First report on classical biological control releases of the larval parasitoid Ganaspis brasiliensis against Drosophila suzukii in northern Italy

Current management strategy of the invasive fruit fly Drosophila suzukii (Matsumura) (Dip-tera: Drosophilidae) exploits different tools but relies mainly on chemical control. In the invaded areas, the local natural enemy community mostly consists of generalist pupal parasitoids unable to control the pest efficiently. Conversely, in the pest native area, there are more specialized sympatric larval parasitoids attacking D. suzukii . Following foreign explorations and quarantine risk assessments, the larval endopara-sitoid Ganaspis brasiliensis (Ihering) (Hymenoptera: Figitidae) was selected as the best candidate for classical biological control programs. In 2021, the first ever propagative biocontrol program using a Japa-nese G1 lineage of G. brasiliensis started in Italy. Here we report the results of the first year of releases in the province of Trento (Northeast Italy), wherein G. brasiliensis was released in 12 locations. Pre-and post-release samplings on fresh and fallen fruits were performed around the release points to assess the recapture rate, the impact of the exotic parasitoid on D. suzukii and its potential interactions with local non-target species. After releases, G. brasiliensis was recovered at 50% of the locations. The exotic parasitoid only emerged from D. suzukii , mostly from fresh fruit still on the plant. Post-overwintering monitoring revealed the presence of a four G. brasiliensis


Introduction
Drosophila suzukii Matsumura (Diptera: Drosophilidae), native to southeast Asia, has gradually invaded North America and Europe (2008), South America (2012) and Africa (2017) becoming a global economic threat to soft-skinned fruit crops (Cini et al. 2012;Asplen et al. 2015; Anfora 2020; Hassani et al. 2020). The control of this highly polyphagous pest primarily relies on local area management, mostly on the use of insecticides (Tait et al. 2021). However, this approach is often not environmentally friendly and economically sustainable because of pest re-infestations occurring shortly after treatment from surrounding vegetation (Tait et al. 2020). On the other hand, an area-wide management approach would deal better with such mobile pest by reducing the total pest population within crop and non-crop areas (Haye et al. 2016).
In this sense, biological control strategies to suppress D. suzukii have been intensively studied, and include the use of predators, parasitoids, pathogens and competitors (Tait et al. 2021). In particular, it has been shown that parasitoids can regulate the population of Drosophilids, either by directly causing high mortality rates (Janssen et al. 1988;Driessen & Hemerik 1991) or by affecting competitive interactions between sibling host species (Boulétreau et al. 1991). In the invaded areas, two cosmopolitan pupal parasitoids have been identified to successfully parasitize D. suzukii: Trichopria drosophilae (Perkins) (Hymenoptera: Diapriidae) and Pachycrepoideus vindemiae (Rondani) (Hymenoptera: Pteromalidae) (Miller et al. 2015), with the first being released in Italy and Mexico and the latter in North America for augmentative biological control (Rossi Stacconi et al. 2019;Gonzalez-Cabrera et al. 2019;Hogg et al. 2022). However, natural parasitism by these generalist pupal parasitoids is below 10% (Lee et al. 2019) and the introduction specialist parasitoids from the native area of D. suzukii for classical biological control remains a preferable option (Lee et al. 2019).
Classical biological control can be an effective strategy to control invasive insect pests (Hajek et al. 2016) and can offer protection to natural ecosystem (van Driesche et al. 2010). Although its application and number of introductions gradually decreased since the 1970s, this method gained higher establishment and success rate in recent days, regaining momentum and trust (Cock et al. 2016). This positive trend is also the result of higher attention given from national and international regulatory bodies into providing appropriate tools for risk assessment, import and release of exotic biocontrol agents (BCA), all necessary steps to limit potential negative impacts on biodiversity in areas of introduction (Barratt et al. 2018).
In its native environment, D. suzukii is regulated by a multitude of natural enemies and, among these, at least 14 parasitoid species   (Girod et al. 2018a, b;Giorgini et al. 2019;Daane et al. 2021). In particular, for G. brasiliensis, five clades (G1 to G5) where identified on a genetic basis (Nomano et al. 2017;Giorgini et al. 2019), with the G3 attacking few hosts in the Drosophila melanogaster Meigen species group and the G1 being considered a specialist on D. suzukii, thus rapidly gaining researchers' interest as a candidate for classical biological control. Adventive populations of L. japonica have been recently identified in northeastern Italy ) and in the northwestern North America (Abram et al. 2020a), whereas the G1 clade of G. brasiliensis was only recorded in British Columbia (Canada) and Washington (USA) (Abram et al. 2020a;Beers et al. 2022).
In 2021, after the evaluation of a comprehensive risk assessment submitted by Italian scientific institutions and phytosanitary services, the Italian Ministry of Ecological Transition authorized the release of G. brasiliensis G1 in seven regions and two autonomous provinces within the frame of a national biological control program (Rossi Stacconi pers. comm.). Here we report the results of the monitoring program following the first-year releases of G. brasiliensis in the province of Trento (Northeastern Italy), where the exotic parasitoid was released at 12 locations.

