Introduction

There are nearly 15,000 known species of jewel beetles (Coleoptera: Buprestidae) worldwide, with 787 species occurring in Canada and the United States (Nelson et al. 2008). Most buprestids feed in stressed, dying, or dead trees though some introduced species, e.g., the emerald ash borer, Agrilus planipennis Fairmaire, which can attack and kill healthy trees (Haack et al. 2002; Cappaert et al. 2005). Only a few buprestid species are considered economically injurious in the United States, e.g., the two-lined chestnut borer, A. bilineatus Weber, and the bronze birch borer, A. anxius Gory; therefore, relatively few taxa have been closely studied (Carlson and Knight 1969). The life history of A. planipennis makes studying them problematic and labor intensive; larvae are cryptic, hidden within leaves, or under bark during development, and most adult stages rest, feed, and mate in tree canopies out of reach of conventional trapping methods (Timms et al. 2006).

The most common detection method for adult wood-boring beetles has been the use of baited (chemical lures) or unbaited passive flight intercept traps (Chénier and Philogène 1989; Leather 2005). Historically, however, these methods result in relatively low yields of buprestids (Chénier and Philogène 1989; McIntosh et al. 2001; De Groot and Nott 2003). Recently, Cerceris fumipennis Say (Hymenoptera: Crabronidae), a native digger wasp, was suggested as a taxaspecific monitoring tool for detection of Buprestidae (Marshall et al. 2005). Female C. fumipennis provision their larvae almost exclusively with paralyzed buprestids in multiple cells within their subterranean nests. The buprestid prey can be collected by intercepting females as they return to their nests. Monitoring C. fumipennis nests has been shown to be an efficient way to collect buprestids and is being proposed as an alternative detection tool for A. planipennis at low densities (Careless 2009).

Consideration of C. fumipennis as a sampling tool requires an understanding of foraging behavior, including factors influencing prey selection. Cerceris fumipennis are known to prey on nearly 90 species of Buprestidae across 12 genera from all four North American subfamilies (Careless 2009). Despite their selectivity for buprestids, prey vary in appearance (e.g., shape, color, metallic luster) with no apparent uniformity at the family level to explain prey choice. Chemical assays are being undertaken to search for non-visual cues for prey location. Curiously, C. fumipennis is known to infrequently take anomalous (nonbuprestid) prey. Documented non-buprestid prey includes small Chrysomelidae (Neochlamisus bebbinae Brown, Bassareus mammifer Newman, Leptinotarsus decemlineata Say), mid-sized Scarabaeidae (Popilia japonica Newman), and Cerambycidae (Saperda discoidea F., Oberea schaumii LeConte) (Rutledge et al. 2011). As more C. fumipennis colonies are observed, it is becoming apparent that capture of these apparently anomalous prey are not isolated incidents and our understanding of prey selection is far from complete.

The overall goal of this study was to census New York Buprestidae using C. fumipennis as a sampling tool. Specific objectives were to: 1) identify beetles targeted by C. fumipennis, 2) compare how catches differed among colonies, 3) elucidate size range of prey, and 4) evaluate sampling efficiency of C. fumipennis relative to historical (i.e., museum) collections.

Table 1.

Cerceris fumipennis colony locations and sampling dates for Buprestidae sampling in four New York counties.

Figure 1.

Aerial images of five Cerceris fumipennis colonies within a 0.5 km radius. (A = Three Rivers Wildlife Management Area, B = RT9, C = RT50, D = WPR, E = LaFargeville Central School). High quality figures are available online.

Materials and Methods

Results

Beetle Sampling

Overall, 1,530 beetles representing 44 species from 13 genera and three families were collected from C. fumipennis during the two field seasons (Table 2). In 2009, 654 beetle captures yielded 32 species of Buprestidae and two species of Chrysomelidae. More intensive weekly sampling in 2010 at all five colonies yielded 876 specimens representing 35 species of Buprestidae, two species of Chrysomelidae, and one species of Cerambycidae.

