Vespa mandarinia Smith 1852 is a semi-specialized predator of other social Hymenoptera and one of the two largest species of Vespa. Several individuals of this predatory wasp were detected in Canada and the United States in 2019, including an entire nest that was located and destroyed on Vancouver Island, British Columbia. The Washington State Department of Agriculture and the United States Department of Agriculture’s Animal and Plant Health Inspection Service have collaborated to survey Washington State for V. mandarinia since 2020, using traps staffed by agency personnel, collaborators from local governments and nongovernmental organizations, and the general public. Trap data and public reports were used to select sites for live-trapping or net surveys, and live hornets captured in these efforts were subsequently collected and fitted with radio tags to locate nests. The survey ultimately led to the discovery of a V. mandarinia nest in October 2020, and three nests in August and September 2021. All of the nests were located within in red alder trees (Alnus rubra), with one just above the ground in a standing dead tree, and the other three in cavities ~2 to 5 meters above the ground in living trees. The number of combs in each nest varied between four and ten, cells between 418 and 1,329, and total hornets per nest between 449 and 1,474 (including immature and mature stages). Together, the four nests indicate an incipient population of V. mandarinia in the Cascadia region, and ongoing action by local, state, provincial, and federal governments, and residents of both countries is required to avoid the establishment of this exotic species in the region.

community science, Eradication, radio-tracking, social Hymenoptera

Vespa mandarinia Smith, 1855, is one of the 22 species of hornet (Vespidae: Vespa) (Smith-Pardo et al. 2020). Like all hornets, V. mandarinia is a eusocial predaceous wasp that feeds on many different types of insects. One aspect of its behavior (shared by the closely related V. soror) is the ability to mount group attacks on other social Hymenoptera, including honey bees (Apis spp.), resulting in rapid destruction of entire colonies of their prey (Matsuura and Sakagami 1973). This behavior makes it a perennial pest of apiaries throughout its native range, and it is a notable predator of Apis mellifera, which lacks effective defensive behaviors against V. mandarinia (Matsuura and Sakagami 1973; Arca et al. 2014; Mattila et al. 2020).

The first verified North American specimens of V. mandarinia outside of ports of entry were collected in British Columbia in 2019. Several wasps were observed that year on Vancouver Island, British Columbia, culminating in the location and eradication of an entire nest in the city of Nanaimo in September 2019 (Bérubé 2020). In December 2019, V. mandarinia workers were found approximately 90 km SW in Blaine, WA. The first confirmed specimen in the United States was a single dead worker that was collected on a porch (Wilson et al. 2020). Two additional specimens that had been collected earlier in the year were subsequently provided to the Washington State Department of Agriculture (WSDA), both of which were associated with beehive attacks – one hornet that had been “hawking” at a hive, and two specimens which were collected from a killed colony. Two colony losses that strongly resembled descriptions of V. mandarinia attacks in Japan (see Matsuura and Sakagami 1973) were also recorded in fall 2019 in the same region, although wasp specimens were not collected at either event. Based on the available evidence, WSDA, the United States Department of Agriculture’s Animal and Plant Health Inspection Service (USDA APHIS), and the British Columbia Ministry of Agriculture implemented a robust trapping program in 2020, with the goal of early detection and eradication of any nascent V. mandarinia populations. The program included trapping by agency personnel and the general public and continuous public outreach to locate hornet populations. Because trapping hornet workers will not lead to eradication, WSDA also developed a protocol for capturing live hornets and tracking them to their nest, which could then be eliminated.

Lethal trapping

The basic survey approach was to use lethal traps and public reports to locate centers of hornet activity, and then shift to capturing live hornets that could be tracked back to nests. A variety of trapping approaches were identified from the literature and the experiences of beekeepers in the hornet’s native range (Tatsuta and Makino 2003; Choi et al. 2012; Paschapur et al. 2022). Although there is considerable diversity in trapping approaches reported in the literature, many utilize a blend of fruit juices and ethanol. WSDA used bottle traps armed with orange juice and rice cooking wine (Makino and Sayama 2005; Okuda et al. 2011) because both components were available at grocery stores and promoted consistency across our collaborator and community science trapping programs. In 2020, traps were constructed from 1.65L (56 oz) clear plastic bottles with 2 cm square openings on 3 sides (Fig. 1). Each bottle contained a bait/killing solution comprising approximately 120 ml (4 oz) of orange juice and 120 ml (4 oz) of rice wine of at least 10% ABV. Captured hornets would subsequently drown in the bait. This approach was repeated in 2021, with the exception that the square-shaped openings were replaced by star-shaped openings due to concerns that the hornets might escape from the traps. This concern was raised when late-season captures of other insects in 2020 were so massive that they formed a solid surface inside the traps, allowing hornets to avoid the killing solution.