Parasitoid rearing and releases
Parasitoids were reared according to the mass rearing protocol specified in Rossi Stacconi et al. (2022). For host, we used laboratory-reared D. suzukii that originated from multiple field collections of live adults at different locations of Trento Province (TN) Italy, during 2020 and 2021. The starting colony of G. brasiliensis derived from wild individuals collected during surveys in Naganuma park (Hachioji, Tokyo, Japan) from 2015 to 2017 (Girod et al. 2018a) and were provided by 3 First report on classical biological control releases of the larval parasitoid Ganaspis… 1 3 Vol.: (0123456789) CABI's Swiss centre (Delémont, Switzerland). Parasitoid releases were performed at 12 locations in the Trento province, Italy (Table 1). Locations were selected according to presence of suitable unmanaged D. suzukii hosts, known presence of the pest and absence of ecological vulnerability constraints (protected areas). The sites were distributed along an altitudinal range (Table 1) and categorized as valley bottom sites (0-249 m a.s.l.), hillside sites (250-699 m a.s.l.) and mountain sites (700 + m a.s.l.). All sites were located at the margins of wooded areas, providing natural corridors for parasitoid movements. Four sites (#1, #2, #10 and #12) were in proximity (< 50 m) of crops considered non-host for D. suzukii (apple orchards and vineyards), four sites (#4, #5, #8 and #11) were in proximity of cherry orchards and four sites (#3, #6, #7 and #9) were in proximity of multiple small fruit orchards (blueberry, blackberry, raspberry and strawberry). At each site, a single release point was selected and parasitoid releases were performed once a week during three consecutive weeks. The dates of the releases were set according with local host plants' phenology (Table 1). Each release consisted of 200 three-day-old adults of G. brasiliensis (sex ratio 50:50 M:F).

Field surveys
The authorization to release G. brasiliensis was received on August 17th 2021 and parasitoid releases started within one week. Field surveys were performed before (1-2 pre-release samplings) and after (six post-release samplings) the first release event (Table 1). At each site, at least one pre-release sampling and five post-release samplings were conducted. Due to the lack of experimental evidence comparing the outcomes of different sampling methods (Abram et al. 2022a), the twelve release locations were divided into two subgroups, establishment sites (n = 3) and specificity sites (n = 9), each adopting a different sampling methodology. The goal of monitoring establishment sites (ESs) was to verify the ability of G. brasiliensis to disperse and reproduce in the new environments after releasing; to this aim, large quantities of fruits were collected from a wide area surrounding the release sites (see below). On the other hand, the goal of monitoring in the specificity sites (SSs) was to assess the host specificity of G. brasiliensis towards D. suzukii and other non-target species, and its interactions with other parasitoids. To this aim, at each site a fruit was collected from few selected host patches and all fly puparia deriving from it were sorted prior to eclosion. In Longitudes and latitudes are provided as degrees, minutes, and seconds format and refers to North and East, respectively. All sites are located in Northeastern Italy Light and dark shadings indicate pre-and post-release sampling events, respectively. SS = specificity site; ES = establishment site 2020 and 2021, monitoring activity was also conducted in multiple control sites (Supplementary  Table S1) where no release of G. brasiliensis was performed. Monitoring in control sites was carried out according to the sampling protocol described for ESs, with the aim to assess the presence of adventive populations of the parasitoid in the area of the study (Trento province, Italy) and supporting results from the pre-release monitoring in ESs and SSs. In the ESs, post-overwintering surveys were conducted on May 27th and June 21th 2022 to check the survival of G. brasiliensis after the cold season.

