Bacterial volatiles elicit differential olfactory responses in insect species from the same and different trophic levels

Insect communities consist of species from several trophic levels that have to forage for suitable resources among and within larger patches of nonresources. To locate their resources, insects use diverse stimuli, including olfactory, visual, acoustic, tactile and gustatory cues. While most research has focused on cues derived from plants and other insects, there is mounting evidence that insects also respond to volatile organic compounds (VOCs) emitted by microorganisms. However, to date little is known about how the olfactory response of insects within and across different trophic levels is affected by bacterial VOCs. In this study, we used Y‐tube bioassays and chemical analysis of VOCs to assess how VOCs emitted by bacteria affect the olfactory response of insects of the same and different trophic levels. Experiments were performed using two aphid species (Amphorophora idaei Börner and Myzus persicae var. nicotianae Blackman), three primary parasitoid species (Aphidius colemani Viereck, A. ervi Haliday, and A. matricariae Viereck), and two hyperparasitoid species (Asaphes suspensus Nees and Dendrocerus aphidum Rondani). Olfactory responses were evaluated for three bacterial strains (Bacillus pumilus ST18.16/133, Curtobacterium sp. ST18.16/085, and Staphylococcus saprophyticus ST18.16/160) that were isolated from the habitat of the insects. Results revealed that insects from all trophic levels responded to bacterial volatiles, but olfactory responses varied between and within trophic levels. All bacteria produced the same set of volatile compounds, but often in different relative concentrations. For 11 of these volatiles we found contrasting correlations between their concentration and the behavior of the primary parasitoids and hyperparasitoids. Furthermore, olfactometer experiments on three of these compounds confirmed the contrasting olfactory responses of primary parasitoids and hyperparasitoids. The potential of these findings for the development of novel semiochemical‐based strategies to improve biological aphid control has been discussed.


Introduction
Insect communities typically consist of species from several different trophic levels that have to forage for suitable resources that are commonly embedded within larger patches of nonresources (Aartsma et al., 2017).Herbivorous insects, for example, need to find their food plants among a diverse array of nonfood plants, whereas predators and parasitoids have to find herbivore-infested plants as well as the actual prey or hosts on them (Aartsma et al., 2019).To locate their resources, insects use a variety of stimuli, including olfactory, visual, acoustic, tactile and gustatory cues (Visser, 1988;Vinson, 1998;Little et al., 2019).The use of olfactory cues during long-range foraging is widely distributed among insects, although the precise cues attracting insects can be expected to differ between trophic levels (Bruce et al., 2005;Webster & Cardé, 2017;Aartsma et al., 2019).Whereas herbivorous insects predominantly use plant volatiles and colors to find suitable host plants during long-range foraging, higher trophic levels such as parasitoids mainly use herbivore-induced plant volatiles (HIPVs) and volatiles from herbivore feces, as well as visual and mechanosensory cues to locate suitable hosts (Wäckers & Lewis, 1994;Fischer et al., 2001;van Oudenhove et al., 2017).
Most research on insect olfactory behavior has focused on volatiles derived from plants and insects (Kaplan, 2012;Meiners & Peri, 2013).However, there is mounting evidence that insects also respond to volatile organic compounds (VOCs) emitted by microorganisms (Leroy et al., 2011a,b;Davis et al., 2013;Dzialo et al., 2017), and that these volatiles play an important role in mediating chemical interactions between plants and insects, as well as among insects (Schulz-Bohm et al., 2017;Weisskopf et al., 2021).For example, it has been shown that insects respond to microbial VOCs to locate suitable food sources, hosts or oviposition sites (Leroy et al., 2011a;Davis et al., 2013;Dzialo et al., 2017).Most research in the field of insect-microbe interactions has focused on yeasts, which have generally been found to attract insects (Becher et al., 2018;Madden et al., 2018).This chemical communication between yeasts and insects is believed to be of mutual benefit, where the insects benefit from the microorganisms signaling suitable sugar resources and the yeasts profit from the insects by being transported to new habitats where they can continue to proliferate or complete their life cycle (Christiaens et al., 2014;Madden et al., 2018).Similarly, an increasing number of studies have shown chemical interactions between insects and bacteria, particularly between bacteria and Diptera, Hymenoptera, Coleoptera, and Orthoptera (Leroy et al., 2011b).
