Key message

  • Drosophila suzukii is an invasive pest without effective natural enemies in invaded regions.

  • Surveys for native parasitoids were conducted from 2013 to 2016 in China.

  • 458 (15 species) and 1354 (6 species) parasitoids were reared from drosophilids in fruit-baited traps, or D. suzukii and Drosophila pulchrella developing in field-collected fruits, respectively.

  • Ganaspis brasiliensis was the dominant species recovered only from field-collected fruits.

  • The narrow host range showed by G. brasiliensis both in the field and in the laboratory makes this species a potential agent for classical biological control.

Introduction

The spotted wing drosophila, Drosophila suzukii (Matsumura) (Diptera: Drosophilidae), is native to the Eastern and Southeastern regions of Asia but has become a serious invasive pest causing considerable economic damage to fruit crops worldwide (Asplen et al. 2015). Current management of D. suzukii in Europe or Americas relies primarily on insecticide applications (Beers et al. 2011; Van Timmeren and Isaacs 2013). However, this approach is often ineffective in preventing all economic damage, in part because there is often a diversity of wild flora neighboring cultivated fields that can serve as refugia for the pest to recolonizing treated crops (Kenis et al. 2016; Klick et al. 2016; Lee et al. 2015). Therefore, there is a critical need to develop additional control tools for this extremely polyphagous and mobile pest, and to include area-wide management strategies that can suppress populations in the entire landscape (Haye et al. 2016; Wang et al. 2016b). From this perspective, classical biological control through the introduction of Asian natural enemies could play a key role in reducing D. suzukii populations in non-cultivated habitats.

A complex of larval and pupal parasitoids plays an important role in the natural suppression of many drosophilid species (Carton et al. 1986), with levels of parasitism often exceeding 50% (Fleury et al. 2009). Unfortunately, for D. suzukii in North America and Europe, most resident larval parasitoids are unable to complete development because of D. suzukii’s stronger cellular immune response than native Drosophilidae species (Chabert et al. 2012; Kacsoh and Schlenke 2012; Poyet et al. 2013). Only a few generalist indigenous parasitoid species such as the pupal parasitoids Trichopria drosophilae (Perkins) (Hymenoptera: Diapriidae) and Pachycrepoideus vindemiae (Rondani) (Hymenoptera: Pteromalidae) have been reported to readily attack D. suzukii in both Europe (Gabarra et al. 2015; Rossi Stacconi et al. 2013, Mazzetto et al. 2016, Knoll et al. 2017) and North America (Miller et al. 2015; Wang et al. 2016a, b). A strain of the larval parasitoid Leptopilina heterotoma (Thomson) (Hymenoptera: Figitidae) in northern Italy has been reported to attack D. suzukii in the laboratory but does not provide adequate control levels (Rossi Stacconi et al. 2015, 2017), while most tested L. heterotoma populations failed to develop from D. suzukii (Chabert et al. 2012; Miller et al. 2015; Mazzetto et al. 2016; Knoll et al. 2017). Overall, extant parasitoids have not provided sufficient control of D. suzukii in invaded regions (Gabarra et al. 2015; Miller et al. 2015; Rossi Stacconi et al. 2015; Wiman et al. 2016).

The lack of effective natural enemies of D. suzukii in Europe and North America led to recent field surveys in Asia for native, co-evolved and specialized parasitoids to be considered for use in classical biological control programs (Daane et al. 2016; Guerrieri et al. 2016; Girod et al. 2018a; Giorgini et al., reported herein). Previously, D. suzukii–parasitoid associations were described in Japan, with 10 parasitoid species reported (Ideo et al. 2008; Mitsui and Kimura 2010; Mitsui et al. 2007) among which they identified a ‘strain’ of Ganaspis xanthopoda (Hymenoptera: Figitidae) being the most active parasitoid and considered a specialist of D. suzukii (Kasuya et al. 2013). Additionally, as a point of clarification, we note that Ganaspis individuals recorded from D. suzukii and previously reported as G. xanthopoda by Kasuya et al. (2013), Mitsui et al. (2007) and Mitsui and Kimura (2010) have been assigned the name Ganaspis brasiliensis (Ihering) by Buffington and Forshage (2016) in their treatment of Ganaspis associated with D. suzukii. This updated name was utilized by Nomano et al. (2017) who grouped G. brasiliensis into five genetic lineages, with just two recorded from D. suzukii. The populations of Ganaspis found to parasitize D. suzukii in China and Japan by Girod et al. (2018a, b, c) are also morphologically attributable to G. brasiliensis.