Fruit collections
In the ESs, multiple fruit collections were conducted on host plants located in an area within 200 m from the release point. To account for spatial variation in host densities and parasitism levels, all plant patches included in the monitoring area and carrying fruit susceptible to D. suzukii attack were sampled and kept individually. In the SSs, fruit collection was carried out from only four pre-selected host plant patches located within 100 m from the release point. For both sampling methodologies, all sampling points were georeferenced using SMASH Digital Field Mapping (HydroloGIS S.r.l, Bolzano, Italy) to calculate their distance from the release site. At each sampling point and fruit quantity permitting, sub-samplings were carried out both from the plants (fresh fruit), randomly picking over the whole vertical height of the canopy of each selected host plant patch, and from the underlying ground (overripe and rotten fruit). Each sub-sample aimed to fill a 280 ml plastic container (Unipak 5011-21, Berry Superfos, Bologna, Italy) corresponding to 50-80 g of fruit, depending on the size and the water content of individual fruit. All sub-samples were incubated for three weeks in separate containers under optimal condition ranges (21-24 °C; 65-75% RH; L:d 16:8 photoperiod). Incubation was carried out in modified plastic containers similar to those described by Abram et al. (2022a). Within each container, fruit was placed over blotting paper covering a layer of sand (1 cm). Such method allowed for a better control over humidity, as sand worked as moisture retainer and prevented formation of liquid pools.

Laboratory handling of collected fruit
Two methodologies were adopted to assess emergence of the different fly and parasitoid species depending on the collection site (ESs or SSs). For fruit collected in the ESs, emerged individuals (flies and parasitoids) from each sub-sample were regularly removed from the container three times per week and stored in 70% ethanol until taxonomic identification. For samples collected in the SSs, Drosophila puparia were collected from containers as they formed. Based on size, color and the unique morphology of the spiracles of D. suzukii puparium (EPPO 2013; Abram et al. 2022a), puparia were divided in D. suzukii and non-D. suzukii and kept in separate containers over moist filter paper until emergence of flies or parasitoids. This method allowed for an accurate estimation of parasitization rate on target and non-target by G. brasiliensis and by other larval and pupal parasitoid species. For both methodologies, at the end of the incubation period, containers were carefully inspected and all remaining uneclosed puparia were collected, re-constituted in water for 24 h and dissected to check for presence of dead hosts or parasitoids.

Data analysis
The number of G. brasiliensis and other parasitoids as well as that of hosts emerged from the samples was recorded. Identification was conducted on a morphological basis using taxonomic keys (

Occurrence of G. brasiliensis
No G. brasiliensis emerged from fruit collected in the control sites (Supplementary Table S1) or at the release sites during pre-release sampling (Tables. 2, 3). During post-release sampling, a total of 87 G. brasiliensis individuals was obtained from fruit collected at six release locations (#1, #4, #8, #9, #10 and #12). (Fig. 1 (Fig. 2, 3, 4). In the SSs, all fruit samples containing G. brasiliensis were collected in the proximity of the release point (0-10 m), representing 18% of   brasiliensis was present, apparent parasitism was on average 2.3% (0.2-13.1), accounting for 17.9% (span: 1.8-67.8%) of the total apparent parasitism recorded in those samples (G. brasiliensis + other parasitoids). When G. brasiliensis parasitism occurred, L. japonica was always present and more abundant than G. brasiliensis. Local larval parasitoid species never eclosed from plant-sampled fruit and were absent in the two ground-sampled fruit samples attacked by G. brasiliensis (Tables 2, 3). When dissecting D. suzukii uneclosed puparia (< 0.7% of the total puparia), no sign of aborted parasitism was recorded.

