Discussion
Infochemical cues play an important and crucial role in parasitoid host location and selection (Tumlinson, Lewis, et al., 1993; Vet & Dicke, 1992; Vinson, 1976). Vet and Dicke (1992) framed the use of these cues by parasitoids as a tradeoff between reliability and detectability, which they called the reliability-detectability problem. Host infochemical cues were hypothesized to be a highly reliable indicator of a host’s presence, but due to their localized nature would not be detectable over long distances. Conversely, infochemical cues from the plant were hypothesized to be highly detectable but poor indicators of host presence and thus unreliable, unless they were induced by herbivory. Vet and Dicke (1992) also hypothesized that when a parasitoid that had one or most hosts that fed on a single plant species, plant infochemicals could become both reliable and detectable. In a third scenario in which the parasitoid specializes on a host species that feeds upon multiple plant species, host infochemical cues were hypothesized to be highly reliable while herbivore induced synomones were not. Here, however, the synomones from each plant species could be learned and associated with the host, thus compensating for the low detectability of the host kairomones. Finally in a fourth scenario , in which both the parasitoid and its hosts were extreme generalists, Vet and Dicke (1992) hypothesized that infochemical cues would not be used. Although Vet and Dicke (1992) acknowledged each of these four scenarios were extremes on a continuum, and that intermediates could be found in a nature, they did not specifically discuss cases that involved oligophagous parasitoids and their hosts. This is the situation forEadya parasitoids, which attack multiple related beetle hosts that feed on multiple related Eucalyptus plants. Thus, the question remains as to how these parasitoids might use infochemicals to locate their hosts.
For parasitoids that specialize on one host and rely heavily on host infochemical cues, we would expect to see some degree of coevolution leading to cospeciation between parasitoids and their hosts (Page, 2003). As Eadya are host-flexible (Peixoto et al., 2018), there would be less expectation of coevolution and cospeciation with their beetle hosts. Indeed, our data did not support cospeciation in the event-based analysis, but rather provided robust evidence for duplication and host-switching with duplication being the dominant type of evolutionary event. Although the distance-based analysis demonstrated strong congruence between parasitoid and beetle phylogenies indicating cospeciation, further exploration of the data suggested that cophylogenetic signal was restricted to the interactions betweenE. annleckieae and Pst. cloelia (Hutchinson et al., 2017). This large disparity in cophylogentic signal between E. annleckieae and the rest of Eadya can be attributed to sampling error. Peixoto et al. (2018) reared E. annleckieae from P. charybdis, Pst. cloelia, Pst. selmani, and Pst. nobilitata(Erichson, 1842) but DNA was only successfully extracted for Pst. cloelia. As a phylogram is required to calculate the distance matrices for the PACo distance-based analysis, only Pst. cloelia could be included, creating a false one-to-one relationship between E. annleckieae and Pst. cloelia . Thus, E. annleckieae is the only parasitoid species with a single host association in the distance-based analysis, which most likely biased the PACo analysis. Although we cannot rule out the use of host infochemical cues inEadya as these were not directly tested here, the lack of evidence for cospeciation between these parasitoids and their beetle hosts as well as their known host flexibility, suggests these cues are unlikely to be the dominant driver in the location and selection of a suitable host.
Although Vet and Dicke (1992) discussed plant synomones and hypothesized how the compounds are used in a tritrophic context, this was only considered from a bottom-up perspective in which the herbivore host breadth solely influences whether plant infochemicals could reliably be used by the parasitoid to locate its herbivore host. For parasitoids that specialize on a single host, the low reliability of plant cues discussed in Vet and Dicke (1992) likely holds true, as in most cases these signals cannot guarantee the presence of a suitable host unless they are herbivore-induced volatiles (de Moraes et al., 1998). However, this may not be the case for less specialized parasitoids that can survive on a wider range of hosts, which in turn utilize a range of closely related plant species. Here, plant infochemicals, even if not herbivore-induced, may be both reliable and detectable for parasitoids if they specialize on a single host tree species, or a distinct infochemical signature, as long as they can utilize multiple hosts associated with that species/infochemical signature. This may allow parasitoids to reduce competition with related parasitoid species through plant specialization rather than host specialization. Here we find evidence that Eucalyptus infochemicals are species-specific, corresponding to specific species of Eadya and their beetle hosts (Fig. 5A & D). However, since we found no evidence of cospeciation between the parasitoids and their hosts we suggest that host selection in Eadya is influenced by Eucalyptus phytochemistry irrespective of the specific beetle host used. Although, several limitations to our study restrict us from considering this conclusion definitive. 1) Euc. globulus was the intended target plant of this study, with Euc. sp. 1 being a volunteer in a single plantation, and therefore was not widely sampled. 2) Sampling was limited to a small area in western Victoria and does not represent the endemic range of Euc. globulus nor that of the beetle or parasitoid species. 3) As only juvenile Euc. globulus plantations were targeted, the full range of Eucalyptus utilized byEadya is still unknown.
