A Comparison of Girdled Ash Detection Trees and Baited Artificial Traps for \u003cI\u003eAgrilus...

$: A Comparison of Girdled Ash Detection Trees and Baited Artificial Traps for \u003cI\u003eAgrilus planipennis\u003c/I\u003e (Coleoptera: Buprestidae) Detection$
SAMPLING

A Comparison of Girdled Ash Detection Trees and Baited ArtificialTraps for Agrilus planipennis (Coleoptera: Buprestidae) Detection

RODRIGO J. MERCADER,1 DEBORAH G. MCCULLOUGH,2,3 AND JOHN M. BEDFORD4

Environ. Entomol. 42(5): 1027Ð1039 (2013); DOI: http://dx.doi.org/10.1603/EN12334

ABSTRACT Detection of newly established populations of Agrilus planipennis Fairmaire, the mostdestructive forest insect to invade the United States, remains challenging. Regulatory agenciescurrently rely on artiÞcial traps, consisting of baited three-sided panels suspended in the canopy ofash (Fraxinus spp.) trees. Detection trees represent another survey option. Ash trees are girdled inspring to attract ovipositing A. planipennis females then debarked in fall to assess larval presence anddensity. From 2008Ð2010, systematic grids of detection trees and artiÞcial traps were established acrossa 390-km2 area for the SLow Ash Mortality pilot project. We compared probabilities of detectionassociated with detection trees and artiÞcial traps along varying A. planipennis density proxiesestimated as distance-weighted averages of larval counts (detection trees) or adult captures (traps)within 800 m of each detection tree or trap. Detection trees were consistently more likely to bepositive, that is, detect A. planipennis, than traps in all three years, even when traps were placed incanopies of detection trees. Probability of detection with a single detection tree was �50% whendensity proxies for the area were �5 larvae per detection tree, while the probability of detection withan artiÞcial trap placed in the same area was �35%, even when density proxies exceeded 25 larvae perdetection tree. At very low densities of �5 larvae per detection tree, using three detection trees wouldincrease detection probabilities to 90%, while Þve artiÞcial traps would increase the detectionprobability only to 40%.

KEY WORDS invasive forest pest, Fraxinus spp., detectability, Manuka oil, survey method

More than 450 species of non-native forest insects areknown to be established in the United States (Aukemaet al. 2010). Approximately 15% of these alien speciescause signiÞcant damage to trees in forest or urbanforest settings in the United States (Aukema et al.2010), costing landowners, municipalities, and plant-based industries billions of dollars annually (Kovacs etal. 2010, 2011; Aukema et al. 2011). Nearly all non-native phytophagous insects arrive in new habitatsthrough accidental transportation with commodities,baggage, or as hitchhikers. Many efforts, includingpathway analyses, preshipping treatment, quaran-tines, and inspections, have been implemented to pre-vent new introductions of non-native organisms intothe United States and other countries. These effortsare largely offset, however, by burgeoning rates ofinternational trade and travel (Aukema et al. 2010),making it likely that introductions of non-native, po-tentially invasive insects will continue (National Re-

search Council 2002, Work et al. 2005, Brockerhoff etal. 2006, McCullough et al. 2006).

Developing effective detection and monitoringtools is a high priority when a new and potentiallyinvasive insect pest is identiÞed. Early detection al-lows implementation of management actions duringthe initial stages of establishment that can slow growthand spread of the pest population, potentially decreas-ing damage and possibly providing an opportunity foreradication (Shigesada and Kawasaki 1997, Myers etal. 2000, Liebhold and Tobin 2008, Suckling et al. 2012,Tobin et al. 2013). Tools capable of detecting low pestdensities are also necessary for accurately delimitingthe extent of an infestation and measuring effects ofmanagement activities. Ideally, a detection and mon-itoring tool should enable program managers to re-duce uncertainty regarding the presence and densityof a pest at an acceptable cost. Accuracy and precisionof different detection tools can vary with pest density,however, affecting the costÐbeneÞt relationship. Inaddition, not all detection tools may be available orappropriate for use under all circumstances.

For insects that produce long-range sex or aggre-gation pheromones, identiÞcation of active com-pounds can facilitate production of highly attractivelures. Pheromone lures for numerous Lepidopteranfoliage feeders and some scolytine bark beetles, forexample, are commercially available, enabling baited

1 Department of Biology, 1700 SW College Ave., Washburn Uni-versity, Topeka, KS 66621.

2 Department of Entomology and Department of Forestry, 243Natural Sciences Bldg., 288 Farm Lane, Michigan State University,East Lansing, MI 48824.

3 Corresponding author, e-mail: [email protected] Michigan Department of Agriculture and Rural Development, 525

West Allegan St, Lansing, MI 48909.

0046-225X/13/1027Ð1039$04.00/0 � 2013 Entomological Society of America

traps to be used locally or distributed across largegeographic regions to detect or track infestations (e.g.,Borden 1989, El-Sayed 2011).

Detection of invasive insects that do not producelong-range pheromones presents a greater challenge.This situation is exempliÞed by the emerald ash borer,Agrilus planipennis Fairmaire, an invasive pest nativeto Asia Þrst identiÞed in Detroit, MI, and Windsor,ON, Canada, in 2002 (Cappaert et al. 2005). In itsnative range, A. planipennis is considered a secondarypest, colonizing stressed trees (Yu 1992, Poland andMcCullough 2006). North American ash species, how-ever, lack a co-evolutionary history withA.planipennisand even healthy trees are colonized and killed (Cap-paert et al. 2005, Wei et al. 2007, Anulewicz et al. 2008,Rebek et al. 2008). To date,A. planipennis populationsare established in 20 states and two Canadian prov-inces (EAB.info 2013), and tens of millions of ash(Fraxinus spp.) trees in forests and urban settings havebeen killed (EAB.info 2013). Recent analyses indicateA. plannipennis has become the most costly forestinsect to ever invade the United States (Kovacs et al.2010, Aukema et al. 2011).

Like its North American congeners, A. planipennisdoes not produce long-range sex or aggregation pher-omones, although short-range or contact pheromonesmay be involved in mating behavior (Lelito et al. 2009;Silk et al. 2009, 2011; Ryall et al. 2012). Adult A. pla-nipennis rely on visual cues and host volatiles to locateash trees and potential mates (RodriguezÐSaona et al.2007, Crook et al. 2009, Crook and Mastro 2010). Visualsurveys to identify infested ash trees, used extensivelyafter the initial identiÞcation of A. planipennis, areproblematic. Infested trees typically exhibit few ex-ternal symptoms for at least 4 yr, until larval densitiesbuild to levels high enough to disrupt nutrient andwater transport (Cappaert et al. 2005, Siegert et al.2010, Tluczek et al. 2011).

Currently, the tools used most commonly for A.planipennis detection and monitoring include girdledash trees and artiÞcial traps baited with ash volatiles.AdultA. planipennis are attracted to and preferentiallyoviposit on ash trees stressed by girdling (McCulloughet al. 2009a,b), likely in response to stress-inducedchanges in volatile emissions (RodriguezÐSaona et al.2006) and perhaps light reßectance (Bartels et al.2008). Grids of ash trees girdled in spring, then de-barked in fall or winter to assess A. planipennis larvalpresence, for example, “detection trees,” have beenused for surveys in Michigan, Ohio, and other statessince 2005 (Rauscher 2006, Hunt 2007, Poland andMcCullough 2010, SLAMEAB.info 2012). In previousstudies, A. planipennis larval densities were consis-tently higher in girdled trees than in trees stressed bywounding or herbicides, or baited with volatile at-tractants (McCullough et al. 2009a,b; Tluczek et al.2011). Although girdled trees are effective for A. pla-nipennis detection or monitoring, debarking girdledtrees can be labor intensive and trees suitable forgirdling may not always be available or accessible.

