On July 15, 2019, we first observed numerous parasitoids walking on and flying in the vicinity of raspberry (Rubus idaeus ‘Rudi’) fruits in an experimental plot heavily infested by D. suzukii in Agassiz, British Columbia, Canada (GPS: 49°14'34.8"N, 121°45'18.0"W). Remarkably, we collected > 100 parasitoid specimens by mouth aspirator in less than 45 min of effort by two individuals (PKA, personal observations). We also collected 20 raspberry fruits presumed to be infested by D. suzukii, which were divided between two ventilated plastic containers (0.25 L) with filter paper to observe whether any parasitoids successfully emerged. A subset of field-collected parasitoid individuals were exposed to 2^nd and 3^rd-instar D. suzukii larvae (approx. 50 per vial) in two Drosophila rearing vials containing artificial diet (Formula 4–24 medium, Carolina Biological Supply Co. Burlington, NC) to informally test whether or not they were capable of successfully parasitizing D. suzukii larvae. Parasitoids emerged both from field-collected D. suzukii-infested raspberries and lab-reared D. suzukii larvae in artificial diet exposed to field-collected parasitoid adults (PKA, personal observations). This was surprising, given the known low developmental success of North American larval parasitoids attacking D. suzukii (Lee et al. 2019). This observation motivated an attempt to obtain a preliminary species-level identification. A 683 bp region of the cytochrome oxidase I (COI) gene of one aspirator-collected parasitoid specimen was barcoded and compared against existing records in GenBank using a nucleotide BLAST (NCBI, 2019). This sequence showed 98.67 % identity with GenBank Accession AB546875.1, which is a sequence of L. japonica from Japan (Novkovic et al. 2016). Specimens from the same date and location that were collected by mouth aspirator (n = 24), emerged from raspberry fruit (n = 3), and that emerged from lab-reared D. suzukii larvae (n = 6), were all identified as L. japonica (by MLB) based on morphology (see Identification of parasitoid specimens below for details on morphological determinations). This initial find prompted a re-examination of parasitoid specimens (n = 14) emerging from 60 D. suzukii-infested strawberry (Fragaria × ananassa) fruits (total n = 375 emerging D. suzukii) collected near Langley, British Columbia in two commercial fields (GPS: 49°07'34.7"N, 122°32'45.2"W and 49°08'07.8"N, 122°33'42.1"W) between September 20 and 27, 2016. Based on morphology, the parasitoid specimens were also confirmed to be L. japonica, suggesting that this species has been present in British Columbia since at least 2016.

Motivated by the unexpected detection of L. japonica in British Columbia, we next conducted additional collections in 2019 to determine how widespread in coastal British Columbia L. japonica might be, and whether any other native or exotic species of parasitoids might also be present. From August to October 2019, parasitoid specimens were obtained from 54 opportunistic collection samples from 23 different sites in coastal British Columbia (the lower mainland and Vancouver Island) (Fig. 1). Parasitoid specimens (n = 525) were collected from plants and areas known to be infested by D. suzukii with one of three sampling methods: (1) mouth aspiration of parasitoids seen flying or resting on the foliage or fruits of Himalayan blackberry, Rubus armeniacus (n = 34 samples) or cultivated blackberry, Rubus fruticosis ‘Triple Crown’ (n = 2 samples); (2) collection of R. armeniacus fruits (approx. 20–60 berries per sample) into ventilated plastic rearing containers (volume: 0.5 or 1.0 L) lined with filter paper to rear out parasitoids and host Drosophilidae in the laboratory (16:8 hour light:dark, 24 ± 1 °C) (n = 14); (3) separation of parasitoid by-catch from plastic traps baited with apple cider vinegar placed in hedgerows on a university experimental farm to collect D. suzukii (n = 2); and (4) collection of R. armeniacus berries followed by exposure outside in a plastic trap for 6 days, and then rearing out insects in glass jars in the laboratory (n = 2). Collections from R. armeniacus were in suburban or natural areas along roadways or in parks. For a subset of 14 R. armeniacus fruit collections from which at least one parasitoid emerged, emerging Drosophilidae (n = 1,637) were also counted and identified. All identifiable emerged Drosophilidae were either D. suzukii (85.8 %) or D. melanogaster (0.6 %). Remaining flies (13.6 %) could not be identified conclusively due to deterioration or mold in containers. Full information on collection sites, dates, and methods, and insect emergence if applicable, can be found in the Suppl. material 1. Due to the opportunistic nature of our sampling in space and time, and unknown levels of parasitoid and host mortality in rearing, we did not calculate or present percent parasitism levels. All collected parasitoids were preserved in 95 % EtOH for subsequent morphological and molecular identification.