Figure 1. 

A bottle trap used to survey for V. mandarinia by the Washington State Department of Agriculture, collaborating agencies, and community scientists.

Traps maintained by WSDA were mostly placed in Whatcom County in northwestern Washington State and maintained from June through November, covering the entire presumed area of potential hornet occupancy. In 2020, three traps were placed per square km, within a 2 km radius from any detection made through May 2020. Trap density was decreased to two traps/km² at a 4 km radius, and to one trap/km² at an 8 km radius. The maximum distance for the 2020 trapping areas was based on the maximum foraging distance of 8 km reported by Matsuura and Sakagami (1973). Traps were placed on branches in trees at ca. 2-2.2 meters high, depending on site characteristics and trapper height. In 2021, the trap density was reduced to 1/km² throughout the trapping area, but the geographic area of trap coverage was expanded. A series of traps testing alternative lures (e.g., isobutanol, 2-methyl butanol, and acetic acid; Landolt and Zhang 2016) was also maintained at eight sites in 2020 during the trapping season. Those results are not analyzed in this paper, but hornets caught using experimental lures are included in the results reported here.

Traps maintained by WSDA were inspected every 7 to 10 days, and each trap action (e.g., trap placement, service, specimen collected) was recorded using a smartphone. A short trap check interval was selected in part because the traps lacked a preservative, with longer trap return intervals potentially compromising DNA and morphological analysis of trap contents, in addition to being unpleasant to service. The short interval also allowed the agency to rapidly respond to captures and begin attempts to collect live specimens to track to a nest. In 2020, all trap contents were strained, sorted for field detection of V. mandarinia, and all bycatch collected into a sample bottle for further analysis. In 2021, contents were sorted in the field to detect V. mandarinia, with one trap randomly selected daily by each trapper for analysis of bycatch. Because trap contents were not returned to the lab in 2021, the agency initiated a quality control program by placing a preserved hornet in a random trap in each trapper’s field area once a week to ensure suspect hornets were reported.

WSDA also recruited other agencies, nongovernmental organizations, and private residents (i.e., community scientists) to employ the same trapping protocol and expand the program more broadly in the state. This approach was also used in British Columbia, with community scientists staffing most traps. Instructions for constructing and servicing traps were provided on the WSDA website, and trap locations were logged by participants via an ArcGIS web app. These traps were located opportunistically and at the convenience of the participants, who were asked to check traps weekly. In 2020 the agency requested participants to send all trap contents to the WSDA Entomology Laboratory for analysis, in part to ensure hornets were not missed, and in part to analyze bycatch for impacts on non-hornet species. In 2021 the agency only requested trap contents from captures of potential hornets. In all, 1,681 traps were deployed in 2020 (797 WSDA, 263 collaborating agencies, 621 community science), and 1,648 total traps were deployed in 2021 (868 WSDA, 370 collaborating agencies, 410 community science) (Fig. 2).

Figure 2. 

A map of bottle traps maintained by WSDA B map of bottle traps maintained by private residents and collaborating agencies. (Trap sites in British Columbia were not typically logged on the website and are thus underrepresented in these maps.) Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.

Live trapping

An array of live traps was deployed near any hornet detections. Two live trap designs were used. One was based on a modified bottle trap incorporating a screen to separate the hornets from the attractant/kill solution. We also used translucent unitraps, again with a screen to isolate hornets from the killing solution. The unitraps were further modified by adding additional screened holes to the upper section to reduce heat and fumes inside of the trap, and potentially increase the chemical plume released from the baits (Fig. 3). Once deployed, live traps were checked daily, six days a week.

Figure 3. 

Two styles of live traps used to collect V. mandarinia specimens for subsequent tracking back to an active nest.