Beginning in 2008, regulatory agencies have reliedon artiÞcial traps baited with lures containing volatile

compounds associated with ash foliage or bark andwood for A. planipennis surveys (U.S. Department ofAgricultureÐAnimal and Plant Health Inspection Ser-vice [USDAÐAPHIS] 2008, 2010). Numerous Þeldstudies have compared trap designs, colors, and luresto see which combination captured the greatest num-ber of A. planipennis beetles (Francese et al. 2005,2010; Marshall et al. 2009, 2010a; Crook and Mastro2010; Poland et al. 2011). Traps currently used by thenational A. planipennis program in the United Statesconsist of three-sided plastic prisms in speciÞc shadesof purple or green attractive to A. planipennis (Crooket al. 2009, Francese et al. 2010). Traps are assembledon-site, coated with a clear sticky substance to capturebeetles, then placed in the canopy of ash trees.

Relatively few trials have attempted to compare theefÞcacy of both girdled ash trees and artiÞcial traps. Ina 16-ha forested site with a very low A. planipennisdensity,McCulloughet al. (2011) found larvae in100%of the girdled ash trees when they were debarked infall, while only 25% of the purple prism traps baitedwith Manuka oil and suspended in ash trees captureda single A. planipennis adult. Other studies recordedcaptures of adultA.planipennisbeetles on sticky bandswrapped around the circumference of girdled or un-girdled ash trees and purple or green prism traps(Marshall et al. 2009, 2010b). Results were affected byvarious factors including tree size or exposure to sun,A. planipennis population levels, trap height, and com-pounds used as lures. When girdled trees were de-barked, however, larvae were consistently presenteven in trees that captured no adult beetles on stickybands (Marshall et al. 2009, 2010b).

Much is still unknown about the relative effective-ness of artiÞcial traps and girdled detection trees atvaryingA. planipennis densities or whether these toolscan accurately estimateA.planipennispopulation den-sities. Data collected from 2008 to 2010 as part of theSLow Ash Mortality (SLAM) pilot project (SLAMEAB.info 2012) provided a unique opportunity to measuredetection trees and artiÞcial traps across a wide rangeof A. planipennis population levels. The SLAM pilotproject was a large-scale operational program thatinvolved personnel from universities, and state andfederal agencies. Objectives included development,implementation, and measurement of an integratedmanagement strategy to determine whether EAB pop-ulation growth or the progression of ash mortalitycould be delayed in a localized outlier infestation. Amajor aspect of the SLAM operation involved inten-sively monitoringA.planipennis annually using girdledash detection trees and artiÞcial traps deployed sys-tematically across the project area that encom-passed nearly 400 km2. Our speciÞc goals in thisarticle were to use data from the SLAM pilot project

to compare the ability of the detection trees andartiÞcial traps to 1) detect A. planipennis and 2)

estimate local population densities across the SLAMproject area.

1028 ENVIRONMENTAL ENTOMOLOGY Vol. 42, no. 5

Materials and Methods

In September 2007, an isolated A. planipennis infes-tation was identiÞed near Moran, Mackinac Co., in theUpper Peninsula of Michigan, when larvae were foundon a girdled ash tree debarked by state surveyors.Additional ash trees in the immediate area were sub-sequently felled and debarked, resulting in identiÞ-cation of 10 more infested trees. Ash trees in the areaappeared healthy and nonsymptomatic, but examina-tion of cross-sections from infested trees indicated theA. planipennis infestation likely originated at least 4 yrearlier. This relatively recent and localized infestationwas subsequently selected as the focus of the SLAMpilot project to measure the effectiveness of integrat-ing management options to reduce the progression ofash mortality in the area. Activities undertaken as partof the SLAM pilot project included extensive ash in-ventories and establishment of long-term plots tomonitor ash condition, along with intensive surveysof A. planipennis (Poland and McCullough 2010,SLAMEAB.info 2012). Cooperators in the SLAM pilotproject established an irregularly shaped project areaencompassing 390 km2, which extended 20Ð29 kmnorth from the Mackinac Bridge and 15Ð22 km west ofLake Huron in 2008. The SLAM project area encom-passed a mixture of private and federal forest lands,and rural, residential, and municipal areas.Detection Trees. In spring 2008, systematic grids of

girdled ash detection trees and artiÞcial traps (seebelow) were established at three densities (Table 1)to surveyA. planipennis distribution across the projectarea. The dense survey grid (16 detection trees per 2.6km2) was centered on the area where infested treeswere identiÞed in 2007 and was surrounded by thecoarse survey grid (four detection trees per 2.6 km2)(Table 1). The standard survey grid (one detectiontree per 2.6 km2) (Table 1) extended from the coarsegrid to the perimeter of the project area. Numerousgrid cells were excluded from the A. planipennis sur-

vey because no ash trees were present or, in a fewcases, because survey crews were unable to securelandowner permission for access.

From 12 May to 30 May 2008, in total, 383 girdleddetection trees were established; 56% were in thedense grid, 26% were in the coarse grid, and 18% werein the standard grid (Table 1). Crews were directed toselect trees that were 10Ð15 cm diameter at breastheight (DBH; measured 1.3 m above ground). Thissize was considered optimal to eliminate the risk ofsmaller trees breaking during high winds and to avoidthe excessive amount of time required to debark andexamine larger trees. Because A. planipennis beetleactivity is consistently higher in sunny areas than inshade (Yu 1992; McCullough et al. 2009a,b), crewswere encouraged to preferentially select open-growntrees, or trees growing along roadways, the edge ofwoodlands, or in canopy gaps. Trees were girdledusing drawknives to remove the outer bark andphloem from a 15- to 30-cm-wide band around thecircumference of the trunk, �1.1Ð1.3 m above ground(McCullough and Siegert 2008). Global positioningsystem coordinates of each tree were recorded. De-tection trees could not be established in some gridcells located in areas designated as protected wilder-ness or as critical habitat for a federally listed endan-gered dragonßy.

Detection trees were felled and examined from 15September to 19 December 2008. Surveyors carefullyremoved outer bark on the trunk and any branches �5cm in diameter with drawknives. Thin layers ofphloem were gently shaved away until sapwood wasexposed. Larval counts were recorded for each de-tection tree. Representative samples of larvae werecollected from infested trees and placed into vials withethanol. If an intact larva could not be recovered, atleast one suspect gallery was collected, bagged, andsaved for subsequent inspection. Debarked trees andlimbs in forested areas or along secondary roads werebucked into small sections to speed decomposition. Indeveloped or residential areas, logs and tops werecollected and transported to a disposal yard.

