Physiological flight potential of adult beetles

Physiological flight capabilities of fed and unfed, and mated and unmated beetles of both sexes were evaluated by Taylor et al. (Reference Taylor, Bauer, Poland and Windell2010) using computer-monitored flight mills in a laboratory setting. In general, male EAB spent more time flying and flew further than unmated females. Mated adult females, however, are of more interest because they are most likely to expand the range of a population through their dispersal, reproductive potential and reduced Allee effects (Liebhold and Tobin Reference Liebhold and Tobin2008; Poland and McCullough Reference Poland and McCullough2014). Mated females allowed to feed on ash leaves between flight periods flew further and longer per day in the flight mill studies than unmated females (almost 2.5 times farther) or males (Taylor et al. Reference Taylor, Bauer, Poland and Windell2010). Mated females flew an average of 1.3 km per day and 12% flew more than 5 km in four days, with a maximum distance of 7.2 km over the four-day period. Many adult beetles only flew a few minutes per day, but of those that flew longer, total flight distances ranged from 71 m to 2.3 km for fed, six-day-old females compared with a range of 53 m to 5.2 km for males (Taylor et al. Reference Taylor, Bauer, Poland and Windell2010). These results suggest mated females may play a major role in the rate and pattern of EAB spread, particularly if flight mill recordings underestimate flight potential (Taylor et al. Reference Taylor, Bauer, Poland and Windell2010).

However, while flight mill studies can provide an estimate of the physiological capabilities of adult EAB and are useful for comparing EAB of different sexes, ages, and conditions (e.g., fed and unfed), these results cannot be directly extrapolated to field settings (Taylor et al. Reference Taylor, Bauer, Poland and Windell2010). Insects cannot necessarily be assumed to fly in a straight line and landing rates or dispersal behaviour is almost certainly affected by various biotic and abiotic factors, including the physiological state of the beetles and environmental conditions. For example, adult beetles are consistently attracted to stressed trees (McCullough et al. Reference McCullough, Poland, Anulewicz and Cappaert2009a, Reference McCullough, Poland and Cappaert2009b, Reference McCullough, Siegert, Poland, Pierce and Ahn2011) and to trees exposed to full or nearly full sun (Yu Reference Yu1992; McCullough et al. Reference McCullough, Poland, Anulewicz and Cappaert2009a). Host quality and availability of phloem for developing larvae vary as an infestation runs its course, altering development rates (Tluczek et al. Reference Tluczek, McCullough and Poland2011), but whether larval density or competition affects flight potential of adults is unknown. A high density of ash or even a single stressed ash growing in the immediate area of adult emergence may retain beetles that might otherwise disperse (Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b). Other variables such as disturbance by potential predators, wind, or other weather conditions, could act to trigger dispersal but, to date, have not been evaluated.

Despite these limitations, the flight mill data suggest beetle dispersal could lead to a fairly rapid spread rate. Assuming a completely random walk model and a 4-km flight potential, Taylor et al. (Reference Taylor, Bauer, Poland and Windell2010) estimated that ~1% of females would be displaced roughly 500 m while flying. Given that beetles are unlikely to engage in completely random walks and the ability of females to fly more than 4 km over their lifetime, it seems likely that a few mated females will fly at least 500 m and possibly further. Despite the difficulty of finding newly infested trees with low densities of EAB larvae, field studies confirm this suggestion (see below), and indicate that a non-trivial number of mated females are engaging in long-distance directed flights (Mercader et al. Reference Mercader, Siegert and McCullough2012).

Dispersal observed under field conditions

Estimates of local EAB dispersal and spread have been acquired from intensive field studies that involved felling and debarking up to 220 ash trees per site to assess presence and density of the larval progeny of adults that emerged from a known origin (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2009; Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b). This method provides a reliable estimate of dispersal patterns within the sampled area but the expense and logistical difficulties of collecting these data limit the spatial extent of the area that can be evaluated. Thus, the probability of finding isolated trees colonised by long-distance dispersers is minimal (Mercader et al. Reference Mercader, Siegert and McCullough2012).

