Yosra Chabaane, Diane Laplanche, Ted C. J. Turlings and Gaylord A. Desurmont*

Transcription

1 Journal of Ecology 2015, 103, doi: / Impact of exotic insect herbivores on native tritrophic interactions: a case study of the African cotton leafworm, Spodoptera littoralis and insects associated with the field mustard Brassica rapa Yosra Chabaane, Diane Laplanche, Ted C. J. Turlings and Gaylord A. Desurmont* Institute of Biology, University of Neuch^atel, Neuch^atel, Switzerland Summary 1. When exotic herbivores invade new environments, they have the potential to interfere with native predator-prey relationships. This interference can be indirect, via changes induced in native host plants, and may have cascading consequences at the community level. Here, we investigate the impact of the presence of an exotic generalist insect herbivore, the African cotton leafworm Spodoptera littoralis, on the outcome of native tritrophic interactions between the plant Brassica rapa, the herbivore Pieris brassicae and its parasitoid Cotesia glomerata. 2. In olfactometer choice tests, plants damaged by S. littoralis and plants damaged by both S. littoralis and P. brassicae were consistently less attractive to the parasitoid than plants damaged by P. brassicae alone. Chemical analyses revealed that B. rapa volatiles typically induced by S. littoralis dominated the headspace in case of dual herbivore infestation. 3. In behavioural observations in petri dishes, C. glomerata wasps mistakenly attacked S. littoralis larvae significantly more often when P. brassicae was present, and attacks on both herbivores were comparable in terms of time (attack duration) and energy investment (number of eggs injected). Successful parasitism of S. littoralis was never observed, but larvae exposed to C. glomerata for 24 h exhibited reduced survivorship, possibly as a result of failed parasitism attempts. 4. In tents with herbivore-infested B. rapa plants, the presence of S. littoralis did not have an effect on the parasitism rates of P. brassicae by C. glomerata, regardless of whether the two species were on the same or on different plants. Field experiments in newly invaded environments are necessary to evaluate the realized impact of S. littoralis. 5. Synthesis. Our study illustrates that exotic herbivores can impact native tritrophic interactions associated with Brassica rapa, even if they cannot be used as prey by native natural enemies. The mechanisms behind such effects, in particular chemical interference with foraging cues via changes in herbivore-induced plant volatiles, have the potential to be quite general, and their long-term effects on native communities should not be underestimated. Key-words: exotic herbivore, HIPVs, invasion ecology, multitrophic interactions, non-consumptive effects, parasitoid foraging behaviour Introduction Invasive species are considered a major driver of biodiversity loss worldwide (Wilcove et al. 1998; Didham et al. 2005). As they invade new environments, exotic species interact with native species, creating new links in native food webs (Verhoeven et al. 2009; Harvey et al. 2010). These new associations can impact the strength of existing links (White, Wilson & *Correspondence author. gaylord.desurmont@unine.ch Clarke 2006), with effects rippling through trophic levels (Mooney & Cleland 2001; Fortuna, Vet & Harvey 2012). A number of these ecological effects have been documented, including changes in predator prey dynamics (Hoogendoorn & Heimpel 2002) and local extinctions (Cronin & Haynes 2004). In the context of plant insect interactions, investigations of the ecological impact of exotic species have focused on a number of topics, including but not limited to i) the use of exotic plants by native herbivores (Harvey et al. 2010) and its consequences for the third trophic level (Fortuna, Vet & 2014 The Authors. Journal of Ecology 2014 British Ecological Society

2 110 Y. Chabaane et al. Harvey 2012; Harvey & Fortuna 2012), ii) the impact of exotic herbivores on native plants (Kenis et al. 2009; Gandhi & Herms 2010) and iii) evaluation of the direct and indirect impacts of introduced biological control agents in novel environments (Louda et al. 2003; Pearson & Callaway 2005). Compared to these three major research foci, studies exploring the impact of exotic herbivores on native tritrophic interactions between native plants, herbivores and natural enemies are scarce (White, Wilson & Clarke 2006; Gandhi & Herms 2010). They have mostly focused on the location and use of exotic prey by generalist native natural enemies (Carlsson, Sarnelle & Strayer 2009; Berkvens et al. 2010; Sloggett 2010) or on consequences of shared natural enemies between native and exotic herbivores on the predator prey dynamics in native food webs (Settle & Wilson 1990; Redman & Scriber 2000). The guiding principle of these studies is that some native natural enemies can use exotic herbivores as prey or host and that these newly formed associations have consequences at the community level (Timms, Walker & Smith 2012). Here, we advance the idea that exotic herbivores can impact native tritrophic interactions even if they cannot be used as prey or hosts by native natural enemies. Firstly, exotic herbivores may interfere with the foraging behaviour of natural enemies, for example via changes in herbivoreinduced plant volatiles used by natural enemies to locate their prey or host from a distance (Harvey & Fortuna 2012). Such interferences may occur if plants damaged by the unsuitable invasive herbivore attract the native predator or parasitoid or if plants that have been damaged simultaneously by the exotic and the native herbivore lose their attractiveness (Shiojiri et al. 2002). Secondly, after landing on an infested plant, natural enemies may waste time and energy by attacking unsuitable exotic herbivores. This can be particularly problematic if the exotic herbivore is toxic for the attacker (Suttle & Hoddle 2006) or cannot be successfully parasitized in the case of