Similar methods were used to establish girdled de-tection trees in 2009 and 2010, although the portion ofthe project area surveyed intensively (e.g., the densegrid) increased. From 11 May to 20 May 2009, in total,579 girdled detection trees were established in theSLAM project area; 70% were in the dense survey grid,18% were in the coarse grid, and 12% were in thestandard survey grid (Table 1). In 2010, crews estab-lished a total of 748 detection trees from 3 May to 4June, with 65, 30, and 5% in the dense, coarse grid, andstandard grids, respectively (Table 1). Quality controlmeasures implemented in 2009 and 2010 included in-spection of up to 10% of the detection trees by super-visors, to ensure tree selection, girdling, and data col-lection were completed appropriately. Detectiontrees were debarked to assess presence of A. plani-pennis larvae from 15 September to 19 December 2009and from 13 September to 29 October 2010.Artificial Traps.ArtiÞcial traps and lures used in the

SLAM pilot project each year were provided by the

Table 1. Area and number of grid cells, girdled detection trees,and artificial traps, in the dense, coarse, and standard systematicsurvey grids used to monitor A. planipennis in the SLAM pilotproject area from 2008 to 2010

Survey grids 2008 2009 2010

DenseTotal area (km2) 58 78 108Grid cell size (ha) 16.2 16.2 16.2No. grid cells 363 489 684No. detection trees 215 406 483No. artiÞcial traps 223 269 418

CoarseTotal area (km2) 86 79 152Grid cell size (ha) 64.7 64.7 64.7No. grid cells 136 125 242No. detection trees 99 101 226No. artiÞcial traps 2 25 44

StandardTotal area (km2) 247 233 124Grid cell size (ha) 259 259 259No. grid cells 110 104 58No. detection trees 69 72 39No. artiÞcial traps 4 37 13

October 2013 MERCADER ET AL.: ASH DETECTION TREES AND ARTIFICIAL TRAPS 1029

USDAÐAPHIS. Design, lures, and placement of theartiÞcial traps followed guidelines issued by USDAÐAPHIS for the national A. planipennis detection pro-gram (USDAÐAPHIS 2008). Each artiÞcial trap con-sisted of a three-sided prism constructed from purplecorrugated plastic (4 mm in thickness). The threepanels were each �60 cm tall by 40 cm wide and thesurface of the panels was coated with clear Tanglefoot.A lure with Manuka oil, released from a pouch at 50mg/d (Synergy Semiochemicals Corp., Burnaby, B.C.,Canada), was attached to each trap. Compounds inManuka oil, derived from the New Zealand tea tree,Leptospermum scoparium J. R. and C. Forst (Myrta-ceae), are similar to compounds associated with ashwood and bark. Traps baited with Manuka oil lureshave been show to be attractive to A. planipennis(Crook et al. 2008; Grant et al. 2010, 2011; Poland et al.2011). Baited traps were suspended in the canopy, atleast 1.5 m aboveground, on the sunny or exposed sideof ash trees.

Survey crews established a total of 229 artiÞcialtraps from 2 June to 20 June 2008 in the standard,coarse, and dense trapping grids (1, 4, or 16 traps per2.6 km2, respectively) (Table 1). Crews placed 223 ofthe artiÞcial traps in the dense survey grid (97%); 171of those traps (75%) were placed in the canopy ofgirdled detection trees (Table 1). In the coarse andstandard grids, artiÞcial traps were established only ingrid cells where ash trees were present but could notbe used as girdled detection trees (e.g., protectedwilderness). Global positioning system coordinates ofall artiÞcial traps were recorded. Traps were collectedfrom 18 August to 12 September 2008. Suspect A.planipennis beetles were removed from traps, placedinto individual vials with ethanol, labeled, and re-turned to Lansing, MI, for identiÞcation by the SurveyProgram Manager, MI Department of Agriculture. Afew traps were blown out of trees and could not berecovered.

Survey crews installed 331 artiÞcial traps from 1June to 19 June 2009 and 475 artiÞcial traps between7 June and 25 June 2010 (Table 1). The same three-sided prism trap design was used as in 2008, but in 2009and 2010, traps were baited with a lure consisting of an80:20 ratio of Manuka and Phoebe oils, emitted frompouches at 50 mg per day. Like Manuka oil, Phoebe oil,derived from the Brazilian walnut tree, Phoebe porosaMez (Lauraceae), is attractive to A. planipennis bee-tles (Grant et al. 2011, Poland et al. 2011). In 2009, 81%of the traps were placed in the dense survey grid while8 and 11% of the traps were placed in the coarse andstandard grids, respectively. Thirty percent of thetraps in the dense grid were placed in the canopy ofdetection trees. When artiÞcial traps were not placedin the canopy of a detection tree, surveyors prefer-entially selected ash trees �20 cm DBH, growing inopen areas or along edges. In 2010, no artiÞcial trapswere placed on detection trees and when possible,traps were placed in ash trees at least 50 m from adetection tree. In 2010, 88% of the artiÞcial traps wereplaced in the dense survey grids, 9% in the coarse grid,and 3% in the standard grid (Table 1). As in 2008,

artiÞcial traps were the only survey tool used in gridcells that fell within designated wilderness or pro-tected areas. ArtiÞcial traps and all suspect beetlescaptured on traps were collected from 17 August to 12September 2009 and 16 August to 10 September 2010.Analysis of Paired Detection Trees and ArtificialTraps.Using data from the 177 artiÞcial traps placed ondetection trees in 2008 and the 99 artiÞcial traps placedon detection trees in 2009, we contrasted the proba-bilities that an artiÞcial trap placed on a detection treeand the detection tree itself would be detected asinfested (i.e., “positive”) using a generalized linearmixed model with a logit link in the glmmPQL func-tion in the MASS package for the statistical package R(R Development Core Team 2010). For these analy-ses, the paired detection tree and artiÞcial trap wereconsidered the experimental unit and the pairedgroup was treated as a random effect. This enabled usto avoid potentially confounded effects between bee-tles attracted to detection trees and beetles capturedon artiÞcial traps. We also included an estimate of themean local A. planipennis population, as describedbelow, using larval counts from detection trees. Thetotal number of larvae counted on each detection treewas used for this and subsequent analyses because ofthe uniform size of the detection trees. We similarlyused the total number of adults captured on eachartiÞcial trap because of the uniform size of the traps.On average (�SEM), DBH of the detection trees was12.1 � 0.11 cm, corresponding to �2.8 m2 of phloemsurface area (McCullough and Siegert 2007), whilesurface area of each trap was 0.72 m2.Probability of Detection Estimates. In 2009, 232

artiÞcial traps were placed in ash trees that were notgirdled and in 2010, all of the artiÞcial traps wereplaced on nongirdled ash trees. Larval counts andnumber of adult beetles from unpaired detection treesand artiÞcial traps, respectively, were used to estimatethe mean local A. planipennis population presentwithin 800 m of each artiÞcial trap and detection treeseparately for 2009 and 2010. To derive these esti-mates, we Þrst calculated the Euclidean distance be-tween all detection trees and artiÞcial traps. Subse-quently, the detection trees or artiÞcial traps locatedwithin an 800 m radius of each artiÞcial trap or de-tection tree were used to calculate a weighted esti-mate of the mean number of larvae (detection trees)or adult beetles (artiÞcial traps) in the area surround-ing a given detection tree or artiÞcial trap. Mean den-sity estimates were weighted using the distance of thedetection trees or artiÞcial traps to the point of inter-est. SpeciÞcally, the weight of each detection tree orartiÞcial trap was estimated as 1 � (distance of de-tection tree or artiÞcial trap/distance of all detectiontrees or traps). For all estimates, the value of the focaldetection tree or artiÞcial trap (e.g., the detection treeor artiÞcial trap for which the neighborhood larvaldensity or adult density was being estimated) wasexcluded from the analysis. Only detection trees andartiÞcial traps with at least 10 detection trees or arti-Þcial traps located within 800 m were used for anal-yses.


Logistic regressions were performed in R (R Devel-opment Core Team 2010) to determine the relationshipbetween A. planipenis density and the likelihood of A.planipenis detection by a detection tree or artiÞcial trap(e.g., at least one larva recorded from the detection treeor one beetle recovered from the artiÞcial trap). Subse-quently, odds ratios determined through logistic regres-sions from the 2010 results were used to calculate theprobability that a detection tree or artiÞcial trap wasdetected as positive following the relationship probabil-ity � odds/(odds � 1). To illustrate the effect of in-creasing or decreasing the number of detection trees orartiÞcial traps used, we adjusted the detection probabil-ities from 2010 across a range of A. planipennis larvaldensities. The probability was adjusted following thesimple relationship: adjusted probability � 1 � (1 �probability of detection)number of DTs or ATs.

