Changing Host Plants Causes Structural Differences in the Parasitoid Complex of the Monophagous Moth Yponomeuta evonymella, but Does Not Improve Survival Rate

Recently in Poland, cases of host expansion have frequently been observed in the typically monophagous bird-cherry ermine moth (Yponomeuta evonymella), which has moved from its native host plant, bird cherry (Prunus padus), to a new, widely distributed plant that is invasive in Europe, black cherry (P. serotina). We attempted to verify the reasons behind this host change in the context of the enemy-free space hypothesis by focusing on parasitoids attacking larval Y. evonymella on one of three host plant variants: The primary host, P. padus; initially P. padus and later P. serotina (P. padus/P. serotina); or the new host, P. serotina. This experiment investigated if changing the host plant could be beneficial to Y. evonymella in terms of escaping from harmful parasitoids and improving survival rate. We identified nine species of parasitoids that attack larval Y. evonymella, and we found that the number of parasitoid species showed a downward trend from the primary host plant to the P. padus/P. serotina combination to the new host plant alone. We observed a significant difference among variants in relation to the percentage of cocoons killed by specific parasitoids, but no effects of non-specific parasitoids or other factors. Total mortality did not significantly differ (ca. 37%) among larval rearing variants. Changing the host plant caused differences in the structure of the parasitoid complex of Y. evonymella but did not improve its survival rate. This study does not indicate that the host expansion of Y. evonymella is associated with the enemy-free space hypothesis; we therefore discuss alternative scenarios that may be more likely.


Introduction
Plants and the herbivores that feed on their tissues and sap have coevolved, creating a relationship wherein plants produce new defences and herbivores attempt to mitigate those defences to further feed on the plants. Generalists, which include polyphagous and oligophagous insects, feed on a wide range of hosts because they are not specialised to feed on a particular range of plants, whereas specialists are insects that are monophagous. When choosing their host, monophagous insects search for the best food quality and the best environment in terms of the survival of their offspring (see diversity of natural enemies of European species of the genus Yponomeuta provides information concerning 60 species [30]. The main parasitoids of Y. evonymella in Europe are Ageniaspis fuscicollis Dalman (Hymenoptera; Encyrtidae), and Diadegma armillata Grav. (Hymenoptera; Ichneumonidae), whereas in Korea the most important parasitoid species are D. armillata, Herpestomus brunnicornis Grav. (Hymenoptera; Ichneumonidae), and Zenillia dolosa Meig. (Diptera; Tachinidae) [29].
In this study, we investigated the parasitoid complex of Y. evonymella in light of its recently discovered colonisation of a new host plant. The aim of this study was to determine if larval stage-specific parasitism is the main reason for the observed change of host plant exhibited by Y. evonymella. Employing an experiment based on three variants of larval feeding-on the primary host P. padus, partly on P. padus and later on P. serotina (P. padus/P. serotina), or on the new host P. serotina-we attempted to explain this larval behaviour by posing a question: Is the observed host expansion of Y. evonymella associated with the enemy-free space hypothesis-does Y. evonymella feed on P. serotina to avoid parasitoids?

Insects
This study was carried out on a herbivorous insect, the bird-cherry ermine moth Y. evonymella (L.) (Lepidoptera; Yponomeutidae), which primarily feeds on bird cherry P. padus L. [19] but is lately more often found on invasive non-native black cherry P. serotina Ehrh. [22]. During this study, we used larvae, pupae, and adults.

Site and Plant Material
Larval stage-specific parasitism of Y. evonymella was studied in Poland in 2014 in Kobylepole Forest (Babki Forest District; 52 • 38'61" N and 17 • 04'54" E). The research was carried out on two species of undergrowth shrubs: Black cherry (P. serotina Ehrh.; non-native species in Europe) and bird cherry (P. padus L.; native species in Europe). Our previous field investigations conducted in Kobylepole Forest were indicative of the abundant presence of both species of Prunus in the forest's undergrowth layer. Both investigated understory shrubs grow in mid-fertile soil under the canopy of coniferous forests and mixed forests consisting of various tree species (mainly Pinus sylvestris L. mixed with Quercus robur L.). In the late summer of 2013, 24 typical shrubs (12 of P. padus and 12 of P. serotina) growing in temperate shade (under the tree canopy; approximately 50% full sunlight) were randomly chosen so that their canopy dimensions, ages, and heights (3-5 m) were similar.

