Northeastern Naturalist NENHC-40 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 22001155 NORTHEASTERN NATURALIST 22(1):NENHC-40V–oNl.E 2N2H, NCo-5. 51 Long-Term Effects of Imidacloprid on Eastern Hemlock Canopy Arthropod Biodiversity in New England Wing Yi Kung1, Kelli Hoover1, Richard Cowles2, and R. Talbot Trotter III3,* Abstract - The systemic insecticide imidacloprid is commonly used to protect trees against attack by the Adelges tsugae (Hemlock Woolly Adelgid [HWA]), an invasive pest that threatens Tsuga canadensis (Eastern Hemlock) and T. caroliniana (Carolina Hemlock) in eastern North America. Although there have been some studies documenting the short-term (1–3 years) impact of imidacloprid on non-target arthropods in hemlock systems, almost nothing is known about the impact over longer time scales. Here, using a set of trees which were experimentally treated 3 and 9 years prior to this study, we found that while the impact of imidacloprid on HWA may be approaching the limits of detection and efficacy on trees treated 9 years ago, there is still an intermittently detectable impact on HWA density. Similarly, 9 years after application there is a subtle but detectable increase in arthropod richness and a shift in canopy-arthropod community composition. Results from the 3-year treated trees were, however, ambiguous, but may be the result of detectable cross-contamination of insecticide among trees. Introduction Adelges tsugae Annand (Hemiptera: Adelgidae; Hemlock Woolly Adelgid [HWA]), is an introduced insect pest that poses a serious threat to Tsuga canadensis L. Carriere (Eastern Hemlock) and Tsuga caroliniana (Carolina Hemlock) in eastern North America. These tree species provide a unique ecological niche for a wide diversity of flora and fauna (Ingwell et al. 2012, Jordan and Sharp 1967, Tingley et al. 2002), and are associated with changes in fish-community structure in riparian systems (Snyder et al. 2002). Stands infested with HWA commonly experience high rates of hemlock mortality, and stand structure can change rapidly (Orwig and Foster 1998, Orwig et al. 2002). In Connecticut, HWA has been present since at least 1985 and has led to hemlock-mortality rates as high as 95% (Orwig et al. 2002; Preisser et al. 2008, 2011), and the loss of this foundation species is likely to impact a diverse community of organisms associated with hemlock ecosystems (Ingwell et al. 2012). Efforts to manage the impacts of HWA have included biological control (Onken and Reardan 2011), identification of naturally occurring resistance (Ingwell et al. 2009), the development of hybrids (Montgomery et al. 2009), silvicultural methods (Fajvan 2008), and the use of systemic insecticides (Coots 2012, Cowles and Lagalante 2009, Cowles et al. 2006). Currently, the only readily available and proven 1Pennsylvania State University, College of Agricultural Sciences, Department of Entomology, University Park, PA 16802. 2Connecticut Agricultural Experiment Station, Windsor, CT 06095. 3USDA Forest Service, Northern Research Station, Hamden, CT 06513. *Corresponding author - rttrotter@fs.fed.us. Manuscript Editor: David Orwig Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-41 method for protecting individual trees is the application of insecticides, including imidacloprid. Commercial formulations labeled for use on hemlocks for the control of HWA facilitates the broad use of this insecticide, and its ease of application, low mammalian toxicity, and strong binding affinity to organic matter (Silcox 2002) have made it a valuable short-term solution for HWA infestations. Imidacloprid can, however, persist in the environment and potentially affect a broad range of arthropods (Kreutzweiser et al. 2008, Suchail et al. 2001), raising concerns about the understudied potential for long-term impacts on non-target organisms. Previous studies have documented impacts on non-target arthropods a year after application (Dilling et al. 2009), and have shown imidacloprid can remain effective against HWA up to 4 years post-application (Eisenback et al. 2014). Past research has also found that this insecticide can be detected in plant tissues up to 8 years postapplication (Cowles and Lagalante 2009). Yet, little is known about the long-term impacts of imidacloprid applications on non-target hemlock-canopy arthropods. In this study, we sought to take advantage of trees treated with imidacloprid in an early HWA-control study to evaluate the long-term impacts of its application. Using trees treated 3 and 9 years prior to this study, we addressed 3 key questions. First, is imidacloprid (and the metabolite olefin) still detectable in treated trees 3 and 9 years post-application? Second, is there a detectable effect of imidacloprid application on HWA 3 and 9 years post application? Third, is there evidence of effects on the alpha (within