In 2009, 30% of the artiÞcial traps in the dense trapgrid (delineated by a 2.4-km radius around the epi-center of the infestation) were suspended in the can-opies of detection trees, while detection trees weredistributed across the project area. These conditionsresulted in an underrepresentation of artiÞcial trapsthat met our minimum requirement of having at least10 detection trees or artiÞcial traps located within800 m in areas distant to the core of the infestationwhere A. planipennis densities were generally low(Tables 1 and 2). Given this situation, the 2009 datawere excluded from the analysis described below.Proportion of Positive Detection Trees and Artifi-cial Traps.A grid of 9,752 equidistant points was over-laid on the 2010 SLAM pilot project area, effectivelydividing the project area into 200 by 200 m cells. Meancounts of larvae from detection trees and adult beetlesfrom artiÞcial traps were estimated for each grid pointusing all detection trees and artiÞcial traps within800 m of each point in the grid. The proportion of

detection trees and artiÞcial traps detected as positivewithin 800 m of each grid point was also estimated.Because of the extensive size of the project area andthe patchy nature of ash distribution, some grid pointsfell within areas where ash trees did not occur, so nodetection trees or artiÞcial traps were established inthese areas. To ensure our estimates were reasonable,we analyzed 561 grid points, all of which had at least10 detection trees and 10 artiÞcial traps within 800 m.We then contrasted the proportion of detection treesand artiÞcial traps detected as positive using localadult and larvalA. planipennis counts, as well as directcomparisons within grid points. In addition, the stan-dard deviation in larval and adult beetle counts amongdetection trees and artiÞcial traps was estimated foreach grid point. These estimates were used to deter-mine the coefÞcient of variation for larval and adultbeetle counts to provide comparable estimates of vari-ation for artiÞcial traps and detection trees. CoefÞ-cients of variation were calculated as the standarddeviation of the number of larvae or adults detectedrecorded per grid point divided by the mean numberof larvae or adults detected across the project area.

Results

Girdled detection trees were more likely to be de-tected as positive forA. planipennis than artiÞcial trapsplaced on the same trees in 2008 and again in 2009. In2008, only six (3.5%) of the 171 artiÞcial traps placedon girdled detection trees captured an A. planipennisbeetle, but 18 (10.5%) of those trees had one or moreA. planipennis larvae. Similarly, in 2009, larvae werepresent on 17 of the 99 detection trees (17.2%) thatalso had an artiÞcial trap, but only four (4.0%) of the99 traps placed on those detection trees captured abeetle.Results fromaGLMMconÞrmedthat theprob-

Table 2. Total number of unpaired detection trees and artificial traps and percentage detected as positive for A. planipennis by distanceclass from the epicenter of the infestation in 2008 to 2010

Distance class (m)Total detection trees Total artiÞcial traps

% positive detectiontrees

% positive artiÞcial traps

2008 2009 2010 2008 2009 2010 2008 2009 2010 2008 2009 2010

0Ð499 0 23 2 0 7 1 NA 95.7 100 NA 85.7 100500Ð999 4 40 5 0 8 6 50 80 100 NA 75 66.71,000Ð1,499 4 20 26 0 14 2 25 45 96.2 NA 64.3 501,500Ð1,999 5 16 10 7 13 7 20 43.8 80 0 57.1 71.42,000Ð2,499 3 12 19 3 14 8 0 50 84.2 0 50 62.52,500Ð2,999 4 23 28 6 17 21 0 21.7 67.9 0 29.4 52.43,000Ð3,499 12 26 33 5 17 22 8.3 23.1 57.6 0 29.4 31.83,500Ð3,999 9 28 47 4 18 27 0 14.3 48.9 0 33.3 29.64,000Ð4,499 14 28 62 1 17 47 0 17.9 37.1 0 5.9 19.14,500Ð4,999 15 35 59 2 13 48 0 8.6 32.2 0 14.3 18.85,000Ð5,499 15 16 42 0 9 38 0 6.3 28.6 NA 0 15.85,500Ð5,999 17 19 43 0 2 29 0 5.3 25.6 NA 0 20.76,000Ð6,499 15 14 35 0 3 18 0 0 17.1 NA 0 5.66,500Ð6,999 8 8 18 0 1 5 0 0 11.1 NA 0 07,000Ð7,499 4 5 12 2 3 3 0 0 25 0 0 07,500Ð7,999 4 6 10 0 3 6 0 0 0 NA 0 16.78,000Ð8,499 6 6 14 0 4 5 0 0 14.3 NA 0 08,500Ð8,999 10 8 22 0 3 2 0 12.5 13.6 NA 0 09,000Ð9,499 6 10 17 2 3 2 0 0 5.9 0 0 09,500Ð1,000 5 4 19 0 2 7 0 25 0 NA 0 0�10,000 52 128 225 37 61 171 1.9 2.3 7.6 0 0 1.2


ability of detection was signiÞcantly higher for girdleddetection trees than artiÞcial traps in 2008 (t170 ��6.49; P� 0.001) and 2009 (t97 � 3.78; P� 0.001). In2009, there was a positive and signiÞcant interactionbetween the estimated localA.planipennisdensity andthe detection tool (detection tree or artiÞcial trap)(t97 � 2.78; P � 0.007). This result indicates thatdifferences between the detection trees and artiÞcialtraps were more pronounced when local A. planipen-nis densities were higher, likely because of both de-tection trees and artiÞcial traps having a relatively lowprobability of detection in areas where A. planipennisdensities were extremely low.Probability of Detection Estimates.Mean larval es-

timates for the800-mradius surroundingeachartiÞcialtrap or detection tree ranged from 0 to 42.5 larvae perdetection tree in 2009 and 0Ð52.6 larvae per detectiontree in 2010, with average counts (�SEM) of 4.84 �0.01 larvae and 9.15 � 0.51 larvae per detection tree in2009 and 2010, respectively. Mean adult beetle esti-mates for the 800-m radius surrounding each artiÞcialtrap or detection tree ranged from 0 to 11.4 beetles pertrap, with an average (�SEM) of 2.86 � 0.02 beetlesin 2009 and 0Ð2.70 beetles per trap, averaging 0.23 �0.002 beetles in 2010. When data from 2009 and 2010were analyzed together (including only unpaired de-tection trees and artiÞcial traps), there was no inter-action between the local A. planipennis density andthe year in which sampling was undertaken (z ��1.36; P � 0.17). However, an interaction was ob-served between the detection tool and the year inwhich sampling took place (z � 3.4; P � 0.001). In2009, fewer unpaired artiÞcial traps and detectiontrees were used for density estimates and analysis thanin 2010. Even in 2009, however, we observed signiÞ-cant effects of local A. planipennis density (z� 4.013;P� 0.001), detection tool (z� �2.11;P� 0.0348), andtheir interaction (z � 1.997; P � 0.0458) (Table 2).The interaction likely resulted from a difference be-tween detection trees and artiÞcial traps present onlyat higher A. planipennis densities, reßecting the con-centration of detection trees and artiÞcial traps in theepicenter of the A. planipennis infestation and theunderrepresentation, particularly of artiÞcial traps,near the edge of the project area (Table 2).

In 2010, when artiÞcial traps were placed only onungirdled ash trees, a larger number of traps had atleast 10 detection trees within 800 m (295 and 232,respectively), increasing the power of our analysescompared with 2009. In 2010, the effect of local A.planipennisdensity on the probability of detection wasonce again signiÞcant (z� 3.83; P� 0.001), as was theeffect of detection tool (z � 2.925; P � 0.003). How-ever, unlike 2009, there was no signiÞcant interactionbetween localA. planipennisdensity and the detectiontool (z � 1.168; P � 0.242).