Study Design
To identify and estimate the number of natural parasitoids of Y. evonymella in this field experiment, we used larvae in the L1 stage of development (before their emergence from under the domed refuges covering egg batches), which we collected in the autumn of 2013, because searching for larvae was simpler in the leafless period. Twenty-four shoots (10-15 cm) with egg batches collected from shrubs of P. padus growing in Kobylepole Forest were overwintered in external controlled conditions by placing the shoots vertically in a pot filled with sand and covering them with a plastic net that guaranteed good ventilation and protection from insectivorous birds. In the spring, the collected shoots containing egg batches were attached randomly with coated wire to places on shrubs of P. padus or P. serotina where eggs are usually laid (short lateral shoots), at the time when larvae of this species begin to feed (mid-April 2014). Moreover, during a field inventory process held in Kobylepole Forest (beginning of May 2014), we marked 12 places where we noticed that larval Y. evonymella that had originally been grazing on P. padus had spontaneously moved to shrubs of P. serotina in their close vicinity. In summary, we created three different variants of larval rearing: On the primary host P. padus, initially on P. padus and later on P. serotina (P. padus/P. serotina), and on the new host P. serotina alone.
When the first larvae began to pupate (beginning of June), the pupae and individual larvae of the last stadium (5th instar) were collected and moved to the laboratory. Shoots with aggregates of pupae were cut out of the shrubs and placed separately in one-litre plastic containers. Larvae that did not manage to pupate in the field had access to leaves of the proper shrub species. The vast majority pupated within 24 h, although some larvae did not pupate in the entire time of the experiment, which is a peculiar defence mechanism of this species, as some larvae entwine their cocoons with those of the colony to the end of their life in an effort to protect themselves. As the containers were filled with collected material, we placed them in a room with stable temperature and light conditions (22 • C, day:night cycle of 16 h:8 h). Every container was checked daily to observe and record not only the emergence of mature forms of Y. evonymella individuals but also the emergence of parasitoids. Moths that had just emerged were removed the same day, sex was determined, and further tests-irrelevant to the current study-were conducted. The parasitoids that emerged were kept for further identification. We also determined that there were a number of cocoons from which moths did not emerge because of other factors (e.g., a moth being blocked in the process of leaving the pupa, viruses, and pathogens).
When we observed that Dipteran larvae emerged from pupal Y. evonymella, these pupae were moved to another separate vessel to avoid accidental infection of healthy pupae. Similarly, when polyembrionic A. fuscicollis was observed, pupae with visible marks showing emergence of this parasitoid were taken out of the container, but they were not included in this study because they are egg parasitoids.

Statistical Analyses
The percentage of parasitism of each parasitoid species was calculated by dividing the number of individual pupae killed by the parasitoid by the number of pupae collected for the host variant in nature. The average parasitism of each parasitoid species was calculated by averaging the percentage of parasitism of each parasitoid species for each host-plant variant. The total mortality was obtained by adding the percentage of parasitism of each parasitoid species to the percentage of pupae killed by other factors. To evaluate the importance of each parasitoid, we identified significant differences in percentage of parasitism using the Kruskal-Wallis Test and Dunn's Test (post-hoc; p < 0.05). The average total parasitism was calculated by adding the percentage of parasitism of each parasitoid. A standard deviation (SD) was given for each mean value. All calculations were performed using JMP Pro 13.0 software (SAS Institute Inc., Cary, NC, USA).

Results
The structure of the Y. evonymella parasitoid complex and the richness of the various species were analysed in three different variants of larval rearing. In this study, we found and identified nine species of parasitoids attacking larval Y. evonymella ( Table 1). The main group of insects parasitizing Y. evonymella are individuals from the order Hymenoptera, which caused more mortality than those from Diptera. We found the following parasitic species: Bactromyia aurulenta (Meig.) (Diptera; Tachinidae) and Baryscapus evonymellae (Bché.) (Hymenoptera; Eulophidae), as well as six taxa from the family Ichneumonidae (Hymenoptera), including Agrypon anxium (Wesm.), D. armillata (Grav.), H. brunnicornis (Grav.), Itoplectis maculator (F.), Lissonota sp. Grav., Dirophanes invisor (Thunb.), and Pimpla turionellae L. Table 1. Percentage of parasitism by natural enemies reared from pupal Yponomeuta evonymella, which were in turn reared on Prunus padus, P. serotina or a mix of both P. padus and P. serotina shrubs (see Section 2: Materials and Methods). Means in the same row followed by different letters indicate significant differences based on Dunn's test (p < 0.05). Bold values indicate p < 0.05. A standard deviation (SD) is given for each mean value.