We identified a diverse array of plant metabolites that could accurately distinguish between the two groups of Eadya (Figure 5D, Table S3). Among top contributing compounds to axes 1 & 2, we identified two general groups of compounds: (1) fatty acid derivatives (Docosanoic acid, phenylmethyl ester (C106), 4H-1-Benzopyran-4-one, 5-hydroxy-7-methoxy-2-(4-methoxyphenyl)-6,8-dimethyl- (C105), and Octadecanoic acid, 2,3-dihydroxypropyl ester) and (2) compounds with known larvicide activity (agarospirol, ascaridole epoxide, alpha-pinene, alpha-phellandrene, phytol, and eucalyptol) (de Castro et al., 2016; Johnson & Singh, 2017; Kaczmarek, Wrońska, Kazek, & Boguś, 2020; Mohamed & Jong, 2014)
Docosanoic acid has the potential to function as either a kairomone or synomone in both insects and plants. Docosanoic acid is a derivative ofEucalyptus cuticular wax, a known Coleopteran attractant, and changes in Docosanoic acid correspond with larval maturation in certain insects (Courtney, Lassak, & Speirs, 1983; Gosney et al., 2016; Kaczmarek et al., 2020; Sarkar & Barik, 2015). Docosanoic acid is commercially manufactured and presents as a strong candidate for attractants for Eadya in New Zealand P. charybdisbiological control efforts (T. Withers, Todoroki, Allen, Pugh, & Gresham, 2019) with low potential for toxic effects on non-target insects (Kaczmarek et al., 2020; Sarkar & Barik, 2015). Although we can discern which Eucalyptus infochemicals are associated with host use and that compounds can separate the two groups of Eadya , future work should aim at directly testing for attractant responses across the genus Eadya . Further work to identify compounds that attract host flexible parasitoids should consider volatile compounds that are present in both the host and associated plant species, with foci on fatty acid derivatives based on this study’s results.
We did not find evidence for an herbivore-induced response, as no overall pattern could be discerned between herbivore damaged and undamaged E. globulus leaves from the PCA (Fig. 5B). However, these inconclusive results may be due to the methodology of our study. As herbivore damaged leaves were collected from the field, it was not possible to determine when the leaves were damaged. The production of volatile compounds is metabolically expensive (Cipollini, Walters, & Voelckel, 2018) and is likely not maintained indefinitely after herbivore damage. Based on our data, we were unable to determine whether herbivore damage induced the production of volatile compounds inEuc. globulus , but cannot rule out the possibility until a study can be undertaken using methodology better suited for this question (Materić et al., 2015).
With this evidence we propose an amendment to Vet and Dicke’s (1992) reliability-detectability hypothesis to include how infochemicals are used when oligophagous parasitoids utilize oligophagous hosts. In this ecological context, we expect little to no response to host kairomones and no evidence of cospeciation between the parasitoid and their hosts, but a strong response to plant synomones and potentially herbivore-induced plant synomones. Although host kairomones are the most reliable indicators of host presence, especially for specialists, plant synonomes, whether induced or not, may play a larger role for oligophagous parasitoids. The ability to successfully utilize multiple hosts is an adaptive advantage in instances when hosts have a patchy distribution or small population size, or when multiple parasitoids are competing for a limited number of hosts (Price, 1971). Further, if hosts feed on multiple plants, specializing on one plant where multiple plant species are present may limit competition between parastioids.Eadya annleckieae and E. daenerys ’ wide niche breadth and preference for hosts that feed on plants with different chemoprofiles may aid in reducing competition between the two species. This study greatly expands our understanding of the complex interactions within the tri-trophic Eadya system and how they utilize plant infochemicals for host location and potentially to limit intra-specific competition. To further understand these interactions a more robust taxon sampling ofEucalyptus and Eadya across their native range is required to definitively establish the adaptive advantages of host flexibility and the use of plant synomones for host location in similar tri-trophic systems. Subsequently no-choice competition assays between Eadyaspecies using distinct infochemical signatures should be conducted under a controlled environment to establish the competitive advantages of plant specialization.