Among the different stink bug-associated odours tested in the olfactometer, only those from plants bearing an egg mass of H. halys or N. viridula elicited positive attraction in T. mitsukurii females. These results validated our hypothesis that a more in-depth evaluation of the parasitoid host specificity, through chemical ecology investigation, would further restrict the parasitoid host range that was depicted by physiological host range assays. Indeed, several non-coevolved hosts, although accepted under laboratory simplified conditions (i.e., no-choice and paired choice black-box tests), would be hardly located in the field due to lack of suitable host-associated odours. This interpretation conceptually agrees with results from prior studies on T. japonicus, another H. halys exotic biocontrol agent that is currently released in Italy [28]. In fact, T. japonicus displayed a relatively wide physiological host range in laboratory choice and no-choice assays [6,8,29], but a more restricted oligophagy in olfactometer or in field conditions [44,53].
Our results also indicate that soybean plants exposed to H. halys feeding and egg deposition can emit volatile molecules that are detected by the coevolved parasitoid T. mitsukurii. This response is consistent among residence time and first choice data. Volatile emission induced by host oviposition represents an indirect defence for the plants, as already demonstrated in different systems involving Trissolcus egg parasitoids [44,45,48], and is an exceptionally reliable signal of the presence of target hosts in the canopy [9,11]. Induced plant volatiles act on long-distance range and are easily detected by the egg parasitoids, allowing them to rapidly locate the host that is suitable only for a short period of time [45,54]. The ability of T. mitsukurii to exploit induced plant volatiles for locating H. halys, although it is investigated here for the first time, was recently hypothesized following results of open field surveys in northeast Italy or France [41,55]. For instance, higher parasitization of H. halys by T. mitsukurii was detected in naturally-laid eggs compared to sentinel eggs [55]. In other two studies, T. mitsukurii exhibited remarkable discovery efficiency of eggs laid on plant tissues, with 27.3 to 46.5 % of parasitised egg masses [35,41].
Noteworthy is also the positive attraction of T. mitsukurii towards plants bearing eggs of non-coevolved N. viridula. Our result, together with the fact that T. mitsukurii positively responds to tracks of N. viridula females [56], would suggest high ability to locate eggs in the field. Indeed, the parasitoid is considered a main enemy of N. viridula in Japanese areas where the stink bug has established [37]. Surprisingly, despite this remarked discovery ability, the emergence rate displayed by T. mitsukurii was null or very low, in the case of the adventive Italian population [35,43], and moderate (e.g., ~40%) for Japanese strains [57,58]. This mismatch between the positive response of T. mitsukurii to cues associated with N. viridula and the low suitability of this host for parasitoid development is partially consistent with lack of coevolution, as N. viridula is of Ethiopian-South Mediterranean origin [59]. However, this does not explain why the parasitoid responds to cues associated with the novel host. A hypothetical explanation for this could be that N. viridula induces in soybean a defensive response that might be similar to that induced by other herbivores, like Nezara antennata Scott, native to Eastern Asia [51] and listed as host of T. mitsukurii [40]. It is known that N. viridula and N. antennata share common volatile compounds [47], however, whether they also induce similar plant responses is unknown.
Previous choice and no-choice bioassays underlined high acceptance and suitability of D. baccarum for T. mitsukurii [43]. Surprisingly, our data did not reveal any behavioural response of this parasitoid to odours from soybean plants bearing egg masses of D. baccarum. Dolycoris baccarum has a wide distribution in the Paleartic region [51], occurring also in Asia where it is a pest of several crops including soybean [60,61]. Although D. baccarum is listed in the host range of T. mitsukurii [40], there are no quantitative data related to the prevalence rate of this parasitoid. On the other hand, naturally-laid and sentinel egg masses of D. baccarum are highly parasitized in Chinese orchards by other parasitoids, i.e., T. japonicus [29]. Similarly, in Korea and Japan, Trissolcus nigripedius Nakagawa and Telenomus gifuensis Ashmead (both Hymenoptera: Scelionidae) are commonly found parasitizing D. baccarum eggs laid in crop field, including soybean [62], sometimes demonstrating high prevalence [63]. Therefore, while some parasitoid species seem to effectively track host eggs of D. baccarum in the field, the same cannot be demonstrated for T. mitsukurii, suggesting that parasitization by this species might be occasional.
In no-choice black box experiments, E. ventralis eggs represented a poorly suitable host for T. mitsukurii, as parasitoids failed to develop inside its eggs [43]. The lack of response towards odours from plants bearing an egg mass of E. ventralis in our experiments is consistent with the very low suitability of this species [43]. The stink bug is of Paleartic origin and is only marginally present in the native area of T. mitsukurii [52]. Therefore, the risk that under field conditions T. mitsukurii would parasitize E. ventralis appears quite low.