tree) and beta (among trees) diversity and community structure of the canopy arthropods found in Eastern Hemlock 3 and 9 years after insecticide application? Providing this information may facilitate long-term planning and strategies based on the use of this widely used insecticide. Methods Study history/tree selection The hemlock trees used in this study were originally selected as part of several previous studies designed to evaluate the efficacy of imidacloprid as a systemic insecticide to control HWA in Eastern Hemlock (Cowles et al. 2006). The first of these was established in 2002, when 28 HWA-infested Eastern Hemlock trees in Shenipsit State Forest (Stafford, CT, 41.96322N, 72.40436W) were used to evaluate the efficacy of multiple methods and seasons of imidacloprid application. Twentyfour trees were treated with imidacloprid using one of several application methods, the remaining 4 trees were used as controls (full description available in Cowles et al. 2006). In this original study, trees were treated in the fall of 2002 or the spring of 2003; however, seasonality was not found to play a significant role in insecticidal efficacy (Cowles et al. 2006), and so we consider the trees as a single cohort. In the spring and summer of 2011, we sampled 18 of the original study trees and 22 interspersed untreated trees (two of which were original control trees) of similar size and canopy position. For our samples, we collected canopy arthropods and foliage for insecticide-residue analysis, and undertook HWA-population surveys as described below. The time interval between insecticide treatments made to these trees and our sampling represents a unique opportunity to evaluate the long-term impacts of imidacloprid on non-target canopy arthropods. Northeastern Naturalist NENHC-42 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 A second group of trees was treated 3 years prior to the current study as part of an evaluation of multiple insecticides (including imidacloprid) for use in controlling HWA. The experimental trees were located in a windbreak at the Connecticut Agricultural Experiment Station Griswold Research Center (Voluntown, CT; 41.560197N, 71.876225W) (R. Cowles, unpubl. data). We selected trees from this group that were treated in 2008 with a single application of imidacloprid. The treatment consisted of 1 g of active ingredient per 2.5 cm DBH, injected into the soil within 30 cm of the base of the trunk using a Kioritz soil injector (Yamabiko Corp, Ome, Japan). Control trees were selected from the untreated trees within the same windbreak. To reduce the potential for contamination by applications made to nearby trees (trees within the windbreak were within 2–3 m of neighboring trees), we did not select untreated trees adjacent to trees which received chemical treatment, producing a minimum treated-control inter-tree distance of 6 m. We further limited study trees to those trees with adequate foliage for complete sampling. The selection criteria used yielded 15 treated trees and 14 control trees. The 2 sets of study trees treated in 2002 and 2008 are hereafter referred to as the 3-year and 9-year trees. Imidacloprid analyses In October 2011, we quantified imidacloprid and olefin levels in the needles from each of the treated and untreated trees in the 3- and 9-year groups. We collected 4 branches, one from each of 4 orthogonal directions, from each tree. The four branches were then combined and dried in paper bags for 2 months at room temperature (~24 °C). We then dislodged the dried needles from the branches, and ground 5–10 g of needles using a Wiley mill and a #40 screen. We added 5 ml of methanol to 0.5 g of the ground foliage and agitated the mixture on a platform rocker for 24 hours. Samples were then passed through a 0.2-μm filter, and a 0.5-ml aliquot was placed in an amber vial and sent to Dr. Anthony Lagalante of Villanova University, Villanova, PA, where it was analyzed for imidacloprid and olefin concentrations using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Additional details regarding these methods are available in Eisenback et al. (2010). HWA survey In January 2012, we assessed HWA density (sistens generation) on both the 9-year and 3-year trees. We sampled HWAs at this time because HWA matures and feeds during the winter, so individuals sampled in mid-winter had been exposed to the foliage and insecticide concentrations found in the foliage sampled the previous fall. Because HWA populations on a tree are often heterogeneously distributed (Evans and Gregoire 2007), we collected 5 randomly selected branches within reach of the ground from each tree. For each branch, we counted both dead and live HWA on 5 of the outermost stems (first bud-scale scar to current bud) with live buds and recorded the length of the foliage surveyed. Using this portion of the