Ten additional beetle species were captured in 2010 that were not collected in 2009 despite considerable overlap of species capture. Over the two years of study, five new prey species for C. fumipennis in the family Buprestidae were recorded: Agrilus cuprescens (Ménétriés), Agrilus pensus (Horn), Buprestis nutalli (Kirby), Chrysobothris scabripennis (Gory and Laporte), Dicerca pugionata (Germar). The capture records for the 44 beetle species collected, including the number collected by county and known larval and adult hosts, are provided in Appendix A.

Table 2.

Summary data by site for beetles collected at New York Cerceris fumipennis colonies in 2009 and 2010.

Figure 2.

Cerceris fumipennis beetle catches by genus in 2009 and 2010 from five colonies in New York State. Nonbuprestid genera are indicated by an asterisk. High quality figures are available online.

Figure 3.

Weekly phenologies of four abundant species of Buprestidae collected from five Cerceris fumipennis colonies in New York State in 2010. Sampling began 27 June and ended 21 August. High quality figures are available online.

Unique (sole record) beetle species were captured in all counties except Onondaga (Three Rivers Wildlife Management Area, Table 2). Of the 13 genera captured, the genus Dicerca was collected most frequently by C. fumipennis (57.3%), almost 4× as often as the second most captured genus, Agrilus (15.6%) (Figure 2). The genera Buprestis and Dicerca contained the largest sized species. In each county, at least half of the catches were from these two genera combined (50–83% of total catches) and overall they accounted for 68% of all beetles caught. The three non-buprestid genera were infrequently caught (0.6%).

Comparison of adjacent sampling sites

Two wasp colonies, RT9 and RT50, were located approximately 2.7 km apart in Saratoga County. Almost twice the number of beetles were collected at RT9 (n = 226) compared to RT50 (n = 125) in 2010. Composition of beetle catches also differed between the sites. Of the 26 beetle species collected at both sites, 12 were commonly collected from both (Figure 4). RT9 had higher species richness (23 versus 15) and diversity (H' = 2.23 versus 1.79), and C. fumipennis from this colony collected 11 unique beetle species, compared to only three unique species at RT50. Novel species accounted for 11.5% of captures at RT9 and 6.4% at RT50.

Figure 4.

Beetle captures at A) RT9 and B) RT50 Cerceris fumipennis colonies in Saratoga County in 2010. Shaded numbers indicate species caught at both sites. High quality figures are available online.

Table 3.

A comparison of buprestids collected from five Cerceris fumipennis colonies in New York State in 2009 and 2010 to buprestid collections from four New York State museums. Museum collection dates ranged from 1902 to present.

Table 4.

Average lengths of the largest and smallest beetles recorded from collection and museum records in the censused counties. The six species shaded in gray represent the three smallest and three largest beetle species caught by Cerceris fumipennis in this study.

Prey range

All beetles caught by C. fumipennis fell within a range of 4.1–18.9 mm in length (Table 4). The longest and broadest beetle caught by C. fumipennis, Buprestis maculativentris, measured 18.9 mm in length and 5.6 mm wide at the thorax (where C. fumipennis hold beetles during flight). Mean size measurements of 40 beetle species caught by C. fumipennis were plotted along with the 22 species identified in the museum records that were not caught to assess distribution of prey sizes (Figure 5A). Six buprestid species from museum collections were outside of the recorded prey size range. The four smallest buprestids ranged from 2.6–3.8 mm long, and the two largest beetles, both in the genus Chalcophora, were 19.6 and 24.0 mm long and 0.3–1.3 mm wider than the largest beetle collected from C. fumipennis, Buprestis maculativentris. The smallest and largest beetles from museum collections that were not caught by C. fumipennis fell outside of the 5^th and 95^th percentiles for length of all C. fumipennis-caught beetles measured. Activity periods of adult buprestids were also taken into account by looking at dates on which uncaught museum specimens were collected. Records of uncaught species indicate that 18 out of 22 species were historically collected outside of the C. fumipennis flight, and thus monitoring, period (Figure 5B).

Figure 5.