We previously reported on the presence and identity of parasitoid species in the genus Asobara (Hymenoptera: Braconidae), collected from Yunnan Province, China, that were associated with drosophilids; seven parasitoid species were identified including five new species (Guerrieri et al. 2016). In South Korean collections, six larval parasitoid species (Asobara brevicauda van Achterberg and Guerrieri, A. japonica, A. leveri (Nixon), L. j. japonica, L. japonica formosana Novković and Kimura, and G. brasiliensis) and one pupal parasitoid (T. drosophilae) were recorded from D. suzukii (Daane et al. 2016). Interestingly, A. brevicauda, L. j. japonica, and G. brasiliensis emerged only from D. suzukii from field-collected fruits but not from traps baited with uninfested fruit (which yielded predominately other Drosophila species) (Daane et al. 2016). To complement these recent explorations and to look for parasitoid species with potentially higher efficacy and specificity to D. suzukii, we conducted surveys in the Yunnan Province, China, from 2013 to 2016. Here, we report on (1) parasitoid species collected using banana-baited traps and through direct sampling of D. suzukii host plant fruits; (2) genetic analyses of G. brasiliensis collected; and (3) quarantine test of host specificity of G. brasiliensis with 10 different drosophila species. Our findings in Yunnan expand the current knowledge on the diversity, geographic distribution and host range of parasitoid species attacking drosophilids, and specifically D. suzukii in Asia.

Materials and methods

Parasitoid collections

Surveys for D. suzukii parasitoids were conducted in summers from 2013 to 2016 in different locations of Yunnan Province, China (Table 1), using either banana-baited traps placed near natural vegetation or cultivated fields, or collections of fruits from natural vegetation including known or presumed host plants of D. suzukii. This region in southern China is part of the presumptive native range of D. suzukii and the closely related species Drosophila pulchrella Tan, Hsu & Sheng (Takamori et al. 2006; Zhao et al. 2017), both characterized by a serrated ovipositor that allows them to penetrate the intact skin of fruits.

Table 1 Sampling locations for drosophilid parasitoids and collection methods during 2013–2016 in Yunnan Province, China
Full size table

Banana-baited traps were used from 2013 to 2015. The traps were made of plastic food boxes (10 × 15 × 30 cm) with 0.5-cm holes along the side for ventilation and provisioned with sliced sections of banana for fruit fly egg deposition (developing into fresh larvae available for parasitization). At each of the seven sampled sites, 4–11 traps were placed in a linear transect at distances of ~ 100 m from each other. After 7 days, traps were collected and transferred to a laboratory (Yunnan Academy of Agricultural Science), where the ventilation holes were covered with organdy and the traps were held at 25 ± 3 °C and then observed daily for fly or parasitoid emergence. Emerged flies and parasitoids were collected and immediately killed in absolute ethanol and preserved at − 20 °C for subsequent species identification.

Field collection of fruits was adopted in 2015 (limited collections) and 2016 to get more precise information on the association among host plant-Drosophila-parasitoid species, and on the level of field parasitism. Fruits were collected in July 2016 from plants of Sambucus adnata Wallich ex de Candolle, Rubus foliosus Weihe, Rubus niveus Thunberg, and Fragaria moupinensis Cardot (Table 1) and taken to a laboratory (Yunnan Academy of Agricultural Science) where they were stored at 25 ± 3 °C in plastic boxes, as described previously, and sorted by host plant, collection date and site. The fruit were checked daily under a binocular microscope (60 ×) for a period of 7–10 days after field collection for Drosophila puparia. When discovered, the D. suzukii and D. pulchrella puparia were separated from those of other drosophilids by the characteristic shape of the two everted anterior spiracles (tubes with finger-like projections). It was not possible, however, to sort the puparia of D. suzukii from those of D. pulchrella, based on puparium morphology (Hauser 2011). For this reason, we refer to collected puparia of these two fly species as D. suzukii-like puparia. These latter represented almost all (~ 99%) puparia found in collected fruits. Groups of 10–20 collected puparia were isolated in glass vials (1 × 7.5 cm) that were sealed with a cotton plug and provisioned with moistened filter paper strip to prevent desiccation (but trying to avoid water condensation or deposition). Adult fly and parasitoid emergence was checked every 2–3 days for a period of 1 month (either at the laboratory of the Yunnan Academy of Agricultural Science or in quarantines at the University of California, Berkeley, USA, and at USDA Agricultural Research Service, Beneficial Insects Introduction Research Unit, Newark, Delaware, USA). Emerged flies and parasitoids were killed in absolute ethanol and stored at − 20 °C for subsequent species identification. Part of the emerged parasitoids were used to establish quarantine colonies at Berkeley laboratory. All puparia were later dissected under a binocular microscope (60 ×) and sorted as: puparia bearing either a fly or parasitoid emergence hole, unhatched puparia bearing a dead fly, unhatched puparia bearing a dead parasitoid (larva, pupa or adult), or unhatched puparia dead for unknown reasons. Percentage parasitism was calculated by dividing the sum of parasitized puparia (bearing a parasitoid emergence hole or containing a dead parasitoid) by the total number of puparia, excluding those categorized as dead for unknown reasons (see Daane et al. 2016). In 2015, berries of Osyris wightiana Wall ex Wight were also collected and emerged adult parasitoids were treated as described above (Table 1); however, fly puparia were not isolated and the parasitism level was not determined. Percentage parasitism between different host plant species was compared by G-test of independence (McDonald 2014).

