Learning can be detrimental for a parasitic wasp

Animals have evolved the capacity to learn, and the conventional view is that learning allows individuals to improve foraging decisions. The parasitoid Telenomus podisi has been shown to parasitize eggs of the exotic stink bug Halyomorpha halys at the same rate as eggs of its coevolved host, Podisus maculiventris, but the parasitoid cannot complete its development in the exotic species. We hypothesized that T. podisi learns to exploit cues from this non-coevolved species, thereby increasing unsuccessful parasitism rates. We conducted bioassays to compare the responses of naïve vs. experienced parasitoids on chemical footprints left by one of the two host species. Both naïve and experienced females showed a higher response to footprints of P. maculiventris than of H. halys. Furthermore, parasitoids that gained an experience on H. halys significantly increased their residence time within the arena and the frequency of re-encounter with the area contaminated by chemical cues. Hence, our study describes detrimental learning where a parasitoid learns to associate chemical cues from an unsuitable host, potentially re-enforcing a reproductive cul-de-sac (evolutionary trap). Maladaptive learning in the T. podisi—H. halys association could have consequences for population dynamics of sympatric native and exotic host species.


Introduction
Animal decision-making, which is involved in processes such as resource and habitat selection, mate choice and progeny allocation, relies on innate behaviour (instinct), stochastic processes, physiological feedbacks (e.g., hormonal signalling) and learning (reviewed in [1]). Insect parasitoids have been used as model systems to explore both proximate and ultimate perspectives of optimal foraging. In order to cope with spatial and temporal variability in resources, parasitic wasps have evolved the capacity to associate host-related chemical cues to host availability and suitability [2][3][4][5][6]. They further consolidate and improve this capacity through learning processes [7][8][9][10], resulting in increased reproductive success [3,[11][12][13]. The probability of including a new stimulus in the behavioural repertoire of a parasitoid female depends on its reliability in host location [2], with oviposition having been shown to consolidate a change in foraging behaviour and host acceptance [3,9,14]. For  chemical cues on the egg surface or those deposited on plant surfaces by gravid females (footprints) to locate hosts in the habitat (reviewed by [15]). Following detection of host chemical cues, experienced scelionid females show stronger arrestment response (increased residence time, slower walking and increased turning tendency) than naïve females, a behaviour that is re-enforced by successful oviposition [16,17]. Biological invasions generate novel interactions which can have negative consequences on populations of native species [18][19][20][21]. This occurs, for example, when an invasive exotic species becomes accepted as a host by native parasitoids but is unsuitable for offspring development. The introduced species then acts as an egg sink [ . Halyomorpha halys is progressively spreading in invaded areas, thereby increasing the probability of native T. podisi encountering this novel but unsuitable host. We examined the learning capacity of T. podisi towards its coevolved host, P. maculiventris, and non-coevolved host, H. halys, and discussed consequences on interacting species.

Hosts and parasitoid
A colony of H. halys reared continuously on raw pumpkin seeds, carrots, green beans, grapes and potted soybean plants, was established from adults collected in Ontario (Canada) in 2012. The P. maculiventris colony was initially established using adults collected in Ontario in 2011 and 2012 and supplemented with bugs from Anatis Bioprotection (Canada). They were fed with live mealworm, Tenebrio molitor L. (Coleoptera: Tenebrionidae) larvae reared in the laboratory, fresh green beans and bean plants. Nymphs were kept in plastic cylinders and fed with mealworm and green beans. For both stink bug species, freshly laid eggs (< 24 h old) were used for the experiments.
The T. podisi colony was established with individuals collected in 2011 and 2012 in Ontario. Adult parasitoids were provided with a 1:1 (vol/vol) honey:water solution rubbed on a small piece of ParaFilm 1 . Each week, 1-2 days-old egg masses of P. maculiventris (stuck on filter paper using Pritt1 stick glue) were exposed to T. podisi females for 24 h to maintain the colony. After emergence, male and female parasitoids were kept together in glass tubes for mating and provided with the honey-water solution. Naïve (i.e. without oviposition experience), 3-8 days-old females were randomly assigned to the different experimental treatments. Females T. podisi from our laboratory colony are synovigenic (ovigeny index < 0.05), typically emerging with very few mature oocytes (mean of approximately 2.0 [36]), and can survive for up to 100 days under laboratory conditions [37]. Unpublished data (M. Gaudreau, pers. com.) further show that 4-8 day-old females have an average of 14.2 mature eggs in their abdomen, similar to the maximum number of mature eggs observed per female throughout their reproductive life.