Number of Parasitoids
Percentage of Parasitism (SD) The composition of parasitoid species was more diverse when larvae fed on P. padus than when they fed on P. serotina. There was a downward trend in the number of parasitoid species present, from the primary host plant variant of larval rearing (P. padus) to the partial variant (P. padus/P. serotina) to the new plant (P. serotina). All nine parasitoid species found were present in samples reared on P. padus, seven were present when larvae were reared partly on P. padus and P. serotina, and five were present when larvae were reared on P. serotina alone. The total mortality for each larval rearing variant, however, was not significantly different, with ca. 37% mortality among larvae feeding on P. padus, 32% among those on the P. padus/P. serotina variant and 38% among those on the P. serotina variant (χ 2 = 0.42; p = 0.81). On larvae reared first on P. padus and then P. serotina, B. evonymellae and P. turionellae were absent. In turn, in larval samples reared only on P. serotina, B. aurulenta, H. brunnicornis, I. maculator, and D. invisor were absent. Bactromyia aurulenta and B. evonymellae were the most abundant general parasitoid on both larvae raised on only P. padus and those raised on a mix of P. padus/P. serotina, whereas in the case of larvae raised on P. serotina alone, the most common parasitoid was A. anxium ( Table 1).
The total percentages of parasite-caused mortality (χ 2 = 2.99; p = 0.22) and mortality due to other non-studied factors (e.g., entomopathogenic fungi and damage by predators) for each larval rearing variant were not significantly different (Table 1). Based on the biology and ecology of the studied parasitoids, we divided them into one group of parasites more specific to the genus Yponomeuta Latreille (including mono-and oligophagous) and another group of non-specific parasitoid species (polyphagous). The specific parasitoid group included B. evonymellae and H. brunnicornis. All other parasitic species found were placed in the non-specific group. A closer look at the results ( Figure 1) showed a statistically significant difference among feeding variants when only regarding the percentage of cocoons killed by specific parasitoids. In comparison, the percentage of cocoons killed by specific parasitoids in the P. serotina variant ("b"; Dunn's test) was lower than the percentage killed in the P. padus alone and the P. padus/P. serotina variants (both "a"). Furthermore, there were no differences among variants regarding the percentage of cocoons killed by non-specific parasitoids or other factors or regarding the number of non-parasitised cocoons.

Host Plant Changes Exhibited by Y. evonymella
In our study, we investigated the connections among a host plant, the monophagous bird-cherry ermine moth (Y. evonymella), and the composition of the moth's parasitoid complex, depending on the larval-host plant variant. We were searching for an explanation of a phenomenon that has been observed in nature over the past ten years: Y. evonymella more and more frequently utilising a new plant, P. serotina [22], a non-native and invasive species [24,31]. Generally, Y. evonymella hatches and feeds on P. padus, but it often also moves to the shoots of P. serotina shrubs, where it may then feed and pupate. A recent study confirmed that Y. evonymella egg clusters are also observed on the stems of P. serotina [23]. In the current study, we determined that utilising this new host does not benefit Y. evonymella in terms of reduced larval stage-specific parasitism. Thus, our study does not support our prediction that larval Y. evonymella feed on P. serotina solely to avoid parasitoids and predators. In a broader sense, the observed host expansion of Y. evonymella does not appear to be associated with the enemy-free space hypothesis [7]. Most likely, the main benefit for Y. evonymella is simply the expansion of its food base.
As insects are strongly dependent on their preferred food [32,33], feeding on the optimal host plant allows them to obtain a larger body size [34]. Choosing a suboptimal host plant could result in a reduction in body mass [3,9,35] and may extend larval development time, thereby increasing larval mortality due to extended exposure to predators and parasitoids [10,12]. In the case of Y. evonymella, however, larvae can feed and complete their development not only on native P. padus leaves but also on P. serotina with no apparent negative effects on growth and development. Leaves of the black cherry are not qualitatively worse as a food source than leaves of the bird cherry [22]. Taking into consideration evolutionary history, which predicts plant defence against a pest, we do not expect that P. serotina does not have pre-adapted defences [15,16]. It cannot be ruled out, however, that P. serotina is a pre-adapted optimal host due to the observed lack of significant differences in the defensive response among hosts [22]. As the defence and attractant signalling of P. serotina can be recognized as those of a host by female moths, P. serotina may already be considered optimal or nearly so, possibly due to a close evolutionary relationship between both shrub species (although the hosts come from different continents). Because feeding on P. serotina does not affect the duration of larval development [27], there is no considerable prolongation of the period that could result in a larger risk from natural enemies [10,36]. Taking the abovementioned results into consideration, the lack of difference in body mass-as well as in duration of larval development-harmonises with the fact that no significant differences in levels of parasitism among feeding variants were found in the current study. The previously observed longer period over which adults continue to emerge from pupae could, however, affect pupal-specific parasitism [22,37], although this possibility cannot yet be supported because pupal-specific parasitism has not yet been studied.
Based on the literature, the impact of alien plants on insects and their host selection preferences is well known, as can be evidenced by the large number of publications in this field [17,[38][39][40][41][42][43]. In our research, we determined that the observed host expansion of Y. evonymella does not appear to be related to an attempt to escape from parasites, which makes alternative scenarios more likely. In situations where the optimal host is a rare species or there is heavy competition for food, the cost of feeding on the optimal host may be large enough for an insect to promote the strategy of increasing the number of potential host plants through the use of new species [4]. The inclusion of a new host plant undertaken by Y. evonymella may also restrain inter-and intraspecific competition [44]. Prunus serotine, in its European range of occurrence (in contrast to P. padus), is attacked by a smaller number of herbivorous species, as evidenced by a lower level of damage to its leaves [45]. Larval movement from P. padus (where they hatch) onto P. serotina during the early stages of larval development is often observed due to extensive defoliation of P. padus shrubs. This presents an opportunity for the moth population to avoid starvation and competition with other folivorous species. Utilising P. serotina may also be beneficial for Y. evonymella as an opportunity to improve geographical and ecological expansion, as it is known that a wider diet allows herbivores to broaden their range [46], and some herbivores benefit from the establishment and spread of exotic plants because these plants increase the amount of resources available for use [47]. In the state forests of Poland, where we first observed larval feeding on P. serotina, invasive black cherry is a widespread species [48]. Simultaneously, native P. padus is an increasingly rare species and is much more demanding in terms of soil humidity and nutrient requirements [49]. In summary, it is likely true that a noticeable abundance of P. serotina, in comparison to the more outcompeted main host plant P. padus, provides an alternative food source in the case of the total defoliation or disappearance of the primary host plant.