Molecular analysis of Ganaspis brasiliensis

Molecular analyses focused exclusively on G. brasiliensis, which was the most abundant parasitoid species emerging from D. suzukii-like puparia. Because Nomano et al. (2017) separated G. brasiliensis into five lineages, we sequenced the cytochrome oxidase subunit I (COI) gene to clarify our Chinese specimens’ genetic identity and extend the knowledge on host range and geographic distribution of different G. brasiliensis lineages. Whole specimens were subjected to genomic DNA extraction using a nondestructive (without grinding the specimen) Chelex-proteinase K protocol (Gebiola et al. 2009). The COI gene of 30 individuals (Table 2) was amplified using the primer combination LCO and HCO (Folmer et al. 1994). Reactions were performed in 20 µl volumes containing 4 µl of 1X GoTaq buffer (Promega Corp., Madison, Wisconsin, USA), 1.6 µl dNTP (2.5 mM each), 1 µl of forward and reverse primer (10 µM each), 0.4 µl GoTaq G2 DNA Polymerase (Promega) (5u/µl), and 2 µl template DNA. Amplifications were achieved using a Bio-Rad thermocycler Mycycler (Bio-Rad, Hercules, California, USA) programmed at 1 min at 94 °C, followed by 40 cycles of 30 s at 94 °C, 90 s at 48 °C, and 60 s at 72 °C, and a final step of 7 min at 72 °C. PCR products were visualized after electrophoresis on 1% agarose gel stained with Gel Red™ (Biotium Inc, Fremont, California, USA) to confirm the amplification. Fragments obtained were sequenced in both sense and antisense directions by adopting EZ-seq standard service (Macrogen Inc., Seoul, South Korea). The chromatograms obtained were viewed and edited in Chromas v.2.6.4 (Technelysium, South Brisbane, Queensland, Australia). COI sequences of G. brasiliensis were deposited in GenBank under the accession numbers MG755073-MG755101 (Table 2), and parasitoid wasps were vouchered at the CNR, Institute for Sustainable Plant Protection, Unit of Portici, Italy. COI sequences were aligned using the Muscle alignment tool in Aliview 1.18.1 (Lasson 2014). Protein coding was checked by translating the sequences into amino acids. No evidence for the presence of pseudogenes (i.e., no stop codons or frame shifts) was detected.

Table 2 Ganaspis brasiliensis specimens reared from D. suzukii-like puparia and used for COI gene sequencing
Full size table

Phylogenetic analysis of COI sequences was performed by Bayesian inference (BI) using MrBayes 3.2.6 (Ronquist et al. 2012). The BI tree was obtained by implementing a different substitution model for each codon position, that is GTR + I + G, GTR + I and GTR + G for codon 1, 2 and 3, respectively. Substitution models were selected by PartitionFinder version 2.1.1 (Lanfear et al. 2012), based on the AICc criterion (AIC for small sample size). BI was conducted according to the best partitioning scheme selected by PartitionFinder using ‘all’ search algorithm with branch lengths linked. Two runs of four Monte Carlo Markov chains (3 ‘heated’ and 1‘cold’) were run in parallel in MrBayes for 5,000,000 generations, and trees were sampled every 1000 generations. Convergence of the separate runs was checked using the average deviation of split frequencies diagnostic (< 0.01) and the PSRF parameter (close to 1.00 for all parameters). The burn-in value was set at 25% of sampled topologies. Trees were imported into the tree editor TreeGraph 2.14.0-771 beta (Stover and Muller 2010) for annotation and layout. COI sequences of G. brasiliensis from Nomano et al. (2017) were added to the ingroup. Ganaspis xanthopoda was used as an outgroup.

Quarantine test of host specificity of G. brasiliensis

A host specificity test using Yunnan-collected G. brasiliensis was conducted under controlled conditions (22 ± 2 °C, 14L: 10D, 40–60% RH) in a quarantine at the University of California, Berkeley. The D. suzukii larval colony was maintained in Petri dishes (1.5 × 8 cm) filled with a cornmeal-based artificial diet and that had been inoculated by exposure to adult flies for a 24-h period in organdy-screened cages (see Wang et al. 2018). Colonies of native non-target species tested were initiated from specimens purchased from the University of California’s Drosophila Stock Center in San Diego, California, where these species were originally collected from different locations in the USA. Nine non-target species were tested: Drosophila simulans Sturtevant, D. melanogaster Meigen, D. persimilis Dobzhansky and Epling, D. pseudoobscura Frolova, D. busckii Coquillett, D. montana Stone, Griffen & Patterson, D. robusta Sturtevant, D. funebris (Fabricius) and Hirtodrosophila duncani Sturtevant (Table S1). Non-target species selection was based on their phylogenetical relationship to D. suzukii: D. simulans and D. melanogaster are closely related to D. suzukii (all belong to the D. melanogaster species group), whereas the other species are more distantly related (Markow and O’Grady 2006). The colony of G. brasiliensis was initiated from the 2016 collections in Yunnan, China, by combining the progeny of 45 female parasitoids identified morphologically as G. brasiliensis and collected at either Chang Chong Shan (Wu Hua District) or Dong Da Cun (Pan Long District) (Table 1). At that time, we had no information on the genetic variability of G. brasiliensis (Nomano et al. 2017), and for this reason, we did not genetically characterize the colony. Subsequent molecular characterization of the original field populations of G. brasiliensis revealed that the colony originated from a mixture of G1 and G3 lineages (lineages described by Nomano et al. 2017) (see below, results of molecular analysis of G. brasiliensis). Since the collection in China, the colony was maintained on D. suzukii larvae for 3–5 generations. Previously, we found that G. brasiliensis prefers to attack young larvae (Wang et al. 2018) and, for this reason, we used 1–2-day-old D. suzukii larvae for the established colony. For the rearing of this parasitoid, 40 larvae were transferred from the Petri dish colony to a small plastic vial (25 × 95 mm) filled with diet and immediately exposed to two mated, 3–4-day-old female G. brasiliensis for a 2-day period. Vials of exposed larvae were then held for the emergence of adult flies or parasitoids, with emerged adult parasitoids transferred to plastic vials supplied with 50% honey water streaked on the vial plug; the adults were later used for colony maintenance or quarantine trials.