Mercader et al. (Reference Mercader, Siegert, Liebhold and McCullough2009) assessed local dispersal of a single generation of EAB emerging from firewood at two sites by intensively sampling ash trees at known distances from the origin of each infestation. Sampling involved debarking either the entire tree or multiple areas on the trunk and large branches of each tree to record larval stage and density per m² of exposed phloem. Both sites had relatively high densities of ash trees growing in a generally linear spatial arrangement, which allowed for a precise estimate of local dispersal based on the observed distribution of larvae. Distribution of the larval cohort representing progeny of the dispersing EAB adults (i.e., EAB larval density by distance) was well described by a negative exponential function. About 90% of the larvae (88.9 and 90.3% at the two sites, respectively) were on trees within 100 m of the origin (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2009). The spatial distribution of ash in these sites, however, may have effectively functioned as a corridor, potentially facilitating tree to tree movement of dispersing beetles (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011a).

Siegert et al. (Reference Siegert, McCullough, Williams, Fraser and Poland2010b) similarly felled and debarked ash trees to assess EAB dispersal at two sites with isolated satellite EAB populations discovered as a result of regulatory officials tracking nursery trees that were shipped from an infested nursery. The sites were identified one and three years after the nursery trees were planted. In contrast to the relatively linear arrangement of ash trees in Mercader et al. (Reference Mercader, Siegert, Liebhold and McCullough2009), ash density and distribution varied in these heterogeneous sites that included urban, residential, and wooded areas. Ash trees at both sites were systematically sampled within an 800-m radius of the origin to locate larval galleries created by the progeny of adult EAB beetles that dispersed from the planted nursery trees. At the one-year site, dispersing adults colonised trees out to 638 m from the origin, with average dispersal of about 360 m (Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b). At the three-year site, dispersal averaged about 320 m and trees out to 540 m from the origin were colonised over the three-year period (Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b). Several site-related factors that may have influenced EAB dispersal at the two sites, such as wind direction, abundance of ash stems and phloem area, distance from the epicentre, and land-use type (i.e., wooded, residential, agricultural, or urban), were evaluated. Abundance of ash phloem within 200 m of the origin strongly affected EAB dispersal, but this effect diminished rapidly with distance (Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b). In other words, within 200 m of the origin, beetles were most likely to colonise trees in areas where ash phloem was plentiful compared with areas with scarce ash phloem, but beyond 200 m, the abundance of ash phloem had little influence on adult dispersal or host selection. Other site-related factors had no discernable effect on larval distribution or density.

Reconstruction of the initial spread

Dendrochronology, the study of tree rings, was applied to recreate the onset and progression of EAB-induced ash mortality at several isolated satellite EAB infestations and across southeastern Michigan, United States of America where EAB initially became established (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). Ash trees were sampled along a systematic network of points across roughly 1.5 million ha (>15 000 km²) in southeastern Michigan, encompassing the primary EAB population. Dendrochronological methods were used to determine the date of EAB-induced ash mortality, which served as a proxy to draw inferences about the spatio-temporal dynamics of the EAB invasion, including EAB spread and dispersal that occurred before this pest was detected in 2002. Reconstruction of the progression of ash mortality and EAB spread was possible due to several factors specific to this system, including the distinct larval galleries, lack of host resistance, relatively rapid host mortality, and nearly ubiquitous distribution of green ash (F. pennsylvanica Marshall) in southeastern Michigan.