foliage limited sampling to live stems on only the most recent growth. We pooled values for each tree to produce a tree-level HWA density (HWA/cm) value. Because HWA densities were low in the winter of 2011/2012, we repeated HWA surveys in February 2013. However on these branches, we counted only live HWA, Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-43 which produce conspicuous wax-like secretions and are the survivors of the overwintering sistens generation. Arthropod-community sampling We conducted canopy-arthropod surveys in mid-summer (9–17 August 2011). Arthropods were collected from 2 branches of each tree using 1 of the following 2 methods. We sampled the first branch using a standard beat-sheet method in which a 1-m2 sheet (Bioquip Products Ripstop Beating Sheet, Rancho Dominguez, CA) was placed under a branch with approximately 1 m2 of foliage. We rapidly struck the branch 20 times with a 0.5-m piece of 1-inch PVC pipe to dislodge arthropods. There was a 3-cm hole at the center of the beat sheet with a 3 cm x 7 cm collection vial suspended below into which arthropods could slide. We used an aspirator to collect arthropods that did not quickly slide into this central vial. We sampled the second branch using a method that mirrored the first, except that prior to beating the branch, we fogged the branch with a knockdown fogging agent containing 1% pyrethrin (Pyrocide®100). The fog was applied using a commercially available home-owner-style propane-powered fogger (Fountainhead Group Inc., New York Mills, NY, models Burgess and Black Flag) modified with the addition of 61 cm of 10 cm-diameter flexible corrugated aluminum tube (dryer venting) with a 90° vertical bend that directed the fog upwards towards the branch. We added the fogging agent to augment the standard beat-sheet method which might otherwise have failed to capture more active, winged arthropods. We used pyrethrin because it has low mammalian toxicity and relatively short persistence (Crosby 1995, Kaneko 2011). Sampled branches were within 3 m of ground level. Because estimates of species richness are highly dependent on sampling effort (Schoener 1976, Trotter and Whitham 2011), we omitted from the analyses data from trees with canopies from which we were only able to sample a single branch, and so obtained a sample size of 15 treated and 14 control trees in the 3-year group. Each of the 18 treated and 22 control trees in the 9-year group yielded complete samples. Arthropod identification We identified arthropods to order and family and, when feasible, genus and species; however, we conducted our analyses using individuals sorted into “morphospecies” based on morphological similarities. Previous work by Oliver and Beattie (1996) and Siemann et al. (1998) have shown that a morphospecies approach is acceptable for the description of community composition and species richness. We stored arthropods in 70% ETOH and used representative specimens from each morphospecies to build a reference collection, which is archived in the Frost Entomological Museum at Pennsylvania State University (University Park, PA). Psocodea and Acari specimens were highly abundant, and identification to species, particularly with immature specimens, was not always possible. Consequently, we pooled juveniles in these taxa, with the recognition that species richness estimates for these groups is likely conservative. Northeastern Naturalist NENHC-44 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 Statistical analyses Because the 3-year and 9-year trees were separated by ~63 km, we expected a high level of variation in community composition which would be unrelated to imidacloprid, but rather to geographic differences. Thus, we limited our analyses of HWA densities and arthropod community composition to comparisons of trees within locations (i.e., within the 3- and 9-year groups). We compared imidacloprid concentrations in needles between treated trees in the 3-year and 9-year groups of trees using a Student’s t-test (R Core Team 2014). Comparisons of HWA densities between treated and control trees within the 3- and 9-year groups were made using the non-parametric Wilcoxon ranked-sum test (R Core Team 2014). Three measures of arthropod community structure—abundance, morphospecies richness, and community composition—were compared between treated and control trees within the 3- and 9-year tree groups. We compared individual abundance and morphospecies richness between the treated and control trees using the Wilcoxon ranked-sum test, and our results are presented using medians and an interquartile range (distance between 25th and 75th percentile) as measures of central tendency and dispersion. Abundance and morphospecies richness are dependent on sampling effort, so we generated rarefaction curves to qualitatively compare morphospecies richness (beta