Tracking

Live hornets captured in traps or by net were chilled on ice and affixed with a radio tag for tracking back to a nest. Bluetooth tags constructed at the University of Washington, modified from tags developed to monitor bumble bee activity, were used for the first two tracking attempts. Subsequent tracking efforts employed VHF radio tags produced by Lotek (in 2020) or Advanced Telemetry Systems (in 2021). The initial attempt to glue a tag to a captive hornet, following the approach used by Iyer et al. (2019, 2020) was not successful. Bluetooth and VHF tags were subsequently attached by gluing them to dental floss or Teflon thread, and then looping this around the petiole of the insect and tightening it against the body (Fig. 4). After affixing the tag to a captive hornet, it was provided with commercially-produced jam or jelly, allowed to feed, and then followed with antenna-enhanced cell phones (for the Bluetooth tags) or commercially-produced receivers and Yagi antennas (for VHF tags) once the hornet commenced flying (Fig. 4).

Figure 4. 

A antenna for Bluetooth tracking tag after Iyer et al. (2019) B V. mandarinia with affixed Bluetooth tag C V. mandarinia with ATS T-15 tag affixed D VHF receiver and Yagi antenna (Advanced Telemetry Systems).

Four hornets were found and one photograph was submitted by the public in 2019. In 2020, 15 hornets were captured in WSDA or community science traps, one was collected in a net by a WSDA entomologist, one was collected in a net by a community member, and 17 confirmed hornet reports were received through photographs or dead specimens. In 2021, four hornets were collected in traps, three were captured by WSDA entomologists, one was captured by a community member, and three were reported or found dead by community members (Table 1, Fig. 5).

Figure 5. 

Map of confirmed Vespa mandarinia specimens and nests recorded from British Columbia and Washington State in 2020–2021. Details on the nest collected in British Columbia (not shown) in 2019 can be found in Bérubé, 2020. A single male specimen found in 2021 further south in Washington State was unrelated to these sightings and is also not included here. Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.

Only a small amount of these detections led to opportunities to find nests. A homeowner report in late September 2020 was followed by a site visit by a WSDA entomologist, who was able to capture a foraging hornet with a net. The following day a Bluetooth tag was glued to the mesosoma of the hornet, but it failed to fly. It was initially supposed that the wings were glued together while affixing the tag, but later examination showed that the wings were intact. We suspect instead that the hornet may have been chilled for too long. The homeowner captured another hornet the following week, which was successfully affixed with a Bluetooth tag by gluing it to floss and tying it around the hornet’s petiole. The hornet successfully recovered and flew, but left the range of the antennas within an hour and was not detected again. Four hornets were collected alive in nearby traps the following week, two of which were affixed with VHF tags (NanoPin, Lotek Inc.). One of these was subsequently tracked for approximately 240 m from the release site to the first V. mandarinia nest located in the United States, and the second in North America. The nest was found in a cavity within a living red alder (Alnus rubra Bong.), approximately 2.4 m above the ground (Fig. 6a). A few more hornets were subsequently found in the vicinity of that nest after the nest was removed, but no more hornet nests were detected in 2020.

Figure 6. 

A nest 1 (23 October 2020) B nest 2 (25 August 2021) C nest 3 (11 September 2021) D nest 4 (23 September 2021). Arrows indicate location of nest entrance.

In 2021, several reports of hornets were received from a site located near the US/Canada border. Upon intensive trap placement and survey of the area, WSDA entomologists captured two hornets, which were subsequently tagged with VHF tags (T15, Advanced Telemetry Systems). One of these was successfully followed for about 625 m to the second nest detected in the US. Nest 2 was found within a completely dead alder tree, with the entrance at roughly ground level and combs extending both into the tree and below ground (Fig. 6b). Soon after the detection and removal of the second nest, a homeowner captured a hornet using a makeshift trap, which was subsequently tagged and tracked 490 m from the release point to the third nest in the US, and fourth in North America. A public report also led to the capture of a fifth living hornet in 2021, which was tagged and tracked for 650 m to the fourth nest. Nests 3 and 4 were, like the first nest in 2020, located within red alder trees well above ground (nest 3 – ~2.2 m, and nest 4 – ~ 5.5 m; Fig. 6c, d).