parasitoids. Reduced foraging efficiency and wasteful, unsuccessful attacks may add up and negatively affect the fitness of natural enemies (Hoogendoorn & Heimpel 2002), potentially leading to weakened native predator prey interactions. This in turn may result in outbreaks of native herbivores or local extinctions of natural enemies (Cronin & Haynes 2004). In this study, we investigated the effects of the presence of an exotic herbivore, the African cotton leafworm Spodoptera littoralis, on a well-known Eurasian tritrophic system: the plant Brassica rapa, the herbivore Pieris brassicae and the parasitoid Cotesia glomerata. The field mustard Brassica rapa is an important food source for Pieris brassicae, particularly in Western Europe where B. rapa grows earlier in the Spring than most wild crucifers and can support the development of the first generation of P. brassicae caterpillars, which typically occurs in April May (Fei, Gols & Harvey 2014). Cotesia glomerata is the main parasitoid of P. brassicae caterpillars and is known to be attracted to the odours of Pieris-infested Brassica plants (Hare 2011; Fei, Gols & Harvey 2014). S. littoralis is an extremely polyphagous noctuid native to Africa, where it is considered a major pest of several crops and vegetables (Pimentel 2002; El-Sayed et al. 2009), including crucifers (CABI Invasive Species Compedium: datasheet/51070). It is known to be a range expanding species, with migratory populations invading the southern part of Europe every year. In the context of global climate warming, S. littoralis can be expected to keep moving up north in Western and Central Europe each passing year, posing a real threat for European agroecoystems. Because of its extremely broad host range, S. littoralis has the potential to impact numerous existing herbivores natural enemies interactions. We focused our study on the consequences of the presence of S. littoralis on the foraging behaviour and reproductive output of C. glomerata. Specifically, we addressed the following questions: 1. Does feeding by S. littoralis larvae alter the blend of volatiles induced by B. rapa plants in case of single and dual (i.e. S. littoralis + P. brassicae) herbivory and does it affect plant attractiveness to C. glomerata? 2. Does C. glomerata attack S. littoralis in presence or absence of P. brassicae, how do attacks on S. littoralis and P. brassicae compare in terms of time and energy investment and can C. glomerata successfully parasitize S. littoralis? 3. Does the presence of S. littoralis have an impact on the realized fitness of C. glomerata? We investigated the first question using chemical analyses of the volatiles emitted by singly and dually infested B. rapa plants and behavioural olfactometer tests with C. glomerata wasps. The second question was investigated using observations of attacks of both herbivores in petri dish bioassays and by subsequently following the development of parasitized S. littoralis caterpillars. The third question was addressed in a manipulative experiment where the parasitism of P. brassicae caterpillars by C. glomerata wasps was measured in tents containing B. rapa plants, in presence and absence of S. littoralis. Materials and methods INSECT AND PLANT MATERIAL Plants used in the study came from a wild accession of Brassica rapa whose seeds were collected in 2009 and 2012 near Maarsen, the Netherlands. Plants were grown in controlled growth chambers under 16/8 L:D light regime at 25 C, light intensity lmol. For the olfactometer tests, plants were grown in cylindrical plastic pots, whose dimensions were 4*10 cm for olfactometer tests/chemical analyses and 11*13 cm for tent experiments, with fertilized commercial soil (Ricoter Aussaaterde, Aarberg, Switzerland). Plants were watered every other day without supplemental nutrients. Plants used for experiments had six to eight fully expanded leaves for the tent experiments (four to 5 weeks old) and three to five for the olfactometer tests (approx. 3 weeks old). Pieris brassicae caterpillars originated from a laboratory rearing maintained in our laboratory on Brassica rapa (for oviposition) and B. oleracea plants (for larval development). The rearing was initiated with individuals from Switzerland collected in the field in locations containing various Brassica species. Spodoptera littoralis caterpillars came from eggs provided weekly by Syngenta (Stein, Switzerland) and were fed with an artificial diet until they were needed for

3 Exotic herbivores disrupt tritrophic interactions 111 experimental purposes. Cotesia glomerata parasitoids originated from a laboratory colony maintained on P. brassicae larvae in the laboratory. Parasitoid cocoons were collected after emergence from parasitized caterpillars and were placed in a cage without any host or plant material. Newly emerged adults were provided water and droplets of honey and were kept in an incubator at 25 C for 48 h, then in an incubator at 13 C. Mated naive females that were 2- to 4-week-old wasps were used for all experiments. CHEMICAL ANALYSES To identify and quantify the blends of volatile organic compounds (VOCs) emitted by undamaged B. rapa plants or plants damaged by P. Brassicae, S. littoralis or both herbivores simultaneously, potted plants (n = plants per treatment) were placed in a VOC collection setup (Ton et al. 2007) for 5 h. Herbivore-damaged plants were prepared by infesting the plants with 20 first instar P. brassicae larvae, 30 second instar S. littoralis or both, placed randomly on the leaves. With these numbers of larvae, we obtained comparable amounts of damage on B. rapa leaves after 24 h during preliminary experiments. Caterpillars were not removed from the plants during volatile collection. VOCs were collected using a trapping filter containing 25 mg of mesh SuperQ absorbent. Before use, trapping filters were cleaned with 300 ll of methylene chloride (HPLC grade). After each collection, VOCs were extracted from the filters with 150 ll of methylene chloride. Two internal standards (200 ng