diversity or species turnover) in the 3-year and 9-year trees using EstimateS version 9 (Colwell 2013) with 1000 randomizations. Differences in community composition between treated and control trees were graphically depicted using non-metric multidimensional scaling (NMDS) based on Sorensen Bray-Curtis distances (McCune and Grace 2002) using random starting coordinates in 250 runs. We employed the non-parametric multi-response permutation procedure (MRPP) to statistically compare dissimilarities in community composition (using Sorensen Bray-Curtis distances) between treated and control trees within the 3-year and 9-year groups (McCune and Grace 2002). Indicator-species analysis was conducted to identify morphospecies that had a high abundance and fidelity associated with treated or control trees using 4999 permutations. NMDS, MRPP, and indicator-species analyses were conducted using PC-ORD (McCune and Mefford 2006). Post hoc analyses Eight treated trees in the 9-year group did not have detectable levels of imidacloprid, but 2 control trees in the 3-year group did, thus, we repeated the previously described comparisons and statistical analyses as post-hoc tests, using the presence/ absence of imidacloprid as an a-posteriori grouping variable. Results Imidacloprid analyses Within the 9-year group, 10 of the 18 treated trees had detectable levels of imidacloprid, while in the 3-year group all of the treated trees, and 2 of the untreated control trees had detectable levels of imidacloprid. We removed these 2 Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-45 contaminated control trees from analyses of community composition and HWA abundance, resulting in a reduction in the number of control trees from 14 to 12. As we expected, imidacloprid-residue levels were higher in the more recently treated 3-year trees compared to the 9-year treated trees (t = 6.2695, df = 14, P = 0.00002; Table 1). HWA surveys In January 2012, trees treated with imidacloprid 3 years prior to the study had lower total HWA densities than control trees (W = 2, P < 0.0001; Fig. 1A). However, Table 1. Imidacloprid levels detected in treated and control trees treated 9 and 3 years prior. Mean imidacloprid concentration (ppb) ± SE Treated tree Control tree 9-year trees 3.76 ± 5.04 (n = 18) 0.00 ± 0.00 (n = 22) 3-year trees 122.82 ± 73.41 (n = 15) 0.00 ± 0.00 (n = 12) Figure 1. Boxand- whisker plots showing the median (dividing line in box), quartile (upper and lower ends of box), 10th and 90th percentiles (ends of lines), and outliers (dots) for: (A) HWA densities (both live and dead HWA combined) in the 3-year and 9-year groups in January 2012, and (B) Live HWA densities in the 3-year and 9-year groups in February 2013. The symbol * indicates statistical significance (P < 0.05) using a Wilcoxon ranked-sum test. Northeastern Naturalist NENHC-46 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 comparison of the treated and control trees in the 9-year group revealed no statistically significant differences in HWA density (W = 146, P = 0.1598), though the trends were similar to those observed in the 3-year group. When we surveyed the trees in February 2013, the overall density of live HWA was higher in the untreated trees in both the 3- and 9-year groups (3-year W = 11, P < 0.0001; 9-year W = 124, P = 0.0457; Fig. 1B). Arthropod community structure Within the 3-year group, treated trees had a median of 78 arthropods (interquartile range = 75, n = 15) from a median of 30 morphospecies (interquartile range = 20), while control trees had a median of 103 arthropods (interquartile range = 85.5, n = 12) from a median of 32.5 morphospecies (interquartile range = 15.75). Neither richness nor abundance differed significantly (Wilcoxon: richness W = 75, P = 0.479; abundance W = 71, P = 0.373; Fig. 2A, B), indicating no differences in median alpha diversity between control trees and trees treated with imidacloprid 3 years prior. Figure 2. Boxand- whisker plots showing the median (dividing line in box), quartile (upper and lower ends of box), 10th and 90th percentiles (ends of lines), and outliers (dots) for: arthropod abundance (top) and richness (bottom) in treated and control trees in the 3-year (left) and 9-year (right) groups. The symbol * indicates statistical significance (P < 0.05) using a Wilcoxon ranked-sum test. Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-47 Sampled trees in the 9-year post-treatment group did however, exhibit a difference in median morphospecies richness, though the directionality was unexpected. Among the 9-year post-imidacloprid-application trees, treated trees had a median of 33 arthropod morphospecies (interquartile range = 13, n = 18); control trees yielded a median of 29 morphospecies (interquartile range = 7.25, n = 22, W = 274.5, P = 0.0385; Fig. 2C, D). The median arthropod abundance did not differ between treated (120, interquartile range = 90) and control trees (86.5, interquartile range = 43) in the 9-year group (W = 254.5, P = 0.128). Beta diversity (variation in species composition from one tree to the next) was also very similar between treated and control trees in both the 3-year and 9-year groups (Fig. 3), and the curves indicated the sample effort/ intensity was similar between the groups at both locations. Figure 3. Species- accumulation curves rarified by sample unit (trees) for t r e a ted and control trees in the 3-year (A) and 9-year (B) groups. Vertical lines represent the 95% confidence intervals. Northeastern Naturalist NENHC-48 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 Figure 4. NMDS ordinations based on canopy arthropods in treated and control trees in the 3- (A) and 9-year (B) groups. Comparison of community composition among the 3-year trees by ordination (NMDS, 2 dimensions) showed substantial overlap in community structure (Fig. 4A) between treated and control trees, with no statistical difference (MRPP P = 0.374). However, indicator-species analysis in the 3-year group identified 1 morphospecies with a statistically significant association with control trees (Table 2). In contrast to the findings in the 3-year group, community structure among the 9-year trees differed between treated and control groups (MRPP P = 0.041), though ordination suggested the communities’ composition included substantial overlap (NMDS, 2 dimensions; Fig. 4B). Indicator-species analysis also highlighted differences in the communities; treated trees yielded 5 indicator morphospecies, and control trees had 4 (Table 2). Post hoc analyses The lack of a difference in median alpha diversity between treated and control trees in the 3-year group initially appeared to be in conflict with previously published Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-49 data (Dilling et al. 2009) as did the directionality of the observed differences in alpha diversity in the 9-year group. As mentioned previously, analysis of foliage from the sampled trees revealed that in the 3-year group, 2 untreated trees had detectable levels of insecticide, while imidacloprid was not detected in 8 of the 9-year treated trees. To address this issue, we re-analyzed the data a posteriori with trees assigned to groups based on the presence or absence of detectable imidacloprid. Using these a posteriori groups, HWA densities were significantly lower in trees with detectable imidacloprid levels in both the 3-year and 9-year groups in both 2012 (Wilcoxon: 3-year P < 0.0001, 9-year P < 0.0356) and 2013 (Wilcoxon: 3-year P < 0.0001, 9-year P = 0.0076) (Fig. 5A, B). The 3-year trees yielded a median of 30 (interquartile range = 19; n = 17) morphospecies and 78 (interquartile range = 64) individual arthropods on trees with imidacloprid. Trees without imidacloprid had a median of 32.5 (interquartile range = 15.75; n = 12) morphospecies and 103 (interquartile range = 85.5) individual arthropods (Fig. 6A, B). However, similar to the results of our a priori analyses, there were no statistically significant differences in canopy-arthropod morphospecies abundance (W = 86, P = 0.4919) or richness (W = 80.5, P = 0.3524). Arthropod abundances among the 9-year trees with and without detectable levels of imidacloprid remained statistically indistinguishable when grouped a posteriori. Trees with detectable imidacloprid yielded a median of 120 (interquartile range = 59.5; n = 10) arthropods; trees without detectable imidacloprid yielded a median of 88 (interquartile range = 51; n = 30) arthropods (W = 199, P = 0.1297; Fig. 6C). The difference in morphospecies richness between treated and control 9-year trees detected using the a priori groups remained when based on the a posteriori grouping. Trees with detectable levels of imidacloprid had a median of 36.5 (interquartile range = 9.5) morphospecies, and trees without detectable imidacloprid had a median of 29.5 (interquartile range = 8.25) morphospecies (W = 231, P = 0.0118; Fig. 6D), indicating an increase in median alpha-level diversity associated with detectable imidacloprid. Table 2. Indicator species for treated and control trees treated 9 and 3 years prior. Indicator value is a composite of the value of % perfect indication based on abundance and fidelity. Mean IV is from randomized groups. Morphopecies Observed indicator (ID number) value (IV) Mean IV (± SD) P-value Imidacloprid Application Araneae (107) 41.7 17.1 (± 6.64) 0.011 Control 3-year Lepidoptera (03) 34.0 19.1 (± 5.43) 0.023 Treated 9-year Araneae (034) 39.1 17.6 (± 5.48) 0.006 Treated 9-year Araneae (115) 26.1 13.6 (± 5.08) 0.048 Treated 9-year Araneae (147) 38.2 20.9 (± 6.06) 0.016 Treated 9-year Araneae (162) 31.3 16.5 (± 5.43) 0.015 Treated 9-year Psocoptera (13) 50.6 36.5 (± 6.35) 0.034 Control 9-Year