The non-target test consisted of no-choice exposures to determine whether G. brasiliensis could attack and develop from the drosophilid species tested. For each replicate, 20 host larvae were carefully transferred under a stereomicroscope using a soft and fine brush from the Petri dish colonies to a plastic vial (25 × 95 mm) filled with 1 cm cornmeal diet and then exposed to a single mated, 3–4-day-old naïve female wasp for a 24-h period. Our preliminary observations showed negligible mortality caused by the transfer. Exposed fly larvae were held in these vials until the emergence of flies and wasps. There were 24 or 25 replicates for each fly species, except for H. duncani (20 replicates due to the shortage of host larvae) and D. robusta (30 replicates) (i.e., 20–30 female wasps tested for each fly species). The tests were conducted in five consecutive days, and there were five positive control replicates (each consisting of 20 D. suzukii larvae similarly exposed to G. brasiliensis) for each testing date and totally five negative control replicates (unexposed fly larvae held under the same conditions) for each tested fly species. After insect emergence ceased, all dead pupae were reconstituted in water for 1 day and then dissected under a microscope to determine the presence or absence of recognizable fly or parasitoid cadavers (e.g., pharate adults).

The ‘Success rate of Parasitism’ (SP), which measures the probability that a parasitized host will give rise to an adult wasp, was estimated as pi/(T − di), where pi = the number of emerged parasitoids, T = the number of emerging flies in the absence of the parasitoid, and di = the number of emerged flies in the presence of the parasitoid (Chabert et al. 2012). If pi>(T − di), we set SP = 1 (Chabert et al. 2012). We included developed but dead flies and wasps from the dissection of dead pupae to estimate the total number of emerged flies or wasps for a more precise estimate of SP (Kaçar et al. 2017). To examine possible encapsulation of the parasitoid eggs or larvae by the hosts, all emerged flies were checked for the presence of black capsules inside the fly’s abdomen (Chabert et al. 2012; Wang et al. 2016b). All positive controls (totally 25 replicates) for D. suzukii were pooled. The number of emerged wasps and SP were compared among the 10 different drosophilid species using one-way ANOVA. Prior to the analyses, all percentage data were arcsine, square root transformed to normalize the variance after checking the normality of residuals and homoscedasticity with Shapiro’s and Bartlett’s test. We also compared the numbers of developed adult flies in the presence (treatment) and absence (negative control) of parasitoids. Mean values among different treatments were separated using Tukey’s HSD. All analyses were conducted using JMP® Pro 13 (SAS, Cary, NC).

Results

Parasitoids from fruit-baited traps

A total of 458 adult parasitoids of drosophilid flies were collected from fruit-baited traps from 2013 to 2015, with majority being Braconidae (49.56%), followed by Figitidae (37.55%), Diapriidae (7.42%), and Pteromalidae (5.46%) (Table S2). Seven Asobara species were collected (Tables 3, S2–S5), among which Asobara mesocauda van Achterberg and Guerrieri (36.46%) and A. brevicauda (9.17%) were the most abundant. Only a few individuals of other Asobara species were collected. Four genera of Figitidae were collected; the majority of specimens reared belonged to Leptopilina, with L. j. japonica resulting the most abundant (14.41%), followed by L. decemflagella Lue & Buffington (5.46%). From these fruit-baited traps, Ganaspis species were the least represented figitids (0.22%) with only a single individual of G. xanthopoda collected (Tables 3, S2–S5); other figitids collected included two new species: one belonging to an undescribed genus related to Leptopilina (on the basis of our unpublished morphological and molecular analysis) (11.35%) and one to the genus Leptolamina (6.11%) (Tables 3, S2–S5). As for the new genus, 50 of 52 individuals collected were females, suggesting a possible thelytokous reproduction. The Diapriidae and Pteromalidae collected were T. drosophilae and P. vindemiae, respectively. Identification of a sub-sample of 1707 drosophilid flies emerged from the fruit-baited traps found only 18 D. suzukii (1.05%) and four D. pulchrella (0.23%) (Table S6).

