.-J., and M. S. McClure. 2000. Seasonal syn- Kunkel, B. A., D. W. Held, and D. A. Potter. 2001. Lethal chrony of life cycles between the exotic predator, and sublethal effects of bendiocarb, halofenozide, and Pseudoscymnus tsugae (Coleoptera: Coccinellidae) and imidacloprid on Harpalus pennsylvanicus (Coleoptera: its prey, the hemlock woolly adelgid Adelges tsugae (Ho- Carabidae) following different modes of exposure in turf- moptera: Adelgidae). Agric. For. Entomol. 2: 241Ð251. grass. J. Econ. Entomol. 94: 60Ð67. Cheah, C.A.S.-J., and M. S. McClure. 2002. Pseudoscymnus Lamb, A. B., S. M. Salom, and L. T. Kok. 2005. Survival and tsugae in Connecticut forests: the Þrst Þve years, pp. reproduction of Laricobius nigrinus, a predator of hem- 150Ð165. In B. Onken, R. Reardon, and J. Lashomb [eds.], lock woolly adelgid, Adelges tsugae in Þeld cages. Biol. Hemlock woolly adelgid in the eastern United States Control 32: 200Ð207. Symposium. U.S. Dep. Agric. For. Serv. Publ., East Bruns- Lamb, A. B., S. M. Salom, L. T. Kok, and D. L. Mausel. 2006. wick, NJ. ConÞned Þeld release of Laricobius nigrinus (Coleoptera: Cheah, C.A.S.-J., M. A. Mayer, D. Palmer, T. Scudder, and R. Derodontidae), a predator of the hemlock woolly adel- Chianese. 2005. Assessments of biological control of gid, Adelges tsugae (Hemiptera: Adelgidae), in Virginia. hemlock woolly adelgid with Sasajiscymnus tsugae in Can. J. For. Res. 36: 369Ð375. Connecticut and New Jersey, pp. 116Ð130. In B. Onken Lind, R. J., M. Clough, F. Earley, S. Wonnacott, and S. and R. Reardon [eds.], Third Symposium on Hemlock Reynolds. 1999. Characterization of the relationship be- Woolly Adelgid in the Eastern United States. U.S. Dep. tween binding sites for imidacloprid and other nicotinic Agric. For. Serv. Publ. FHTET-2005-01, Asheville, NC. ligands in insects. Pestic. Sci. 55: 1029Ð1031. Cole, P. G., and P. A. Horne. 2006. The impact of aphicide Lundgren, J. G., and R. N. Wiedenmann. 2002. Coleopter- drenches on Micromus tasmaniae (Walker) (Neuroptera: an-speciÞc Cry3Bb toxin from transgenic corn pollen Hemerobiidae) and the implications for pest control in does not affect the Þtness of a nontarget species, Co- lettuce crops. Aust. J. Entomol. 45: 244Ð248. leomegilla maculata DeGeer (Coleoptera: Coccinelli- Conway, H. E., T. J. Kring, and R. McNew. 2003. Effect of dae). Environ. Entomol. 31: 1213Ð1218. imidacloprid on wing formation in the cotton (Ho- Lundgren, J. G., and R. N. Wiedenmann. 2005. Tritrophic moptera: ). Fla. Entomol. 86: 474Ð476. interactions among Bt (Cry3Bb1) corn, aphid prey, and Cowles, R. S., M. E. Montgomery, and C. A. Cheah. 2006. the predator (Coleoptera: Coc- Activity and residues of imidacloprid applied to soil cinellidae). Environ. Entomol. 34: 1621Ð1625. and tree trunks to control hemlock woolly adelgid Mausel, D. L. 2007. Release and monitoring of Laricobius (Hemiptera: Adelgidae) in forests. J. Econ. Entomol. 99: nigrinus (Coleoptera: Derodontidae) for biological con- 1258Ð1267. trol of the hemlock woolly adelgid in the eastern US. Decourtye, A., J. Devillers, S. Cluzeau, M. Charreton, and Ph.D. dissertation, Virginia Polytechnic Institute and M.-H. Pham-Dele`gue. 2004. Effects of imidacloprid and State University, Blacksburg, VA. deltamethrin on associative learning in honeybees under Mausel, D. L., S. M. Salom, L. T. Kok, and G. Davis. 2010. semi-Þeld and laboratory conditions. Ecotoxicol. Envi- Establishment of the hemlock woolly adelgid predator, ron. Saf. 57: 410Ð419. Laricobius nigrinus (Coleoptera: Derondontidae), in the Doccola, J. J., P. M. Wild, I. Ramasamy, P. Castillo, and C. eastern United States. Environ. Entomol. 39: 440Ð448. Taylor. 