Wolbachia increases the susceptibility of a parasitoid wasp to hyperparasitism

Heritable endosymbiotic bacteria are extremely widespread among insects, and their presence may have an important impact on their host ecology and evolution. The symbiotic bacterium Wolbachia benefits from strategies that increase the number of infected individuals in the host population, especially the number of females that pass the bacterium on to their offspring. There are many mechanisms by which Wolbachia enhances its transmission through generations, including parasitic phenotypes that manipulate the host reproductive system by inducing cytoplasmic incompatibility, male-killing, feminization or parthenogenesis (O'Neill et al., 1997). The study of the population dynamics of the wRi strain infecting Californian populations of the fruit fly Drosophila simulans (Turelli and Hoffmann, 1991, 1995) provides a classic example of the successful and rapid spread of such manipulative Wolbachia.

Other Wolbachia strains are mutualistic, boosting their host fitness by, for example, improving the host’s ability to overcome stress due to environmental pressures or poor diet (Zug and Hammerstein, 2015). More recently, Wolbachia has attracted wide interest for its ability to increase host resistance to parasite and pathogen infection. Studies have found Wolbachia-infected Drosophila to be more resistant to viral (Hedges et al., 2008; Teixeira et al., 2008) and bacterial infections (Ye et al., 2013), or parasitoid attacks (Hsiao, 1996) than their Wolbachia-free counterparts. The ability to improve host resistance is, however, not pervasive across all host–Wolbachia–parasite interactions. Thus, the presence of Wolbachia in D. simulans did not always improve the flies' resistance to the fungal pathogen Beauveria bassiana (Fytrou et al., 2006), diverse viruses (Martinez et al., 2014; Osborne et al., 2009) or bacteria (Wong et al., 2011). Furthermore, although Martinez et al. (2012) found no effect of wRi in flies parasitized by the parasitoid wasp Leptopilina boulardi, Fytrou et al. (2006) showed that the same endosymbiotic bacterial strain increased the susceptibility of the flies to the closely related parasitoid wasp L. heterotoma. Thus, the role of Wolbachia in the susceptibility to other parasites appears extremely variable between host–Wolbachia–parasite systems, and may depend on the Wolbachia strain and the host genotype or species (Bordenstein et al., 2003; Hornett et al., 2008).

Current research mainly focuses on fly and mosquito (Diptera) host species, because of the utility of Drosophila as a model system and the potential for using Wolbachia in the control of vector-borne diseases of concern to humans. These studies are mostly laboratory based, with just a few using natural host populations (Skelton et al., 2016; Zele et al., 2014), and only some with parasitoids rather than pathogen infection (Fytrou et al., 2006; Hsiao, 1996; Martinez et al., 2012; Xie et al., 2014). In order to understand the complex role of endosymbionts, such studies should also be conducted under natural conditions, and in a broad range of host taxa. There have been just a few isolated studies of Wolbachia–pathogen interactions outside of Diptera (Isopods: Braquart-Varnier et al., 2015; Lepidoptera: Graham et al., 2012; Coleoptera: Hsiao, 1996). Our study is the first exploration of the effect of Wolbachia on the relationship between a Hymenoptera host and parasite in a natural population.

We present an analysis of the association of Wolbachia with its host, the parasitoid wasp Hyposoter horticola (Gravenhorst) (Hymenoptera: Ichneumonidae: Campoplaginae). This wasp is a specialist parasitoid of the Glanville fritillary butterfly, Melitaea cinxia (L.) (Lepidoptera: Nymphalidae) (Shaw et al., 2009). The butterfly is widespread across Eurasia. The study area, Åland, is a Finnish archipelago in the Baltic Sea, where the butterfly lives as a classical metapopulation in a 50 by 70 km fragmented landscape (Hanski, 2011). The wasp occupies the entire host metapopulation (Couchoux et al., 2016). About half of the wasp population is infected by a unique Wolbachia strain, wHho (Duplouy et al., 2015). It is not clear yet how the bacterium is maintained throughout the wasp population. Duplouy et al. (2015) have shown that the transmission rate of the bacterium is high but not perfect. The bacterium has no apparent effect on egg-load, longevity and metabolism of the host wasp, and the sex ratio of the wasp population is not female-skewed. We investigated the potential effect of wHho on the interaction between the wasp H. horticola and its specialist hyperparasitoid Mesochorus cf. stigmaticus (Brischke) (Hymenoptera: Ichneumonidae: Mesochorinae). There is a negative correlation between the prevalence of wHho-infected wasps and the prevalence of hyperparasitism in the landscape in Åland (Duplouy et al., 2015). This pattern of association, if causal, could arise in two ways (Fig. 1). (i) Wolbachia may increase host resistance to hyperparasitism, leading to a low density of the specialist hyperparasitoid where Wolbachia is common. As the hyperparasitoid is present at some density throughout the landscape, this would suggest that while beneficial with respect to hyperparasitism, the Wolbachia infection should have other costs. (ii) Wolbachia may decrease host resistance because individuals that are hyperparasitized do not survive to transmit the symbiont. Therefore, the frequency of Wolbachia-infected individuals would be low where the hyperparasitoid is common. Under this scenario, Wolbachia infection should be beneficial in some way that counterbalances the cost of increased susceptibility to parasitism. We conducted a manipulative field experiment to distinguish between these alternative hypotheses under natural conditions.