Mean size of beetle species caught by Cerceris fumipennis (diamonds) and species that were not caught (circles). Species not caught were compiled by reviewing buprestid collections from four New York State museums. A) Includes all measured beetles. B) After removal of museum specimens whose flight periods do not overlap with the Cerceris fumipennis flight period. Note: circles/diamonds overlap for several species. High quality figures are available online.

Figure 6.

Ratio of large and small beetles captured by three size classes of Cerceris fumipennis females. Numbers within bars are numbers of beetles (total n = 46) measured within the different size classes captured by Cerceris fumipennis (total n = 46) in each of the C. fumipennis size classes. Large beetles were Buprestis and Dicerca (n = 33), and small beetles were Agrilus and Chrysobothris (n = 13). Beetle genera were placed into size categories based on average length measurements (see text). High quality figures are available online.

Discussion

Cerceris fumipennis captured more buprestid individuals (1,520) and more species of Buprestidae (41) over these two short sampling seasons than any other single North American study found in an extensive search of the published literature. This survey is part of a larger northeastern biosurveillance project that will be the largest buprestid survey in North America to date (M. Bohne, USDA Forest Service Durham, personal communication). Overall, 16% of species caught in New York were new prey records, and with the inclusion of these seven new records, C. fumipennis is now known to prey on > 100 species of beetles. Five of the new prey records were Buprestidae (A. cuprescens, A. pensus, B. nutalli, C. scabripennis, D. pugionata). Oberea schaumii LeConte (Coleoptera: Cerambycidae) is a new record as well and only the second species of this beetle family caught by C. fumipennis. This novel prey family has never been recorded for a crabronid wasp prior to 2010 (Rutledge et al. 2011). A new chrysomelid species, B. mammifer, was also captured in addition to Neochlamisus bebbianae (Brown), a known prey species (Rutledge et al. 2011).

Though much is still unknown about how and why prey are selected, this study hints at some aspects of prey preference. Dicerca divaricata, a larger buprestid, was caught most frequently (37% of catches). This beetle uses 11 genera of trees as hosts and is well distributed within the United States (Nelson 1975; Nelson et al. 2008). Though potentially common, it is questionable as to whether the high catch frequency accurately reflects abundance at sampling sites or preferential predation. Nevertheless, C. fumipennis may prefer this species because of its larger size when compared to most of the other buprestid prey. A majority of species captured (70%) were collected 10 or fewer times and so are either rare in the landscape, less desirable to C. fumipennis, or possibly more cryptic (i.e., harder to find).

There was a great deal of variation in beetle species composition by county, as evidenced by the differing catch frequencies. Differences in forest abundance and composition likely affect wood-boring beetle abundance and composition. Dicerca divaricata was the most collected beetle in all counties except Jefferson County, where D. lurida was caught far more often than D. divaricata. One reason may be a greater number of D. lurida hosts, i.e., more hickories and oaks. Kurczewski and Miller (1984) documented another New York C. fumipennis colony near the town of Auburn where Agrilus anxius was caught most frequently followed by D. lurida in the apparent absence of D. divaricata. This could be an example of prey “switching,” as proposed by Murdoch (1969), where the most common species is preyed upon with a disproportionately high frequency over a less common species.

From a buprestid sampling standpoint, it would be optimal if available tree species are visited by C. fumipennis without preference. The most commonly caught species, D. divaricata, uses a diverse array of tree species as hosts (Appendix A). Though most Buprestidae are phloem feeders beneath tree bark, some of the smaller beetles in the tribe Trachyini are exclusively leaf miners. In this study, only two species of Trachyini in the genus Brachys, both leafminers of trees, were preyed on by C. fumipennis. An examination of museum records revealed other species of Trachyini that mine non-woody plants are present in the censused counties, but these were not collected by C. fumipennis. This may reflect a preference by C. fumipennis to forage in trees over lowergrowing, non-woody vegetation.