Siegert et al. (Reference Siegert, McCullough, Liebhold and Telewski2014) determined that ash trees were killed by EAB as early as 1997, indicating that EAB was present in North America by at least 1993. Early spread of EAB (i.e., progression of ash mortality) progressively radiated from the epicentre of the infestation at 3.8 km per year initially, but increased abruptly to roughly 13 km per year after a few years, as an increasing number of satellite infestations became established, expanded, and coalesced with each other and the primary EAB population. Whether the propensity of adult EAB to engage in long-distance dispersal was related to EAB density remains unknown, but little evidence currently exists to support a density-dependent relationship. Rather, biphasic range expansion like that exhibited by EAB is typical of invasive species with stratified dispersal in which new satellite populations become established close to the edge of the primary population and coalescence occurs in a short time (Shigesada and Kawasaki Reference Shigesada and Kawasaki1997).

The number of new satellite populations increased annually as the primary EAB population increased in size, with an average increase of 7.4 satellite populations per year over six years (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). Specifically, ⩽5 new satellite populations formed in 1998 and 1999, but that number steadily increased each year to over 30 new satellite populations formed in 2002 (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). Distance between the primary EAB infestation and new satellite populations averaged 24.5 km but ranged from <10 km to over 70 km. Approximately 70% of new satellite populations became established within 30 km of the primary EAB population. Early satellite populations became established an average distance of 8.2 km, with a maximum distance of 12.9 km from the edge of the primary population (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). Thereafter, however, distances of new satellite populations averaged 27.4 km from the primary EAB population, with average maximum distances of 64.7 km (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). These patterns likely reflect a combination of natural EAB dispersal and anthropogenic movement of EAB-infested ash material across the southeastern Michigan study area before the detection and identification of EAB in North America in 2002. As noted by the authors, regulatory and outreach efforts have increased public awareness about the risk of transporting EAB in infested ash material, presumably reducing the likelihood of new satellite EAB populations becoming established (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). Previously unknown infestations, however, continue to be detected every year (EAB 2014).

Emerald ash borer spread models

Stratified dispersal characterises spread of many invasive species and may greatly influence the rate of radial expansion in a biological invasion (e.g., Moody and Mack Reference Moody and Mack1988; Okubo et al. Reference Okubo, Maini, Williamson and Murray1989; Andow et al. Reference Andow, Kareiva, Levin and Okubo1990; Shigesada and Kawasaki Reference Shigesada and Kawasaki1997). Since long-distance dispersal events inherently play an integral role in this process and are often driven by anthropogenic factors (Johnson and Padilla Reference Johnson and Padilla1996; Suarez et al. Reference Suarez, Holway and Case2001; Gilbert et al. Reference Gilbert, Grégoire, Freise and Heitland2004; Leal et al. Reference Leal, Allen, Humble, Sela and Uzunovic2010; United States Department of Agriculture Animal and Plant Health Inspection Service 2010), several efforts have been made to incorporate human-mediated, socio-economic factors in regional spread models of various invasive species, such as through gravity models (e.g., Bossenbroek et al. Reference Bossenbroek, Kraft and Nekola2001; Leung et al. Reference Leung, Bossenbroek and Lodge2006; Bigsby et al. Reference Bigsby, Tobin and Sills2011). Predictions of spread along pathways influenced by anthropogenic factors, however, may be critically associated with high levels of uncertainty. Recent approaches that account for and attempt to quantify this uncertainty may provide more realistic predictions of human-mediated, long-distance dispersal events (Yemshanov et al. Reference Yemshanov, Koch, Lyons, Ducey and Koehler2011, Reference Yemshanov, Koch, Ducey, Siltanen, Wilson and Koehler2013; Koch et al. Reference Koch, Yemshanov and Haack2013).