Specimens from both initial and more extensive collections in British Columbia were sent to MLB after initial COI barcoding (by SJP and CC) yielded results consistent with barcode sequences on GenBank under the name L. japonica (see above). Upon further examination, it was determined that not only was L. japonica present in the samples but so too were G. brasiliensis, an undetermined species of Asobara, and P. vindemmiae (Figs 2–7). Specimens of L. japonica and G. brasiliensis were vacuum dried and card mounted; the Asobara and Pachycrepoideus specimens were prepared using HMDS (Heraty and Hawks 1998) and point mounted. Specimens are deposited in the insect collection of the National Museum of Natural History (USNM; Smithsonian Institution, Washington DC) and the Canadian National Collection of Insects, Arachnids, and Nematodes (Ottawa, Ontario, Canada).

The morphology-based research in Novković et al. (2011), Buffington and Forshage (2016) and Lue et al. (2016), as well as specimens in the USNM, were used to confirm the identities of the Figitidae. While this research was being conducted, it became clear that specimens of Leptopilina submitted to MLB for identification in 2016 were misidentified as Leptopilina heterotoma (Thompson). At that time, the USNM lacked specimens of L. japonica, making a direct comparison impossible. Recently, the USNM has received voucher specimens from the original paper describing L. japonica, as well as additional material from China and Korea. This made direct comparisons possible, and the identity of the 2016 specimens was corrected.

In terms of Figitidae attacking D. suzukii, L. japonica and G. brasiliensis can be difficult to distinguish (Figs 2, 3). Careful examination of the posterior aspect of the metapleuron reveals a complete patch of setae in G. brasiliensis (Fig. 4a), while the same location in L. japonica is glabrous (Fig. 5a). Additionally, the metasoma setal hair ring is complete in G. brasiliensis (Fig. 4b) and dorsally incomplete in L. japonica (Fig. 5b). Lastly, the metapleuron, in lateral view, is simple in G. brasiliensis (Fig. 4c); in L. japonica, a longitudinal carina transverses the middle of the metapleuron (Fig. 5c). Specimens of Asobara, Pachycrepoideus, and Figitidae are relatively easy to differentiate from one another given their overall dissimilar shape and color (Figs 6, 7).

The single Pteromalidae specimen was identified and confirmed using Bouček and Heydon (1997) and by comparison with authoritatively identified specimens in the USNM. The specimens of Asobara were sight identified to genus; Wharton (1980), Vet et al. (1984), and Nomano et al. (2015) were consulted in an effort to identify the specimens to species.

A subset of the Figitidae and Braconidae specimens were sequenced for the COI ‘barcode’ region; as only one Pachycrepoideus was collected, no barcode was generated for this species. DNA was extracted using the QIAamp DNA Micro Kit (Qiagen). PCR amplifications of COI were carried out using LCO1490 (Folmer et al. 1994) and C1-N-2191 (Simon et al. 1994) with a Bio-Rad T100 thermal cycler under the following cycling conditions: initial denaturation for 7 min at 95 °C, 35 cycles of 30 s at 95 °C, 30 s at 50 °C, 45 s at 72 °C, and a final extension for 1.5 min at 72 °C. PCR products were cleaned with ExoSAP-IT (Affymetrix), sequenced with BigDye v.3.1 (Life Technologies) and run on an ABI 3730xl automated DNA sequencer (Applied Biosystems). Contigs were assembled and edited in Geneious Prime 2020.04 (Biomatters). Sequences used in the study have been deposited in GenBank (Accession nos. MT559416–MT559426).