Results from 2010 showed that the probability ofdetection observed for detection trees and artiÞcialtraps improved as the local A. planipennis densityincreased, particularly for detection trees (Fig. 1).For example, the probability of detection with de-tection trees was �50% in areas where an average of

only Þve larvae per tree were found when the localdetection trees were debarked. In contrast, evenwhen average larval counts exceeded 25 larvae pertree, the probability of detection with artiÞcial trapswas �35%. An important consideration when inter-preting these results is that local A. planipennis lar-val density was estimated from girdled trees, nothealthy trees. Therefore, the A. planipennis larvalcounts on the X-axis of Fig. 1 and other Þgures likelyoverestimate the local population and should beinterpreted as a relative density estimates, ratherthan absolute densities.

Figure 2 illustrates the increased probability of de-tection if one to Þve detection trees or artiÞcial trapsare used in a given locale. When the local A. plani-pennis population is very low, for example, �5 larvaeper tree, using three detection trees increases theprobability of detection from 57 to 90% and if Þvedetection trees are used, the likelihood of detectionapproaches 100% (Fig. 2a). Once larval counts average15 larvae per tree, the probability of detection with asingle detection tree exceeds 95% and using additionaldetection trees would have little beneÞt (Fig. 2a). Forthe artiÞcial traps, when larval counts in the areaaverage Þve larvae per tree, the probability of detec-tion goes from 10% with a single artiÞcial trap up to40% if Þve artiÞcial traps are used (Fig. 2b), which isstill lower than the detection probability associatedwith a single detection tree (Fig. 2a). At an average of15 larvae per tree, using Þve artiÞcial traps instead ofa single artiÞcial trap increases the probability of de-tection from �20 to 62% (Fig. 2b).Proportion of Positive Detection Trees and Artifi-cial Traps. Figure 3 presents a direct contrast of theproportion of positive detection trees to artiÞcial traps in2010. More than half the points fell below the 1 to 1 line,indicating a higher proportion of detection trees weredetected as positive compared with artiÞcial traps. Thisdifference was also reßected in the standardized mean

Fig. 1. Relationship between average A. planipennis lar-val counts from local detection trees (within 800 m) and theprobability of a detection tree or artiÞcial panel trap beingdetected as positive based on 2010 survey data.


difference between the proportion of positive detectiontreesandartiÞcial traps,whichwasfairlyhigh(0.78).Thestandardized mean difference was calculated as: (Meanvalue for detection trees � Mean value for artiÞcialtraps)/Combined standard deviation.

Therelationshipbetweentheaverage larvalcountsonlocal detection trees and the proportion of detectiontrees and artiÞcial traps that detected A. planipennis in2010 is shown in Fig. 4a. Estimates of the local A. plani-pennis population was strongly related to the likelihoodthat a detection tree in the area would be infested (R2 �0.69). Figure 4b presents the relationship between theaverage number of adultA. planipennis beetles capturedon artiÞcial traps in the local area and the proportion ofdetection trees and artiÞcial traps that were positive. Onaverage, less than two beetles were captured per trap,andthehighestnumberofbeetlesonasingle trap in2010was 18. The relationship between A. planipennis adult

counts on traps and the proportion of positive detectiontrees was fairly weak (Fig. 4b), which reßects the highincidenceofpositivedetection trees inareaswhere trapsdidnotcapturea singleadultbeetle(Fig. 4b).Asimilarlyweakrelationshipwasobservedbetweenthemeannum-ber of larvae recorded per detection tree and the meannumber of adults captured per artiÞcial trap per gridpoint (Fig. 5).

CoefÞcients of variation, which represent the stan-dard deviation scaled by the mean, indicated a highlevel of variation for both detection trees and artiÞcialtraps whenA. planipennis larval counts were relativelylow (Fig. 6). CoefÞcients of variation can be roughlyinterpreted as noise to signal ratios, indicating that atlow population densities, the noise reached or ap-proached four times the value of the signal. Theseresults indicate that within a relatively small areabounded by an 800-m radius, densities recorded from

Fig. 2. Relationship between average A. planipennis larval counts estimated from detection trees within 800 m, and theprobability of detecting anA. planipennis infestation for (a) detection trees, and (b) artiÞcial traps, when one to Þve detectiontrees or artiÞcial traps are used, based on 2010 survey data.


detection trees or artiÞcial traps were highly variablewhere A. planipennis densities were low. Attemptingto extrapolate local population levels based on resultsfrom one or a few detection trees or artiÞcial traps atthis spatial scale, therefore, is not justiÞed when A.planipennisdensities are low.Therewasnodiscerniblerelationship between the coefÞcients of variation be-tween detection trees and artiÞcial traps (Pearsoncorrelation � �0.01; P � 0.88) (Fig. 6c).

Discussion

The success of efforts to reduce the spread of in-vasive species through the use of quarantines or activemanagement relies on the ability to effectively detect,

delimit, and monitor infestations. Early detection cangreatly increase the success of efforts to eradicate,contain, or manage invasive pests (Suckling et al. 2012,Tobin et al. 2013), including actions to reduce thespread and population growth of A. planipennis (Mer-cader et al. 2011a,b). Furthermore, early detectionprovides municipal foresters with time to secure fundsfor insecticide treatment or staged removals of ashtrees on public land, rather than reactively removingdead and dying trees. For example, in a recent simu-lationencompassinga10-yrperiod, costsof treatingupto 50% of urban ash trees annually with a highly ef-fective systemic insecticide were substantially lowerthan removing trees as they declined (McCulloughand Mercader 2012). Unfortunately, budgets usually

Fig. 3. Comparison between the proportion of detection trees and artiÞcial traps detected as positive for A. planipennisper grid point in the 2010 survey. Black line represents the 1 to 1 line. (n � 560 grid points).

Fig. 4. Relationship between the proportion of detection trees and artiÞcial traps detected as positive for A. planipennisper grid point as a function of (a) average larval counts, or (b) adult beetle captures in the 2010 survey. (n� 560 grid points).


limit the efforts that can be expended in surveys, andmanagers must carefully consider the relative cost-effectiveness of survey tools.

Placing a baited artiÞcial trap into the canopy of anash tree would presumably require less labor thandebarking a girdled ash tree, translating to a lowerprice per unit. In practice, however, the time and laborrequired to set andrecoveranartiÞcial trapandcollectsuspect beetles is often relatively similar to the timeneeded for an experienced surveyor to girdle andsubsequently debark a detection tree, assuming thedetection tree is �20 cm in diameter (McCullough etal. 2011). In this operational project, detection treeswere substantially more likely to detectA. planipennisthan artiÞcial traps in every analysis we performed,especially when the larval A. planipennis populationwas relatively low. Of particular relevance is thehigher likelihood of detection when the use of mul-tiple detection trees or artiÞcial traps was consideredor when the proportion of positive detections wasconsidered. These results show that even if more ar-tiÞcial traps than detection trees are used in an area,the detection trees will be more likely to detect an A.planipennis infestation than the artiÞcial traps. Forexample, the probability of detection associated witha single detection tree was 80% if larval counts indetection trees within an 800-m radius averaged 10larvae (�3.5 larvae/m2 of phloem surface). At thissame density, a single artiÞcial trap corresponded to adetection probability of only 15%. Using Þve artiÞcialtraps would increase the detection probability to 50%,which is still substantially lower than the probabilityassociated with a single detection tree.

The patchy distribution of A. planipennis within aninfested site reßects the tendency of A. planipennis todisperse short distances when ash trees are immedi-ately available (Mercader et al. 2009, Siegert et al.2010), the typically clumped distribution of ash treesin forested or rural settings, and the attraction of A.planipennis females to stressed ash trees (McCulloughet al. 2009a,b; Siegert et al. 2010; Tluczek et al. 2011).

