Visual signs and symptoms of infestation

One survey technique is to inspect trees for signs and symptoms of EAB infestation. Signs include exit holes and galleries observed through cracks in the bark (de Groot et al. Reference de Groot, Biggs, Lyons, Scarr, Czwerwinski and Evans2006; Lyons et al. Reference Lyons, Caister, de Groot, Hamilton, Marchant and Scarr2007). Symptoms can include woodpecker feeding, canopy thinning and decline, and production of epicormic shoots. Two studies quantified relationships among crown decline and thinning and the number of EAB exit holes and woodpecker attacks (Anulewicz et al. Reference Anulewicz, McCullough and Cappaert2007; Smitley et al. Reference Smitley, Davis and Rebek2008). In one of the studies, the number of EAB exit holes explained approximately half of the variation in canopy thinning (Smitley et al. Reference Smitley, Davis and Rebek2008). Peters et al. (Reference Peters, Iverson and Sydnor2009) developed a rating system to classify tree health based on per cent light transmission, the number of epicormic shoots, and the number of exit holes on affected trees. In another study, as the density of EAB galleries increased so did the amount of a heavier carbon isotope in the foliage, indicative of the stress caused by the infestation (Flower et al. Reference Flower, Knight, Rebbeck and Gonzalez-Meler2013). Thus, it is possible to relate visual signs and symptoms to the presence of EAB infestation, with some indication of EAB density within a given tree (Table 1).

Table 1

Description of sampling tools and techniques for emerald ash borer (EAB), Agrilus planipennis.

Counting of exit holes and rating crown condition may be easy and relatively inexpensive, however there are significant drawbacks with relying on visual signs and symptoms to detect EAB. First, signs and symptoms are typically not apparent until infestations are moderate to high in intensity. In some cases, numerous exit holes were present on ash trees with little or no apparent dieback (Anulewicz et al. Reference Anulewicz, McCullough and Cappaert2007). Similarly, in Peters et al. (Reference Peters, Iverson and Sydnor2009), moderate to high populations of EAB occurred before epicormic shoots were apparent. Knight et al. (Reference Knight, Brown and Long2013) reported that the first EAB exit holes would not be detected until a minimum of two years after initial infestation. We have also observed numerous trees displaying no signs or symptoms that were subsequently determined to be infested (Ryall et al. Reference Ryall, Fidgen and Turgeon2011). An additional concern is that EAB infestation levels may have been underestimated in previous work because exit holes were counted only on the main stem (Smitley et al. Reference Smitley, Davis and Rebek2008; Peters et al. Reference Peters, Iverson and Sydnor2009; Knight et al. Reference Knight, Brown and Long2013); this approach would fail to detect early infestation, which typically begins in the crown of larger trees (Haack et al. Reference Haack, Jendek, Liu, Marchant, Petrice and Poland2002; Herms et al. Reference Herms, McCullough and Smitley2004; Cappaert et al. Reference Cappaert, McCullough, Poland and Siegert2005; Ryall et al. Reference Ryall, Fidgen and Turgeon2011). Finally, in Flower et al. (Reference Flower, Knight, Rebbeck and Gonzalez-Meler2013), carbon isotope composition did not change significantly until EAB density was high. Moreover, sampling of carbon isotope composition is too time consuming and expensive to be feasible as a survey method. Another major problem is that ash trees can exhibit symptoms of stress that are not related to EAB. For instance, both Anulewicz et al. (Reference Anulewicz, McCullough and Cappaert2007) and Smitley et al. (Reference Smitley, Davis and Rebek2008) reported trees with few or no EAB exit holes but with high levels of dieback from other causes. Thus, in the absence of a sign of EAB (exit hole, adult, larva, or gallery), it is not possible to confirm EAB infestation based on visual symptoms. Therefore, currently available methods based on visual signs and symptoms alone are both inadequate for early detection and not necessarily diagnostic of EAB.