Thysanoptera (06) 59.7 43.8 (± 5.87) 0.016 Control 9-year Lepidoptera (14) 31.5 17.4 (± 5.48) 0.023 Control 9-year Araneae (241) 52.2 28.0 (± 6.54) 0.005 Control 9-year Northeastern Naturalist NENHC-50 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 Surprisingly, for both the 3-year and 9-year trees, a post-hoc analysis of community composition did not detect a significant difference between the a posteriori tree groups (MRPP: P = 0.241 and P = 0.160, respectively). Discussion These data show that nearly a decade after the application of imidacloprid to Eastern Hemlock trees, the insecticide was still detectable among many of the treated trees, and that there was still a significant, but intermittent impact on HWA populations. The detection of a difference in HWA densitiy in February 2013 (nearly 11 years after its application), and the failure to detect a difference the previous year suggest that the concentrations of imidacloprid within the trees may be at the threshold of detectable efficacy. Figure 5. Box-andwhisker plots showing the median (dividing line in box), quartile (upper and lower ends of box), 10th and 90th percentiles (ends of lines), and outliers (dots) for: HWA densities (both live and dead HWA combined) in 2012 (A), and 2013 (B) on trees treated 3-years and 9-years prior. The symbol * indicates statistical significance (P less than 0.05) using a Wilcoxon ranked-sum test. Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-51 The data also suggest that 9 years after application, the use of imidacloprid is associated with a change in species richness and community composition among canopy arthropods. The directionality of the shift in richness is, however, rather surprising with a statistically higher median alpha diversity associated both with treated trees as a group, and with those trees in which imidacloprid was found at detectable concentrations. The reasons for this pattern are not known, and merit further study. In addition to shifts in median morphospecies richness, the 9-year group also provided indications of differences in community structure as shown by the ordination. However, those differences may be subtle, and based on the loss of statistical significance associated with the reduction in sample sizes, these shifts in community structure may also, at 9-years post-pesticide application, be approaching the limits of detection. It is interesting to note that analysis by LC-MS/MS did not detect imidacloprid in 8 of the 9-year post-treatment trees, though there were differences in the composition of canopy-arthropod communities revealed by MRPP analysis. Additionally, several species may serve as indicators of imidacloprid treatment. In combination, these data raise the possibility that the sensitivity of the insect communities to imidacloprid in trees may exceed the sensitivity of the chemical analyses used to detect Figure 6. Boxand- whisker plots showing the median (dividing line in box), quartile (upper and lower ends of box), 10th and 90th percentiles (ends of lines), and outliers (dots) for: Arthropod abundance (top) and richness (bottom) in trees with and without detectable imidacloprid in trees treated 3-years (left) and 9-years (right) prior to sampling. Statistically significant differences are indicated by *. Northeastern Naturalist NENHC-52 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 it. The data also suggest that there is a strong need to evaluate the long-term impacts of the use of imidacloprid on arthropod communities because this study found an unexpected increase in richness associated with imidacloprid-treated trees. Whether this pattern is generalizable and the ecological importance of these changes remain unknown. In addition to the patterns found in the 9-year post-application trees, our data yielded a second surprise when analyses of the 3-year treated groups indicated differences in HWA density but not in community composition. A previous study by Dilling et al. (2009), in which arthropods were sampled approximately 1 year after imidacloprid application, found substantial impacts on both alpha diversity and community structure. We detected no differences only 3 years after application. Though interesting, our data have critical limitations, specifically the high potential for contamination by both imidacloprid and other insecticides within the control trees. The trees within the 3-year study group are arranged in a windbreak with 2–3 m separating each tree, a spacing which results in trees with interdigitated canopies, and a high potential for root grafting in this shallow-rooted species (Eckenwalder 2009, Fowells 1965, Frothingham 1915). The short distances also create a high potential for lateral movement of imidacloprid through the soil (Felsot et al. 1998, Gupta et al. 2002). Both of these conditions could result in the