Increased host diversity is linked to increased herbivore diversity in other systems (Gaston 1992), and the opposite is also true with increased host abundance being linked to higher numbers of herbivores (Dempster 1971). The results of this study corroborate the tenet that increased tree diversity will result in increased beetle diversity, as the RT9 colony adjacent to the tree nursery with its variety of ornamental trees in Saratoga County had the most beetle species (29) over the two-year study. The highest numbers of conifer dwelling buprestid species (7) were also captured at this colony. This was more than LaFargeville Central School (Jefferson County) and WPR (Westchester County), where there were few or no conifers, based on a perusal of digital images, and the fact that only four conifer feeding species were caught. The second most species rich site was WPR, with 26 beetle species. Using the assumed relationship stated above, the Saratoga site, with higher numbers of conifer and deciduous feeding beetles, likely had the most species rich forest followed by Westchester County. Evaluation of this relationship was beyond the scope of this study, however, vegetative sampling near C. fumipennis colonies is strongly encouraged in future studies of this type to elucidate relationships between beetle species diversity and tree diversity.

Weekly sampling of beetles allowed construction of phenologies and provided insights into buprestid life histories as well as C. fumipennis foraging behavior. The most puzzling aspect of the phenological trends (Figure 5) was the sudden drop in beetle catches across all sites in the second week. An investigation of weather data excluded poor climactic conditions as a possible explanation as C. fumipennis foraging activity drops during cool, windy, cloudy, or rainy days (Careless 2009; W. Hellman, personal observation). Another unexplained phenomenon was the dwindling catches in the last four weeks. Whether beetle numbers decreased, C. fumipennis foraging declined, or a combination of both occurred remains to be documented. One possibility is that female C. fumipennis were foraging less in order to defend the nest from nest parasites (A. Hook, St. Edwards University, personal communication).

Species composition and capture frequency varied between relatively close colonies as well. RT9 and RT50 were only a few kilometers apart; however, beetle species and numbers were dissimilar between the two locations. Of the 26 species caught, only 12 were common to both sites, even though > 88% of catches were from the common species. The fact that both colonies were located on tree farms with various and presumably different ornamental tree species likely factored into these differences.

These data suggest a finite range of prey length for C. fumipennis of 4–19 mm, although weight may be equally, if not more, important. Smaller buprestids (< 4 mm) may avoid capture due to C. fumipennis being unable to manipulate and carry them or due to their less attractive size. The four smallest beetles measured from museum records were all < 4 mm and are not known prey items of C. fumipennis. Smaller beetles can weigh 1/20^th of the larger beetles, and so it would be inefficient for females to expend the energy to catch them if larger prey were available. Optimal foraging theory (MacArthur and Pianka 1966) would predict a greater payoff by preferentially targeting larger beetles up to a certain threshold where costs exceed reward to reduce the number of foraging flights C. fumipennis must take to provision a single cell. Nest excavation reported previously in the literature also supports optimal foraging theory and the idea that biomass influences provisioning. At one site, three relatively larger Dicerca beetles were used to provision a single C. fumipennis larval cell, while cells containing smaller Agrilus contained up to 51 beetles (Hook and Evans 1991).

Prey choice recorded in this study was overwhelmingly in favor of larger beetles (Figure 6). Other sphecoid wasps are known to select prey based on size with a marked preference for larger prey (Elgar and Jebb 1999, Polidori et al. 2005). In accordance with optimal forage predictions, the majority of prey captured by C. fumipennis (66%) were larger beetles from the genera Dicerca and Buprestis, with > 37% of all captures being D. divaricata. According to optimal foraging theory, D. divaricata would be a logical prey item, as it is both large (the second largest beetle preyed upon by C. fumipennis) and common (museum records report D. divaricata presence in each censused county and provide over 60 unique collection locations from other counties within New York State). Only small C. fumipennis preyed frequently on smaller beetles, a trend found in studies of other predatory wasps (Gwynne and Dodson 1983; Polidori et al. 2005). The size variation in female C. fumipennis may prevent smaller C. fumipennis from preying on larger beetles, but it has been suggested that the presence of smaller C. fumipennis may reduce intraspecific competition for large prey and this may, over time, promote speciation through niche partitioning (Stubblefield et al. 1993; Swanson et al. 2003; Woo et al. 2008).