Use of girdled trap trees

Girdling is a process where ash trees are intentionally damaged to artificially induce stress and thus increase their attractiveness to adult EAB. Several studies demonstrated that girdled trees attract more adults, or have higher larval densities, or both, as compared with trees subjected to other treatments (e.g., wounded, stressed with methyl jasmonate, treated with herbicide, cutting into trap logs, or untreated) (McCullough et al. Reference McCullough, Poland, Anulewicz and Cappaert2009a, Reference McCullough, Poland and Cappaert2009b, Reference McCullough, Siegert, Poland, Pierce and Ahn2011; Tluczek et al. Reference Tluczek, McCullough and Poland2011; Fierke et al. Reference Fierke, Whitmore, Foelker, Vandenberg and Carlson2013). Girdling can thus potentially increase the likelihood of detecting EAB infestation (Table 1). Girdled trap trees have been deployed in systematic grids to delimit EAB populations and within-tree larval densities have been used to quantify the spatial distribution of the infestation (Fierke et al. Reference Fierke, Whitmore, Foelker, Vandenberg and Carlson2013; Mercader et al. Reference Mercader, McCullough and Bedford2013). Mercader et al. (Reference Mercader, McCullough and Bedford2013) also reported that girdled trap trees were more likely to detect EAB populations than artificial traps (discussed below). Girdling of trees is, however, a destructive sampling method, requiring trees to be felled and then peeled after girdling. Deployment of this method into a detection or delimitation survey is reliant on the availability of ash trees of the appropriate size, age, and location. Moreover, cost-benefit analysis based on detection rates and the resources required to implement a girdling or trap-tree programme have not been conducted.

Sub-sampling protocols

Direct sampling of wood-boring insects is labour-intensive, therefore it is important to develop sub-sampling protocols, which focus sampling efforts on portions of the tree to maximise detection rates or optimise density estimates. To develop such protocols, it is necessary to first determine where infestations begin within a given tree (Speight Reference Speight2005). Knowledge of the distribution of EAB larvae within ash trees, especially during the initial stages of infestation, is therefore critical for developing sub-sampling techniques for early detection. A number of studies reported that most of the EAB galleries occurred in regions of the stem ~7–12 cm in diameter, either within naturally infested trees (Timms et al. Reference Timms, Smith and de Groot2006; Tluczek et al. Reference Tluczek, McCullough and Poland2011) or on trees which had been girdled to induce EAB attack (Marshall et al. Reference Marshall, Storer, Fraser, Beachy and Mastro2009, Reference Marshall, Storer, Fraser and Mastro2010a; Tluczek et al. Reference Tluczek, McCullough and Poland2011). In a fourth study on girdled trees, sampling from an intermediate diameter of 17 cm was recommended to maximise probability of detection (Foelker et al. Reference Foelker, Vandenberg, Whitmore and Fierke2013); however, trees in that study were slightly larger (15–25 cm diameter at breast height (DBH)) than those sampled in the other studies (<15–20 cm DBH). Based on these studies of within-tree distribution of larval galleries, sub-sampling approaches have been suggested (Table 1). For example, by targeting sampling efforts on the 8–12 cm diameter size range of a tree, Marshall et al. (Reference Marshall, Storer, Fraser and Mastro2010a) found it would be necessary to peel <45% of the tree to detect >50% of the larvae. Moreover, if the objective of a sampling programme was the detection per se of EAB infestation, then peeling of bark could stop after the first discovery of a larval gallery, thus further reducing sampling effort.

