Different ovipositor characteristics among parasitoid species that share similar niches are associated with different wasp life histories and selective pressures. The length of wasp ovipositors, for example, can determine the accessibility of hosts that feed at different depths within food substrates. Two parasitoids, Ganaspis brasiliensis and Leptopilina japonica (Hymenoptera, Figitidae), which attack Drosophila suzukii (Diptera, Drosophilidae) in their native range, have been investigated for their suitability for the global biological control of the small fruit pest. Despite their sympatry in microhabitat, the parasitoids have differing host ranges, and D. suzukii parasitism rates by each parasitoid species appear to depend on the fruit species occupied by the host species. Adventive populations of both parasitoids have been detected in the Pacific Northwest of Canada and the United States where they can be found parasitizing D. suzukii larvae in crop and non-crop fruits. We dissected and measured the ovipositors of parasitoids reared from three species of fresh fruits at three sites in southwestern British Columbia, Canada, and investigated the influence of parasitoid species, fruit type, and collection site on ovipositor characteristics. We found that ovipositor length differed markedly between the two parasitoid species and between sites while ovipositor width, and stoutness, differed only between the two parasitoid species, but did not vary among sites or fruit hosts. We discuss how ovipositor morphology traits could be associated with differences in life history and host ranges in the two parasitoid species.

biological control, competition, Ganaspis brasiliensis, Leptopilina japonica, morphology, niche partitioning, spotted-wing drosophila

Parasitoid reproductive success depends on access to and successful exploitation of available hosts. For parasitoid Hymenoptera, the morphology of ovipositors used to lay their eggs into or onto hosts, is critical for the effective exploitation of hosts within structural niches (Le Ralec et al. 1996; Quicke et al. 1999; Vilhelmsen and Turrisi 2011). One way that insect herbivores can reduce attack by parasitoids is by feeding deep within the confines of plant structures like galls, fruits, or tree trunks. Select parasitoid species are able to circumvent this type of host defence by accessing their hosts earlier in host development in smaller plant structures (a phenological adaptation, e.g. Weis and Abrahamson 1985), while others overcome this defence using long ovipositors that can access hosts deep inside larger plant structures (a morphological adaptation, e.g. Weis and Abrahamson 1985). Hosts can further escape pressures imposed by previously associated parasitoids by occupying larger plant structures like cultivated fruits (e.g. Feder 1995) or feeding on earlier phenological stages of their host plants (e.g. ripening as opposed to rotting fruit). Some of these shifts in host feeding behaviour could engender costs for the host, which would be balanced against the benefit of decreased parasitism risk (Gratton and Welter 1999). In addition, these differences in feeding niche could hypothetically spur the evolution of modified ovipositors in parasitoids that allow them to effectively access hosts in these novel environments (Vermeij 1999; Sivinski et al. 2001; Sivinski and Aluja 2003).

Constraints imposed by parasitoid ovipositor length can affect the success of biological control of non-native pests such as the olive fruit fly Bactrocera oleae Rossi (Diptera: Tephritidae) in the California olive agricultural system (Sime et al. 2007; Wang et al. 2008, 2009). Here, larger fruits are preferred by growers and this bias in cultivar choice has been shown to prevent certain parasitoids (Hymenoptera: Braconidae) with shorter ovipositors from accessing their hosts (Sime et al. 2007; Wang et al. 2008, 2009). Parasitoid species with shorter ovipositors are less effective at controlling the pest larvae in larger fruits, while those species with longer ovipositors can access host larvae more effectively in larger fruits (Sime et al. 2007; Wang et al. 2008, 2009). Parasitoids in this system with longer ovipositors also have broader host ranges; a consideration that complicates the biological control of B. oleae in an agricultural system that selects for larger fruits (Sime et al. 2007; Wang et al. 2008, 2009). Thus, the success of a biological control programme could be reduced by physical barriers to parasitism.