presence of imidacloprid in varying concentrations within the control trees. It should also be noted that other insecticides were evaluated in the original study (e.g., dinotefuran); we did not test for the presence of these compounds and so their presence as a result of lateral contamination from nearby applications is unknown. Similarly, the study trees are located along the edge of an active experimental agricultural field used to grow products including fruit trees and corn, and contamination of our study trees from insecticide applications made to the adjacent field cannot be ruled out. We suggest that these factors may have influenced the lack of pattern in the canopy arthropod communities in our study trees. Overall, our data indicate the use of imidacloprid for the control of HWA in Eastern Hemlock stands has the potential to provide protection from these invasive species for up to a decade, though it is worth noting that the HWA densities observed during this study were quite low, and the efficacy of the concentrations within the trees subject to high HWA pressures is not known. Although the duration of the impact may provide land-managers with options and flexibility regarding the timing and frequency of imidacloprid applications, our data also suggest that the legacy effects of the use of this systemic insecticide can be longlasting for both the target species and non-target communities. Replication of this study is needed to further examine the stability, repeatability, and ecological meaning of the changes in arthropod richness and community composition found in Eastern Hemlocks nearly a decade after treatment with imidacloprid. Disclaimer: The use of trade, firm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement or approval by the US Department of Agriculture or the Forest Service of any product or service to the exclusion of others that may be suitable. Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-53 Literature Cited Colwell, R.K. 2013. EstimateS: Statistical estimation of species richness and shared species from samples. Version 9. Available online at http://www.purl.oclc.org/estimates. Accessed 23 July 2014. Coots, C.I. 2012. Spatial, temporal, and tri-trophic distribution of imidacloprid, olefin, and 5-hydroxy and their effect on Hemlock Woolly Adelgid, Adelges tsugae Annad (Hemiptera: Adelgidae). Ph.D. Dissertation. University of Tennessee, Knoxville, TN. Cowles, R.S., and A.F. Lagalante. 2009. Activity and persistence of systemic insecticides for managing Hemlock Woolly Adelgids. Proceedings of the 20th US Department of Agriculture Interagency Research Forum on Invasive Species. USDA Forest Service General Technical Report NRS-P-51:17–18. Northern Research Station, Newtown Square, PA. 114 pp. Cowles, R.S., M.E. Montgomery, and C. Cheah. 2006. Activity and residues of imidacloprid applied to soil and tree trunks to control Hemlock Woolly Adelgid (Hemiptera: Adelgidae) in forests. Journal of Economic Entomology 99:1258–1 267. Crosby, D.G. 1995. Environmental fate of pyrethrins. Pp. 194–213, In J.E. Casida and G.B. Quistand (Eds.). Pyrethrum Flowers: Production, Chemistry, Toxicology, and Uses. Oxford University Press, New York, NY. 356 pp. Dilling, C., P. Lambdin, J. Grant, and R. Rhea. 2009. Community response of insects associated with Eastern Hemlock to imidacloprid and horticultural oil treatments. Environmental Entomology 38:53–66. Eckenwalder, J.E. 2009. Conifers of the World: The Complete Reference. Timber Press Portland, OR. 720 pp. Eisenback, B.M., S.M. Salom, L.T. Kok, and A.F. Lagalante. 2010. Lethal and sublethal effects of imidacloprid on Hemlock Woolly Adelgid (Hemiptera: Adelgidae) and two introduced predator species. Journal of Economic Entomology 103 :1222–1234. Eisenback, B.M., S.M. Salom, L.T. Kok, and A.F. Lagalante. 2014. Impacts of trunk and soil injections of low rates of imidacloprid on Hemlock Woolly Adelgid (Hemiptera: Adelgidae) and Eastern Hemlock (Pinales: Pinaceae) health. Journal of Economic Entomology 107:250–258. Evans, A.M., and T.G. Gregoire. 2007. The tree-crown distribution of Hemlock Woolly Adelgid, Adelges tsugae (Hem., Adelgidae) from randomized branch sampling. Journal of Applied Entomology 131:26–33. Fajvan, M.A. 2008. The role of silvicultural thinning in eastern forests threatened by Hemlock Woolly Adelgid (Adelges tsugae). USDA Forest Service General Technical Report PNW-GTR 733:247-256. Pacific Northwest Research Station, Portland, OR. 306 pp. Felsot, A.S., W. Cone, J. Yu, and J.R. Ruppert. 1998. Distribution of imidacloprid in soil following subsurface drip chemigation. Bulletin of Environmental Contamination and Toxicology 60:363–370. Fowells, H.A. 1965. Silvics of Forest Trees of the United States. Agriculture Handbook 271. US Department of Agriculture Forest Service, Washington, DC. 762 pp. Frothingham, E.H. 1915. The Eastern Hemlock. Agricultural Bulletin 152. US Department of Agriculture Forest Service, Washington, DC. 31 pp. Gupta, S., V.T. Gajbhiye, and N.P. Agnihotri. 