of n-octane and nonyl acetate in 10 ll methylene chloride) were added to each sample. VOCs were analysed with an Agilent 6890 gas chromatograph with a flame ionization detector. A 2 ll aliquot of each sample was injected in the pulsed splitless mode onto a nonpolar column (HP-1 ms, 30 m, 0.25 mm ID, 0.25 lm film thickness, Agilent J&W Scientific, USA). Helium was used as carrier gas at constant pressure (15 psi). After injection, temperature was maintained at 40 C for 3 min, then increased to 100 C at8 C per min and then to 220 C at5 C per min. The quantities of the major components of the blends were estimated based on the peak areas of the compounds compared to the peak areas of the internal standards. Compounds were identified by comparing the spectra obtained from the samples with those from a reference database (NIST mass spectral library). OLFACTOMETER TESTS The preferences of C. glomerata females for different odours were measured in two separate experiments using a 4-arm olfactometer setting (Turlings, Davison & Tamo 2004), investigating the attractiveness of single herbivory by S. littoralis and dual herbivory by S. littoralis and P. brassicae compared to plants damaged by P. brassicae alone, respectively. In this setting, wasps were given the choice between 4 odour sources (=treatments) contained in separated glass bottles. Individual air flows were connected to each odour source and all converged to a central glass piece, where the wasps were released (D Alessandro & Turlings 2005). After 30 min, wasps were recollected and the treatment they chose was recorded. An olfactometer test (=1 replicate) consisted of five consecutive releases of 5 wasps for the 4-arm olfactometer tests. Plants were changed and glassware was cleaned between replicates, and 5 replicates were conducted for each experiment. Thus, the behaviour of 125 individual wasps was observed for each experiment. In the first experiment, we compared the attractiveness of the four following treatments: plant infested by P. brassicae, plant infested by S. littoralis, undamaged plant and blank odour source (=empty bottle). In the second experiment, we compared the attractiveness of the following treatments: plant infested by P. brassicae, plant infested by P. brassicae and S. littoralis simultaneously, undamaged plant and blank odour source. The herbivore-infested plant treatments were prepared the same way as they were prepared for the chemical analyses described above. OBSERVATIONS OF COTESIA GLOMERATA ATTACKS ON PIERIS BRASSICAE AND SPODOPTERA LITTORALIS To determine whether C. glomerata wasps would attack S. littoralis caterpillars in presence or in absence of their host P. brassicae, and whether C. glomerata could successfully parasitize S. littoralis, behavioural observations were conducted in glass or plastic petri dishes (10 and 9 cm diameter, respectively) with a water-moistened filter paper. Individual C. glomerata wasps were released in petri dishes containing a leaf disc (5.5 cm diameter) of B. rapa var. pekiniensis (Chinese cabbage) and a droplet of honey as food for the parasitoid, as well as one of the following three herbivore treatments: five 1st instar P. brassicae caterpillars (n = 24), five 2nd instar S. littoralis caterpillars (n = 26) or five 1st instar P. brassicae plus five 2nd instar S. littoralis caterpillars (n = 22). Wasps were observed for 30 min, and the number of attacks on each caterpillar species was recorded. We considered that a parasitoid attacked a caterpillar every time it inserted its ovipositor inside the body of a caterpillar for at least 1 s. After the behavioural observations, wasps were kept together with the herbivores for 24 h. Then, S. littoralis caterpillars were removed from the petri dishes and transferred to individual plastic containers (5 7 cm diameter) with artificial diet, and their development was monitored until they died or pupated. Parasitism was considered successful whenever C. glomerata larvae emerged from a S. littoralis caterpillar. In parallel, the development of 2nd instar S. littoralis caterpillars, placed in the same petri dish setting for 24 h but not exposed to C. glomerata wasps, was monitored as a control (n = 20). At the end of the experiment, survivorship of S. littoralis larvae was calculated as the number of caterpillars that reached pupation divided by the number of caterpillars that survived the initial 24 h in petri dishes. Thus, mortality that occurred in the petri dishes during the initial 24 h, due to cannibalism or other factors, was not included to calculate S. littoralis survivorship. To determine differences in time and energy investment associated with attacks of P. brassicae and S. littoralis caterpillars, additional behavioural observations were conducted in a similar petri dish setup. For these observations, the duration of individual attacks on both herbivores was recorded. The attacked caterpillar was dissected immediately after the end of the observation period, and the number of eggs of C. glomerata found in the haemolymph of the caterpillar was counted. These observations were done for 20 C. glomerata wasps, and single attacks on P. brassicae and S. littoralis were observed for each wasp tested. To avoid potential effects due to the order in which the herbivores were presented to the parasitoid, half of the wasps were presented P. brassicae first and then S. littoralis, and the second half were presented S. littoralis first and then P. brassicae. Thus, a total of 20 attacks were monitored, and 20 caterpillars were dissected for each herbivore species. IMPACT OF THE PRESENCE OF SPODOPTERA LITTORALIS ON THE REALIZED FITNESS OF COTESIA GLOMERATA The goal of this experiment was to measure the impact of the presence of S. littoralis on the parasitism rate of C. glomerata on