Not surprisingly, in a relatively new infestation, thelikelihood that any one ash tree will be infested is low(Mercader et al. 2012), and in the absence of a strongradius of attraction, any detection tool can be ex-pected to yield variable results. This was observed forboth artiÞcial traps and detection trees in terms of theproportion of positive detections and in the variabilityof estimates of the local population density based onlarval counts from detection trees or adult beetle cap-tures on artiÞcial traps. This high variability indicatesefforts to estimate the A. planipennis infestation levelacross a 2-km2 area (e.g., bounded by an 800-m radius)based on counts from only a few detection trees orartiÞcial traps placed within that area should be madecautiously. Moreover, the A. planipennis density esti-mates in this study were generated from the detectiontools themselves. Given the markedly greater attrac-tion of egg-laying A. planipennis females to ash treesstressedbygirdling than tohealthy trees(McCulloughet al. 2009a,b), the larval counts from the detectiontrees almost certainly overestimate the true popula-tion density in the local area. Therefore, detectiontrees as well as artiÞcial traps are both probably moresensitive than the results presented here indicate.

Despite the likely underestimation of the sensitivityof detection trees, the estimated probability of detec-tion with detection trees presented here was consid-erably higher than would be expected if nongirdledash were used for surveys (Mercader et al. 2012). Incontrast, the probability of A. planipennis detectionwith artiÞcial traps in our project area was substan-tially lower than the probability of detecting A. pla-nipennis by simply debarking an ungirdled ash tree(Mercader et al. 2012). As noted above, the sensitivityof artiÞcial traps was probably underestimated in thisstudy, and these results should not be interpreted asevidence that artiÞcial traps are always less sensitivethan debarking ungirdled trees. These results do, how-ever, stress the relatively high sensitivity of detectiontrees compared with other currently available detec-tion tools. Owing to the comparatively low probability

Fig. 5. Relationship between the mean number ofA. planipennis larvae in detection trees and the mean number of adultscaptured on artiÞcial traps per grid point in 2010. (n � 560 grid points).


of detection with artiÞcial traps, increasing the num-ber of artiÞcial traps up to Þve did not increase theprobability that an infestation would be detected tothe level of a detection tree, even in areas with mod-erately high A. planipennis populations. Differencesbetween detection trees and artiÞcial traps were lesspronounced at very low A. planipennis densities, re-ßecting the generally poor detection ability of bothsurvey tools. These results highlight the fact that al-though detection trees are considerably more likely todetect infestations than ungirdled trees or artiÞcialtraps, detecting new infestations of A. planipenniswhere densities are very low remains difÞcult.

The use of detection trees for A. planipennis detec-tion or survey can be limited by the availability of treessuitable for girdling. The size of detection trees usedin the SLAM project (10Ð20 cm in DBH) reßects theneed to select a tree that can be efÞciently girdled,felled, and debarked, but also of sufÞcient size toremain attractive to egg-laying A. planipennis femalesduring the summer (McCullough and Siegert 2008).Trees meeting these criteria may be unavailable orinaccessible for use as detection trees, particularly inurban areas and when multi-year surveys are neces-sary. Sentinel ash trees, for example, trees in large potsor planted speciÞcally for A. planipennis detection,have been used occasionally in operational programs(Sargent et al. 2010). ArtiÞcial traps, in contrast, arenot destructive and can be used in sensitive areas orin locales where ash trees are scarce. The traps maygenerate interest among residents or visitors, provid-ing an opportunity for outreach efforts to increaseawareness of A. planipennis.

Results from artiÞcial traps, however, should beinterpreted cautiously. While a positive result clearlyindicates the presence of anA. planipennis infestation,a negative result is likely to occur even when moder-ately high infestations are present. Any measurementof detection and survey tools must include consider-ation of the opportunity costs associated with falsenegatives. Numerous Þeld trials have been conductedto measure trap designs and lures for A. planipennis,but many of these occurred in sites with a moderateor even heavy infestation (e.g., Crook et al. 2008, 2012;Grant et al. 2010, 2011; Ryall et al. 2012). Multiplestudies in many sites have shown an alternative arti-Þcial trap design, termed the double-decker trap (Mc-Cullough and Poland 2009), is consistently more ef-fective for A. planipennis detection in areas with lowor ultra-low population densities than baited canopytraps (Marshall et al. 2010a,b; McCullough et al. 2011;Poland et al. 2011). A sampling method that involvesdebarking two branches to assess A. planipennis larvalpresence was measured on open-grown urban ashtrees (24Ð34 cm in DBH) and may provide a usefultool for municipal foresters (Ryall et al. 2011). Effortsto further assessA. planipennis attraction to other hostvolatiles or short-range pheromones (Silk et al. 2009,2011; Ryall 2012) are continuing and may eventuallylead to improved lures. Thorough analyses that incor-porate costs of failing to detect new A. planipennisinfestations are needed, however, to more fully mea-

Fig. 6. CoefÞcients of variation for A. planipennis lar-val counts or adult beetles captured in artiÞcial traps pergrid point in 2010. These are represented as (a) the co-efÞcient of variation of the larvae counted per detectiontree, (b) the coefÞcient of variation of adult captures perartiÞcial trap, or (c) a regression of the coefÞcient ofvariation values for the larval counts per detection tree andadult captures per artiÞcial trap. (n� 389, 226, and 226 gridpoints, respectively).


sure the efÞcacy not only of individual detection toolsbut also the relative cost-effectiveness of combininggirdled detection trees with various traps and lures.

Acknowledgments

We thank our colleagues and members of the SLAM Man-agement Board, particularly Steven Katovich, Therese Po-land, and Noel Schneeberger (USDA Forest Service), An-drew Storer (Michigan Technological University), BrendaOwen (Michigan Association of Timbermen), Robert Heyd(Michigan Department of Natural Resources), and KenRauscher (Michigan Department of Agriculture). Polandand Katovich provided helpful comments on an earlier draftof this manuscript. We also thank Gabriel Carballo, TravisPerkins, and Amos Ziegler (Michigan State University[MSU]) for their assistance with data collected for the SLAMpilot project. We appreciate the efforts of all the surveyorswho collected the SLAM data, often in challenging and dif-Þcult conditions. Funding for the SLAM project and ourwork was provided by the USDA Forest Service, the Amer-ican Recovery and Reinvestment Act, and USDAÐAPHIS.

References Cited

Anulewicz, A. C., D. G. McCullough, D. L. Cappaert, andT.M. Poland. 2008. Host range of the emerald ash borer(Agrilus planipennis Fairmaire) (Coleoptera: Bupresti-dae) in North America: results of multiple-choice Þeldexperiments. Environ. Entomol. 37: 230Ð241.

Aukema, J. E., D. G. McCullough, B. Von Holle, A. M.Liebhold, K. Britton, and S. J. Frankel. 2010. Historicalaccumulation of non-indigenous forest pests in the con-tinental US. BioScience 60: 886Ð897.

Aukema, J. E., B. Leung, K. Kovacs, C. Chivers, K. O. Britton,J. Englin, S. J. Frankel, R. G. Haight, T. P. Holmes, A.Liebhold, et al. 2011. Economic impacts of non-nativeforest insects in the continental United States. PLoS ONE6: e24587.

Bartels, D., D. Williams, J. Ellenwood, and F. Sapio. 2008.Accuracy assessment of remote sensing imagery for map-ping hardwood tree and emerald ash borer-stressed ashtrees, pp. 63Ð65. In:V. Mastro, D. Lance, R. Reardon, andG. Parra (eds.), Emerald Ash Borer Research and Tech-nology Development Meeting, 23Ð24 October 2007, Pitts-burgh, PA. USDA Forest Service, FHTET-2008Ð07. Mor-gantown, WV.

Borden, J. H. 1989. Semiochemicals and bark beetle popu-lations: exploitation of natural phenomena by pest man-agement strategists. Holarctic Ecol. 12: 501Ð510.