Recent research testing a large set of bacteria, isolated from diverse origins (i.e., different insect species, honeydew and aphid mummies), for their ability to mediate insect behavior has shown that insect olfactory responses to bacterial volatiles can differ among bacterial species, varying between attraction and repellence (Goelen et al., 2020a(Goelen et al., , 2020b)).Additionally, it has been found that volatiles from bacteria found in aphids and their honey-dew may inform the foraging behavior of mutualistic insects such as ants (Fischer et al., 2015).This indicates that insects can respond strongly to volatiles produced by bacteria that live on, in, or near food sources, hosts or prey (Leroy et al., 2011a,b).Furthermore, recent Ytube olfactometer bioassays suggest that insect responses to bacterial volatiles differ across trophic levels (Goelen et al., 2020a).However, as this study was only performed with two insect species (i.e., the primary parasitoid Aphidius colemani Viereck (Hymenoptera: Braconidae) and the hyperparasitoid Dendrocerus aphidum Rondani (Hymenoptera: Megasplidae)), specificity of olfactory response to bacterial VOCs within and across multiple trophic levels remains largely unknown.
The objective of this study was to investigate the role of bacterial VOCs in mediating the olfactory response of insects from different trophic levels.Specifically, we asked whether VOCs emitted by bacteria elicit the same olfactory responses in insects from the same and different trophic levels.Furthermore, we asked whether insect olfactory responses can be related to specific compounds within the bacterial VOC blends.To answer these questions, we here used a study system with different species of the aphid-primary parasitoid-hyperparasitoid food web.This food web has been used as a model system in ecological studies, partly because of the economic worldwide importance of aphids as pests and their natural enemies as biocontrol agents on a variety of agricultural crops (van Emden & Harrington., 2007;Dedryver et al., 2010;van Lenteren, 2012), but also because of the relative ease of rearing aphids, their primary parasitoids, and hyperparasitoids.Olfactory response was evaluated for VOC blends from three bacterial strains isolated from two aphid species and one parasitoid species that were collected from the habitat of the studied insects, and which were previously shown to affect the olfactory behavior of either the primary parasitoid A. colemani or the hyperparasitoid D. aphidum.The composition of the volatile blends was analyzed using gas chromatography-mass spectrometry (GC-MS) to identify compounds which may be correlated with insect olfactory response.Finally, for a number of compounds, another set of olfactory bioassays was performed to confirm their role in the olfactory behavior of the insects.

Study organisms
Experiments were performed using seven insect species belonging to three different trophic levels associated with the cultivation of sweet pepper (Capsicum annuum) or raspberry (Rubus idaeus).These included two aphid species, three primary parasitoids and two hyperparasitoids.Specifically, experiments were conducted using apterous adults of Amphorophora idaei Börner and Myzus persicae var.nicotianae Blackman (both Hemiptera: Aphididae); adult females of their primary parasitoids A. colemani, A. ervi Haliday, and A. matricariae Viereck (Hymenoptera: Braconidae); and adult females of the hyperparasitoids Asaphes suspensus Nees (Hymenoptera: Pteromalidae) and D. aphidum.The European large raspberry aphid A. idaei is a specialist on R. idaeus plants (Blackman et al., 1977), while M. persicae var.nicotianae is a subspecies of the green peach aphid M. persicae, one of the most polyphagous aphid species.The subspecies M. persicae var.nicotianae is found mainly on tobacco and sweet pepper, although it can survive and reproduce on a relatively wide range of plant species (Clements et al., 2000).Aphidius spp.are generalist aphid parasitoids, commonly used in commercial augmentative biological control (Yano, 2006).Dendrocerus aphidum and A. suspensus are generalist, secondary idiobiont ectoparasitoids attacking prepupal and pupal stages of primary aphid parasitoids, such as Aphidius spp., inside mummified aphids (Walker & Cameron, 1981;Höller et al., 1993).