Table 3 Composition of drosophila parasitoids collected by banana-baited traps during 2013–2015 in Yunnan Province, China
Full size table

Parasitoids from soft berries

From fruit collections, 14,183 D. suzukii-like puparia were recovered during July 2016: 737 from R. foliosus, 4504 from R. niveus, 2456 from F. moupinensis, and 6486 from S. adnata (Tables 4, S7). Overall, 60.9% of these puparia were classified as dead for unknown reasons. This high mortality could have been caused by various factors, including impairment of puparia during collection, sub-optimal environmental conditions during transport from China to USA quarantine laboratories, some dehydration, and host killing by parasitoids (ovipositor sting not followed by egg laying, although, having found a very low number of pupal parasitoids, this factor may have affected the mortality only marginally). As it was not possible to understand the cause of single puparia mortality and if parasitized or not, puparia dead for unknown reasons were excluded from calculation of percentage parasitism. Across all sample dates and sites, 48.9% of the remaining 5550 puparia were parasitized. Percentage parasitism (Table 4) did not vary substantially among puparia collected from berries of R. foliosus (22.35%), R. niveus (18.81%), and F. moupinensis (19.75%), whereas puparia collected from S. adnata had the highest percentage parasitism at 63.46% (G-test of independence, G = 1000.36, df = 3, P < 0.0001). A similar result was obtained when the analysis was restricted to location 11 where most of the puparia were collected and three host plants (R. foliosus, F. moupinensis, and S. adnata) occurred simultaneously (G-test of independence, G = 735.24, df = 2, P < 0.0001). Finally, percentage parasitism differed between location 10 (20.58%), where only Rubus berries were found, and location 11 (53.94%), characterized by the occurrence of berries of different botanical origin (G-test of independence, G = 300.19, df = 2, P < 0.0001).

Table 4 Numbers of fruits and hosts (Drosophila suzukii and D. pulchrella) puparia collected and percentage parasitism on different host fruits from different locations in 2016 in Yunnan Province, China
Full size table

The majority of collected adult parasitoids were figitids (G. brasiliensis at 65.4%, L. j. japonica at 32.9%), with braconids (Asobara spp. at 0.6%) and diapriids (T. drosophilae at 0.6%) occurring in only a few samples (Table S8). Three different Asobara species were found: A. mesocauda from R. niveus and F. moupinensis, A. leveri from R. niveus, and A. unicolorata van Achterberg and Guerrieri from S. adnata. Trichopria drosophilae emerged only from puparia collected from S. adnata. Host fruit species influenced the ratio between G. brasiliensis and L. j. japonica (Fig. 1), with G. brasiliensis more abundant than L. j. japonica on R. foliosus (80% vs 20%), F. moupinensis (96.8% vs 2.7%) and S. adnata (64.7% vs 34.4%), whereas L. j. japonica was dominant over G. brasiliensis on R. niveus (71.4% vs. 25.7%) (G-test of independence, G = 204.75, df = 3, P < 0.0001) (Fig. 1, Table S8). Ganaspis brasiliensis was the most abundant species collected, reaching a parasitism rate of 31.54% on average, ranging from 4.31% (R. niveus) to 40.42% (S. adnata). Parasitism rate by L. j. japonica was 16.18% on average, ranging from 4.47% (R. foliosus) to 21.52% (S. adnata) (Fig. 1). As for O. wightiana berries (collected in 2015), only two G. brasiliensis and one L. decemflagella emerged from infested fruits.

Fig. 1
figure 1

a Composition of the parasitoid complex and b parasitism rate of hosts (Drosophila suzukii and D. pulchrella) on four different host fruits. Others include the parasitoids Asobara mesocauda, A. unicolorata, A. leveri, and Trichopria drosophilae

Full size image

Taxonomic identification of flies emerging from unparasitized puparia revealed the occurrence of both D. suzukii and D. pulchrella. Drosophila suzukii was the less frequent of the two (on average 12.4%) (Table S9), with no significant difference among host fruits (Fisher’s exact Test of independence, df = 3, P = 0.154).

Molecular analysis of G. brasiliensis

Trimmed COI sequences of G. brasiliensis showed a fragment size of 635 nucleotides and 68 polymorphic sites. Alignment was straightforward with no frame shifts, nonsense codons, insertions or deletions identified in any sequence. BI analysis produced a tree (Fig. 2) showing that G. brasiliensis from Yunnan clustered in two groups corresponding to the G1 lineage (specialist on D. suzukii) and the G3 lineage (generalist on drosophilids), as described by Nomano et al. (2017). Of the 30 individuals analyzed, 27 (77%) were in the G1 and 7 (23%) in the G3 group. The G1–G5 lineages were highly supported (posterior probabilities 0.98, 1, 0.99, 1, and 0.93 for G1, G2, G3, G4, and G5 lineages, respectively).