All nest measurements and life stages are presented in Tables 2–6. The nests varied in the number of combs and total cells, ranging from 4 to 10 combs (Figs 7–8; Tables 2–4), and 418 to 1,329 cells (Tables 2–6). Nest 1, found late in 2020, contained 76 queens, and another 25 emerged while the nest was being studied. It was not possible to identify the foundress, which may have already died. Nine males were also recovered from nest 1. Only a single queen was found in each of the other nests. In each of these, the queen was not found until the nest was extracted, and in each case was the last adult hornet to be collected in the nest. The only other nest that contained adult males was nest 3 (the smallest), of which approximately 37% of adult hornets were males (Table 4). All capped cells from nest 4 were opened, and found to contain 23 male pupae, 112 female pupae, and 126 pre-pupae.

Figure 7. 

Top – combs from nest 1, 23 Oct 2020; Bottom – combs from nest 2, 25 August 2021. Numbers next to the combs indicate the position in the nest (smaller numbers are older and higher in the nest). Comb characteristics are recorded in Tables 2–5.

Figure 8. 

Top – combs from nest 3, 11 September 2021; Bottom – combs from nest 4, 23 September 2021. Numbers next to the combs indicate the position in the nest (smaller numbers are older and higher in the nest). Comb characteristics are recorded in Tables 2–5.

A subsample of 366 hornets comprising 15 males, 249 workers, and 102 queens was measured from the four nests. Worker mass ranged from 0.36 g to 1.41 g, and males from 0.82 g to 1.3 g. Queens ranged between 1.84 g and 2.88 g (Fig. 9). The three foundress queens massed 2.7 g (nest 2), and 2.16 g (nests 3 and 4). Length of males ranged from 27.8 to 35.19 mm, workers from 21.83 to 37.08 mm, and queens from 36.93 to 44.15 mm. Mesosomal width of males was between 7.3 to 8.99 mm, workers from 6.2 to 9.1 mm, and queens from 8.94 to 10.96 mm. While queen mass was always appreciably larger than workers and males, there was slight overlap at the extremes of length and mesosomal width for a few individuals.

Figure 9. 

Body measurements of V. mandarinia collected from four nests in North America. The three foundress queens are represented with solid circles.

Other insects located in the nest included species of Staphylinidae, Elateridae, and Cantharidae commonly associated with decaying tree environments. Two species of flies were common in two nests, and have been recorded in other Vespidae nests (Table 7). Other invertebrates (e.g., Acari, Collembola, Annelida) were also collected in the Berlese funnels but were not enumerated.

The nests located in the Pacific Northwest were generally somewhat smaller than those described by Matsuura and Sakagami (1973). The nest removed on Vancouver Island in mid-September 2019 contained approximately 400 cells and 200 adult hornets and was located in a subterranean burrow (Bérubé 2020). The largest nest reported here, collected in late August 2021, contained 1,329 cells, which is close to the size of the smallest nest (collected in December) reported by Matsuura and Sakagami (1973). It is possible that the August nest would have continued to grow and may have approached the more typical size reported for nests in Japan. However, nests described by Matsuura and Sakagami (1973) were collected from south-central Japan, and more southerly nests may be able to expand more quickly to large sizes. Although details about nests from northern parts of the hornet’s range are sparse, two nests collected on Hokkaido (Yamane and Makino 1977) each comprised five combs, with 675 cells (col. Aug 26, 1973) and 1141 cells (col. Sep 15, 1976). The nests removed in Washington State and British Columbia seem to be in accord with these northern records, so it is also possible that the nests we removed were typically sized for this latitude and climate.

One factor that may have impacted nest size and shape, particularly for nests 1, 3, and 4, was the constraining geometry of the tree cavity they were located within. The shape of the combs mirrored the internal shape of the cavities, and it is possible that workers could not use the space as efficiently as a nest in excavated soil. Indeed, the nest with the greatest number of cells reported here was nest 2, collected in August and the only one of the Washington nests not wholly confined to a tree cavity. Nest 3 seemed exceptionally small and contained a high proportion of males early in the season. The high number of males so early in the season is suggestive of inbreeding effects causing the production of diploid males (Van Wilgenburg et al. 2006; Darrouzet et al. 2015), which may have contributed to the very small nest size. Body measurements of all castes indicate that the largest workers and smallest queens could be confused if mesosomal width or body length is the only characteristic available for distinguishing between them, a phenomenon observed in other Vespidae (O’Donnell 1998).