Production of VOCs
For production of VOCs, the procedure of Goelen et al. (2020a) was followed.In brief, cryopreserved stock cultures were plated on tryptic soy agar (TSA; Oxoid) and incubated at 25°C for 24 h.Subsequently, strains were restreaked on the same medium and incubated at 25°C for another 24 h.Next, a single colony was inoculated in 10 mL TSB and incubated overnight at 25°C at 120 r/min.After incubation, cells were washed twice and diluted in sterile physiological water (0.9% NaCl) until an optical density (OD 600 nm) of 1 was reached.Next, 1.5 mL of the cell suspension was inoculated in a 250 mL-Erlenmeyer flask containing 150 mL filter-sterilized GYP25 medium, a medium of 5% w/v glucose (Sigma-Aldrich), 0.5% w/v peptone (Bac-toTMPeptone; BD Biosciences) and 0.25% w/v yeast extract (Sigma-Aldrich).The medium represents a general medium for bacterial growth and was selected to ensure abundant bacterial growth and VOC production, while the noninoculated medium had no significant effect on insect olfactory response (Goelen et al., 2020a(Goelen et al., , 2020b)).Flasks were sealed with sterile silicone plugs and incubated for 48 h at 25°C at 120 r/min.Fermentations were set up in triplicate, and noninoculated, blank GYP25 medium was included as a negative control (also in triplicate; sterility of the medium was confirmed following incubation by plating on TSB).After incubation, the media were centrifuged for 15 min at 10 000× g.Subsequently, collected supernatants were filter-sterilized to obtain cell-free test media containing the VOCs.The cell-free samples were then stored in small aliquots in sterile, amber glass vials at −20°C until further use.

Olfactometer bioassays
To investigate the olfactory response of the insects, a Y-tube olfactometer bioassay was performed as described previously (Goelen et al., 2020a).The Y-tube (stem: 20 cm; arms: 12 cm with a 60°angle at the Y-junction; inner diameter: 1.5 cm) was put on a table that was homogeneously illuminated by four 24W T5 TLfluorescent tubes (16 × 549 mm, 1350 Lumen, 5500K, True-Light ® , Naturalite Benelux) with a 96% color representation of true day light at a height of 45 cm.Furthermore, the Y-tube was mounted at a 20°incline to stimulate movement of the insects toward the bifurcation, and charcoal-filtered air was led through each arm of the Y-tube at a speed of 400 mL/min.To eliminate any visual cues, the olfactometer was fully enclosed with white curtains.Olfactory response was investigated by loading 150 μL of the cell-free cultivation medium onto a filter paper (diameter: 37 mm; Macherey-Nagel, Düren, Germany) and subsequently placing it in one of the odor chambers, while a second filter paper loaded with 150 μL of the control GYP25 medium was placed in the second odor chamber.Insects were tested in 24 cohorts of five adult individuals and were released at the base of the stem section of the olfactometer.Olfactory response was evaluated 10 min after parasitoid or hyperparasitoid release (Goelen et al., 2020a), or 20 min after aphid release, which was found to be the optimal time point of evaluation in preliminary experiments.Although different set-ups can be used to assess the olfactory behavior of aphids, previous studies have identified glass Y-tubes to not only be suited to assess the olfactory response of primary and secondary parasitoids, but also aphids (Fernández-Grandon et al., 2013;Wilberts et al., 2022).Individuals that had passed a set line in one of the olfactometer arms (1 cm from the Y-junction) at the time of evaluation were considered to have chosen the odor source presented by that olfactometer arm (Goelen et al., 2020a).All other insects were considered nonresponding individuals and were eliminated from statistical analysis.For every bioassay, new individuals were used.After every two releases, the filter papers in both odor chambers were replaced with fresh filter papers with 150 μL cell-free cultivation medium or control medium.Furthermore, to compensate for unforeseen asymmetry in the setup, we swapped the odor chambers after releasing six cohorts.At the same time, the Y-tube glassware was replaced by clean glassware.At the end of the bioassay, all olfactometer parts were cleaned with tap water, distilled water, acetone, and finally pentane, after which the different parts were put in an oven at 150°C until the next day.All bioassays were conducted at 23 ± 1°C and 65% ± 5% RH between 08:00 and 17:00.As the VOC composition of all three biological replicates was highly similar (see further), olfactory response was determined for one of the three biological replicates.