The behaviour of host larvae can also influence the success of parasitism, by evading detection by parasitoids or by allowing larvae to escape parasitism when first encountered by a parasitoid (Gross 1993; Kacsoh et al. 2013; Robertson et al. 2013; Singh et al. 2015). Larvae living in semi-concealed environments may escape parasitism by diving deeper into their developmental substrate (van Lenteren et al. 1998). In order to parasitize mobile hosts that attempt to escape parasitism, Figitidae (Hymenoptera) that parasitize semi-concealed dipterous larvae have an ovipositor clip (Buffington 2007), a morphological feature that restrains host larvae prior to their envenomation (van Lenteren et al. 1998). Drosophila Fallén (Diptera: Drosophilidae) larvae have differing foraging behaviours within their developmental substrate where some (e.g. D. melanogaster Meigen) feed relatively close to the surface while others (e.g. D. suzukii Matsumura), that typically develop in deeper substrates such as ripening fruits, dive deeper into the substrate more frequently and for longer (Kim et al. 2017). Differential diving behaviours of Drosophila larvae could have implications for parasitoid access to hosts.

Two parasitoid wasp species, Leptopilina japonica Novković & Kimura and Ganaspis brasiliensis Ihering (Hymenoptera: Figitidae), have been proposed as candidate classical biological control agents for the small fruit pest D. suzukii (Lee et al. 2019). These parasitoids have recently been detected in North America, specifically in southwestern British Columbia (BC), Canada (Abram et al. 2020) and northwestern Washington State, USA (Beers et al. 2022). Leptopilina japonica has the broader host range of the two parasitoids and successfully parasitizes several drosophilid species in four genera in laboratory trials (Kimura and Novković 2015; Girod et al. 2018b; Daane et al. 2021). The lineage of G. brasiliensis recorded in BC and Washington State (“G1”; Beers et al. 2022) has a narrower host range, primarily attacking flies in the subgenus Sophophora Sturtevant in laboratory trials (Girod et al. 2018b; Giorgini et al. 2019; Daane et al. 2021). These parasitoids are adventive in the fruit growing regions of southwestern BC but have evidently not dispersed into the province’s interior (Abram et al. 2022a, b).

The sympatry of parasitoids exploiting the same host in the same habitat can lead to competition between parasitoids, which typically manifests as either competition between adult parasitoids for mating or oviposition resources (extrinsic competition) and/or competition between immature parasitoids competing for host resources (intrinsic competition) (Harvey et al. 2013; Ode et al. 2022). The inability of a developing parasitoid to compete in intrinsic competition within the host can be mitigated by extrinsic competition by the adult, and vice versa (Hood et al. 2021). Extrinsic and intrinsic competition may influence niche partitioning by parasitoids that exploit the same host in the same habitat by influencing their exploitation of hosts in unrecognized microhabitats (e.g. Heatwole et al. 1964; Heatwole and Davis 1965). The competitive interactions between L. japonica and G. brasiliensis are beginning to be characterized. While G. brasiliensis avoids hyperparasitizing hosts (Wang et al. 2019), L. japonica does not avoid such intrinsic competition (Böttinger et al. 2019) and is a superior intrinsic competitor in hyperparasitized larvae than is G. brasiliensis if hyperparasitism occurs within 24 hours of the initial parasitism event (Wang et al. 2019). Therefore, extrinsic competition by G. brasiliensis through the exploitation of host larvae in media that is not accessible to ovipositing L. japonica could facilitate host partitioning in this system.

In a field survey of fresh fruit collections in southwestern BC, L. japonica and G. brasiliensis made up 67.2% and 32.0% of the larval parasitoid community, respectively (Abram et al. 2022a). In the adventive range, L. japonica appears to cause higher percent parasitism of D. suzukii in larger fruits (e.g. strawberries, Fragaria spp.; blackberries, Rubus spp.) whereas G. brasiliensis appears to impose higher percent parasitism than L. japonica in smaller fruits (e.g. elderberries, Sambucus spp.; Abram et al. 2022a). In fresh fruit collections in China, G. brasiliensis was better able to parasitize D. suzukii in smaller fruits than in larger fruits with a parasitism rate that was higher in Sambucus adnata Wallich than in Rubus foliosis Weihe or Rubus niveus Thunberg, respectively (Giorgini et al. 2019). Abram et al. (2022a) observed differences in larval D. suzukii attack by adventive parasitoids that were similar to those observed in California olives (Sime et al. 2007; Wang et al. 2008, 2009), where higher parasitism rates were observed in parasitoid species with broader host ranges, and these tended to be associated with larger fruits. Additionally, Fellin et al. (2023) found that the G1 lineage of G. brasiliensis that was released to control D. suzukii in Italy have significantly shorter ovipositors than the adventive Italian L. japonica. It is possible that different ovipositor characteristics may affect host-choice and facilitate coexistence through niche partitioning (Price 1972) in the D. suzukii biological control system.