Leptopilina japonica was the most common parasitoid species found in our collections, representing 95.8 % (503/525) of all parasitoids we collected and identified in British Columbia in 2019. This species was present across the sampled area (Fig. 1) and was collected with all three sampling methods (traps, aspirator collections, rearing from infested fruit). Ganaspis brasiliensis was the next most common species (2.7 % of all specimens) and was present at three sites, which were all in the eastern part of the sampled geographic range (Fig. 1). This species was collected both with aspirators while foraging on berries and emerging from D. suzukii-infested berries. Somewhat surprisingly, no native Leptopilina spp. were found in our samples in 2019.

Results from DNA barcoding of a small subset of Figitidae specimens was consistent with all morphologically-based identifications. COI sequences from L. japonica showed high similarity to specimens previously collected in China and Japan (> 98 % identity). All four of the G. brasiliensis specimens from British Columbia that we barcoded clustered within the “G1” clade (100 % identity), which has previously been found in South Korea, China, and Japan (Nomano et al. 2017; Giorgini et al. 2019) and appears to be the most host-specific lineage of G. brasiliensis (Nomano et al. 2017; Girod et al. 2018b). Further studies are needed to determine the levels of genetic diversity of L. japonica and G. brasiliensis in North America and potential source populations of these parasitoids in Asia.

We emphasize that because our sampling took place within a restricted time window and set of habitats, our data are likely not representative of the two adventive figitid parasitoid species’ relative abundances or impact. They simply demonstrate that these two species are present in the south coastal region of British Columbia. More widespread, season-long surveys in several habitats are needed to determine the relative abundance and distribution of each species. Similarly, repeated fruit collections containing known life stage distributions of hosts, coupled with population modelling approaches such as stage-structured matrix models (e.g., Wiman et al. 2014, 2016), will be necessary to estimate the population-level impact of these adventive exotic parasitoids attacking D. suzukii – and possibly other, non-target Drosophilidae – in North America.

One species of Asobara (1.5 % of all parasitoids in collections) was collected with aspirators from berries (n = 3) and reared from D. suzukii-infested blackberries (n = 5) from two sites. Although nearly all of the Drosophilidae emerging from these collections were D. suzukii, we cannot exclude the possibility that the five Asobara that emerged from berries could have been associated with D. melanogaster, which was present at very low levels in our samples (see above). Exposure of three female Asobara to lab-reared D. suzukii larvae in artificial diet (see methods in “Preliminary parasitoid collections” above) only yielded a single male offspring (PKA, personal observations). While the field collections and single lab observation provides a proof-of-concept that this Asobara species can attack and develop in D. suzukii, at this time the association of this Asobara species with D. suzukii should be considered tentative. As far as is known, species of Asobara are solitary koinobiont endoparasitoids of cyclorrhaphous flies, mostly Drosophilidae (Wharton 1997, Yu et al. 2016).

Two species of Asobara, A. tabida (Nees) and A. fungicola (Ashmead), are known from the Nearctic Region, and both species have been reported from British Columbia (Yu et al. 2016). One of us (RRK) initially identified the Asobara we found associated with D. suzukii in British Columbia as A. tabida using Wharton (1980), but A. tabida and A. fungicola are the only species included in the key. Several species of Asobara, including A. tabida and A. fungicola, occur in multiple zoogeographic regions (Yu et al. 2016). Therefore, it is possible that other Asobara species morphologically similar to A. tabida and A. fungicola occur in the Nearctic Region but are commonly misidentified as the aforementioned species. For example, Asobara rufescens (Förster) and A. tabida are extremely similar morphologically; Vet et al. (1984) considered them sibling species and identified subtle differences in color, shape of antennal flagellomeres 1 and 2, tarsal claw length, and the basitarsus. Asobara rufescens (Förster, 1863) is currently known only from the Palearctic Region but has been considered a junior synonym of A. tabida (Nees, 1834), which occurs in the Nearctic, Palearctic, Oriental, and Oceanic regions. Thus, if A. rufescens occurs in the Nearctic Region it could be easily misidentified as A. tabida, especially when using identification tools that were based on region or published during the period of time when A. rufescens was a junior synonym of A. tabida. Recent efforts to accurately identify species of Asobara have relied on COI DNA barcoding along with authoritative identifications based on morphological features (Nomano et al. 2015, Guerrieri et al. 2016).