2002. Leaching behavior of imidacloprid formulations in soil. Bulletin of Environmental Contamination and Toxicology 68:502–508. Ingwell, L., J. Brady, M. Fitzpatrick, B. Maynard, R. Casagrande, and E.L. Preisser. 2009. Intraspecific variation in Tsuga canadensis foliar chemistry. Northeastern Naturalist 16:585–594. Northeastern Naturalist NENHC-54 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 Vol. 22, No. 1 Ingwell, L.L., M.R. Miller-Pierce, R.T. Trotter III, and E.L. Preisser. 2012. Vegetation and invertebrate-community response to Eastern Hemlock decline in southern New England. Northeastern Naturalist 19:541–558. Jordan, J.S., and W.M. Sharp. 1967. Seeding and planting hemlock for Ruffed Grouse cover. USDA Forest Service Research paper NE-83:1–7. Northeastern Forest Experiment Station, Upper Darby, PA. Kaneko, H. 2011. Pyrethroids: Mammalian metabolism and toxicity. Journal of Agricultural and Food Chemistry 59:2786–2791. Kreutzweiser, D.P., K.P. Good, D.T. Chartrand, T.A. Scarr, and D.G. Thompson. 2008. Toxicity of the systemic insecticide, imidacloprid, to forest-stream insects and microbial communities. Bulletin of Environmental Contamination and Toxicology 80:211–4. McCune, B., and J.B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, OR. 300 pp. McCune, B. and M.J. Mefford. 2006. Multivariate analyses of Ecological Data. Version 5.31. MjM Software, Gleneden Beach, OR. Montgomery, M.E., S.E. Bentz, and R.T. Olsen. 2009. Evaluation of hemlock (Tsuga) species and hybrids for resistance to Adelges tsugae (Hemiptera: Adelgidae) using artificial infestation. Journal of Economic Entomology 102:1247–1254. Oliver, I., and A.J. Beattie. 1996. Invertebrate morphospecies as a surrogate for species: A case study. Conservation Biology 10:99–109. Onken, B., and R. Reardon. 2011. Implementation and status of biological control of the Hemlock Woolly Adelgid. US Forest Service Forest Health Technology Enterprise Team FHTET-2011-04. Morgantown, WV. 230 pp. Orwig, D.A., and D.R. Foster. 1998. Forest response to the introduced Hemlock Woolly Adelgid in southern New England, USA. Journal of the Torrey Botanical Society 11:437–452. Orwig, D.A., D.R. Foster, and D.L. Mausel. 2002. Landscape patterns of hemlock decline in New England due to the introduced Hemlock Woolly Adelgid. Journal of Biogeography 29:1475–1487. Preisser, E.L., A.G. Lodge, D.A. Orwig, and J.S. Elkinton. 2008. Range expansion and population dynamics of co-occuring invasive herbivores. Biological Invasions 10:201–213. Preisser, E.L., M.R. Miller-Pierce, J. Vansant, and D.A. Orwig. 2011. Eastern Hemlock (Tsuga canadensis) regeneration in the presence of Hemlock Woolly Adelgid (Adelgids tsugae) and Elongate Hemlock Scale (Fiorinia externa). Canadian Journal of Forest Research 41:2433–2439. R Core Team. 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at http://www.R-project. org. Accessed 1 October 2014. Schoener, T.W. 1976. The species-area relationship within archipelagos: Models and evidence from island landbirds. Pp. 629–642, In H.J. Frith and J.H. Calaby (Eds.). 16th International Ornothological Conference. Australian Academy of Science, Canberra, Australia. 765 pp. Siemann, E., D. Tilman, J. Haarstad, and M. Ritchie. 1998. Experimental test of the dependence of arthropod diversity on plant diversity. American Naturalist 152:738–750. Silcox, C.A. 2002. Using imidacloprid to control Hemlock Woolly Adelgid. Pp. 280–287, In B.Onken, R.Reardon, and J. Lashomb (Eds.). Proceedings: Hemlock Woolly Adelgid in the Eastern United States Symposium. The State University of New Jersey Rutgers, NJ. Available online at http://www.fs.fed.us/na/morgantown/fhp/hwa/pub/proceedings/ imidacloprid.pdf. Accessed 1 October 2014. Northeastern Naturalist Vol. 22, No. 1 W.Y. Kung, K. Hoover, R. Cowles, and R.T. Trotter III 2015 NENHC-55 Snyder, C.M., X. Zhang, and L.J. Wysocki. 2002. Influence of Eastern Hemlock (Tsuga canadensis) forests on aquatic invertebrate assemblages in headwater streams. Canadian Journal of Fisheries and Aquatic Sciences 59:262–275. Suchail, S., D. Guez, and L.P. Belzunces. 2001. Discrepancy between acute and chronic toxicity induced by imidacloprid and its metabolites in Apis mellifera. Environmental Toxicology and Chemistry 20:2482–2486. Tingley, M.W., D.A. Orwig, R. Field, and G. Motzkin. 2002. Avaian response to removal of a forest dominant: Consequences of Hemlock Woolly Adelgid infestation. Journal of Biogeography 29:1505–1516. Trotter III, R.T., and T.G.Witham. 2011. The roles of climate variability and habitat heterogeneity in structuring a forested system: An integrated accumulation function of species richness. Journal of Sustainable Forestry 30:721–735.