4 112 Y. Chabaane et al. P. brassicae. This experiment was carried out in tents (length: 92 cm, width: 46 cm, height: 47 cm). Each tent contained eight B. rapa plants and received one of the three following herbivore treatments: Pieris infestation, separate infestation and mixed infestation (Fig. S1 in Supporting Information). In all treatments, four plants were randomly chosen and infested with P. brassicae caterpillars. In the Pieris infestation treatment, the other plants remained uninfested. In the separate and mixed infestation treatments, S. littorallis caterpillars were added either on the four plants that were not infested by P. brassicae (separate infestation treatment) or on the same plants as P. brassicae (mixed infestation treatment). A first series of replicates was carried out with 10 larvae (1st instar P. brassicae or 2nd instar S. littoralis) per plant infested (n = 5 tents per treatment), then a second series with 20 larvae per plant infested (n = 7 tents per treatment). After 24 hours of infestation, 4 mated naive females and 4 males of C. glomerata were released into each cage. In order to be sure that parasitoids always had access to food during the experiment, a petri dish with droplets of honey was placed in each tent. One day after introduction of the parasitoids, P. brassicae larvae were recovered and monitored until parasitoid emergence or pupation. Because P. brassicae larvae sometimes escaped or could not be recovered from the tents, the total number of larvae recovered, rather than the number of larvae placed in the tents was used to calculate C. glomerata parasitism rates at the end of the experiment. Caterpillars that were recovered but died before parasitoid emergence or pupation were excluded from the calculation of parasitism rates. STATISTICAL ANALYSIS Regarding the results of the chemical analyses, the amounts of plant volatiles belonging to each category identified (green leaf volatiles, terpenoids, isothiocyanates and alkanes) were compared among treatments using the nonparametric Wilcoxon rank sum statistical procedure (a = 0.05, JMP9). Results of the olfactometer tests were analysed using one-way ANOVA (a = 0.05), and treatments were compared using the all-pairwise Tukey Kramer HSD procedure (JMP9). The number of attacks observed on P. brassicae and S. littoralis during 30 min behavioural observations in petri dishes was compared using one-way ANOVA (a = 0.05, JMP9) for each herbivore species. Duration of the attacks and number of eggs injected during an attack were compared between the two herbivores in two ways: firstly by using a paired t-test (JMP9), each wasp tested representing a pair of observations; secondly by running a two-way ANOVA (a = 0.05, JMP9) testing the effects of herbivore species and herbivore order (P. brassicae then S. littoralis, or S. littoralis then P. brassicae), and the interaction between these two terms on the variables measured. For the tent experiments, the effects of herbivore density, treatment and the interaction of herbivore density and treatment on C. glomerata parasitism rates were analysed using two-way ANOVA (a = 0.05, JMP9), and differences between means were analysed using all-pairwise Tukey Kramer HSD procedure. Results OLFACTOMETER TESTS AND CHEMICAL ANALYSIS There were significant differences in attractiveness to C. glomerata wasps among treatments for both experiments, the first one investigating the effects of single S. littoralis herbivory (F 3,96 = 60.2, P < ) and the second one the (a) Fig 1. Percentage of Cotesia glomerata females attracted towards different treatments in a 4-arm olfactometer. (a) effect of single herbivory by Spodoptera littoralis on C. glomerata preferences. (b) effect of dual herbivory Pieris brassicae + S. littoralis on C. glomerata preferences. Treatments represent the following: Blank = empty odour source, Ctrl = non infested plant, Inf P = Pieris-infested plant, Inf S = Spodoptera-infested plant, Inf P + S = Pieris + Spodopterainfested plants. Treatments followed by a different letter are significantly different (One-way ANOVA, a = 0.05, JMP 9). effects of dual herbivory by S. littoralis and P. brassicae (F 3,96 = 18.7, P < ). In the first experiment, plants damaged by P. brassicae were the most attractive treatment while plants damaged by S. littoralis, undamaged plants and a blank odour source showed comparably low attractiveness (Fig. 1a). In the second experiment, plants damaged by P. brassicae were again the most attractive treatment, while plants damaged by both herbivores and blank odour source had comparably low attractiveness. Wasps never chose the undamaged plant during the second experiment, and this treatment had the lowest attractiveness (Fig. 1b). CHEMICAL ANALYSES The volatiles collected and identified in plants subjected to the different treatments (undamaged, infested by P. brassicae, infested by S. littoralis and dual herbivory) were classified in four categories, which correspond to the main classes of herbivore-induced plant volatiles documented in brassicaceae (Ahuja, Rohloff & Bones 2010; Pierre et al. 2011): green leaf volatiles, isothiocyanates, terpenoids and alkanes (Fig. 2). The identity of each consistently detected compound found in each class is given in Table S1. Other compounds, including aromatics or ketones, were either rarely detected or were also present in empty odour sources and were considered contaminants. There were significantly more green leaf volatiles produced in the S. littoralis and dual herbivory treatments than in the P. brassicae and control treatments. Isothiocyanates were also more abundant in the S. littoralis and dual herbivory treatments, they were intermediate in the P. brassicae treatment, and were considerably lower in the control treatment. Alkanes were significantly more abundant in the (b)