Several researchers have used a variety of regulatory or biological data in efforts to model EAB spread at local and regional scales. Regional models were usually based on the rate of new county or sub-county EAB detections over time (e.g., Muirhead et al. Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006; Kovacs et al. Reference Kovacs, Haight, McCullough, Mercader, Siegert and Liebhold2010, Reference Kovacs, Mercader, Haight, Siegert, McCullough and Liebhold2011; Prasad et al. Reference Prasad, Iverson, Peters, Bossenbroek, Matthews and Sydnor2010), while local models often incorporated EAB population dynamics, ash density and distribution, and functions to describe short-distance dispersal of adult beetles (e.g., BenDor et al. Reference BenDor, Metcalf, Fontenot, Sangunett and Hannon2006; Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011a, Reference Mercader, Siegert, Liebhold and McCullough2011b; McCullough and Mercader Reference McCullough and Mercader2012; Kovacs et al. Reference Kovacs, Haight, Mercader and McCullough2014). Long-distance movement of EAB is usually addressed only in regional models, where pathways of potential spread may be weighted by demographic or anthropogenic factors.

In one of the first attempts to model EAB regionally, Muirhead et al. (Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006) incorporated local diffusion of the EAB population and a long-distance dispersal estimate. Local diffusion was modelled with a standard negative exponential function based on sub-county (i.e., township or municipal) detections of EAB in the United States of America and Canada between 2002 and 2004. A long-distance dispersal estimate, based on human population size as a surrogate of human activities that could transport EAB from infested to non-infested areas, was additionally incorporated into their model (Muirhead et al. Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006). A gravity model predicting firewood movement from infested areas to new areas based on number of campsites and distance from the known EAB infested area was also included (Muirhead et al. Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006). As more became known about EAB biology and detection technology progressed, however, it became clear that the detection maps used as the basis for the Muirhead et al. (Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006) model were inaccurate. Many of the infestations identified between 2002 and 2004 represent improved and more extensive detection efforts rather than new infestations, a pattern confirmed by dendrochronological data (Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014). Despite these issues, observations derived from the model presented by Muirhead et al. (Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006), such as the importance of satellite populations and artificial movement, remain valid.

Another effort to model regional EAB spread also incorporated short-distance and long-distance dispersal parameters, including EAB and ash abundance, major roads and traffic density, campground size and usage, distance from the core infestation, variables related to the wood products industry, and human population density (Iverson et al. Reference Iverson, Prasad, Bossenbroek, Sydnor and Schwartz2010; Prasad et al. Reference Prasad, Iverson, Peters, Bossenbroek, Matthews and Sydnor2010). Diffusion was modelled using a fat-tailed power function (Schwartz Reference Schwartz1992). An EAB spread rate of about 20 km per year, based on the authors’ approximation of the progression of ash mortality and preliminary flight mill data (Taylor et al. Reference Taylor, Bauer, Miller and Haack2005, Reference Taylor, Poland, Bauer, Windell and Kautz2007), was used to predict short-distance flight. Long-distance dispersal was modelled separately using the same diffusion model by adjusting the upper dispersal limit to 400 km. This model was parameterised using approximated patterns of new EAB detections observed in the region and weighted using factors that might contribute to artificial movement (e.g., road traffic, wood products industries, human population density, and campground presence) (Prasad et al. Reference Prasad, Iverson, Peters, Bossenbroek, Matthews and Sydnor2010). This modelling approach once again highlighted the importance of satellite populations and the potential importance of road networks for artificial movement (Prasad et al. Reference Prasad, Iverson, Peters, Bossenbroek, Matthews and Sydnor2010).