Anecdotal evidence from earlier studies suggested that EAB infestation begins in the crown of larger trees (Haack et al. Reference Haack, Jendek, Liu, Marchant, Petrice and Poland2002; Herms et al. Reference Herms, McCullough and Smitley2004; Cappaert et al. Reference Cappaert, McCullough, Poland and Siegert2005). Ryall et al. (Reference Ryall, Fidgen and Turgeon2011) found that early infestation in trees 20–40 cm DBH was often detected in branches 6–10 cm in diameter, corresponding with the size range previously reported from smaller trees (Timms et al. Reference Timms, Smith and de Groot2006; Marshall et al. Reference Marshall, Storer, Fraser, Beachy and Mastro2009, Reference Marshall, Storer, Fraser and Mastro2010a; Tluczek et al. Reference Tluczek, McCullough and Poland2011). Based on that finding, they developed a technique to sample branches from the crown of mature open-grown ash trees. This technique provides a high rate of detection (74%) of low-density asymptomatic infestations and does not require felling of the tree. Next, Legg et al. (Reference Legg, Fidgen and Ryall2014) developed random resampling models to estimate the number of trees necessary to sample using branch sampling to provide a high probability of detecting EAB, based on sampling for infestations of varying sizes (e.g., 1% of trees infested); this enables detection surveys to be conducted with a known probability of detection. However, while the branch sampling method is simple, it is considered to be labour-intensive by some managers (K.R., personal observation). Moreover, sampling protocols for trees outside the size range studied by Ryall et al. (Reference Ryall, Fidgen and Turgeon2011) (i.e., >20–40 cm in diameter) do not exist nor do protocols for trees without large-enough branches to sample. It is also necessary to develop methods to accurately estimate EAB density within infested trees to inform management decisions. Only two studies have attempted to quantify within-tree EAB density: Marshall et al. (Reference Marshall, Storer, Fraser and Mastro2010a) reported that EAB larval density in the 8–12 cm diameter region of the stem was related to density within the whole tree. In contrast, Foelker et al. (Reference Foelker, Vandenberg, Whitmore and Fierke2013) recommended a 1 m sub-sample be taken from 1 to 2 m height on the bole to provide a good prediction of whole-tree larval density. More research is clearly needed to develop a sampling protocol that minimises sampling effort while maximising detection rates and providing reasonable estimates of density.

Development of artificial traps and attractant lures

Species-specific traps are an important sampling tool for detection and delimitation of insect pests. Extensive research has been conducted with the objective of developing an effective trap and lure combination for attracting and capturing flying EAB adults (Table 1). Much of the literature on EAB trapping has been summarised in three other reviews (Crook and Mastro Reference Crook and Mastro2010; Herms and McCullough Reference Herms and McCullough2014; Silk and Ryall Reference Silk and Ryall2014). A separate review would be necessary to provide a synthesis of results of all trapping studies due to the large number of trap, colour, and lure combinations. Only brief overview of research investigating trapping tools and techniques is provided here.