5 Exotic herbivores disrupt tritrophic interactions 113 A G T I A T G I T I A G T I A G Control Pieris brassicae Spodoptera littoralis Pieris brassicae + Spodoptera littoralis Fig 2. Visual representation of the volatiles blends emitted by Brassica rapa plants in response to different treatments: undamaged plants (control), plants infested by Pieris brassicae, plants infested by Spodoptera littoralis and plants under simultaneous dual infestation (P. brassicae + S. littoralis). The areas of the circles are proportionate to the total quantity of volatiles emitted (ng internal standard equivalents). Letters indicate class of compounds: green leaf volatiles (G), isothiocyanates (I), terpenoids (T), alkanes (A). P. brassicae treatment, were intermediate in S. littoralis and dual infestation treatments, and were lower in the control treatment. There were no significant differences in the amounts of terpenoids and other compounds among treatments, although there was a trend in reduction of emission of terpenoids in the S. littoralis and dual infestation treatments (Fig. S2). The area damaged by P. brassicae and S. littoralis was not significantly different in case of single herbivory and was significantly higher in case of dual herbivory (Fig. S3). (a) OBSERVATIONS OF COTESIA GLOMERATA ATTACKS ON PIERIS BRASSICAE AND SPODOPTERA LITTORALIS During 30-min observations in petri dishes in the presence of P. brassicae, S. littoralis or both, the total number of C. glomerata attacks was higher when both herbivores were present than when single herbivores were present (F 2,69 = 7.41, P = 0.001) (Fig. 3a). C. glomerata was observed attacking S. littoralis both in presence and in absence of P. brassicae, and attacks were significantly more frequent when P. brassicae was present ( vs , F 1,46 = 7.7, P = 0.01) (Fig. 3a). Attacks on P. brassicae were similarly frequent in presence and in absence of S. littoralis (F 1,48 = 2.77, P = 0.1). The survivorship until pupation of S. littoralis larvae was the highest when they had not been exposed to C. glomerata for 24 hours, intermediate when they had been exposed to C. glomerata in absence of P. brassicae and the lowest when they had been exposed to C. glomerata in presence of P. brassicae (Fig. 3b). Cotesia glomerata did not develop successfully in any of the S. littoralis larvae exposed to the parasitoid (i.e. no C. glomerata larvae emerged from the S. littoralis larvae monitored) during this experiment. Further observations of attacks by C. glomerata on both herbivores revealed that there were no differences between the duration of attacks on P. brassicae and S. littoralis, regardless of the order of attack (t = 2.1, P = 0.1 for the order P. brassicae then S. littoralis, t = 1.2, P = 0.28 for the order S. littoralis then P. brassicae, paired t-test JMP9). In accordance, there were no differences in the number of eggs injected during attacks on both herbivores, regardless of the order of attack (t = 0.7, P = 0.5 for the order P. brassicae then S. littoralis, t = 1.6, P = 0.1 for the order S. littoralis (b) Fig 3. (a) observed Cotesia glomerata attacks (i.e. parasitization attempts)(mean SE) on Pieris brassicae (PB) and Spodoptera littoralis (SL) in petri dishes with one or both herbivores. Asterisk indicates that attacks on S. littoralis are significantly more numerous in presence of P. brassicae (one-way ANOVA, a = 0.05). (b) Survivorship until pupation (mean SE) of S. littoralis without exposure to Cotesia glomerata (Control) or after being exposed 24 h to C. glomerata, in absence of P. brassicae (Cotesia) or in presence of P. brassicae (Cotesia + PB). Different letters indicate significant differences between the means (one-way ANOVA, a = 0.05, JMP9). then P. brassicae). A two-way ANOVA conducted with the same dataset gave similar results: there was no effect of the species attacked (P. brassicae or S. littoralis), the order of

6 114 Y. Chabaane et al. (a) (b) Fig 4. (a) Duration (seconds, mean SE) and (b) number of eggs injected (mean SE) per attack during ovipositions of Cotesia glomerata in Pieris brassicae and Spodoptera littoralis. There were no significant differences between the means (One-way ANOVA, a = 0.05, JMP9). attack (P. brassicae then S. littoralis, or S. littoralis then P. brassicae) or the interaction between these two factors, on the duration of C. glomerata attacks ( vs s on P. brassicae and S. littoralis, respectively, F 3,24 = 2.3, P = 0.1) and the number of eggs injected per attack ( vs eggs on P. brassicae and S. littoralis, respectively, F 3,34 = 0.8, P = 0.8) (Fig. 4). We did not find a correlation between the duration of attack and the number of eggs injected for P. brassicae (F 1,13 = 0.4, P = 0.5) and S. littoralis (F 1,10 = 0.1, P = 0.8). IMPACT OF THE PRESENCE OF SPODOPTERA LITTORALIS ON COTESIA GLOMERATA REALIZED FITNESS In tents containing B. rapa plants and, either, P. brassicae alone (pieris treatment), P. brassicae and S. littoralis on separate plants (separate treatment) or P. brassicae and S. littoralis on the same plants (mixed treatment), the mean percentage recovery of caterpillars was and the mean percentage mortality (i.e. caterpillars that died before parasitoid emergence of pupation) was , and there was no difference in recovery nor mortality percentages among treatments (Ps > 0.05). There was a significant effect of herbivore density on parasitism rate by C. glomerata (F 1,27 = 10.14, P = 0.003): parasitism was higher in tents containing fewer hosts ( and per cent parasitism in tents with 40 and 80 hosts, respectively). However, there was no significant effect of treatment (F 2,27 = 1.0, P = 0.4), nor an interaction between treatment and density (F 2,27 = 0.37, P = 0.69) on percentage parasitism (Fig. 5). Discussion Invasions by exotic insect herbivores can have a multitude of direct and indirect effects on native ecosystems at the organismal and community level (Mooney & Cleland 2001; White, Fig 5. Cotesia glomerata parasitism (%, mean SE) on its host Pieris brassicae, in tents containing 40 or 80 hosts, in absence of Spodoptera (Pieris treatment) or in presence of the same number of Spodoptera littoralis, placed on the same plants as P. brassicae (Mixed infestation Treatment) or on separate plants (Separate infestation). There were no significant differences among treatments for each host density (two-way ANOVA, a = 0.05, JMP9). Wilson & Clarke 2006; Kenis