Fig. 2
figure 2

Bayesian phylogenetic tree for COI gene sequences of Ganaspis brasiliensis. Individuals sequenced in this work are reported in bold and italicized (see Table 2 for details; individuals retrieved from the NCBI database and occurring in Nomano et al. (2017) are identified by the GenBank accession number). Acronyms indicate the drosophilid host and fruit host from which each individual sample originated. Drosophila: Dalb (D. albomicans), Dbiz (D. bizonata), Ddar (D. daruma), Deug (D. eugracilis), Dfic (D. ficusphila), Dlut (D. lutescens), Dsuz (D. suzukii), Dsuz/pul (D. suzukii and D. pulchrella), Scor (Scaptodrosophila coracina) and unk (unknown). Host fruits: Fmou (Fragaria moupinensis), Owig (Osyris wightiana), Sadn (Sambucus adnata), Rfol (Rubus foliosus) and Rniv (Rubus niveus). Geographic origin of each individual sample is also reported. G1–G5 indicate G. brasiliensis lineages following Nomano et al. (2017). Posterior probabilities ≥ 0.60 are shown above the branches. Ganaspis xanthopoda AB624300 was used as outgroup. Scale bar indicates 0.02 changes per nucleotide position

Full size image

Host specificity of G. brasiliensis

Ganaspis brasiliensis successfully developed from D. suzukii, D. simulans, and D. melanogaster and the number of offspring that developed and the SP were not significantly different among these three species (Fig. 3). Host species affected the number of offspring developed (F9,236 = 18.6, P < 0.001) and SP (F9,236 = 27.9, P < 0.001) (Fig. 3). Additionally, a few adult parasitoids developed from D. montana or D. persimilis, but there was no significant difference in the number of adult flies developed in the presence and absence of the parasitoid among all these seven non-target species (Fig. S1). In total, only three out of 292 D. simulans adults and one out of 323 D. suzukii adults contained black capsules, while all emerged flies of other species (totally 2370) did not contain them.

Fig. 3
figure 3

a Phylogenetic tree of the host species tested in the quarantine experiment. The phylogenetic relationship was re-structured based on available COI gene sequences from NCBI database. b Number of offspring produced, and c success rate of parasitism by Ganaspis brasiliensis when tested with different Drosophila host species. Values are mean ± SE, and bars bearing different letters are significantly different (Tukey’s HSD, P < 0.05). Number of replicates is reported in brackets after the name of each Drosophila species

Full size image

Discussion

Our surveys in Yunnan, China, discovered 15 parasitoid species reared from drosophilids in fruit-baited traps, but only six parasitoid species from field-collected fruits. This difference is also reflected in the species composition of the parasitoid complex, in the species prevalence and host–parasitoid associations. Asobara species dominated (49.56%) the parasitoid complex from the fruit-baited traps, whereas G. brasiliensis was dominant in field-collected fruits (65.4%). Leptopilina j. japonica was present in both sampling methods (14.41% in fruit-baited traps and 32.9% in fruit sampling). The most apparent difference between these sampling methods is that D. suzukii and D. pulchrella were the only drosophilid species emerging from fresh fruits, whereas they were largely absent from fruit-baited traps. This suggests that the fruit-bait attracted more diverse drosophilid species and consequently a higher diversity of parasitoids, but direct sampling of ripening and fresh fruits collected parasitoids specifically associated with D. suzukii and its close relative D. pulchrella. Similarly to our findings, a contemporary survey of ours carried out in China and Japan during 2015–2017 (Girod et al. 2018a) discovered eight larval parasitoid species emerging form fresh fruits infested by larvae of D. suzukii and the closely related species D. pulchrella and D. subpulchrella; among the collected larval parasitoids, G. brasiliensis was the most abundant, followed by L. japonica. It should be noted that Girod et al. (2018a) referred to G. brasiliensis as ‘Ganaspis cf. brasiliensis’; specimens from the project presented here were compared to those of Girod et al. (2018a), and indeed, all specimens conferred to G. brasiliensis.

Among the Asobara species, A. japonica was rarely collected in Yunnan, China, whereas it was the most common generalist parasitoid of drosophilids collected from fruit-baited traps in Japan and South Korea (Daane et al. 2016; Ideo et al. 2008; Mitsui and Kimura 2010; Mitsui et al. 2007). Our findings represent the first report of A. japonica from China. Among the species, we collected by fresh fruits infested by D. suzukii or D. pulchrella, Asobara mesocauda and A. unicolorata are currently known only from Yunnan Province (Guerrieri et al. 2016); A. leveri was originally described from the Fiji Islands (Nixon 1939) and had previously been recorded from China (Hubei Province), Japan and South Korea (Guerrieri et al. 2016; Nomano et al. 2015) and was associated with D. suzukii in South Korea (Daane et al. 2016). Asobara mesocauda, Asobara pleuralis (Ashmead), and Asobara sp. TK1 (maybe A. triangulata sensu Guerrieri et al. 2016) were also isolated by infested fresh fruits by Girod et al. (2018a). In our collections, A. brevicauda was only recovered from fruit-baited traps, although it was collected from D. suzukii in South Korea (Daane et al. 2016). The large presence of Asobara species in the fruit-baited traps and its paucity in fruit collections suggests a greater propensity of these braconid parasitoids to attack host larvae developing in decaying fruits rather than fresh ones as is the case of D. suzukii and D. pulchrella.