Brockerhoff. E. G., D. C. Jones, M. O. Kimberley, D. M.Suckling, and T. Donaldson. 2006. Nationwide surveyfor invasive wood-boring and bark beetles (Coleoptera)using trapsbaitedwithpheromonesandkairomones.Can.J. For. Res. 228: 234Ð240.

Cappaert, D., D. G. McCullough, T. M. Poland, and N. W.Siegert. 2005. Emerald ash borer in North America: aresearch and regulatory challenge. Am. Entomol. 51: 152Ð165.

Crook, D. J., and V. C. Mastro. 2010. Chemical ecology ofthe emerald ash borer,Agrilus planipennis. J. Chem. Ecol.36: 101Ð112.

Crook, D., A. Khrimian, J. A. Francese, I. Fraser, T. M.Poland, A. J. Sawyer, and V. C. Mastro. 2008. Develop-ment of a host-based semiochemical lure for trappingemerald ash borer, Agrilus planipennis (Coleoptera: Bu-prestidae). Environ. Entomol. 37: 356Ð365.

Crook, D. J., J. A. Francese, K. E. Zylstra, I. Fraser, A. J.Sawyer, D. W. Bartels, D. R. Lance, and V. C. Mastro.2009. Laboratory and Þeld response of the emerald ashborer (Coleoptera: Buprestidae), to selected regions ofthe electromagnetic spectrum. J. Econ. Entomol. 102:2160Ð2169.

Crook, D. J., A. Khrimian, A. Cosse, I. Fraser, and V. C.Mastro. 2012. Inßuence of trap color and host volatileson capture of the emerald ash borer (Coleoptera: Bu-prestidae). J. Econ. Entomol. 105: 429Ð437.

EAB.info. 2013. Emerald ash borer national Web site.(http://www.emeraldashborer.info).

El-Sayed, A. M. 2011. The pherobase: database of insect pher-omones and semiochemicals. (http://www.pherobase.com).

Francese, J. A., V. C. Mastro, J. B. Oliver, D. R. Lance, N.Youssef, and S. G. Lavallee. 2005. Evaluation of colorsfor trapping Agrilus planipennis (Coleoptera: Bupresti-dae). J. Entomol. Sci. 40: 93Ð95.

Francese, J. A., D. J. Crook, I. Fraser, D. R. Lance, A. J.Sawyer, and V. C. Mastro. 2010. Optimization of trapcolor for emerald ash borer (Coleoptera: Buprestidae). J.Econ. Entomol. 103: 1235Ð1241.

Grant, G. G., K. L. Ryall, D. B. Lyons, andM.M. Abou–Zaid.2010. Differential response of male and female emeraldash borers (Col., Buprestidae) to (Z)-3-hexenol andManuka oil. J. Appl. Entomol. 134: 26Ð33.

Grant, G. G., T. M. Poland, T. Ciaramitaro, D. B. Lyons, andG. C. Jones. 2011. Comparison of male and female em-erald ash borer (Coleoptera: Buprestidae) responses toPhoebe oil and (Z)-3-hexenol lures in light green prismtraps. J. Econ. Entomol. 104: 173Ð179.

Hunt, L. 2007. Emerald ash borer state update: Ohio, p. 2. InV. Mastro, D. Lance, R. Reardon, and G. Parra (eds.),Proceedings of the Emerald Ash Borer and Asian Long-horned Beetle Research and Technology DevelopmentMeeting, Cincinnati, OH, 29 Oct.Ð2 Nov.2006. USDA For-est Service, FHTET-2007Ð04, Morgantown, WV.

Kovacs, K. F., R. G. Haight, D. G. McCullough, R. J. Mer-cader, andN.W.Siegert. 2010. Cost of potential emeraldash borer damage in U.S. communities, 2009Ð2019. Ecol.Econ. 69: 569Ð578.

Kovacs, K., R. J. Mercader, R. Haight, N. Siegert, D. G.McCullough, and A. Liebhold. 2011. The inßuence ofsatellite populations of emerald ash borer on projectedeconomic damage in U.S. communities, 2010Ð2020. En-viron. Manage. 92: 2170Ð2181.

Lelito, J. P., K. Boroczky, T. H. Jones, I. Fraser, V. C. Mastro,J. H. Tumlinson, and T. C. Baker. 2009. Behavioral ev-idence for a contact sex pheromone component ofthe emerald ash borer, Agrilus planipennis Fairmaire.J. Chem. Ecol. 35: 104Ð110.

Liebhold, A. M., and P. C. Tobin. 2008. Population ecologyof insect invasions and their management. Annu. Rev.Entomol. 53: 387Ð408.

Marshall, J. M., A. J. Storer, I. Fraser, J. A. Beachy, and V. C.Mastro. 2009. Effectiveness of differing trap types forthe detection of emerald ash borer (Coleoptera: Bupres-tidae). Environ. Entomol. 38: 1226Ð1234.

Marshall, J.M.,A. J. Storer, I.Fraser, andV.C.Mastro. 2010a.EfÞcacy of trap and lure types for detection of Agrilusplanipennis (Col., Buprestidae) at low density. J. Appl.Entomol. 134: 296Ð302.

Marshall, J. M., A. J. Storer, I. Fraser, and V. C. Mastro.2010b. Multi-state comparison of detection tools at lowemerald ash borer densities. In V. Mastro, D. Lance, R.Reardon, and G. Parra (eds.), Proceedings of the EmeraldAsh Borer and Asian Longhorned Beetle Research and


Technology Development Meeting, Pittsburgh, PA,19Ð21 October 2009. USDA Forest Service, FHTET-2010Ð01. Morgantown, WV.

McCullough, D. G., and N. W. Siegert. 2007. Estimatingpotential emerald ash borer (Agrilus planipennis Fair-maire) populations using ash inventory data. J. Econ.Entomol. 100: 1577Ð1586.

McCullough, D. G., and N. W. Siegert. 2008. Using girdledtrees effectively for emerald ash borer detection, delim-itation and survey. (http://www.emeraldashborer.info).

McCullough, D. G., and T. M. Poland. 2009. Using double-decker traps to detect emerald ash borer. (http://www.emeraldashborer.info).

McCullough, D. G., and R. J. Mercader. 2012. SLAM in anurban forest: evaluation of potential strategies to SlowAsh Mortality caused by emerald ash borer (Agrilus pla-nipennis). Int. J. Pest Manage. 58: 9Ð23.

McCullough, D. G., T. T. Work, J. F. Cavey, A. T. Liebhold,and D. Marshall. 2006. Interceptions of nonindigenousplant pests at U.S. ports of entry and border crossings overa 17 year period. Biol. Invasions 8: 611Ð630.

McCullough, D. G., T. M. Poland, A. C. Anulewicz, and D.Cappaert. 2009a. Emerald ash borer (Coleoptera: Bu-prestidae) attraction to stressed or baited ash trees. En-viron. Entomol. 38: 1668Ð1679.

McCullough, D. G., T. M. Poland, D. Cappaert, and A. C.Anulewicz. 2009b. Emerald ash borer (Agrilus planipen-nis) attraction to ash trees stressed by girdling, herbicideand wounding. Can. J. For. Res. 39: 1331Ð1345.

McCullough, D. G., N. W. Siegert, T. M. Poland, S. J. Pierce,and S. Z. Ahn. 2011. Effects of trap type, placement andash distribution on emerald ash borer captures in a lowdensity site. Environ. Entomol. 40: 1239Ð1252.

Mercader, R., N. W. Siegert, A. M. Liebhold, and D. G.McCullough. 2009. Dispersal of the emerald ash borer,Agrilus planipennis, in newly colonized sites. Agric. For.Entomol. 11: 421Ð424.