Mortality of Larvae and Structure of Parasitoid Complex
Yponomeuta evonymella is a species with a large range of occurrence (from Europe to East Asia), and its parasitoid complex appears to differ among geographic regions. Observations in Scandinavia have reported that Yponomeuta populations display regular cycles of intensification in occurrence, with strong gradations [18]. The collapse of a mass outbreak is typically caused mainly by parasitoids [50]. The transitional periods between outbreaks are characterised by approximately 50% parasitisation of the population in Central Europe [21] and approximately 30% in East Asian populations [29]. The results of our research show that a Polish population of Y. evonymella was in a transitional period between outbreaks, as indicated by the observed total larval parasitism (ca. 37%). The nine identified species of parasitoids found attacking our studied population of Y. evonymella represent ca. 15% of all species currently known to attack Y. evonymella [30].
Our study showed that, when rearing larvae on P. padus alone or on a mix of P. padus and P. serotina, the most important species of parasitoid are B. aurulenta and B. evonymellae. In contrast, when larvae fed on only P. serotina, the most common general parasitoid was A. anxium.
Agrypon anxium (syn. Anomalon anxium) is a polyphagous larval/pupal parasitoid. In the past, it did not have a meaningful influence on the population of Y. evonymella in Poland [51] and induced parasitism rates of less than 0.5% in Asian populations [29]. In the current study, mean parasitism by this parasitoid was visibly greatest in larvae reared on P. serotina (more than 15-fold higher than in the P. padus variant). The possible reason for such a high percentage of parasitism in this variant is puzzling. As A. anxium is a polyphagous species, and many other species of herbivorous insects have adapted to grazing on the leaves of the alien P. serotina [9,52,53], it cannot be excluded that A. anxium is the main specialist parasitoid of any of these insect species.
Herpestomus brunnicornis is an endophagous parasitoid of Yponomeuta species, attacking mainly fifth instar, prepupal, or pupal hosts [54], although it also appears to attack the younger host stages (mid-instar host larvae) of Y. malinellus Zell. [37]. Herpestomus brunnicornis is widely distributed in Europe and causes low levels of parasitism in Y. evonymella [18]. In the current study, this parasitoid exhibited similar parasitism levels (ca. 2.3%) in larvae feeding on P. padus alone and those feeding on P. padus/P. serotina, but it was absent in larvae feeding on P. serotina alone. In a study [29] conducted in Korea, H. brunnicornis caused relatively high parasitism (6.8%) of pupal Y. evonymella, but it only caused 0.3% of parasitism in larvae.
Baryscapus evonymellae (syn. Tetrastichus evonymellae) is distributed in Nearctic and Palearctic regions. It is an important endoparasitoid of Yponomeuta spp. [51,55] and can be called a hyperparasitoid because it is also an endoparasitoid of D. armillata (Hymenoptera, Ichneumonidae; [56]. Diadegma armillata is a polyphagous, abundant, and widespread endoparasitoid that is a major parasite of Yponomeutidae in Europe [18,57], although it also uses other Lepidoptera (e.g., Tortricidae, Pyralidae, and Coleophoridae). It attacks mainly larvae and occasionally pupae, and it is commonly a victim of hyperparasitism, which lowers its parasitism efficiency [56]. In related research by Lee and Pemberton [29], the levels of parasitism in Y. evonymella on P. padus were low (average 2.8% for larvae). We observed a greater share of this species in relation to total parasitism, as well as a downward trend-although insignificant-in parasitism when the food plant shifted from the original to the new host.
Itoplectis maculator (syn. Ichneumonid maculator) is a polyphagous endoparasitoid that typically attacks different small members of Lepidoptera (e.g., Operophtera brumata L.), Coleoptera, Hymenoptera and Diptera [58]. It is distributed across Europe and North Africa, and it has been introduced in North America [59]. In the current study, this species was not observed in the parasitoid complex of larvae that were reared on P. serotina.
All species in the genus Lissonota are oligophagous, koinobiont endoparasitoids of larval Lepidopterans [60]. These parasitoids are distributed world-wide, but the greatest diversity of Lissonota species occurs in the Old World, where the taxonomy of this genus has been comparatively better studied [61].
The tachinid B. aurulenta is a polyphagous, larval parasitoid that attacks many species in the genus Yponomeuta, but it has also been recorded on 19 other species [51]. In a Korean study, B. aurulenta caused low parasitism (averaging 1%) in larval Y. evonymella on P. padus [29], and, in Poland, it historically has no major impact on mortality [51]. In our study, the effect of this species in larvae reared on P. padus alone (ca. 6.9%) and on a mix of P. padus/P. serotina (ca. 12%) was meaningful; however, in the P. serotina alone variant, the effect was absent.