Hyposoter horticola is a specialist solitary egg–larval parasitoid of the Glanville fritillary butterfly, M. cinxia (Lei et al., 1997; Shaw et al., 2009). The univoltine host butterfly lays eggs in clutches on host food plants in June. The caterpillars live in gregarious family groups, overwintering in silken nests (Kuussaari et al., 2004). The wasp H. horticola parasitizes about a third of the host caterpillars in each host nest in the Åland Islands (Montovan et al., 2015). The hyperparasitoid M. cf. stigmaticus is a specialist solitary parasitoid of endoparasitoids in M. cinxia caterpillars (Shaw et al., 2009). It is present throughout the Åland Islands, parasitizing 20–60% of H. horticola in many places (Nair et al., 2016). Extremely rarely it also parasitizes Cotesia melitaearum, which is the other specialist endoparasitoid of M. cinxia caterpillars (van Nouhuys and Hanski, 2005). In neighboring Estonia, the butterfly and H. horticola are present but both Wolbachia and the hyperparasitoid are absent (Duplouy et al., 2015; Montovan et al., 2015).

For the experiment, we used H. horticola reared from host caterpillars that were naturally parasitized in the Åland Islands, Finland, and Saarema, Estonia, in summer 2014. These were collected as parasitized caterpillars, and reared in the laboratory until the parasitoid wasps pupated. Upon reaching adulthood, the wasps were maintained under uniform conditions in individual 100 ml vials in an incubator (12 h:12 h light:dark and 18/10°C day/night temperature) and fed honey water (1:3) daily. Once mature (10 days old), the virgin female wasps were offered ∼10 day old M. cinxia clutches from laboratory-reared butterflies originating in the Åland Islands (see Couchoux et al., 2015, for detailed methods of wasp and butterfly rearing and oviposition). Parasitism of each egg cluster was observed. The wasps were unmated because we are unable to make them mate in the laboratory. Because of haplodiploidy, unmated diploid H. horticola is a haplodiploid Hymenoptera, so unmated mothers produce haploid male offspring through arrhenotokous parthenogenesis (Normark, 2003). Wolbachia-infected females transmit the infection to both sons and daughters (Duplouy et al., 2015). The infection status of individual H. horticola females was unknown until after the experiment, but it was assumed that about half the individuals from Åland were infected (Duplouy et al., 2015). After hatching, the parasitized clutches (N=29) were reared until the caterpillars reached the second instar. Several individuals from each clutch were dissected to make sure that the clutch had been successfully parasitized. Groups of 40 caterpillars (large clutches were split) were placed on potted Veronica spicata or Plantago lanceolata, which are the host plants for M. cinxia (Kuussaari et al., 2004), to make 50 nests. After the caterpillars had built a gregarious silken nest on the plants they were placed in M. cinxia habitat patches in Åland, where they were exposed to natural hyperparasitism by M. cf. stigmaticus. The nests (N=45, some natural mortality occurred in the field) were brought back to the lab when they had reached the diapause stage. Caterpillars were then dissected under a microscope to determine which had been parasitized by H. horticola. Each parasitoid larva was then dissected to identify individuals that were hyperparasitized by M. cf. stigmaticus.

The butterfly and the parasitoid wasps are not classified as threatened species in the sampled regions and hence no permits are required for their collection.