The objective of this study was to quantify differences in ovipositor characteristics of L. japonica and G. brasiliensis in BC, Canada. First, based on previously described associations between host range breadth and ovipositor length in other parasitoid guilds, we predicted that L. japonica would have longer ovipositors than G. brasiliensis, as longer ovipositors may be more efficient for indiscriminately parasitizing larvae regardless of their species, or level of prior parasitism. Additionally, we predicted that the apparently more specialized G. brasiliensis would have significantly stouter ovipositors (ovipositor width/ovipositor length) than do L. japonica. Stouter ovipositors may be better able to puncture fruits at an earlier stage of ripeness with D. suzukii developing inside; thus, further reducing the probability of mortality through lethal intra- and interspecific competition not only spatially (within a fruit) but also temporally (over the ripening period of a fruit).

We demonstrated that ovipositor length, width, and stoutness differed between L. japonica and G. brasiliensis in southwestern BC. Our findings are consistent with our predictions that G. brasiliensis have significantly shorter and stouter ovipositors than do L. japonica. Additionally, our findings align with those of Fellin et al. (2023) as G. brasiliensis in both studies had shorter ovipositors than L. japonica.

The absolute difference in ovipositor lengths between the G. brasiliensis and L. japonica is most relevant for determining how far their ovipositors could penetrate into substrates harboring their shared hosts, regardless of the intraspecific relationship between hind tibia length and ovipositor length. Our findings align with those in the California olive biocontrol system (Sime et al. 2007; Wang et al. 2008, 2009); in both systems multiple parasitoids attack hosts in the same developmental substrate (Giorgini et al. 2019; Wang et al. 2021) and those parasitoids with broader host ranges have longer ovipositors than do those parasitoids with more restricted host ranges that have shorter ovipositors.

The intraspecific variation in G. brasiliensis and L. japonica ovipositor characteristics is associated with variation in intraspecific body size. This may be driven by environmental factors such as fruit characteristics or rearing temperature. Although fruit type did not appear to influence ovipositor characteristics in our study, fruit characteristics (size, weight, depth, ripeness, etc.) may influence host size (Poças et al. 2022) and host size can affect parasitoid size (Nicol and Mackauer 1999; Harvey et al. 2004). Additionally, temperature variation can cause size variation in insects (Atkinson 1994) and thus could be a source of variation in our results.

The interspecific variation in ovipositor length that we quantified here may be associated with different host searching behaviours (e.g. Compton et al. 2009). The oviposition behaviour of a subset of Leptopilina and Ganaspis species were described by Vet and Bakker (1985); they showed that the Leptopilina species in their study search for hosts using a “walk and probe” method where they walk over likely host substrate while constantly probing it with their ovipositors and rarely stopping. Conversely, Ganaspis species were shown to search for hosts using a “walk then probe” method where they take prolonged walks then stop and stand still for extended time periods before probing (Vet and Bakker 1985). The apparently more methodical host searching technique used by the Ganaspis species, when compared to the Leptopilina species described by Vet and Bakker (1985) may be related to their ability to sense hosts, likely through host movement, within the substrate. Vet and van Alphen (1985) assessed the importance of vibrotaxis (host vibrational cues), ovipositor searching, and antennal searching in a suite of Leptopilina and Ganaspis species and found that the Ganaspis investigated relied entirely on vibrotaxis. Conversely, the Leptopilina species investigated by Vet and van Alphen relied primarily on ovipositor searching, while vibrotaxis and antennal searching were used to limited and varying degrees depending on the Leptopilina species. Assuming that, like its congeners (Vet and van Alphen 1985), G. brasiliensis relies on vibrotaxis to locate high quality (i.e. unparasitized, early instar) hosts in their oviposition substrate, we speculate that G. brasiliensis is less likely to find hosts deeper in their oviposition substrate, so a longer ovipositor may have less utility. Future work should assess the importance of vibrotaxis, ovipositor searching, and antennal searching for L. japonica and G. brasiliensis while searching for D. suzukii larvae.