An upper limit to prey size is reasonable, as very large beetles would be too heavy and broad for C. fumipennis to handle and transport. Only two Chalcophora species that may be too large for predation were identified from museum collections from the censused counties. These beetles were broader (6.0–7.0 mm wide at the thorax) than the broadest of beetles collected from C. fumipennis (D. maculativentris, 5.6 mm). It was not possible to weigh live Chalcophora, but dried specimens are volumetrically much larger than other prey and are expected to be significantly heavier than Dicerca. Cerceris fumipennis was observed to struggle flying with larger D. divaricata, and in some cases it could not take off once it landed.

Another important collection limitation is whether a particular beetle species is active during the C. fumipennis flight period. A number of buprestids may emerge too early or late in the season for C. fumipennis to catch in New York State. Only four of the 22 species not caught by C. fumipennis in New York museum records were noted as being collected during the wasp's flight period. In New York, the earliest wasps emerged around 27 June and had mostly senesced by 21 August. Beetles whose flight periods do not overlap with these dates would not be available as prey for foraging C. fumipennis. Alternatively, it is possible museum collection dates did not provide adequate insights into beetle flight periods, and that these beetles actually do occur during the C. fumipennis flight season (and in fact eight of the 22 species noted as being collected outside of the flight period are already known C. fumipennis prey items). If so, there may be other factors influencing why beetles were not caught, e.g., rare in the landscape, hosts not visited or not present near the C. fumipennis colony.

The factors listed above may influence why the four beetle species that were not excluded by temporal records were not utilized by C. fumipennis (Figure 5B). Pachyschelus laevigatus (Say) (Coleoptera: Buprestidae) and Mastogenius subcyaneus (LeConte) were the smallest buprestids measured in this study (2.6 and 3.0 mm long) and are outside of the prey size range documented by this study. Chalcophora virginiensis (Drury) was the largest buprestid measured (24.0 mm), and though it is active at the same time as C. fumipennis according to museum collection dates, it may be too large to prey upon. The only remaining beetle is Agrilus ruficollis, which was not collected in this study but is a documented prey item of C. fumipennis (Scullen 1965; Evans 1971). This beetle may either be rare in the landscape or its hosts may not be present near the censused colonies.

Twelve genera of Buprestidae drawn from all four buprestid subfamilies are represented in the known prey records of C. fumipennis (Careless 2009). In this study alone, three of the four North American buprestid subfamilies and 10 genera were represented. Nelson et al. (2008) lists 52 buprestid genera known in the United States and Canada, but only 20 have been reported in New York. Of the eight genera that at present are not known to be preyed on by C. fumipennis, only four (Taphrocerus, Pachyscelus, Mastogenus, Chalcophora) were present in museum collection records from censused counties, and representatives of all of these species were outside of the C. fumipennis prey size range. This means C. fumipennis exhibited a samsampling efficiency of approximately 75% across available buprestid genera in New York. Increased sampling efforts may result in the capture of the remaining four genera (Agrilaxia, Xenorhipis, Melanophila, Paragrilus); however, the factors discussed above may also impact whether a particular beetle species is captured or not. From a strictly practical viewpoint, beetles in the genus Agrilus tend to be the most economically important and, fortunately, they are readily detectable by C. fumipennis sampling.

This study provides information critical to using C. fumipennis as a sampling tool for Buprestidae. Other critical information necessary is a preliminary understanding of the C. fumipennis foraging range. It is suggested that prey-host records may be useful for approximating a true maximum forage distance. Many buprestids use a single or limited number of hosts, some of which are rare in some landscapes. Measurement from a C. fumipennis colony to a beetle's host tree would provide a strong inferential reference point to estimate maximum foraging distances. It is also the case that colonies are not always present by an intended sampling site. Because of this, research into developing protocols for translocating C. fumipennis is critical.

This research supports use of C. fumipennis as an effective and efficient tool for sampling Buprestidae. Sampling does not require extensive training or expensive equipment, and because C. fumipennis are not known to sting, there is no associated danger. They can be used to sample buprestid populations, conduct natural history surveys, and detect invasive species.