We extracted DNA from the abdomen of each H. horticola adult female wasp using a Qiagen DNeasy blood and tissues extraction kit, following the manufacturer's protocol (Qiagen^®, USA). The DNA quality was tested by PCR amplification of the mitochondrial COI gene (primer pair LCO/HCO; Folmer et al., 1994). The COI amplicons were sequenced to determine the mitotype of each wasp (C or T; Duplouy et al., 2015). Duplouy et al. (2015) showed that despite wHho transmission rates being imperfect in both matrilines, the T-mitotype is more often found associated to wHho-infected wasps, while the C-mitotype is more common in non-infected wasps. The Wolbachia infection status of each sample was assessed through the amplification of the Wolbachia wsp gene (primer pair 81F/691R; Zhou et al., 1998). Each PCR included both positive and negative controls. Altogether, 25 H. horticola wasps were used and screened for this study (six Wolbachia-infected and 10 non-infected wasps from Aland Islands, and nine non-infected wasps from Estonia). All infected wasps were of T-mitotype, while non-infected wasps from both Åland and Estonia carried the C-mitotype (Duplouy et al., 2015).

Statistical analyses were performed using R (http://www.R-project.org/). To test the effects of country of origin and Wolbachia-infection status of the parasitoid on the proportion of M. cinxia caterpillars parasitized per clutch, we used a cumulative linked model (clm, from ‘ordinal’ and ‘nlme’ libraries in R). The proportion of caterpillars parasitized by H. horticola was considered categorical with 10 categories (x≤10%, 10%<x≤20%, 20%<x≤30%, 30%<x≤40%, 40%<x≤50%, 50%<x≤60%, 60%<x≤70%, 70%<x≤80%, 80%<x≤90%, 90%<x≤100%) to fit the model. We also used a cumulative linked mixed model to test the effects of country of origin and Wolbachia-infection status on the proportion of H. horticola hyperparasitized by M. cf. stigmaticus. Mesochorus cf. stigmaticus tends to hyperparasitize a higher proportion of H. horticola larvae when a higher proportion of them are present in a butterfly clutch (Montovan et al., 2015). Because per-nest rate of parasitism varied, we took this into account in the statistical model by first making a linear model of the proportion of H. horticola larvae hyperparasitized by M. cf. stigmaticus, and the proportion M. cinxia parasitized by H. horticola in the nest. The linear model residuals were then included as categorical data (x<−50%, <−40%, <−20%, <0%, <10%, <20%, <30%, <40%, <50% and <60%) in the cumulative linked mixed model. As several nests placed in the field were parasitized by the same H. horticola wasp, we also included the ID of the H. horticola wasp as a random factor in the model.

Hyposoter horticola from Estonia (without Wolbachia) and Åland (individuals with and without Wolbachia) parasitized the host egg clusters from Åland at a similar rate (23.9% versus 43.2% of hosts per cluster, d.f.=1, P=0.178; Fig. 2). The Wolbachia infection status of the parasitoid H. horticola did not affect its parasitism success in M. cinxia, as Wolbachia-infected and non-infected wasps parasitized the same fraction of caterpillars within M. cinxia clutches (32.4% versus 34.3%, d.f.=1, P=0.904; Fig. 2).

Female M. cf. stigmaticus do not discriminate between M. cinxia caterpillar nests parasitized by Wolbachia-infected or -free H. horticola larvae (P=0.29, Fisher exact test). Of the 45 caterpillar nests placed in the field, 30 were in meadows visited by M. cf. stigmaticus (at least one larva per meadow was found parasitized by M. cf. stigmaticus). We found hyperparasitoid larvae in seven of the nests parasitized by Wolbachia-infected H. horticola, and in 15 of the nests parasitized by Wolbachia-free wasps. In contrast, only one nest parasitized by Wolbachia-infected H. horticola and seven nests parasitized by Wolbachia-free wasps remained undetected by M. cf. stigmaticus. The remaining nests were lost as a result of natural disturbances (e.g. heavy rains or animals).

Wolbachia-free parasitoid larvae from Estonia and from Åland were hyperparasitized by M. cf. stigmaticus at a similar rate (36.1% versus 41.5%, d.f.=1, P=0.645; Fig. 3). In contrast, a larger proportion of the H. horticola larvae from Wolbachia-infected wasps were parasitized by the hyperparasitoid M. cf. stigmaticus (73.9% versus 39.5%, d.f.=1, P=0.0472; Fig. 3). During dissections, we found no evidence of superparasitism, as no H. horticola larva had more than one M. cf. stigmaticus in it. Additionally, all M. cf. stigmaticus larvae found in H. horticola larvae were alive and moving, with no sign of encapsulation at this stage of larval development.