We obtained a 658-bp fragment of the COI DNA barcoding region from one of the specimens RRK tentatively identified as A. tabida. A GenBank nucleotide BLAST search (NCBI 2019) using the 658-bp fragment, as well as a search of all COI barcode records using the Identification System in the Barcode of Life Database (BOLD; Ratnasingham and Hebert 2007), yielded 53 COI sequences we consider conspecific with our Asobara specimen. Of the 41 sequences from GenBank, 39 had 98.79–100 % identity (79–100 % query coverage) with the AsobaraCOI sequence obtained in this research; percent identities for the two most dissimilar sequences that we still consider conspecific were 97.24 % and 97.26 % (100 % and 93 % query coverage, respectively). The 41 sequences from GenBank were mostly from specimens identified as A. rufescens (n = 36) but also included some identified only as Braconidae or Hymenoptera (n = 5). The sequences in BOLD consisted of the 41 recovered via GenBank and an additional 12 sequences we consider conspecific with our Asobara species. Of the 53 sequences recovered, 51 have sequence similarity ranging from 99.18–100 %, and two have sequence similarity ranging from 97.19–97.22 %; all are in BOLD under the name A. rufescens. One sequence in both GenBank and BOLD under the name A. rufescens (GenBank: KT604569.1) appeared to have quality control issues and was not considered in this research. The most similar sequence in GenBank and BOLD that we do not consider conspecific was from a specimen identified as A. tabida, with 93.46 % identity (99 % query coverage) in GenBank and 93.36 % similarity in BOLD relative to our Asobara sequence.

Sequences in BOLD/GenBank identified as A. rufescens were likely determined as such based solely on their high similarity to sequences already in those databases from specimens identified as A. rufescens using morphological features. Notably, COI sequences from putative A. rufescens specimens from Japan (Nomano et al. 2015; GenBank: AB920758.1, AB920759.1, AB920763.1, AB920762.1), identified by M. Kimura using the morphological features from Vet et al. (1984), are likely the basis for all subsequent COI sequence identifications in BOLD/GenBank. If the specimens from Japan are accurately identified as A. rufescens, a specimen of A. rufescens collected in Puslinch, Ontario, Canada on August 14, 2008 (Hebert et al. 2016; GenBank: KR884537.1) would be the first record of A. rufescens in the Western Hemisphere. Further, collection data for all sequences in BOLD/GenBank putatively conspecific with M. Kimura’s A. rufescens reflect that this species was collected in Canada every year from 2008–2019 (except 2011) with occurrences ranging from Forillon National Park, Quebec (Hebert et al. 2016; GenBank: KR420364.1) to Elk Island National Park, Alberta (Hebert et al. 2016; GenBank: KR793444.1). Thus, A. rufescens has possibly occurred in North America since at least 2008, the year D. suzukii was first detected on the continent (Asplen et al. 2015), and ranges from east to west across nearly all of Canada. Thistlewood et al. (2013) reported specimens of Asobara reared from fermenting banana that was artificially infested with D. suzukii and then placed in the field in British Columbia, which they considered likely to be A. tabida. The Asobara specimens reported in Thistlewood et al. (2013) might be conspecific with our Asobara species from British Columbia. If so, it is possible that our Asobara species (possibly A. rufescens) has been associated with D. suzukii since at least 2011; however, this is uncertain regardless of Asobara species identity because the banana baits were infested with Drosophilidae other than D. suzukii after being placed in the field.

Kimura’s identification of A. rufescens from Japan, based on the morphological features identified in Vet et al. (1984) for differentiating A. rufescens and A. tabida (Nomano et al. 2015, M. Kimura personal communication), should be considered tentative. Features used by Vet et al. (1984) lack enough detail to reliably differentiate species of Asobara that are, perhaps, morphologically cryptic (R. Kula personal observation). The features used were differences, “in the shape of the 3^rd and 4^th antennal segment… and in the relative length of the claws,” as well as unspecified differences “in the colour” and “in the basitarsus of the hind legs” (Vet et al. 1984). These features might reflect differences between A. rufescens and A. tabida, but accurate interpretation is difficult because the color differences were not illustrated or specified, the basitarsus difference was not elaborated or quantified, and the shape and length differences in the flagellomeres and tarsal claws, respectively, were not quantified. Line drawings were used to illustrate the relevant portions of the antenna and tarsus for specimens Vet et al. (1984) considered A. rufescens and A. tabida, as well as an A. rufescens × A. tabida hybrid; thus, identification requires interpreting subtle differences in the line drawings by eye and without consideration for potential intraspecific variation. Thus, the features identified in Vet et al. (1984) require additional assessment to determine their utility.