2003. EfÞcacy of arborjet viper microinjections McClure, M. S. 1987. Biology and control of hemlock woolly in the management of hemlock woolly adelgid. J. Arboric. adelgid, pp. 1Ð9. Connecticut Agricultural Experiment 29: 327Ð330. Station, New Haven, CT. Elbert, A., B. Becker, J. Hartwig, and C. Erdelen. 1991. McClure, M. S. 1991. Density-dependent feedback and Imidacloprid- a new systemic insecticide. Pßanzenshutz- population cycles in Adelges tsugae (Homoptera: Adelgi- Nachr. Bayer 44: 113Ð136. dae). Environ. Entomol. 19: 36Ð43. Flowers, R. W., S. M. Salom, and L. T. Kok. 2006. Compet- McClure, M. S., S. M. Salom, and K. S. Shields. 2001. Hem- itive interactions among two specialist predators and a lock woolly adelgid, pp. 14. U.S. Dep. Agric. For. Ser. Publ. generalist predator of hemlock woolly adelgid, Adelges FHTET-2001-03, Morgantown, WV. tsugae (Hemiptera: Adelgidae) in south-western Virginia. Mizell, R. F., and M. C. Sconyers. 1992. Toxicity of imida- Agric. For. Entomol. 8: 253Ð262. cloprid to selected predators in the laboratory. Godman, R. M., and K. Lancaster. 1990. Silvics of North Fla. Entomol. 75: 277Ð280. America. U.S. Dep. Agric. Forest Service, Agriculture Moser, S. E., and J. J. Obrycki. 2009. Non-target effects of Handbook 654, Washington, DC. neonicotinoid seed treatments; mortality of coccinellid Grafton-Cardwell, E. E., and P. Gu. 2003. Conserving veda- larvae related to zoophytophagy. Biol. Control 51: 487Ð lia beetle, Rodolia cardinalis (Coleoptera: Coccinelli- 492. dae), in citrus: a continuing challenge as new insecticides Nauen, R., K. Tietjen, K. Wagner, and A. Elbert. 1998. Ef- gain registration. J. Econ. Entomol. 96: 1388Ð1398. Þcacy of plant metabolites of imidacloprid against Myzus Groot, A. T., and M. Dicke. 2002. Insect-resistant transgenic persicae and Aphis gossypii (Homoptera: Aphididae). Pes- plants in a multi-trophic context. Plant J. 31: 387Ð406. tic. Sci. 52: 53Ð57. 1234 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 103, no. 4

Nauen, R., U. Reckmann, S. Armborst, H. Stupp, and A. Suchail, S. D., G. De Sousa, R. Rahmani, and L. P. Belzunces. Elbert. 1999. Whiteßy-active metabolites of imidaclo- 2004. In vivo distribution and metabolism of 14C-imida- prid: biological efÞcacy and translocation in cotton cloprid in different compartments of Apis mellifera L. Pest plants. Pestic. Sci. 55: 265Ð271. Manag. Sci. 60: 1056Ð1062. Papachristos, D. P., and P. G. Milonas. 2008. Adverse effects Suchail, S., D. Guez, and L. P. Belzunces. 2001. Discrepancy of soil applied insecticides on the predatory coccinellid between acute and chronic toxicity induced by imida- Hippodamia undecimnotata (Coleoptera: Coccinellidae). cloprid and its metabolites in Apis mellifera. Environ. Biol. Control 47: 77Ð81. Toxicol. Chem. 20: 2482Ð2486. Reay, R. S. 1999. Management of eastern hemlock for deer Tingley, M. W., D. A. Orwig, R. Field, and G. Motzkin. 2002. wintering areas, pp. 144Ð147. In K. A. McManus, K. S. Avian response to removal of a forest dominant: conse- Shields, and D. R. Souto [eds.], Proceedings: Symposium quences of hemlock woolly adelgid infestations. J. Bio- on Sustainable Management of Hemlock Ecosystems in geogr. 29: 1505Ð1516. Eastern North America. U.S. Dep. Agric. For. Serv. Publ. Tomizawa, M., and J. Casida. 2003. Selective toxicity of neo- Gen. Tech. Rep. NE-267. nicotinoids attributable to speciÞcity of insect and mam- Rhea, J. R. 1995. Preliminary results for the chemical control malian nicotinic receptors. Annu. Rev. Entomol. 48: 339Ð of hemlock woolly adelgid in ornamental and natural 364. settings, pp. 113Ð125. In S. M. Salom, T. C. Tigner, and Walker, M. K., M.A.W. Stufkens, and A. R. Wallace. 