Discussion
Implementing a propagative (classical) biocontrol program involves efforts that are equally distributed between the pre-and post-release phases. Pre-release work seeks to draft a comprehensive risk assessment for the candidate exotic BCA, considering all positive and negative consequences of its potential introduction in a new environment (Barratt et al. 2010). Postrelease work mainly aims to verify (1) the initial survival and potential establishment of the released BCA and (2) the interactions of the BCA with the target species and with other local organisms (nontargets) (van Driesche et al. 2010). In our post-release study, the results of the first-year monitoring showed that G. brasiliensis can reproduce and overwinter in the release area. In particular, the sites at which the parasitoid was recovered during the post-overwintering surveys are both located at the valley bottom and characterized by a temperate climate with hot summer and no dry season (Beck et al. 2018). This is consistent with the CLIMEX model prediction from Wang et al. (2020), indicating the suitability of most of the temperate European areas for the establishment of the exotic parasitoid. Despite the low number of parasitoids released per site (300 females and 300 males), G. brasiliensis was recovered in 50% of the sites during the six weeks following the first release species, the number of samples collected followed by the number of fly and parasitoid individuals eclosed are indicated in brackets First report on classical biological control releases of the larval parasitoid Ganaspis… event (weeks 34-39) and regardless the altitudinal range. Moreover, post-overwintering monitoring showed that, at least at two locations, G. brasiliensis was able to survive the cold season and to start new generations in spring. These findings strongly suggest a climatic suitability of the Province of Trento for the parasitoid. However, further data need to be collected in the next years, particularly in relation to the ability of the exotic BCA to overwinter in release area (Hokkanen and Sailer 1985 (Quacchia et al. 2008). By 2009, the parasitoid emerged from 4.08% of the collected galls and after seven years from its first release (in 2012) T. sinensis parasitization rate was steadily above 80% (Ferraccini et al. 2019). Two different monitoring approaches were adopted for establishment and specificity sites, allowing for different evaluations on the impact of G. brasiliensis in the release area. Monitoring in the ESs aimed to sample large quantity of fruit from a wide sampling area in order to maximize the probability to recover G. brasiliensis, to assess its spread capacity from the release point and to provide an overview of the host plants exploited by the parasitoid during host search. On the other hand, most of the sites were monitored as SSs sampling a reduced amount of fruit samples in order to evaluate in detail parasitism parameters and host specificity of G. brasiliensis, As expected, the occurrence rate of G. brasiliensis was higher in ESs (66.6% of the sites) than in SSs (44.4% of the sites), although in the SSs were found more fruit samples attacked by the BCA (n. 11/184) than in the ESs (n. 6/229). However, looking at the distribution of the attacked fruit samples, the ESs monitoring approach seems to provide a more accurate representation of the parasitoid movements within the sampled area compared to the SSs approach, showing some parasitoid activity away from the release point. Despite the practical differences, data from the two monitoring approaches were consistent with the fact that G. brasiliensis G1 behaves as a specialist on D. suzukii, never eclosing from puparia of other species (SSs) or from fruit samples infested with other flies than D. suzukii (ESs). This aspect of the biology of G. brasiliensis G1 was already observed in its native range Girod et al. 2018a), in quarantine facility studies (Girod et al. 2018b;Daane et al. 2021) and in field cage releases (Seehausen et al. 2022), but this is the first study reporting the parasitoid's strict association with D. suzukii in natural conditions outside its native range. A second aspect that has been further confirmed by our study, is that G. brasiliensis tends to avoid competition by attacking its hosts mostly within fresh fruit still on the plant (Giorgini et al. 2019), whereas local parasitoids of drosophilids parasitize