Three different types of traps have been developed for sampling of EAB adults: (1) a sticky prism canopy trap (Francese et al. Reference Francese, Mastro, Oliver, Lance, Youssef and Lavallee2005, Reference Francese, Oliver, Fraser, Lance, Youssef and Sawyer2008, Reference Francese, Crook, Fraser, Lance, Sawyer and Mastro2010a, Reference Francese, Fraser, Rietz, Crook, Lance and Mastro2010b, Reference Francese, Rietz, Crook, Fraser, Lance and Mastro2013a, Reference Francese, Rietz and Mastro2013b; Crook et al. Reference Crook, Francese, Zylstra, Fraser, Sawyer and Bartels2009); (2) a double-decker trap (McCullough et al. Reference McCullough, Siegert, Poland, Pierce and Ahn2011; Poland et al. Reference Poland, McCullough and Anulewicz2011; Poland and McCullough Reference Poland and McCullough2014); and (3) a multiple-funnel trap (Francese et al. Reference Francese, Fraser, Lance and Mastro2011, Reference Francese, Rietz, Crook, Fraser, Lance and Mastro2013a, Reference Francese, Rietz and Mastro2013b). Trap colour is clearly important in terms of eliciting trap captures, with green and purple tending to capture the highest number of beetles (Francese et al. Reference Francese, Mastro, Oliver, Lance, Youssef and Lavallee2005, Reference Francese, Oliver, Fraser, Lance, Youssef and Sawyer2008, Reference Francese, Crook, Fraser, Lance, Sawyer and Mastro2010a, Reference Francese, Fraser, Rietz, Crook, Lance and Mastro2010b, Reference Francese, Rietz, Crook, Fraser, Lance and Mastro2013a, Reference Francese, Rietz and Mastro2013b; Crook et al. Reference Crook, Francese, Zylstra, Fraser, Sawyer and Bartels2009). Two types of host volatiles attractive to EAB have been identified and subsequently tested: bark sesquiterpenes (Crook et al. Reference Crook, Khrimian, Francese, Fraser, Poland and Sawyer2008; Crook and Mastro Reference Crook and Mastro2010) and green leaf volatiles, in particular (3Z)-hexenol (de Groot et al. Reference de Groot, Grant, Poland, Scharbach, Buchan and Nott2008; Crook and Mastro Reference Crook and Mastro2010; Grant et al. Reference Grant, Ryall, Lyons and Abou-Zaid2010, Reference Grant, Poland, Ciaramitaro, Lyons and Jones2011; Crook et al. Reference Crook, Khrimian, Cossé, Fraser and Mastro2012). A female-produced pheromone has also been discovered (Bartelt et al. Reference Bartelt, Cossé, Zilkowski and Fraser2007) and demonstrated to increase trap captures and detection rates (Silk et al. Reference Silk, Ryall, Mayo, Lemay, Grant and Crook2011; Ryall et al. Reference Ryall, Silk, Mayo, Crook, Khrimian and Cossé2012, Reference Ryall, Fidgen, Silk and Scarr2013; Silk and Ryall Reference Silk and Ryall2014). Other research has investigated incorporating visual cues into traps, demonstrating increased attraction of male EABs to traps with a dead or “model” EAB pinned onto ash leaflets (Lelito et al. Reference Lelito, Fraser, Mastro, Tumlinson and Baker2008; Domingue et al. Reference Domingue, Lelito, Fraser, Mastro, Tumlinson and Baker2013, Reference Domingue, Lakhtakia, Pulsifer, Hall, Badding and Bischof2014). Finally, the location in the tree where traps are placed is also known to be important. Higher trap captures are made in the mid canopy compared with the lower canopy, particularly for green traps (Francese et al. Reference Francese, Oliver, Fraser, Lance, Youssef and Sawyer2008, Reference Francese, Crook, Fraser, Lance, Sawyer and Mastro2010a, Reference Francese, Fraser, Rietz, Crook, Lance and Mastro2010b; Crook et al. Reference Crook, Francese, Zylstra, Fraser, Sawyer and Bartels2009; Ryall et al. Reference Ryall, Silk, Mayo, Crook, Khrimian and Cossé2012) and in sunny, exposed edges of stands compared with shaded interiors of stands (Francese et al. Reference Francese, Oliver, Fraser, Lance, Youssef and Sawyer2008; McCullough et al. Reference McCullough, Poland, Anulewicz and Cappaert2009a, Reference McCullough, Poland and Cappaert2009b; Marshall et al. Reference Marshall, Storer, Fraser and Mastro2010b). Finally, higher detection rates and trap captures have been reported when traps are deployed on either larger or dominant trees (Marshall et al. Reference Marshall, Storer, Fraser, Beachy and Mastro2009, Reference Marshall, Storer, Fraser and Mastro2010b).

Selection of a trap-lure combination for EAB surveys will depend on detection rates, costs, and ease of trap installation and maintenance, among other factors. The results of studies on detection rates of various trap-lure options have been complex, often contradictory and difficult to compare directly, with no one combination consistently out-performing all other options (e.g., see results of Marshall et al. Reference Marshall, Storer, Fraser, Beachy and Mastro2009, Reference Marshall, Storer, Fraser and Mastro2010b, Reference Marshall, Smith, Mech and Storer2013; McCullough et al. Reference McCullough, Siegert, Poland, Pierce and Ahn2011; Poland et al. Reference Poland, McCullough and Anulewicz2011; Fierke et al. Reference Fierke, Whitmore, Foelker, Vandenberg and Carlson2013; Mercader et al. Reference Mercader, McCullough and Bedford2013; Francese et al. Reference Francese, Rietz, Crook, Fraser, Lance and Mastro2013a; Poland and McCullough Reference Poland and McCullough2014). Most studies have assumed that higher trap captures will translate into higher detection rates when deployed in low-density sites. This may not be true. Indeed, very few studies have provided detection rates per se. Interactions between trap colour, lure combination, trap placement within and among trees, ash host tree density, and varying EAB population density among sites could all be influencing mean trap captures and detection rates.