Treatments
To determine if T. podisi females exhibit learning behaviour the following four treatments were tested: (i) naïve parasitoid females foraging on H. halys traces; (ii) females with experience on H. halys traces and eggs, then tested on H. halys traces; (iii) naïve females foraging on P. maculiventris traces; and (iv) females with experience on P. maculiventris traces and eggs, then tested on P. maculiventris traces. The experiments were conducted from 10:00 to 14:00 at 24±1˚C, 50±5 percent relative humidity, under a 16L:8D photoperiod. Between 39 to 43 replicates were conducted for each treatment.

Obtaining experienced females
Experienced parasitoid females were obtained in the following manner. A female of either H. halys or P. maculiventris in their pre-ovipositional phase (with a physogastric abdomen) was introduced in an experimental arena consisting of a Petri dish (5 cm diam., 1 cm height) placed upside-down on a filter sheath. The Petri dish cover had tissue mesh (0.01 cm holes) to prevent saturation of the atmosphere with volatiles released by the stink bug female. Females walked for 30 minutes on the filter paper to allow contamination with chemical traces. Females that occasionally walked on the Petri dish were discarded. Filter papers soiled by bug's faeces were discarded. Once the stink bug female was removed, a small egg mass (5-6 eggs) of the same species was placed in the middle of the contaminated area. A naïve (i.e. without foraging and oviposition experience) T. podisi female was then introduced in the arena and continuously observed. Once she had oviposited the female was removed, isolated in a 1.5 mL tube for 1 h before being tested as an experienced parasitoid. Females that did not oviposit within 1 h were discarded.

Parasitoid female response to chemical cues
As previously described [38], bioassays were conducted on a large filter paper arena (20 x 20 cm) where parasitoid females could move freely on the surface. A 5 cm diam area at the centre was exposed to a female of H. halys or P. maculiventris in pre-ovipositional phase, as described above. For the assays, a T. podisi female was released in the middle of the contaminated area (without host eggs) and her behaviour recorded with a HDD video camera (Sony HDR-XR 500) placed 40 cm above the arena. The assay stopped when the wasp left the arena or after 10 minutes [38]. Individuals that flew away within 10 s after the release were excluded from the analysis (n = 6). Each arena was used to test 5 females.
We recorded the time spent by the female in the arena (total residence time) and the number of times the parasitoids went back to the contaminated area once left (number of reencounters), indicative of a tortuous path associated with the searching behaviour of a parasitoid female [15,39,40].

Statistical analyses
Generalized linear models (GLMs with Gaussian distribution for time data or Poisson distribution for count data) were fitted to test the effects of host herbivore species (H. halys or P. maculiventris), parasitoid previous experience (experienced or naïve) and their interaction on total parasitoid residence time and number of re-encounters with the contaminated area. Residence time data were subjected to Box-Cox transformation for normalization before the analyses. Significance of the model terms was evaluated by means of F test or Likelihood Ratio Test [41]. Significance of the different variable levels was assessed using the Tukey method for multiple comparisons procedure, adopting a significance level α = 0.05. The potential influence of chemical residuals from a parasitoid female on the behaviour of the following female was evaluated as a random effect in mixed models. This influence was negligible and therefore models with only fixed effect were included. Analyses were conducted under R statistical environment [42].

Results
Total residence time of T. podisi varied depending on the insect species footprint (Gaussian GLM: F (1, 159) = 53.78, P < 0.0001, the previous experience (F (1,158) = 16.83, P < 0.0001) and the interaction of these two factors (F (1, 157) = 4.28, P = 0.04). For the T. podisi-P. maculiventris association, parasitoid females that had previously experienced the chemical traces left by their native host and had been rewarded with an oviposition did not have a greater residence time in the experimental arena than naïve females. In contrast, females with a rewarded experience on the exotic H. halys stayed significantly longer in the arena than naïve wasps when tested on chemical footprints of H. halys (Fig 1A). However, both naïve and experienced parasitoids displayed higher residence time on chemical traces of their coevolved host P. maculiventris than those of the exotic H. halys (Fig 1A).
The frequency of re-encounter with the area contaminated by chemical cues varied depending on the insect species footprint (Poisson GLM: X 2 = 35.65, P < 0.0001), the previous experience (X 2 = 26.39, P < 0.0001) and the interaction of these two factors (X 2 = 12.3, P = 0.0005). Experienced T. podisi females re-entered the contaminated area more frequently than naïve females (Fig 1B). Naïve females tested on H. halys footprints rarely returned to the contaminated area. However, following a rewarded experience on H. halys, the number of re-encounters with this species was similar to the number of re-encounters of naïve females tested on P. maculiventris (Fig 1B). Both naïve and experienced T. podisi females returned more frequently to a patch contaminated by P. maculiventris than H. halys (Fig 1B).