et al. 2009). Studies on the effects of exotic herbivores on organisms belonging to the third trophic level (i.e. natural enemies of plant consumers) have mainly focused on the use of exotic prey by natural enemies and the consequences of shared natural enemies for native predator-prey relationships. However, unsuitable exotic prey or host species may also have an impact on native natural enemies (Hoogendoorn & Heimpel 2002; Suttle & Hoddle 2006). Here, we examined the consequences of the presence of an exotic polyphagous insect, the noctuid S. littoralis, on the interactions between native organisms belonging to three trophic levels: the plant B. rapa, the herbivore P. brassicae and its specialist parastoid C. glomerata. Our results show that B. rapa plants infested by S. littoralis were not attractive to C. glomerata wasps, which exhibited a clear preference for plants damaged by their host P. brassicae. However, in the case of simultaneous herbivory by both P. brassicae and S. littoralis, plants under dual herbivory became considerably less attractive than plants damaged by P. brassicae alone (Fig. 1). This may be explained by the fact that the chemical blend emitted by plants infested by both herbivores resembled the blend emitted after herbivory by S. littoralis alone (Fig. 2). Indeed, the blend emitted in response to S. littoralis and to dual herbivory was mainly composed of green leaf volatiles and isothiocyanates, while the blend emitted in response to P. brassicae had a much higher proportion of terpenoids and alkanes (Fig. 2). In other words, S. littoralis dominated the headspace in terms of induced plant volatiles in case of dual herbivory, thus giving a proximal mechanism for the lack of attractiveness of doubly infested plants to C. glomerata. We cannot entirely rule out the possibility that most of the damage was done by S. littoralis in case of dual infestation, leading to a predominance of S. littoralis induced

7 Exotic herbivores disrupt tritrophic interactions 115 volatiles in the headspace. However, the area damaged by P. brassicae and S. littoralis was not significantly different in case of single herbivory (Fig. S3), decreasing the likelihood of this possibility. Results of our behavioural observations on caterpillarinfested B. rapa leaves revealed that C. glomerata occasionally attempted to parasitize S. littoralis (Fig. 3a). These attacks were significantly more frequent when P. brassicae was present on the same leaf (Fig. 3a). Higher attack rates on an unfamiliar host when in presence of the preferred host have been documented for other parasitoids (Bailey 1989; Field & Darby 1991; Barratt et al. 1997; Kitt & Keller 1998). It is unlikely that plant volatiles played a role in close-range foraging decisions, since the volatile profile of doubly infested plants closely resembled that of plant infested only by S. littoralis. It is more likely that Pieris-specific kairomones affected the increased attempts to parasitize Spodoptera larvae. Indeed, it has been suggested that the presence of host kairomones in the direct vicinity of an unfamiliar host may lead parasitoids to mistakenly attack the unfamiliar host (Van Driesche & Reardon 2004) or may elicit an excitatory state leading the parasitoid to become less choosy and accept a broader range of hosts (Dethier, Solomon & Turner 1965). In our study, S. littoralis (Noctuidae, Lepidoptera) was expected to be unsuitable as a host for C. glomerata, which is only known to parasitize closely related caterpillars from the family Pieridae (Lepidoptera). This was confirmed by our results: C. glomerata offspring never emerged from S. littoralis larvae exposed to C. glomerata females for 24 hours. However, the survivorship of S. littoralis larvae was reduced after exposure to C. glomerata, a trend that was accentuated when P. brassicae was present in the test area (Fig. 4c). This decreased survivorship is likely due to failed parasitism (i.e. parasitoid eggs/larvae failing to successfully develop in the host), and these results correlate nicely with our observations of higher attack rates on S. littoralis when P. brassicae was present in the test area (Fig. 3a). A key result of our study is that attacks on P. brassicae and S. littoralis resulted in the same time (attack duration) and resources (number of eggs injected) invested by C. glomerata females (Fig. 4). In addition, the sequential order in which herbivores were attacked by the parasitoids did not affect attack characteristics: this result seems to indicate that there is no change in C. glomerata behaviour induced by an encounter with its preferred host (Van Driesche & Reardon 2004) and that attacks on S. littoralis are not restricted to na ıve C. glomerata females. Laying eggs in an unsuitable host may directly affect the realized reproductive output of C. glomerata under natural conditions (Hoogendoorn & Heimpel 2002), leading to reduced parasitism rates on its native host caterpillars. The terms ecological trap or evolutionary trap have been employed in conservation biology to designate an habitat that is favoured by animals but results in negative population growth rates (Battin 2004) and in the context of biological invasions for plants that are readily accepted by native herbivores for oviposition despite being suboptimal or unsuitable for their offspring (Casagrande & Dacey 2007; Keeler & Chew 2008; Harvey et al. 2010). This notion can be extended to the third trophic level in the case of unsuitable prey or host items accepted by natural enemies. Such interactions have been reported among native species, with consequences on the reproductive output of parasitoids (Bukovinszky et al. 2012). In our study, however, parasitism rates on P. brassicae larvae were unaffected by the presence of S. littoralis, and the non-host had therefore no effect on the realized fitness of C. glomerata in the chosen experimental setup. There was a trend towards reduced parasitism in presence of the exotic herbivore in tents with 40 host