2007. R. C. Reardon [eds.], Proceedings of the First Hemlock Indirect non-target effects of insecticides on Tasmanian Woolly Adelgid Review. U.S. Dep. Agric. For. Serv. Publ. brown lacewing (Micromus tasmaniae) from feeding on FHTET 96-10, Charlottesville, VA. lettuce aphid (Nasonovia ribisnigri). Biol. Control 43: Ross, R. M., R. M. Bennett, C. D. Snyder, J. A. Young, D. R. 31Ð40. Smith, and D. P. Lemarie. 2003. Inßuence of eastern Wang, A.-H., J.-C. Wu, Y.-S. Yu, J.-L. Liu, J.-F. Yue, and M.-Y. hemlock (Tsuga canadensis) on Þsh community structure Wang. 2005. Selective insecticide-induced stimulation and function in headwater streams of the Delaware River basin. Ecol. Freshw. Fish. 12: 60Ð65. on fecundity and biochemical changes in Tryporyze in- Schmuck, R. R. 1999. No causal relationship between Gau- certulas (Lepidoptera: Pyralidae). J. Econ. Entomol. 98: cho᭨ seed dressing in sunßowers and the French bee 1144Ð1149. syndrome. Pßanzenschutz-Nachr. Bayer 52: 257Ð299. Webb, R. E., J. R. Frank, and M. J. Raupp. 2003. Eastern Scho¨ning, R., and R. Schmuck. 2003. Analytical determina- hemlock recovery from hemlock woolly adelgid damage tion of imidacloprid and relevant metabolite residues by following imidacloprid therapy. J. Arboric. 29: 298Ð301. LC MS/MS. Bull. Insectol. 56: 41Ð50. Yamasaki, M., R. DeGraaf, and J. Lanier. 1999. Wildlife hab- Shields, K. S., R. F. Young, and G. P. Berlyn. 1995. Hemlock itat associations in eastern hemlockÑbirds, smaller mam- woolly adelgid feeding mechanisms, pp. 36Ð41. In S. M. mals, and forest carnivores, pp. 135Ð143. In K. A. Mc Salom, T. C. Tigner, and R. C. Reardon [eds.], Proceed- Manus, K. S. Shields and D. R. Souto [eds.], Proceedings: ings of the First Hemlock Woolly Adelgid Review. U.S. Symposium on Sustainable Management of Hemlock Dep. Agric. For. Serv. Publ. FHTET 96-10, Charlottes- Ecosystems in Eastern North America. U.S. Dep. Agric. ville, VA. For. Serv. Publ. Gen. Tech. Rep. NE-267. Silcox, C. A. 2002. Using imidacloprid to control hemlock Yang, E. C., Y. C. Chuang, Y. L. Chen, and L. H. Chang. 2008. woolly adelgid, pp. 280Ð287. In B. Onken, R. Reardon, and Abnormal foraging behavior induced by sublethal dosage J. Lashomb [eds.], Hemlock woolly adelgid in the eastern of imidacloprid in the honey bee. J. Econ. Entomol. 101: United States Symposium. U.S. Dep. Agric. For. Serv. 1743Ð1748. Publ., East Brunswick, NJ. Young, R. F., K. S. Shields, and G. P. Berlyn. 1995. Hemlock Smith, S. F., and V. A. Krischik. 1999. Effects of systemic woolly adelgid (Homoptera: Adelgidae): stylet bundle imidacloprid on Coleomegilla maculata (Coleoptera: Coc- insertion and feeding sites. Ann. Entomol. Soc. Am. 88: cinellidae). Environ. Entomol. 28: 1189Ð1195. 827Ð835. Snyder, C. D., J. A. Young, D. P. Lemarie, and D. R. Smith. Zilahi-Balogh, G.M.G., L. T. Kok, and S. M. Salom. 2002. 2002. Inßuence of eastern hemlock (Tsuga canadensis) Host speciÞcity of Laricobius nigrinus Fender (Co- forests on aquatic invertebrate assemblages in headwater leoptera: Derodontidae), a potential biological control streams. Can. J. Fish. Aquat. Sci. 59: 262Ð275. agent of the hemlock woolly adelgid, Adelges tsugae An- Stapel, J. O., A. M. Cortesero, and W. J. Lewis. 2000. Dis- nand (Homoptera: Adelgidae). Biol. Control 24: 192Ð198. ruptive sublethal effects of insecticides on biological con- Zilahi-Balogh, G.M.G., L. T. Kok, and S. M. Salom. 2003. trol: altered foraging ability and life span of a parasitoid Development and reproductive biology of Laricobius ni- after feeding on extraßoral nectar of cotton treated with grinus, a potential biological control agent o f Adelges systemic insecticides. Biol. Control 17: 243Ð249. tsugae. Biocontrol 48: 293Ð306. Suchail, S., L. Debrauwer, and L. P. Belzunces. 2003. Me- tabolism of imidacloprid in Apis mellifera. Pest Manag. Sci. 60: 291Ð296. Received 11 August 2009; accepted 11 April 2010.