A recent study found that the Wolbachia strain wHho persists at the intermediate prevalence of 50% in the population of the wasp H. horticola in the Åland Islands, without impacting the host fecundity, longevity or dispersal (Duplouy et al., 2015). Here, we show that wHho increases wasp susceptibility to hyperparasitism. The specialist hyperparasitoid M. cf. stigmaticus is present throughout Åland (van Nouhuys and Hanski, 2002). However, there is a negative association of the prevalence of the hyperparasitoid with Wolbachia infection in the landscape (Duplouy et al., 2015). The results of our study suggest an increased susceptibility of the Wolbachia-infected wasps to hyperparasitism. This could explain the landscape-scale negative association of the two parasites of H. horticola (Fig. 1, hypothesis 2). The persistence of wHho in the host population despite the cost that we have identified suggests that there should be a counterbalancing benefit to the infected individuals.

As a maternally inherited endosymbiont, Wolbachia can promote its own spread and persistence by enhancing the production of females in its host populations. To this end, the bacterium often has strategies to improve the overall fitness of its insect host. This has been found in some parasitoid wasps; for instance, Wolbachia benefits survivorship of the host Encarsia inaron (White et al., 2011). Asobara japonica wasps infected with Wolbachia show more efficient host (D. melanogaster)-searching ability (Furihata et al., 2015), while Wolbachia-infected Anagrus sophiae parasitoid wasps have higher reproductive success than uninfected individuals (Segoli et al., 2013). In another wasp, Asobara tabida, the association with Wolbachia has evolved into complete mutualism; the bacterium is required for the host to complete oogenesis and reproduction (Dedeine et al., 2001, 2004). However, the presence of Wolbachia is not always associated with enhanced host life history traits. For instance, in natural populations of the Drosophila parasitoid Leptopilina heterotoma, Wolbachia infection reduces adult fecundity, survival and mobility (Fleury et al., 2000). Finally, Wolbachia may have no association with measured fitness traits, as was previously found for H. horticola (Duplouy et al., 2015), but is most likely positively linked to another yet-undefined fitness component(s).

A less direct way for Wolbachia to benefit their host fitness is by improving the host resistance to parasites (see Table 1). This is the case for the fruit fly D. melanogaster, in which the infectious dose (ID₅₀) for, and the titer of the West Nile Virus (WNV) are high in Wolbachia-infected flies, suggesting that Wolbachia-infected individuals resist infection by the virus better than non-infected ones (Glaser and Meola, 2010). Such expression of resistance to pathogens is believed to be costly for the host because an individual must maintain a high density of symbionts (Martinez et al., 2015). Thus, the Wolbachia strains present in higher density in D. melanogaster also shorten the flies' lifespan (Chrostek et al., 2013, 2014). Therefore, if the selection pressure from parasites is weak, there is little chance that the Wolbachia strain would spread in its host population (Martinez et al., 2014, 2015). Hence, the presence of Wolbachia is not always only beneficial with respect to immunity (Table 1).

We found that wasp larvae from wHho-infected matrilines are more often parasitized by the hyperparasitoid M. cf. stigmaticus than are larvae from Wolbachia-free matrilines (Fig. 3, P=0.0472). Although our results strongly suggest that Wolbachia increases susceptibility of H. horticola to hyperparasitism, it is possible that wHho-infected and non-infected hosts may differ in ways other than their Wolbachia infection status, which could be related to the host susceptibility to parasitism (Ferreira et al., 2014). However, we know at least that wHho infects individuals throughout the well-mixed host population in Åland. It is found in the two mitotypes (less than 1% divergence between matrilines; Duplouy et al., 2015), across the different haplotypes of H. horticola (based on 14 microsatellite markers, A.D., unpublished observations), and across the landscape (Duplouy et al., 2015), where different haplotypes occur (Nair et al., 2016).

The mechanistic explanation of H. horticola susceptibility to hyperparasitism that is associated with Wolbachia infection remains unknown, but there are several possibilities. A foraging hyperparasitoid must first of all find M. cinxia caterpillar nests parasitized by H. horticola. Herbivory by M. cinxia causes the host plant to release volatile odors that lead H. horticola to their hosts (Castelo et al., 2010; Pinto-Zevallos et al., 2013). Such volatiles can also be attractive to hyperparasitoids (Zhu et al., 2014). While Wolbachia has not yet been found to affect the volatile chemistry of its hosts’ food plant, it has been shown to play a crucial role in the manipulation of other aspects of host food plant physiology, inducing the ‘green-island’ phenotype, allowing a leaf-mining host insect to feed on senescing autumn leaves (Gutzwiller et al., 2015). In our system, the hyperparasitoid detected caterpillar nests parasitized by wHho-infected and non-infected H. horticola wasps equally well, suggesting that the bacterium is not involved in manipulation of the volatile plant chemistry.