Mercader, R. J., N. W. Siegert, A. M. Liebhold, and D. G.McCullough. 2011a. Inßuence of foraging behavior andhost spatial distribution on the localized spread of theemerald ash borer, Agrilus planipennis. Popul. Ecol. 53:271Ð285.

Mercader, R. J., N. W. Siegert, A. M. Liebhold, and D. G.McCullough. 2011b. Simulating the effectiveness ofthree potential management options to slow the spread ofemerald ash borer populations in localized outlier sites.Can. J. For. Res. 41: 254Ð264.

Mercader, R. J., N.W. Siegert, and D. G. McCullough. 2012.Estimating the inßuence of population density and dis-persal behavior on the ability to detect and monitorAgri-lus planipennis (Coleoptera: Buprestidae) populations. J.Econ. Entomol. 105: 272Ð281.

Myers, J.H.,D.Simberloff,A.M.Kuris, and J.R.Carey. 2000.Eradication revisited: dealing with exotic species. TrendsEcol. Evol. 15: 316Ð320.

National Research Council. 2002. Predicting invasions ofnonindigenous plants and plant pests. National AcademyPress, Washington, DC.

Poland, T. M., and D. G. McCullough. 2006. Emerald ashborer: invasion of the urban forest and the threat to NorthAmericaÕs ash resource. J. For. 104: 118Ð124.

Poland, T.M., andD. G.McCullough. 2010. SLAM: a multi-agency pilot project to Slow Ash Mortality caused byemerald ash borer in outlier sites. Newsl. Mich. Entomol.Soc. 55: 4Ð8.

Poland, T. M., D. G. McCullough, and A. C. Anulewicz. 2011.Evaluation of double-decker traps for emerald ash borer(Coleoptera: Buprestidae). J. Econ. Entomol. 104: 517Ð531.

Rauscher, K. 2006. The 2005 Michigan emerald ash borerresponse: an update, p. 1. In V. Mastro, R. Reardon, andG. Parra (eds.), Proceedings of the Emerald Ash BorerResearch and Technology Development Meeting, Pitts-burgh, PA, 26Ð27 September 2005. USDA Forest Service,FHTET-2005Ð16. Morgantown, WV.

R Development Core Team. 2010. R: a language and envi-ronment for statistical computing. Royal Foundation forStatistical Computing, Vienna, Austria. (http://www.R-project.org).

Rebek, E. J., D. A. Herms, and D. A. Smitley. 2008. Inter-speciÞc variation in resistance to emerald ash borer (Co-leoptera: Buprestidae) among North American and Asianash (Fraxinus spp.). Environ. Entomol. 37: 242Ð246.

Rodriguez–Saona, C., T. M. Poland, J. R. Miller, L. L. Stel-inski, G. G. Grant, P. de Groot, L. Buchan, and L. Mac-Donald. 2006. Behavioral and electrophysiological re-sponses of the emerald ash borer, Agrilus planipennis, toinduced volatiles of Manchurian ash, Fraxinus mand-shurica. Chemoecology 16: 75Ð86.

Rodriguez–Saona, C. R., J. R.Miller, T.M. Poland, T.M. Kuhn,G.W.Otis, T.Turk, andN.McKenzie. 2007. Behaviours ofadult emerald ash borer, Agrilus planipennis (Coleoptera:Buprestidae). Great Lakes Entomol. 40: 1Ð16.

Ryall, K. L., J. G. Fidgen, and J. J. Turgeon. 2011. Detect-ability of the emerald ash borer (Coleoptera: Bupresti-dae) in asymptomatic urban trees by using branch sam-ples. Environ. Entomol. 40: 679Ð688.

Ryall, K. L., P. J. Silk, P. May, D. Crook, A. Khrimian, A. A.Cosse, J. Sweeney, andT. Scarr. 2012. Attraction ofAgri-lus planipennis (Coleoptera: Buprestidae) to a volatilepheromone: effects of release rate, host volatile, and trapplacement. Environ. Entomol. 41: 648Ð656.

Sargent, C., M. Raupp, D. Bean, and A. J. Sawyer. 2010.Dispersal of emerald ash borer within an intensively man-aged quarantine zone. Arbor. Urban For. 36: 160Ð163.

Shigesada, N., and K. Kawasaki. 1997. Biological invasions:theory and practice. Oxford University Press, Oxford,United Kingdom.

Siegert, N.W., D. G.McCullough, D.W.Williams, I. Fraser,and T. M. Poland. 2010. Dispersal of Agrilus planipennis(Coleoptera: Buprestidae) from discrete epicenters intwo outlier sites. Environ. Entomol. 39: 253Ð265.

Silk, P. J., K. Ryall, D. B. Lyons, J. Sweeney, and J.Wu. 2009.A contact sex pheromone component of the emerald ashborerAgrilus planipennisFairmaire (Coleoptera: Bupres-tidae). Naturwissenschaften 96: 601Ð608.

Silk, P. J., K. Ryall, P. Mayo, M. A. Lemay, G. Grant, D. Crook,A. Cosse, I. Fraser, J. D. Sweeney, D. B. Lyons, et al. 2011.Evidence for a volatile pheromone in Agrilus planipennisFairmaire (Coleoptera: Buprestidae) that increases attrac-tion to a host foliar volatile. Environ. Entomol. 40: 904Ð916.

SLAMEAB.info. 2012. SLow Ash MortalityÐEmerald AshBorer. (http://www.SLAMEAB.info).

Suckling,D.M.,D.G.McCullough,D.A.Herms,andP.C.Tobin.2012. Combining tactics to exploit Allee effects for eradica-tion of alien insect populations. Environ. Entomol. 105: 1Ð13.

Tluczek, A. R., D. G. McCullough, and T. M. Poland. 2011.Inßuence of host stress on emerald ash borer (Co-leoptera: Buprestidae) adult density, development, anddistribution in Fraxinus pennsylvanica trees. Environ. En-tomol. 40: 357Ð366.

Tobin, P. C., J. M. Kean, D. M. Suckling, D. G. McCullough,D. A. Herms, and L. D. Stringer. 2013. Determinants ofsuccessful eradication. Bio. Invasions (in press).

(USDA–APHIS) U.S. Department of Agriculture–Animaland Plant Health Inspection Service, Plant Protectionand Quarantine. 2008. 2008 Emerald ash borer survey


guidelines. (http://www.nationalplantboard.org/docs/08_EAB_Survey_Guidelines_Þnal.pdf). 14 p.

(USDA–APHIS) U.S. Department of Agriculture–Animal andPlantHealth Inspection Service. 2010. Plant health: emer-ald ash borer. 2010 Emerald ash borer survey guidelines.(http://www.aphis.usda.gov/plant_health/plant_pest_info/emerald_ash_b/index.shtml). 17 p.

Wei, X., Y. Wu, R. Reardon, T. H. Sun, M. Lu, and J. H. Sun.2007. Biology and damage traits of emerald ash borer(Agrilus planipennis Farmaire). Insect Sci. 14: 367Ð373.

Work, T. T., D. G. McCullough, J. F. Cavey, and R. Komsa.2005. Approach rate of nonindigenous insect species intothe United States through cargo pathways. Biol. Invasions7: 323Ð332.

Yu,C.M. 1992. AgrilusmarcopoliObenberger, pp. 400Ð401.InG. R. Xiao (ed.), Forest Insects of China, 2nd ed. ChinaForestry Publishing House, Beijing, China.

Received 30 November 2012; accepted 10 July 2013.


A Comparison of Girdled Ash Detection Trees and Baited Artificial Traps for \u003cI\u003eAgrilus...

Documents

Transcript of A Comparison of Girdled Ash Detection Trees and Baited Artificial Traps for \u003cI\u003eAgrilus...