Among the figitids, L. decemflagella represents the first report from China. This species was only recently described from Eastern North America, but its biology is so far unknown (Lue et al. 2016). Leptopilina decemflagella was frequently collected in fruit-baited traps, but it was never recovered from puparia collected from ripe fruits, suggesting that it is not likely associated with D. suzukii or D. pulchrella. Two new species records of figitids were also collected exclusively in fruit-baited traps and include Leptolamina sp. and a species of an undescribed genus related to Leptopilina. The large number of L. j. japonica collected with fruit-baited traps probably emerged from drosophilid species other than D. suzukii and D. pulchrella. In contrast, G. brasiliensis was isolated exclusively from D. suzukii-like puparia from field-collected fruits. This suggests that L. j. japonica could be a more generalist parasitoid than G. brasiliensis. Indeed, L. j. japonica can form new associations with some Drosophila species without undergoing adaptive changes (Kimura and Novkovic 2015). In fruit samples, G. brasiliensis co-occurred with L. j. japonica but was much more abundant than L. j. japonica on R. foliosus, F. moupinensis and S. adnata (the opposite was observed on R. niveus). Records of G. brasiliensis and L. japonica attacking larvae of D. suzukii, D. pulchrella or D. subpulchrella in fresh fruits have recently been reported from different provinces of China, and in particular from the Yunnan province (Girod et al. 2018a). One population of L. japonica from Yunnan was able to parasitize other two Drosophila species, including D. melanogaster and D. subobscura (the former is closely related and the latter more distantly related to D. suzukii) (Girod et al. 2018b).

The two pupal parasitoids, T. drosophilae and P. vindemiae, were collected from both fruit-baited traps and field-collected fruits, although in low numbers, in part because of the relatively short exposure period of fruit-baited traps and host larvae that had already left the fruits collected in the field.

Our results also indicate an effect of the host plant species on parasitism rate and parasitoid species composition. Overall, we found a considerable rate of parasitism of fly larvae developing in fresh fruits, ranging from 19 to 22% on R. foliosus, R. niveus, and F. moupinensis to 63% on S. adnata. The parasitism activity was almost entirely due to G. brasiliensis and L. j. japonica, with the former being the most active. On R. foliosus and F. moupinensis, 18–19% of puparia were parasitized by G. brasiliensis, whose parasitism rate reached 40% on S. adnata. Leptopilina j. japonica was more active than G. brasiliensis only on R. niveus (12% vs 4% parasitized puparia). The concentrated sampling period (July) may have affected the parasitism rate and the proportion of the host species due to the climate and fruit availability. However, our results are in line with those reported by Girod et al. (2018a) who found the total parasitism rate ranging from 0–54% in China and 0–76% in Japan, depending on the host plant and the time of collection, with G. brasiliensis being the most active larval parasitoid, reaching the maximum parasitism rate of 42% on Prunus cerasoides in Yunnan and 76% on Morus sp. in Japan.

Nomano et al. (2017) recognized five lineages of G. brasiliensis, which differed in geographic distribution and host range: G1, including individuals from Japan parasitizing only D. suzukii; G2, including individuals from a subtropical Japanese isle parasitizing Drosophila ficusphila Kikkawa & Peng; G3, including individuals from temperate regions of Japan and high mountains of Southeast Asia (Indonesia, Malaysia) parasitizing different species of Drosophila, except D. suzukii; G4, including individuals from Indonesia parasitizing Drosophila eugracilis Bock and Wheeler; G5, including individuals from Japan, Taiwan, Hawaii, and Uganda, from unknown host(s). In the laboratory, individuals belonging to G5 from Hawaii and Uganda have shown a poor ability to parasitize D. suzukii (Kacsoh and Schlenke 2012). Phylogenetic analysis of COI sequences revealed that our G. brasiliensis samples were grouped in two lineages: 77% in the G1 lineage and the remaining 23% in the G3 lineage. Their occurrence was sympatric on the same host plants, in accordance with previous findings indicating that G1, G3, and G5 are sympatric in Japan (Nomano et al. 2017). Although morphologically indistinguishable from each other, theses lineages could be a complex of cryptic species. Nomano et al. (2017) found incomplete reproductive isolation between populations of G3 and G5 lineages which group in the same clade. The level of reproductive isolation between the G1 lineage and the other lineages has not been investigated yet. However, the genetic distance (calculated on the COI gene nucleotide sequence) between the G1 individuals and those of other lineages is large enough (5–7% between G1 and G3 individuals) to suggest a reproductive isolation. Also, the higher level of host specificity of G1 lineage in respect to more generalist ones (Nomano et al. 2017; Matsuura et al. 2018; our results presented here) points toward the existence of a complex of species under the name G. brasiliensis. This hypothesis is also supported by the wide geographic distribution of this taxonomic entity (Asia, Central and South America) (Buffington and Forshage 2016). Studies are underway to disentangle the taxonomy of this putative complex of species.