One of the most desired aspects of risk assessment is that the candidate biocontrol agent has no or limited negative effect on beneficials. The fact that plants with A. custos eggs were less preferred by T. mitsukurii females compared to control can be interpreted as a form of repellence towards the treatment odour [19,64] and this may reduce the risk of non-target parasitisation. We may expect that in case of intentional release of T. mitsukurii, the existence of such an ecological barrier would limit encounters with A. custos eggs in the field. The response of T. mitsukurii to A. custos appears at least partly similar to that of T. japonicus, which in no-choice tests parasitized successfully this stink bug predator, but in large cage tests, using plant bearing egg masses, was less preferred compared to H. halys [8,43]. Similarly to herbivorous species, zoophytophagous Heteroptera, including A. custos, can feed on plant tissues to acquire water and nutrients [65]. It was demonstrated that some of these species can induce the activation of defensive signalling pathways in plants, with consequent release of volatile organic compounds, which can inform natural enemies of the ongoing attack [66,67]. Eventually, behavioural responses of natural enemies to such odour souces can vary in different systems. For instance, oviposition by the zoophytophagous Podisus maculiventris (Say) (Hemiptera: Pentatomidae) induces the emission of plant volatiles which attract its coevolved parasitoid Telenomus podisi (Ashmead) (Hymenoptera: Scelionidae) [67]. In another trophic system, oviposition by the predatory P. maculiventris on tomato plants did not attract T. japonicus, although the parasitoid is able to successfully parasitize P. maculiventris eggs and develop inside the host [44,68]. The lack of a coevolutionary history between the two species may explain the inconsistent behaviour of T. japonicus [44]. Arma custos (junior synonym Arma chinensis Fallou [69]) is historically present in T. mitsukurii native area [70], hence we can speculate that the observed avoidance can be interpreted as a coevolutionary adaptation among the tritrophic system, that might prevent the parasitoid to exploit the predator. Although A. custos eggs were highly suitable for T. mitsukurii in no-choice black box tests [43], development of the parasitoid larva is highly risky due to possible predatory (cannibalistic) behaviour of newly hatched A. custos nymphs [70]. Hence, in the first step of the hierarchical process of habitat assessment, the parasitoid could use plant odours to avoid such risky host.
The exploitation of adults-related chemical odours is quite common in egg parasitoids [11,71,72], however in the present study we noticed lack of T. mitsukurii response to females of H. halys. Although it is known that volatile and non-volatile cues from physogastric females may represent reliable information for some Scelionid wasps, kairomones from stink bug females elicit parasitoid responses mainly at short distance [47,73]. For instance, a previous investigation demonstrated that T. japonicus responds to H. halys females only in a “close-distance” olfactometer and not in a “long-distance” olfactometer (as the one we have used here) [44]. In open arena, Trissolcus brochymenae (Ashmead) (Hymenoptera: Scelionidae) responded to Murgantia histrionica (Hahn) (Hemiptera: Pentatomidae) physogastric females but not to males [74]. Concerning T. mitsukurii, female extracts of H. halys seem to elicit a behavioural response of the parasitoid in Petri dish arena, thus under close-distance environment [56].
We did not detect any attraction of T. mitsukurii towards odours from eggs of the tested stink bugs. Kairomones from host eggs are typically present in small amounts, hence their role in host location is mainly expected at short distance (reviewed by [11,75,76]). In olfactometers, even though few species of egg parasitoids (e.g., T. podisi) were demonstrated to respond to odours directly emitted by eggs [77], so far, investigated Trissolcus species did not respond [44] except when a high number of eggs was placed very close to the air hole at the end of the olfactometer arm [74]. Hence, we can hypothesize that such poorly detectable odours would only permit host location at proximity, as shown for T. brochymenae using short-range bioassays in open arenas [74].
In conclusion, while previous host-acceptance investigations revealed a broad host range for T. mitsukurii in Europe [43], present results suggest that host location at long distance would likely favour parasitization of H. halys (or N. viridula), rather than the other stink bugs tested here, in particular the beneficial A. custos. Possibly, the presence of such ecological filter would have positive implications for preventing undesired impact on non-targets in case of intentional release of the parasitoid in biological control programs. On the other hand, there are no physiological impediments for T. mitsukurii to develop in some non-targets (e.g., A. custos), hence when such species co-occur in the field with H. halys, they would likely be more exposed to parasitization, due to occasional encounters during foraging. Hence, the bioassays conducted here with T. mitsukurii following the host range studies [43], highlight the need of a multidisciplinary approach in pre-release risk assessment, where results from each step are part of the puzzle that will allow a reliable field scenario prediction. Dedicated field surveys of native and exotic stink bug eggs in those areas where T. mitsukurii have fortuitously established would likely help in estimating the relevance of such host-parasitoid interactions and provide better support for the definition of a risk assessment document, necessary for licensing parasitoid releases.