Kovacs et al. (Reference Kovacs, Haight, McCullough, Mercader, Siegert and Liebhold2010) developed a regional EAB spread model to estimate potential economic damages posed by treatment or removal and replacement of ash trees on urban land over a 10-year period. The area of interest encompassed a sizeable portion of eastern North America, extending well beyond that considered by previous models (e.g., Muirhead et al. Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006; Prasad et al. Reference Prasad, Iverson, Peters, Bossenbroek, Matthews and Sydnor2010). Potential spread was projected by overlaying a 7046 equidistant point grid over the study area and using a probabilistic model to predict the annual probability that EAB in an infested cell would spread and cause an infestation in a vacant cell at a detectable level. A negative exponential function was used to predict the probability that one cell would lead to an infestation in another cell. The best fitting parameters of the model were selected by contrasting predictions from 250 000 simulations to observed infestations (Kovacs et al. Reference Kovacs, Haight, McCullough, Mercader, Siegert and Liebhold2010). Potential long-distance transport of EAB by humans, however, was not included in the model. If the extreme long-distance satellite EAB infestations detected since March 2009 (e.g., eastern New York, Georgia, Kansas, Colorado, and Minnesota, United States of America) are excluded, the EAB infested area predicted by the Kovacs et al. (Reference Kovacs, Haight, McCullough, Mercader, Siegert and Liebhold2010) model is a fairly accurate depiction of the known EAB infestation to date. A variation of this spread model was subsequently used to evaluate the role of satellite EAB populations on projected economic damages in United States of America communities and the potential economic benefits accrued by managing satellite populations (Kovacs et al. Reference Kovacs, Mercader, Haight, Siegert, McCullough and Liebhold2011). Results showed that reducing the number of new EAB satellites from 2005 to 2010 could have delayed economic costs by $1.0 to $7.4 billion USD, demonstrating the importance of investing in activities to prevent or slow the expansion of new satellite populations.

Given the consistent importance of satellite populations and the need for local EAB management, a clear understanding of the population dynamics and local spread of EAB is essential. BenDor et al. (Reference BenDor, Metcalf, Fontenot, Sangunett and Hannon2006) developed the first EAB spread model to incorporate EAB population dynamics, ash density and distribution, and short-distance dispersal of adult EAB. Little empirical data about the spread and dispersal of EAB were available at the time, however, to parameterise the models. Their model, further highlighted in BenDor and Metcalf (Reference BenDor and Metcalf2006), assumed a small percentage of adults would disperse as far as 240 m in a density-dependent manner. Parameterisation of the model was based on preliminary results of flight mill studies (Bauer et al. Reference Bauer, Miller, Taylor and Haack2004).

More recent work by Mercader et al. (Reference Mercader, Siegert, Liebhold and McCullough2011a) accounted for factors such as initial EAB population density and distribution, ash density and distribution, and changing EAB population dynamics (i.e., one-year versus two-year EAB development) over time and space. This local spread model was rigorously parameterised using empirical data from several field studies in numerous locations, including those reported in Mercader et al. (Reference Mercader, Siegert, Liebhold and McCullough2009) and Siegert et al. (Reference Siegert, McCullough, Williams, Fraser and Poland2010b) (see above). This model illustrated the difficulties associated with attempting to precisely predict the spread of EAB in a given area. Results highlighted the important influence of ash distribution and assumptions regarding the foraging behaviour of egg-laying EAB females on predictions of spread (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011a). Due to the absence of data regarding the propensity for EAB females to disperse beyond 800 m, the model presented by Mercader et al. (Reference Mercader, Siegert, Liebhold and McCullough2011a) excluded long-distance dispersal and underestimated the spread rate of EAB. Adjusting this model to assume a small portion of the population would engage in long-distance dispersal events improved the accuracy of spread predictions (Mercader et al. Reference Mercader, Siegert and McCullough2012). However, much remains unknown about the factors that trigger adult dispersal and the proportion of adults that engage in such behaviour, which continues to challenge our ability to realistically model the natural expansion of EAB populations.