Beyond morphological features, Vet et al. (1984) discerned via olfactometer tests that what they initially considered A. tabida actually consisted of two species attracted to different host habitats. One of the species was attracted to the odor of fermenting fruit, while the other species was attracted to the odor of decaying leaves, thus resulting in a premating reproductive barrier. The authors discerned, through comparison of their Asobara specimens with the holotype of A. rufescens for aforementioned morphological features, that the specimens attracted to decaying leaves were A. rufescens. Thus, the authors regarded A. tabida as a parasitoid of frugivorous Drosophilidae and A. rufescens as a parasitoid of saprophagous Drosophilidae. Mitsui et al. (2007) reported A. tabida and A. rufescens from Drosophilidae, including the former from D. suzukii, infesting clumps of banana placed at field sites; they also reported A. rufescens from Scaptomyza pallida (Zetterstedt) in decaying leaves. However, Chabert et al. (2012) found that A. tabida rarely oviposits into D. suzukii. Other research has shown, based on analysis of COI DNA barcoding data, that other Asobara specimens perceived as conspecific are apparently complexes of morphologically similar species (Nomano et al. 2015, Guerrieri et al. 2016) with different patterns of host use. Thus, additional research is needed to discern the limits of species in Asobara, as well as patterns of host use and factors that affect those patterns.

The COI sequence from our Asobara species from British Columbia was also very similar to COI sequences from putative A. rufescens collected in Germany (BOLD: 99.28–99.53 % similarity; Matthias et al. 2016) and Japan (GenBank: 99.20–99.24 % identity, 94–100 % query coverage; BOLD: 99.18–99.23 % similarity; Nomano et al. 2015). Asobara rufescens has been collected in the Palearctic Region from Japan to the Netherlands (Yu et al. 2016); thus, if the identifications in GenBank and BOLD are accurate, A. rufescens is distributed broadly across the entire Holarctic Region. Conversely, while A. tabida has been reported across the entire Palearctic Region, from Japan and Russia (i.e., Sakhalin Oblast) west to Ireland, it has been reported infrequently in North America (Wharton 1980, Kraaijeveld and van der Wel 1994, Hoang 2002). Interestingly, there are no COI DNA barcode sequences in either GenBank or BOLD from specimens of A. tabida in North America. This raises the possibility that specimens in North America identified previously as A. tabida are actually A. rufescens. The results reported herein demonstrate the need for discerning species complexes within Asobara overall and its implications for interpreting patterns of host use. It would also be useful to sequence nuclear genes in Asobara, as some of the confusion in resolving A. tabida and A. rufescens may be due to hybridization and mitochondrial introgression.

Only a single specimen of P. vindemmiae was found in our collections. Pachycrepoideus vindemmiae is an idiobiont ectoparasitoid recorded as a primary parasitoid of Diptera, Lepidoptera, Hemiptera, and Hymenoptera. Diptera account for most records with 13 families, 33 genera, and 55 species recorded (Noyes 2019). As a facultative hyperparasitoid, hosts include Diptera (Sarcophagidae, Tachinidae) and Hymenoptera (Braconidae, Diapriidae, Encyrtidae, Figitidae, Eulophidae, Pteromalidae). Various factors influence the degree of primary vs. secondary parasitism in Drosophila systems (Bezerra Da Silva et al. 2019, Chen et al. 2015, Goubault et al. 2003, Rossi Stacconi et al. 2015, Wang et al. 2016). Related to the work herein, P. vindemmiae is known as a parasitoid of D. melanogaster, A. tabida and L. heterotoma (Noyes 2019, Philips 1993, Van Alphen and Thunnissen 1982), and also parasitizes conspecifics (Chen et al. 2015). Future studies could investigate whether P. vindemmiae could act as a facultative hyperparasitoid of L. japonica and G. brasiliensis in North America.