herbivores, but this trend was not confirmed with 80 host herbivores (Fig. 5). High variability in parasitoid reproductive success was observed, with values ranging from 0 10 to 100 per cent parasitism for all treatments. Experiments in more realistic settings in terms of foraging space and densities of herbivores may yield different results. In summary, our work demonstrates that exotic insect herbivores unsuitable to native parasitoids have the potential to disrupt existing native parasitoid host interactions via two main mechanisms: interference with the foraging behaviour of the parasitoid and costs associated with unsuccessful attacks of the exotic non-host. In the case of S. littoralis, impact on the interactions between P. brassicae and C. glomerata depends on a critical factor: whether the exotic herbivore infests the plant at the same time as the native herbivore or not. Because plants solely infested by S. littoralis are not attractive to parasitoids and attacks on S. littoralis are scarce in absence of P. brassicae, we can expect S. littoralis to have minimal impact on C. glomerata when infesting B. rapa on its own. On the other hand, plants infested by both herbivores are unattractive to parasitoids, and attacks on S. littoralis are higher in presence of P. brassicae. Thus, C. glomerata reproductive success may be affected and, conversely, P. brassicae may escape parasitism, on doubly infested plants. In a closely related tritrophic system, females of the diamondback moth Plutella xylostella have been shown to prefer plants infested by Pieris rapae for oviposition, possibly because these plants are less attractive to a parasitoid of P. xylostella, Cotesia vestalis (Shiojiri et al. 2002). The field of invasion ecology has grown tremendously over the past two decades, but our understanding of how invasions by exotic insect herbivores may disturb multitrophic interactions in native communities remains limited. Here, we show that non-host exotic herbivores can affect native parasitoids and that some of these effects are chemically mediated through altered emissions of herbivore-induced plant volatiles. It should be noted that there are other ways exotic herbivores may indirectly affect native organisms belonging to the third trophic level. For example, exotic herbivores may induce direct plant defences, affecting the dynamics of colonization or the performance of native herbivores and, indirectly, their natural enemies. Moreover, direct interferences between native and exotic herbivores infesting the same plant may also lead to behavioural changes in the native herbivore and affect the foraging efficiency of natural enemies. To reliably estimate the consequences of these disruptive effects at the population

8 116 Y. Chabaane et al. level, similar empirical studies will have to be conducted in the field. Logically this cannot be done in areas where a potential invader does not occur, but such experiments could be conducted in areas recently invaded by S. littoralis such as the southern regions of Italy and Spain. Acknowledgements We thank Gregory Roeder and Nathalie Veyrat for their help with the chemistry analyses. Tom de Jong provided Brassica seeds and Nicole van Dam provided helpful comments on the manuscript. Louisa Nelle and Frank Lachmuth helped with growing plants and rearing insects. The authors are grateful to the Swiss National Science Foundation, which funded this work in the context of the EUROCORES Programme EuroVOL (project InvaVOL) of the European Science Foundation. References Ahuja, I., Rohloff, J. & Bones, A.M. (2010) Defence mechanisms of Brassicaceae: implications for plant-insect interactions and potential for integrated pest management. A review. Agronomy for Sustainable Development, 30, Bailey, P.T. (1989) The millipede parasitoid Pelidnoptera nigripennis (F.) (Diptera: Sciomyzidae) for the biological control of the millipede Ommatoiulus moreleti (Lucas) (Diplopoda: Julida: Julidae) in Australia. Bulletin of Entomological Research, 79, Barratt, B., Evans, A., Ferguson, C., Barker, G., McNeill, M. & Phillips, C. (1997) Laboratory nontarget host range of the introduced parasitoids Microctonus aethiopoides and M. hyperodae (Hymenoptera: Braconidae) compared with field parasitism in New Zealand. Environmental Entomology, 26, Battin, J. (2004) When good animals love bad habitats: ecological traps and the conservation of animal populations. Conservation Biology, 18, Berkvens, N., Moens, J., Berkvens, D., Samih, M.A., Tirry, L. & De Clercq, P. (2010) Dinocampus coccinellae as a parasitoid of the invasive ladybird Harmonia axyridis in Europe. Biological Control, 53, Bukovinszky, T., Poelman, E.H., Kamp, A., Hemerik, L., Prekatsakis, G. & Dicke, M. (2012) Plants under multiple herbivory: consequences for parasitoid search behaviour and foraging efficiency. Animal Behaviour, 83, Carlsson, N.O., Sarnelle, O. & Strayer, D.L. (2009) Native predators and exotic prey-an acquired taste? Frontiers in Ecology and the Environment, 7, Casagrande, R. & Dacey, J. (2007) Monarch butterfly oviposition on swallowworts (Vincetoxicum spp.). Environmental Entomology, 36, Cronin, J.T. & Haynes, K.J. (2004) An invasive plant promotes unstable hostparasitoid patch dynamics. Ecology, 85, D Alessandro, M. & Turlings, T.C. (2005) In situ modification of herbivoreinduced plant odors: a novel approach to study the attractiveness of volatile organic compounds to parasitic wasps. Chemical senses, 30, Dethier, V., Solomon, R.L. & Turner, L.H. (1965) Sensory input and central excitation and inhibition in the blowfly. Journal of Comparative and Physiological Psychology, 60, 303. Didham, R.K., Tylianakis, J.M., Hutchison, M.A., Ewers, R.M. & Gemmell, N.J. (2005) Are invasive species the drivers of ecological change? Trends in Ecology & Evolution, 20, El-Sayed, A., Suckling, D., Byers, J., Jang, E. & Wearing, C. (2009) Potential of lure and kill in long-term pest management and eradication of invasive species. Journal of Economic Entomology, 