Efforts to model the spread and dispersal of EAB at regional and local scales have shown steady improvement as we have learned more about the insect. While early models were hampered by the scarcity of information on EAB (e.g., BenDor et al. Reference BenDor, Metcalf, Fontenot, Sangunett and Hannon2006; Muirhead et al. Reference Muirhead, Leung, van Overdijk, Kelly, Nandakumar and Marchant2006), they nonetheless were valid attempts at the time and consistently showed the importance of satellite populations and human-mediated movement along various pathways. More recent models predicting local (e.g., Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011a, Reference Mercader, Siegert and McCullough2012) and regional spread of EAB (e.g., Kovacs et al. Reference Kovacs, Haight, McCullough, Mercader, Siegert and Liebhold2010, Reference Kovacs, Mercader, Haight, Siegert, McCullough and Liebhold2011; Prasad et al. Reference Prasad, Iverson, Peters, Bossenbroek, Matthews and Sydnor2010) are improving as more empirical work has been conducted allowing for more realistic parameterisation of the models. Although we currently do not have an accurate model of the natural spread of EAB, available models do provide functional estimates of the local population dynamics of EAB. The modelling approach presented in Mercader et al. (Reference Mercader, Siegert, Liebhold and McCullough2011a) has proven useful for evaluating potential management strategies in urban forests (McCullough and Mercader Reference McCullough and Mercader2012), comparing the effectiveness of three management options (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011b), and assessing the influence of population density and dispersal behaviour on the ability to detect and monitor EAB populations (Mercader et al. Reference Mercader, Siegert and McCullough2012). Spread and dispersal models of EAB in the future will no doubt build upon these earnest, earlier efforts.

Key areas in need of future research

Broad general patterns regarding the dispersal and spread of EAB are beginning to become apparent, but key components of the dispersal patterns of EAB required for management purposes remain obscure. In particular, we need a clearer understanding of (1) the propensity of EAB to engage in long-distance dispersal flights, (2) the influence of landscape features and climate on EAB dispersal, (3) empirical tests of the potential for management to affect local spread of isolated EAB populations, (4) individual movement of adult beetles, and (5) historical and current factors associated with long-distance movement of EAB by humans.

The proclivity for long-distance dispersal events is a fundamental aspect of the dispersal and spread of EAB that remains a significant challenge. Currently, our knowledge is limited to an admission that even when ash trees are immediately available, some gravid females still lay eggs on hosts located at substantially farther distances than the hosts selected by the great majority of females in the population (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2009; Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b). The proportion of females that engage in long-distance flights and the factors that influence or trigger this behaviour, however, are as yet unknown but remain highly relevant to managers attempting to slow expansion of local EAB populations. Some variables, such as ash condition or EAB density, can be altered by available management tactics (e.g., girdling or applying systemic insecticides to selected ash trees). Large-scale field studies indicate such tactics can slow EAB population growth but there is little evidence that they slow spread (McCullough et al. Reference McCullough, Mercader and Siegert2015) and some management options, such as pre-emptive ash removal, could potentially increase the likelihood of long-distance dispersal (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011b). In addition, accurately predicting how landscape features, such as elevation changes or bodies of water, affect EAB spread will likely depend on a better understanding of long-distance dispersal.

Landscape features and ash distribution are expected to play a significant role in the spread of EAB (Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b; Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011a) and presumably climatic variables will also significantly influence dispersal. As our understanding of biological traits of EAB, such as host selection behaviour and its ability to tolerate varying climate regimes, increases, models to project potential EAB distribution should become increasingly accurate (Sobek-Swant et al. Reference Sobek-Swant, Kluza, Cuddingtion and Lyons2012). The rapid regional spread of EAB has positioned this insect in a large variety of environments and, thus, inferring movement from studies conducted in sites in one region may not be appropriate. For instance, how EAB will disperse in mountainous areas, across wide rivers, in areas with very scattered or very small patches of ash is largely unknown. In addition, whether ash along roads or limited to riparian areas facilitates dispersal or accelerates spread rates remains to be sufficiently evaluated. Likewise, in urban areas where EAB infestations result in substantial economic costs (Poland and McCullough Reference Poland and McCullough2006; Kovacs et al. Reference Kovacs, Haight, McCullough, Mercader, Siegert and Liebhold2010, Reference Kovacs, Haight, Mercader and McCullough2014; McCullough and Mercader Reference McCullough and Mercader2012), differences in climate, landscape features, or the dominant species of ash may affect dispersal or spread. Furthermore, in contrast to southeastern Michigan in the 1990s and early 2000s (e.g., Siegert et al. Reference Siegert, McCullough, Liebhold and Telewski2014), people in more recently infested urban areas are less likely to move ash material, due to state and federal regulatory actions and increased public awareness of the risk involved. Therefore, EAB spread at the local level should presumably be dominated by natural dispersal of the insect rather than a combination of natural spread and unchecked anthropogenic movement of EAB-infested ash material to nearby non-infested areas.