The same experimental set-up was used to evaluate insect response to a number of individual volatiles that were correlated with the olfactory response of A. colemani and D. aphidum, i.e., n-hexanol, linalool and acetic acid (see further).Compounds were purchased from Sigma-Aldrich (Saint-Louis, MO, USA), and all had ≥99.0%purity.Insect response was assessed as described previously (Goelen et al., 2021).Compounds were first dissolved in diethylether in three concentrations: 10 μg/mL, 100 μg/mL, and 1 mg/mL.Next, 10 μL was loaded onto filter paper, and 30 s later (allowing the diethyl ether to evaporate) presented in one of the olfactometer odor chambers (resulting in a dose of 100 ng, 1 μg, and 10 μg, respectively).A filter paper with 10 μL of diethyl ether was placed in the other chamber as a control (Goelen et al., 2021).The experiment was performed as described above, with the only exception that in this case 12 cohorts of five adult individuals were tested, and that it was performed for only M. persicae var.nicotianae, A. colemani, and D. aphidum.

Chemical analysis of VOCs
For each biological replicate of the bacterial fermentations and control medium, the VOC composition was analyzed by headspace solid phase micro-extraction gas chromatography followed by mass spectrometry analysis (HS-SPME-GC-MS).Volatile compounds were separated by a Thermo Trace 1300 GC system equipped with a MXT-5 column (30 m length × 0.18 mm inner diameter × 0.18 μm film thickness; Restek, Bellefonte, Pennsylvania, USA) and subsequently detected by an ISQ mass spectrometer.To this end, first 5 mL of each cell-free sample was supplemented with 1.75 g NaCl and subsequently incubated at 60°C under constant agitation in a TriPlus RSH SPME auto sampler (Thermo Fisher Scientific, Waltham, Massachusetts, USA).The HS-SPME volatile collection was conducted using a 50/30 μm DVB/CAR/ PDMS coating fiber (Supelco, Bellefonte, Pennsylvania, USA).After five min of equilibration, the SPME fiber was exposed to the headspace sample for 30 min.The compounds trapped on the fiber were thermally desorbed in the injection port of the chromatograph by heating the fiber for 15 min at 270°C.Further volatile collection and separation conditions were as described by Goelen et al. (2020a).Compounds were identified and quantified as in Reher et al. (2019).Chromatograms were analyzed with amdis v2.71 (Stein, 1999) to deconvolute overlapping peaks.Obtained spectra were subsequently annotated using the NIST MS Search v2.0g software, using the NIST2011, FFNSC and Adams libraries.This yielded a list of 244 tentatively identified compounds across all bacterial VOC blends and the noninoculated GYP25 medium.Peak areas of these compounds were compared to the background signal, which was identified by running a GC-MS with a sample obtained from five milliliters of demineralized water and 1.75 g of NaCl (cfr.procedure outlined above).The background signal was then subtracted from the peak areas of the corresponding tentatively identified compounds.When the difference in peak area fell below a set threshold of 1000, compounds were eliminated from further analysis.This yielded a list of 66 different, tentatively identified compounds.To further extract and integrate the compound elution profiles, a file was used with these 66 tentatively identified compounds containing the expected retention times and spectrum profiles.Extraction was performed for every compound in every chromatogram over a time-restricted window using weighted nonnegative least square analysis (Lawson & Hanson, 1995), and for every compound, the peak areas were computed from the extracted profiles.Finally, relative peak areas were computed as the ratio between the compounds' peak area and the total peak area per sample, and then multiplied by 100 (%).Relative peak areas were then used for further analysis and summarized in Table S1 (Supporting Information).