At present, purple and green canopy prism traps (variously baited) are widely used in operational detection and monitoring surveys, in part because they are simple to install. However, prism traps are sticky, messy to handle, and disposed of after one season. Double decker traps, a variation of the canopy prism trap that is mounted on a fixed pole, may provide higher detection rates in low-density populations (McCullough et al. Reference McCullough, Siegert, Poland, Pierce and Ahn2011; Poland et al. Reference Poland, McCullough and Anulewicz2011; Poland and McCullough Reference Poland and McCullough2014). However, double decker traps use twice the number of prism traps and lures, therefore managers must decide if they want to invest more resources at fewer sites with a greater probability of detection or sample at a greater number of sites but with a lower probability of detection. Multiple funnel traps are a non-sticky and reuseable option to replace prism traps, but they are more expensive to purchase, more difficult to install because of their greater size and weight, and must be stored in the off-season.

Further research is required relating to all trap types on a number of topics including: quantifying timing of initial detection with respect to the duration of infestation, determining false-negative rates of traps, determining the effective trapping radius of various trap-lure combinations, relating trap captures to infestation densities, determining cost/benefits of the different options, and further improving our understanding of host and mate attraction in EAB. Finally, regardless of trap type, the capture of adults on baited traps does not indicate which trees are infested, or to what extent. Capture of an EAB adult on a trap should trigger delimitation surveys, using a variety of methods (traps, visual surveys, branch sampling, girdled trees) to determine the spatial extent of the infestation.

Detection using Cerceris fumipennis

Another potential option that could be implemented into detection surveys for EAB is to monitor predatory wasps that hunt specifically for buprestid beetles. “Biosurveillance” by the buprestid-hunting wasp, C. fumipennis Say (Hymenoptera: Crabronidae), has been investigated in recent research (Careless et al. Reference Careless, Marshall, Gill, Appleton, Favrin and Kimoto2009; Reference Careless, Marshall and Gill2014; Rutledge et al. Reference Rutledge, Fierke, Careless and Worthley2013; Swink et al. Reference Swink, Paiero and Nalepa2013). Beetle prey collected by female wasps can be monitored as the females return to their nest; suspected EAB adults can be taken from the wasp for identification. Indeed, the first detection of EAB in Connecticut was achieved using this method (Rutledge et al. Reference Rutledge, Fierke, Careless and Worthley2013). This wasp collects a high diversity of buprestid species, is geographically distributed across eastern North America, and monitoring of its populations could be incorporated into citizen science programmes. However, flight periods of the wasp only overlap with EAB for a short duration, and use of this method requires finding and time-consuming monitoring of nest colonies.

Remote sensing studies

Remote sensing is the collecting of imagery or other information from a distance, using sensors located on airplanes or satellites to collect data on patterns of energy being reflected from the earth. Research has investigated use of this approach to characterise the distribution of urban ash trees (Souci et al. Reference Souci, Hanou and Puchalski2009), information that is important when developing EAB surveys. Researchers have also attempted to use hyperspectral remote sensing technologies to map ash decline associated with EAB infestation (Pontius et al. Reference Pontius, Martin, Plourde and Hallett2008; Souci et al. Reference Souci, Hanou and Puchalski2009; Zhang et al. Reference Zhang, Hu and Robinson2014). While some studies were able to accurately predict ash decline, it was not specific to EAB and could be an indicator of any number of stressors. In addition, by the time ash decline was detectable, signs and symptoms of EAB were already evident, indicative that moderate to high populations were already present (Pontius et al. Reference Pontius, Martin, Plourde and Hallett2008) and thus limiting the value of this approach for early detection. Moreover, this technology is expensive, not specific to EAB infestation, and not readily available to forest managers.