102, Fei, M., Gols, R. & Harvey, J.A. (2014) Seasonal phenology of interactions involving short-lived annual plants, a multivoltine herbivore and its endoparasitoid wasp. Journal of Animal Ecology, 83, Field, R. & Darby, S. (1991) Host specificity of the parasitoid, Sphecophaga vesparum (Curtis)(Hymenoptera: Ichneumonidae), a potential biological control agent of the social wasps, Vespula germanica (Fabricius) and V. vulgaris (Linnaeus)(Hymenoptera: Vespidae) in Australia. New Zealand Journal of Zoology, 18, Fortuna, T.M., Vet, L.E. & Harvey, J.A. (2012) Effects of an invasive plant on the performance of two parasitoids with different host exploitation strategies. Biological Control, 62, Gandhi, K.J. & Herms, D.A. (2010) Direct and indirect effects of alien insect herbivores on ecological processes and interactions in forests of eastern North America. Biological Invasions, 12, Hare, J.D. (2011) Ecological role of volatiles produced by plants in response to damage by herbivorous insects. Annual Review of Entomology, 56, Harvey, J.A. & Fortuna, T.M. (2012) Chemical and structural effects of invasive plants on herbivore parasitoid/predator interactions in native communities. Entomologia Experimentalis Et Applicata, 144, Harvey, J.A., Biere, A., Fortuna, T., Vet, L.E., Engelkes, T., Morri en, E., Gols, R., Verhoeven, K., Vogel, H. & Macel, M. (2010) Ecological fits, mis-fits and lotteries involving insect herbivores on the invasive plant, Bunias orientalis. Biological Invasions, 12, Hoogendoorn, M. & Heimpel, G.E. (2002) Indirect interactions between an introduced and a native ladybird beetle species mediated by a shared parasitoid. Biological Control, 25, Keeler, M.S. & Chew, F.S. (2008) Escaping an evolutionary trap: preference and performance of a native insect on an exotic invasive host. Oecologia, 156, Kenis, M., Auger-Rozenberg, M.-A., Roques, A., Timms, L., Pere, C., Cock, M.J., Settele, J., Augustin, S. & Lopez-Vaamonde, C. (2009) Ecological effects of invasive alien insects. Biological Invasions, 11, Kitt, J. & Keller, M.A. (1998) Host selection by Aphidius rosae Haliday (Hym., Braconidae) with respect to assessment of host specificity in biological control. Journal of Applied Entomology, 122, Louda, S.M., Pemberton, R., Johnson, M. & Follett, P. (2003) Nontarget effects-the Achilles heel of biological control? Retrospective analyses to reduce risk associated with biocontrol introductions. Annual Review of Entomology, 48, Mooney, H.A. & Cleland, E. (2001) The evolutionary impact of invasive species. Proceedings of the National Academy of Sciences, 98, Pearson, D.E. & Callaway, R.M. (2005) Indirect nontarget effects of host-specific biological control agents: implications for biological control. Biological Control, 35, Pierre, P.S., Jansen, J.J., Hordijk, C.A., van Dam, N.M., Cortesero, A.-M. & Dugravot, S. (2011) Differences in volatile profiles of turnip plants subjected to single and dual herbivory above-and belowground. Journal of Chemical Ecology, 37, Pimentel, D. (2002) Biological Invasions: Economic and Environmental Costs of Alien Plant. Animal, and Microbe Species. CRC, Boca Raton, FL. Redman, A.M. & Scriber, J.M. (2000) Competition between the gypsy moth, Lymantria dispar, and the northern tiger swallowtail, Papilio canadensis: interactions mediated by host plant chemistry, pathogens, and parasitoids. Oecologia, 125, Settle, W. & Wilson, L. (1990) Invasion by the variegated leafhopper and biotic interactions: parasitism, competition, and apparent competition. Ecology, 71, Shiojiri, K., Takabayashi, J., Yano, S. & Takafuji, A. (2002) Oviposition preferences of herbivores are affected by tritrophic interaction webs. Ecology Letters, 5, Sloggett, J.J. (2010) Predation of ladybird beetles by the orb-web spider Araneus diadematus. BioControl, 55, Suttle, K.B. & Hoddle, M.S. (2006) Engineering enemy-free space: an invasive pest that kills its predators. Biological Invasions, 8, Timms, L.L., Walker, S.C. & Smith, S.M. (2012) Establishment and dominance of an introduced herbivore has limited impact on native host-parasitoid food webs. Biological Invasions, 14, Ton, J., D Alessandro, M., Jourdie, V., Jakab, G., Karlen, D., Held, M., Mauch- Mani, B. & Turlings, T.C. (2007) Priming by airborne signals boosts direct and indirect resistance in maize. The Plant Journal, 49, Turlings, T.C., Davison, A. & Tamo, C. (2004) A six-arm olfactometer permitting simultaneous observation of insect attraction and odour trapping. Physiological Entomology, 29, Van Driesche, R.G. & Reardon, R. (2004) Assessing Host Ranges for Parasitoids and Predators Used for Classical Biological Control: A Guide to Best Practice. Forest Health Technology Enterprise Team, Morgantown, West Virginia. Verhoeven, K.J., Biere, A., Harvey, J.A. & Van Der Putten, W.H. (2009) Plant invaders and their novel natural enemies: who is naive? Ecology Letters, 12, White, E.M., Wilson, J.C. & Clarke, A.R. (2006) Biotic indirect effects: a neglected concept in invasion biology. Diversity and Distributions, 12, Wilcove, D.S., Rothstein, D., Dubow, J., Phillips, A. & Losos, E. (1998) Quantifying threats to imperiled species in the United States. BioScience, 48, Received 14 February 2014; accepted 23 July 2014 Handling Editor: Ryan Phillips

9 Exotic herbivores disrupt tritrophic interactions 117 Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. Experimental design used for the experiment evaluating the realized impact of the presence of Spodoptera littoralis on the fitness of Cotesia glomerata in tents. Figure S3. Area damaged by caterpillars (cm 2, mean SE). Brassica rapa plants damaged for h by P. brassicae (PB), S. littoralis (SL), and both P. brassicae and S. littoralis (PB+SL). Table S1. Compounds identified in the blends of undamaged and herbivore-infested B. rapa plants after 5 hours of volatiles collection. Figure S2. Quantity of compounds (mean SE, ng internal standard equivalents) belonging to four classes of plant volatiles emitted by B. rapa plants in response to different treatments.