Leptopilina japonica larvae generally outcompete G. brasiliensis in D. suzukii hosts that have been parasitized by both parasitoids within 24 hours (Wang et al. 2019). Because of this, G. brasiliensis may find competitor free space by attacking D. suzukii larvae that are present in less ripe fruits. A stouter ovipositor may, for example, facilitate oviposition by G. brasiliensis through tougher fruit skins of undamaged fruits as a stouter ovipositor could support puncturing fruit with a higher required penetration force without the ovipositor buckling (Vincent and King 1995; Quicke et al. 1999; Cerkvenik et al. 2017) although Figitids searching for hosts in damaged fruits are known to wander inside of fruits through holes made by other insects (e.g. Guimarães and Zucchi 2004). Abram et al. (2022a) found that parasitism by the parasitoids in southwestern BC typically lagged behind D. suzukii presence in fruits by approximately two weeks, and that L. japonica appeared earlier in the ripening season in collections of R. armeniacus, while G. brasiliensis appeared earlier in the ripening season in collections of S. racemosa. However, Abram et al. (2022a) did not account for variation in fruit ripeness within samples so their dataset is likely inadequate to describe the effect of fruit ripeness on parasitism by either species. In nature, D. suzukii oviposit into and develop from wild fruits of differing size, depth, and ripeness, which can be present simultaneously (e.g. Ulmer et al. 2022).

The relationship between fruit characteristics and the relative abundance of larval parasitoids of D. suzukii has not yet been quantified. The apparent affinity of L. japonica for larger fruits and G. brasiliensis for smaller fruits observed in fruit collections from both China (Giorgini et al. 2019) and BC (Abram et al. 2022a) invites further research. In the laboratory, D. suzukii larvae spend more time diving in fruits than do D. melanogaster Meigen larvae (Kim et al. 2017), which could be a behavioural response by D. suzukii to pressures imposed by the specialist G. brasiliensis and could present an open niche for L. japonica to exploit with its longer ovipositor. Future research should investigate how L. japonica and G. brasiliensis search for host larvae in fruits, the depths at which they parasitize host larvae, and the relationship between parasitoid ovipositor characteristics and fruit characteristics like fruit diameter, depth, and skin thickness. Future work should also focus on the parasitoid communities within D. suzukii infested fruits throughout the fruit’s phenology (from ripening through decomposition) as D. suzukii infestation opens novel niches for other frugivorous insect species (e.g. Chamberlain et al. 2020) and by extension their associated parasitoids.

The narrow host range of G. brasiliensis has led to its recent approval for field release in the United States of America (Beers et al. 2022). Leptopilina japonica, however, has not yet been approved for release in the USA due to its broader host range. This presents an interesting opportunity to identify the realized niche of G. brasiliensis in the presence of the superior intrinsic competitor L. japonica where both species co-occur compared to the realized niche of G. brasiliensis where it is released for biological control. In its introduced range, the poorer intrinsic competitor (s. str. Harvey et al. 2013; Ode et al. 2022) will not be subject to the competitive pressures of L. japonica. This competitive release could impact the parasitoid’s realized niche wherein G. brasiliensis may be better able to make use of its fundamental niche in North America and may result in differing ovipositor characteristics between populations that are and are not in direct competition with L. japonica.

The clear differences in ovipositor length, width, and stoutness between G. brasiliensis and L. japonica demonstrated in our study, and the apparent use of semiochemical cues for competitive avoidance by G. brasiliensis but not L. japonica (Böttinger et al. 2019; Wang et al. 2019) offers an interesting opportunity to investigate the evolutionary drivers of these biological and ecological differences. Such insights are likely important for our understanding of the usefulness of G. brasiliensis and L. japonica for the biological control of D. suzukii globally.

We thank T. Hueppelsheuser, M. Franklin, J. Sherwood, E. Grove, and P. Eraso for their efforts in fruit collection and insect rearing. We thank D. Iritani and Y. Watanabe for their help with microscopy and C.-H. Lue and M. Buffington for their wisdom regarding ovipositor measurements. Finally, we thank C. Cock, T. Nelson, and J. Sherwood for their technical support, D. Ensing for his support with model selection, and W. Wong for his participation in brainstorming ovipositor questions and future research. This research (funding to P.K.A., C.E.M., N.G.E.) is part of Organic Science Cluster 3, led by the Organic Federation of Canada in collaboration with the Organic Agriculture Centre of Canada at Dalhousie University, supported by Agriculture and Agri-Food Canada’s Canadian Agricultural Partnership-AgriScience Program. P.K.A. and C.E.M. were also supported by funding from Agriculture and Agri-Food Canada, A-BASE #2955. N.G.E. was additionally supported by Graduate Student Fellowships from the University of British Columbia-Okanagan.