Delineating an EAB population is exceedingly difficult (McCullough and Siegert Reference McCullough and Siegert2007; Siegert et al. Reference Siegert, McCullough, Williams, Fraser and Poland2010b; Mercader et al. Reference Mercader, Siegert and McCullough2012), in part because of our rudimentary understanding of EAB spread and dispersal, which poses significant limitations on the ability to effectively manage EAB. In particular, it provides a barrier to the ability to optimally allocate the often limited resources available to manage an infestation. For example, attempting to target management to the origin of an infestation can lead to lower efficiency unless the infestation is detected shortly after it becomes established (McCullough and Mercader Reference McCullough and Mercader2012). This is due to the unfortunate reality that most infestations are detected after EAB has already begun to spread from its initial colonisation point. If the dispersal of EAB could be influenced, such as by the use of clusters of girdled ash trees, then targeted management approaches may become realistic. Some evidence exists that the spread could be influenced by the use of girdled trees (e.g., McCullough et al. Reference McCullough, Poland, Anulewicz and Cappaert2009a, Reference McCullough, Poland and Cappaert2009b; Siegert et al. Reference Siegert, McCullough, Poland, Heyd, Pierce and Xue2010a, Reference Siegert, McCullough, Williams, Fraser and Poland2010b) and models suggest this effect would be important (Mercader et al. Reference Mercader, Siegert, Liebhold and McCullough2011a, Reference Mercader, Siegert, Liebhold and McCullough2011b), but empirical results detailing the extent of the effect or the feasibility of this strategy are limited.

Empirical results of the natural movement of EAB in the field are restricted primarily to observations derived from larval counts under limited conditions. Information regarding the movement of adults aside from that implied by the eggs laid by females is absent. For example, flight mill studies suggest the potential for sex-biased dispersal (Taylor et al. Reference Taylor, Bauer, Poland and Windell2010), which has the potential to affect the spread of invasive species (Miller and Inouye Reference Miller and Inouye2013), but has not been documented for EAB in the field. Furthermore, as systemic insecticides used to treat ash trees are often expressed in the leaves, the individual movement of leaf-feeding EAB adults between trees can significantly impact the effectiveness of management using such treatments. Specifically, the movement of adults between trees would greatly increase the probability of adult beetles encountering lethal treatments leading to the need for fewer treatments to be applied (McCullough and Mercader Reference McCullough and Mercader2012) and/or more effective strategic placement of those treatments in management scenarios.

While increased knowledge of EAB dispersal behaviour could substantially improve spread predictions, survey programmes, and management efforts at local levels, understanding the historical and current spread of EAB by humans is also important. Such information would improve large-scale projections of EAB expansion over time, but also serve as a basis for projecting human-assisted movement of future invasive wood-boring or phloem-boring pests. Evaluation of historical EAB data collected by regulatory officials, results from ongoing EAB detection efforts, and the increase in public awareness about EAB should be useful in evaluating how human-assisted transport of EAB and associated spread rates changed over time. Development of pathway-based models (e.g., Colunga-Garcia et al. Reference Colunga-Garcia, Haack, Magarey and Margosian2010; Koch et al. Reference Koch, Yemshanov, Colunga-Garcia, Magarey and Smith2010; Yemshanov et al. Reference Yemshanov, Koch, Ducey, Siltanen, Wilson and Koehler2013), incorporating socio-economic factors such as human population density, domestic transport networks and recreation or economic activities associated with long-distance movement of potentially infested ash material, could improve accuracy of long-term EAB spread predictions and help to target future management and regulatory efforts.