their hosts on decaying fruit. Although chemical ecology studies have not yet been conducted on olfactory preference of G. brasiliensis G1 towards infested and uninfested fruit, such avoidance behavior was already suggested in a recent study considering the competitive parasitization outcomes of three solitary larval parasitoids of drosophilids:; G. brasiliensis G3, L. japonica and Asobara japonica Belokobylskij (Hymenoptera: Braconidae) . The study showed that G. brasiliensis G3 is outcompeted by both the other parasitoids and, in response to that, it discriminates against parasitized hosts. Also, we observed that the ovipositor of the released strain of G. brasiliensis is significantly shorter compared to that of other larval parasitoids recovered from the monitoring (Supplementary Figure S1). Such morphological feature would allow G. brasiliensis to only parasitize host larvae located just below the fruit skin of fresh fruits (i.e., 1st and 2nd instars) and making it extremely difficult for the exotic parasitoid to attack hosts buried in a rotten substrate. In terms of impact on biodiversity, these functional traits (preference towards fresh fruit and ovipositor morphology) exhibited by G. brasiliensis creates a specific trophic niche for the BCA (Cohen 1977) and reduces the risk of negative impacts on non-target organisms, such as local larval parasitoids.
Overall, the result of our monitoring showed that L. japonica was the dominant parasitoid species in all locations where G. brasiliensis was released. This adventive exotic species was reported in Italy (Province of Trento) for the first time in 2019 at one location ) and, since then, increasing catches have been reported. In its native range, L. japonica is frequently found on D. suzukii, which is considered a preferred host (Girod et al. 2018b;Giorgini et al. 2019), and recent studies reported that this parasitoid reproduces on a relatively narrow host range, although not narrow enough to be considered as a candidate for classical biocontrol programs (Girod et al. 2018a;Daane et al. 2021). Likely, the presence of D. suzukii as one of the most abundant species feeding on cultivated and wild fruit have been a key factor contributing to the rapid dispersal and population growth of L. japonica in the Alpine region and a similar trend can be expected for G. brasiliensis. However, population changes induced by biological control programs often require years to reach stable endpoints and the outcomes are function of many dependent and independent variables, including the intrinsic characteristics of the pest, the BCA and the target ecosystem (Hokkanen and Sailer 1985). Abram et al. (2022b) recently reported that in Canada (British Columbia) the two adventive parasitoids, G. brasiliensis and L. japonica, have re-built the same close association with D. suzukii observed in the pest's native range, with the latter being the dominant species. Extensive fruit samples carried out in 2020 showed D. suzukii parasitization by the two parasitoids to be time-structured and widespread across a variety of habitats and host plants. Such observations have been made a short time after the first reports of G. brasiliensis and L. japonica in the area (one and four years, respectively), and most likely the parasitoids' dispersal is still ongoing (Abram et al. 2022b). In the area of the present study, D. suzukii is considered a major pest of soft fruit and, more than a decade after the introduction, its impact still accounts for 9% of the potential revenues, mostly due to costly control measures (De Ros et al. 2020). Realistic expectations for impact of classical biological control in Italy might be that G. brasiliensis will contribute in restoring part of the top-down pest suppression existing in the D. suzukii native area. However, such contribution will likely be difficult to quantify in terms of success or failure, as it is expected to be regionally variable and to add to that provided by other local and adventive biocontrol organisms (Abram et al. 2020b). In this sense, the establishment of stable populations of G. brasiliensis and their relative contribution to D. suzukii larval parasitism would be a first step towards success of the classical biological control program. Once this is accomplished, an evaluation should be done in the long run to assess whether the increased parasitism has resulted in a reduction of management costs.