In quarantine experiments, the studied population of G. brasiliensis from Yunnan, China (comprising G1 and G3 individuals), attempted to attack all tested fly species in artificial diet, which were provided in a simple no-choice test. These tests were conducted in small vial with artificial diet. The results thus reflect largely the suitability of the tested host larvae for the parasitoid’s development (i.e., physiological host range), and the ecological host range could be further narrow. Indeed, quarantine experiments showed that G. brasiliensis successfully developed from D. suzukii and its two closed related hosts (D. simulans and D. melanogaster) but largely failed to develop from other seven tested host species. Nevertheless, rare encapsulated parasitoid eggs were noticed in emerged flies. Host species that are phylogenetically related may be attacked by common parasitoid species as they often share physiological or morphological characteristics that determine their suitability as a host (Desneux et al. 2012). Although both D. melanogaster and D. simulans are physiologically suitable hosts for G. brasiliensis, these drosophilids typically infest overripe or rotting fruits, while in the field D. suzukii exploits ripening fruit (Mitsui et al. 2007), before they are available to D. melanogaster or D. simulans. Indeed, our collections of fresh fruits rarely found drosophilid species other than D. suzukii and D. pulchrella. In addition, as a part of parasitoid host range drivers, host habitat (fresh vs rotten fruits) could be crucial for the host location success of drosophila larval parasitoids that mainly rely on specific host-derived volatile chemical cues (Biondi et al. 2017). In summary, we showed that D. suzukii and D. pulchrella co-occur on fresh fruit in Yunnan Province, where they are attacked by numerous parasitoid species, the most important of which is the more specialized G. brasiliensis, comprising the G1 and G3 lineages.

Geographic variations in the parasitoid virulence may also exist among host populations or lineages of G. brasiliensis. Our field observations and host specificity quarantine experiments suggest that both G1 and G3 lineages can successfully attack D. suzukii. Previously, the G1 lineage was considered a specialist on D. suzukii (Kasuya et al. 2013; Nomano et al. 2017); however, the characterization of different G. brasiliensis lineages in terms of host specialization may be premature based on a few samples from Japan. For example, Girod et al. (2018c) found that different G. brasiliensis geographic populations (two strains from Yunnan and one strain from Japan) successfully attacked D. suzukii developing in ripe blueberries, but rarely accepted fly larvae developing in artificial diet; two of these parasitoid populations (that were not characterized genetically) showed variations of the level of specificity, with the Japanese population being highly specific to D. suzukii, and a Yunnan population being able to parasitize, in addition to D. suzukii, larvae (in artificial diet) of D. melanogaster and rarely D. subobscura (Girod et al. 2018b). Here, we found similar level of host specificity in the tested population of G. brasiliensis from Yunnan. Interestingly, even in South Korea, we collected G. brasiliensis only from field-collected fruits infested by D. suzukii but not from traps baited with uninfested fruit (which yielded predominately other Drosophila species); this parasitoid population was able to develop successfully in larvae of D. suzukii and D. melanogaster in artificial diet (Daane et al. 2016). In another example, Mitsui and Kimura (2010) reported that a strain of Ganaspis from Drosophila lutescens Okada, which was later assigned to the G3 lineage by Nomano et al. (2017), parasitized only drosophilid larvae in fermenting fruits while D. suzukii in ripe fruits were rarely accepted. Our results extend to G3 lineage individuals the ability to parasitize both D. suzukii and D. pulchrella. More recently, Matsuura et al. (2018) suggested a level of specialization of one Japanese D. suzukii-associated type G. brasiliensis (G1 lineage) which attacked D. suzukii larvae in fresh fruits in the tree canopy and rarely larvae in fruits fallen on the ground. This result supports our findings of G. brasiliensis collected only from fresh fruit and never from fruit-baited traps. One explanation of this variation among populations of G. brasiliensis is that each of these lineages may have adaptated to local host species and host habitats (Murata et al. 2009; Rossi Stacconi et al. 2015). A different ability of G1 and G3 lineages of G. brasiliensis in parasitizing D. suzukii could be associated with a diverse capacity to overcome the strong cellular immune response of D. suzukii larvae (Kacsoh and Schlenke 2012; Poyet et al. 2013). Due to their high specificity toward D. suzukii, populations of G. brasiliensis represent good candidates for use in a classical biological control program. That some closely related species (i.e., D. melanogaster and D. simulans) might be attacked could also aid in biological control as these flies could serve as alternative hosts when D. suzukii are scarce. However, G. brasiliensis is unlikely to impact the populations of these non-target hosts because of competition with resident parasitoids that specialize on flies infesting rotted fruit. Detailed evaluations of the G1 and G3 lineages of G. brasiliensis are currently underway to determine their specific performance on D. suzukii, host range, and impact on non-target species with the goal of selecting a possible biological control agent to be introduced in the USA.

Author contribution

EG, MG, KMD, and KAH conceived and designed the project. MG, EG, KMD, and XGW wrote the initial manuscript; all co-authors helped editing the manuscript thereafter. EG, MB, MG, PC, GF, and GAC completed the molecular and taxonomic work; XGW, EH, AB, and KMD conducted the quarantine work. EG, MG, XGW, YW, FSC, HMZ, ZQC, HYC, CXL, KMD, and KAH conducted the field collections in China. All authors read, revised, and approved the manuscript.