Once at a nest, a M. cf. stigmaticus has to find and parasitize H. horticola larvae using its ovipositor to probe inside the M. cinxia caterpillars (A. Reichgelt, Density-dependent aggregation of hyperparasitoid Mesochorus stigmaticus, MSc Thesis, University of Helsinki, 2007). Drosophila larvae are able to evade parasitoid wasps by rolling on their side in response to a stimulus such as cuticle piercing by the parasitoid ovipositor (Hwang et al., 2007; Robertson et al., 2013). If H. horticola larvae, which can move within the host hemolymph, are similarly evasive, then suppression of that behavior due to the presence of Wolbachia could increase their susceptibility to hyperparasitism.

After oviposition, a host may resist parasitism by killing the parasitoid egg or larva (Strand and Pech, 1995). Wolbachia induce upregulation of several host immune genes (Bian et al., 2010; Hughes et al., 2011; Kambris et al., 2010, 2009), potentially priming the immune system to respond strongly to pathogens or parasitoids (but see Bourtzis et al., 2000). Alternatively, Wolbachia may reduce the fitness of invading pathogens by competing for resources (Martinez et al., 2014; Moreira et al., 2009; Osborne et al., 2009). As we found no evidence of encapsulation of M. cf. stigmaticus, we suggest that M. cf. stigmaticus is able to successfully bypass the H. horticola immune system, regardless of the Wolbachia infection status of the host. The mechanisms of Wolbachia-induced protection against parasitism found in arthropods may target only some infection mechanisms, such as those of RNA viruses, but be unable to counteract others, including the virulence mechanisms of the hyperparasitoid M. cf. stigmaticus (Table 1).

Wolbachia is most well known for its ability to manipulate its host reproductive system in a manner that optimizes its transgenerational transmission (Caspari and Watson, 1959). Turelli and Hoffmann (1991, 1995) documented a rapid spread of the cytoplasmic incompatibility (CI)-inducing Wolbachia strain wRi across the Californian populations of D. simulans. Indeed, the wRi strain causes uninfected females to be incompatible with Wolbachia-infected males, thus increasing the reproductive success of the infected female hosts, whose offspring from matings with both infected and uninfected males are viable. Duplouy et al. (2015) reported that population sex-ratio distortions and female-only broods are not observed for H. horticola in the Åland Islands, suggesting that induction of manipulative phenotypes (male-killing, thelytokous parthenogenesis or feminization) is not occurring. However, induction of CI is not ruled out, as neither the occurrence nor the absence of incompatibility between wHho-infected males and non-infected females has yet been described in this system. By inducing CI, wHho could overcome the negative effect of the bacterium on its host’s susceptibility to hyperparasitism and still maintain an intermediate prevalence (∼50%; Duplouy et al., 2015) in the wasp population through a balance of benefits (from CI) and costs (from the increased host susceptibility) to the infected individuals.

Some of the H. horticola used in this experiment were from Estonia, just a few hundred kilometers by sea from Åland, where neither the hyperparasitoid wasp nor Wolbachia is present. It is possible that both the wasp and the bacterium have not yet arrived here. Should the wHho strain colonize the Estonian population, we would expect the infection to spread rapidly to a high prevalence in the absence of the hyperparasitoid wasp.

Selection due to lethal parasites such as parasitoid wasps can be very strong (Haldane, 1992), so one might expect Wolbachia that increase susceptibility to parasites to be rare. However, Wolbachia has been found to occur in several Diptera hosts (Table 1). We have shown that it occurs in a natural Hymenoptera host population under strong and consistent attack by a Hymenoptera hyperparasitoid. To date, the mechanisms behind how Wolbachia affects the relationship of its host with parasitoids or pathogens remain unclear. However, as the growing literature on diverse host–symbiont–pathogen systems suggests, the interaction is unlikely to be highly specific. In our study system, we saw no evidence of an increase of immune response, nor of any other evading mechanisms due to Wolbachia. Thus, the considerable benefit of the Wolbachia infection that counterbalances increased susceptibility to parasitism, which is not correlated with fecundity or longevity (Duplouy et al., 2015), must also not be directly related to resistance to parasitism.

We would like to thank S. Ikonen, A. Oksanen, T. Hämäläinen and T. Nyman for assistance in the field or in the lab, S. C. Wong for advice on the statistical models and I. Hanski and A. N. Volkoff for helpful comments on the manuscript.