Study Limitation
The great impact of a specific site on the level of parasitism and the species composition of parasitoids is well known [64]. The low species richness of the Y. evonymella parasitoids (nine species) observed in this study may be caused by the relatively small research area in an urban impact zone, as well as the fact that individuals were collected in a commercial forest stand with a simplified species structure of trees and shrubs. Despite this limitation, we believe that we correctly selected our research area because it is one of the places where this natural ecological process of host plant change from P. padus to P. serotina was first observed in Y. evonymella, and relatively long ago [22,27].
Furthermore, in one of the three years of previous experiments [22], we found a greater effect of parasitoids on mortality and a significant influence of the larval host plant on the level of adult eclosion, as compared to the present study. In the two remaining years of research, no effect of host species was found, and the percentage of mortality was similar to the results of the present study (approximately 40%). Transitional periods between Y. evonymella outbreaks are characterised by approximately 50% parasitisation of the population in Central Europe [21] and approximately 30% in East Asian populations [29], and our results do not diverge from those findings.
Finally, we would like to note that we were aware that it would be very difficult to perform an experiment to additionally capture the seasonal phenology and stage-specific parasitism of Y. evonymellus. We made a decision to collect all the necessary egg batches from P. padus at the stage before the winter diapause, into which the L1 larvae enter. The search for egg batches on P. padus is much simpler because there are one hundred times more eggs on its branches than on those of P. serotina. In the pilot study, we were able to find only three egg batches (over several days of searching), which was too small a sample size. With this study, we faced a classic trade off: We could perform an experiment by collecting egg batches from P. padus, but we had to remain aware that we would then not be able to understand the variability among all stages of development (and results could potentially be biased, as in the case of the egg-larval parasitoid A. fuscicollis). Thus, when polyembrionic A. fuscicollis was observed, pupae with visible marks showing emergence of this parasitoid were taken out of the container, and they were not included in this study (in the analyses).

Conclusions
In summary, changing the host plant caused differences in the structure of the parasitoid complex of Y. evonymella but did not improve its survival rate. This study indicates that the host expansion exhibited by Y. evonymella is not associated with the enemy-free space hypothesis. We identified nine species of parasitoids that attack larval Y. evonymella and found that the number of parasitoid species trended downwards from the primary host plant to the P. padus/P. serotina combination to the new host plant alone. We observed a significant difference among variants only in relation to the percentage of cocoons killed by specific parasitoids, but no effects of non-specific parasitoids or other factors were observed. Total mortality did not significantly differ (ca. 37%) among larval rearing variants.