It is interesting that all of the nests we located were in tree cavities in alder trees, with three of them high above the ground in still-living trees. Even though this is a small sample size, it is unexpected based on the most comprehensive reports of other nest sites (Matsuura and Sakagami 1973), where less than 16% of nests were reported from any sort of tree cavity. However, other data suggests that nests in cavities are somewhat more frequent (Choi et al. unpublished), and two of the three V. mandarinia nests reported by Yamane and Makino (1977) from Hokkaido were in tree hollows, although no further description is provided. Even so, the proportion of tree cavity nests in our results seems unusual. This could be a result of heavily saturated ground in the study area at the time queens are establishing nests, with more than 650 mm of total precipitation typical during the winter and spring months. Alternatively, nests from 2021 could have been established by queens that successfully dispersed from the 2020 nest after imprinting on the cues of their natal home, biasing nest selection towards alders.

No hornets were detected in British Columbia or Washington State in 2022. It is too early to feel confident that the species has been prevented from establishing, and several years of survey remain to be conducted. Some of the findings described in this paper suggest that small population effects may be impeding establishment, i.e., the unseasonably high number of males observed in nest 3. However, the characteristics of the other three nests, and our observations of foraging behavior and analysis of the local prey base (unpublished) concur with climate modeling (Alaniz et al. 2020; Zhu et al. 2020; Nuñez-Penichet et al. 2021) in suggesting that the region provides viable habitat for this species. Those results, and the recent spread of other hornet species outside of their range in multiple countries, are evidence that Vespidae require further study to either prevent, or mitigate the effects of, future introductions.

This project was the culmination of collaborative work by multiple agencies and numerous members of the public. We are grateful to the Day, Bovenkamp, Shelton, Beard, DeJong, Polinder, Morin and Tompkins, Kreider, Revak, Jordan, Tjoelker, Rodenberg, and Barrett families for their hospitality and property access. We are indebted to Philip Bovencamp and Dean Tjoelker for their hornet-hunting prowess and observations of hornet behavior. Ted McFall, Ruthie Danielsen, and the Mt. Baker Beekeepers Association have been tireless partners in survey, field experiments, and public outreach. We are deeply indebted to the thousands of residents of Washington, British Columbia, and beyond, who have reported observations, facilitated property access, and staffed their own traps throughout this project. Dr. Peter Kennedy, University of Exeter, Cornwall, provided invaluable advice about radio-tagging Vespa species. Dr. Miriam Cooperband shared insights into radio-tagging insects and provided equipment and training leading to our first successful tracking. Shawn Cleveland provided timely instruction on how to best use radio-telemetry equipment to track wily, fast-moving animals. Zach Techner of Cascadia Venom Collection graciously showed us how to vacuum swarming, stinging wasps. We are grateful for the assistance of Jake Bodart, Jessica Rendon, and Austin Johnson (Oregon Department of Agriculture) in tracking nest 2 in 2021. Camilo Acosta, Andrew Bigelow, Ryan Gelwicks, Brian Henderson, Stacy Herron, Soraya Jessa, Diane MacLean, Kristen Mason, Hadley Ocheltree, Romie Pugh, Olivia Schmit, Katie Simon, Luke Turner, and Ciara Varnum-Lowry were tireless field trappers throughout this program. Quinlyn Baine, Chanda Bartholomew, Warren Hellman, Brandy Kamakawiwo’ole, Wade Petersen, and Angela Yoder sorted the contents of more than 18,000 hornet traps in the first year of this project. Dr. Dina Roberts helped rear hornet queens and suggested approaches for measuring hornet nests. We are grateful to Rod Rood and Dr. Brian Brown for providing identifications of beetles and flies (respectively) collected from the hornet nests. We are very thankful for the guidance and support of Anne LeBrun, Dr. Todd Gilligan, Josie Ryan, and Dr. Tim St. Germain (USDA-APHIS) throughout this project. This project was funded via the USDA-APHIS Plant Protection Act (PPA) Section 7721 and by the State of Washington. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture or the Washington State Department of Agriculture. USDA is an equal opportunity provider and employer.