Discussion
Our findings indicate that T. podisi females exhibit increased foraging behaviour following experience with chemical traces and oviposition in both its coevolved and novel host, probably due to associative learning [11,43]. This capacity permits a female parasitoid to capture and retain information about host availability in the habitat and to adjust her foraging behaviour accordingly. However, the value of learning and its consequences on the reproductive success of T. podisi are opposite when exploiting a suitable (P. maculiventris) vs. unsuitable (H. halys) host. Exploiting H. halys eggs is maladaptive for T. podisi because it incurs significant costs to foraging females and leads to a cul-de-sac for their progeny [27]. Telenomus podisi females are not attracted to plants infested by H. halys under laboratory conditions [44], whereas T. podisi was the most common parasitoid sampled from H. halys eggs under field conditions in Canada [33]. Halyomorpha halys not only represents an egg sink for T. podisi, but also a 'time sink' (sensu [27]) since females increase time foraging in areas contaminated by the unsuitable host. The waste of time is further amplified when females protect the egg mass (patch guarding) from competitors and predators during several hours following oviposition [45].
Maladaptive learning has been reported in a number of social hymenopteran insects (bees and wasps) when copying the foraging decisions of conspecifics leads to the exploitation of already deprived food resources ( [46] and references therein). It was also suggested [47] that new genetic combinations following hybridization in fishes and birds could negatively impact

PLOS ONE
learning capacities, potentially leading to postzygotic reproductive isolation. Maladaptive learning has also been reported in parasitoid species used as biological control agents. When reared on alternative host species, artificial diets or artificial rearing units, parasitoids may partly loose their capacity to find and exploit the natural pest species [48,49]. For example, it was shown [50] that prior experience of Exeristes roborator (Hymenoptera: Ichneumonidae) when reared in an artificial arena significantly altered the behavioural response of parasitoid females to their host, the European pine shoot moth, Rhyacionia buoliana (Lepidoptera: Tortricidae), when released in forests. The present study documents an original case of detrimental learning where a parasitoid learns to associate chemical cues from an unsuitable host, thereby re-enforcing a reproductive cul-de-sac (evolutionary trap).
Both naïve and experienced T. podisi females showed a higher response to chemical traces of P. maculiventris than of H. halys. This pattern was expected because derived stimuli from coevolved hosts are likely to evoke strong innate responses by naïve individuals; indeed the innate response can be stronger than the learned response in the location and acceptance of highly suitable hosts [51]. However, we cannot exclude an effect of the rearing host (P. maculiventris), as shown in other parasitoid species [52][53][54]. For instance, similar residence time between naïve vs. experienced wasps when exposed to P. maculiventris chemical traces could be partly explained by T. podisi females having already gained experience when developing in and emerging from P. maculiventris eggs.
Associative learning of cues from a novel but unsuitable host would also exacerbate the negative effects of the evolutionary trap on T. podisi populations. It may contribute to modify the community structure in areas invaded by H. halys through direct and indirect ecological effects [27,55]. Recent exposure to H. halys may lead to an increase in T. podisi's rate of parasitism on the invasive host (host switching), and a consequent decrease in parasitism of indigenous pentatomid species (apparent predation/parasitism; (+, -) type interaction). Additional research is required to determine the extent to which detrimental learning would affect the population dynamics of native stink bugs and parasitoids. We only tested females 1 h after their experience with host chemical cues and it would be important to determine how long they exhibit such learned behaviour under natural conditions, considering ecological factors such as the relative densities of native and exotic stink bugs and the persistence of the chemical cues in the footprints.
This original case of maladaptive learning arises from a situation where a native parasitoid encounters a new potential host species as a result of a biological invasion. On one hand, there is no operational ecological filter that stops host location and acceptance and, on the other hand, there is a strong physiological filter that prevents parasitoid development [27,32]. Accordingly, a parasitoid could escape such an evolutionary trap by evolving (i) behavioural capacities to prevent acceptance of an unsuitable resource or (ii) physiological capacities to successfully reproduce in the novel host species [27,56].