Statistical analysis
Insect olfactory response to the bacterial VOC blends was analyzed using a Generalized Linear Mixed Model (GLMM) based on a binomial distribution with a logit link function (logistic regression) using bacterial strain as fixed factor (performed in R using the glmer function from the lme4 package).Each release of one cohort of five insects served as a replicate.To prevent pseudoreplication and to adjust for overdispersion, the release of each cohort was included in the model as a random factor.The number of insects choosing either the control or the bacterial VOCs in each cohort was entered as a response variable.Insect preference was examined by testing the null hypothesis (H o ) that insects showed no preference for any olfactometer arm (i.e., 50 : 50 response) by testing H o : logit = 0, which equals a 50 : 50 distribution.Results were presented by calculating the preference index (PI) by dividing the difference between the number of insects choosing the bacterial odors and the number of insects choosing the control odors, by the total number of responding insects.Furthermore, an analysis of variance Type III Wald's chi-square test was performed on the GLMM to determine if there was an overall difference between the olfactory responses for the different VOC blends.A GLMM was also used to determine whether insect species or trophic level affected the olfactory response of the insects.Once again the number of insects choosing either the control or treatment side in each cohort was included as a dependent variable, and insect species or trophic level were included as fixed factors.Finally, interaction GLMM models were created to determine the interaction effect of insect species or trophic level, and the different VOC blends on the olfactory response of the insects.The number of insects choosing either the control or treatment side in each cohort was included as a dependent variable, while the interaction between either insect species and the different VOC blends, or insect trophic level and the different VOC blends were included as fixed factors.The release of each cohort was included in the model as a random factor.To analyze the effect of compound, the dose of each compound and the insect species on the insects' olfactory response to the individual compounds presented (n-hexanol, linalool and acetic acid), an analysis of variance Type III Wald's chisquare test was performed.We included the investigated compounds, doses, and insect species as fixed factors in a full factorial model and the number of insects choosing either the control or treatment side in each cohort as a dependent variable.Furthermore, the release of each cohort was included in the model as a random factor.
Differences in VOC composition of the bacterial VOC blends and control medium were visualized by constructing a heatmap from the strain × relative peak area matrix of calculated Z-scores using the heatmap.2function in the gplots package V3.1.0 in R (Warnes et al., 2016).For each compound in each sample, the Z-scores were calculated by subtracting the mean relative peak area of all samples for the respective compound from the relative peak area of that compound in the sample, and dividing that by the respective standard deviation.Furthermore, a nonmetric multidimensional scaling (NMDS) plot was made from the strain × relative peak area matrix by using a Bray-Curtis distance matrix in the Vegan package in R (Oksanen et al., 2017).Additionally, a permutational multivariate analysis of variance (perMANOVA) was carried out to test for significant differences in the chemical composition of the VOC blends produced by the tested strains, based on 1000 permutations.The analysis was performed using the adonis function from the Vegan package in R. Finally, differences in the relative peak area of different compounds were calculated by first checking for normality using a Levene's test.When the normality assumption was violated, a Kruskal-Wallis one-way analysis of variance (ANOVA), followed by a Dunn's test was performed.When the normality assumption was met, a univariate ANOVA followed by a Tukey's HSD test was performed.
Pearson correlation tests were performed to investigate correlations between insect olfactory response and the VOC composition.Calculations were performed using the data obtained from the biological replicate which was used in the olfactometer experiments and for the VOC analysis.For each insect species and each VOC blend tested, we computed the individual PI values for each of the 24 cohorts of five insects, and correlated this to the relative peak area of each of the volatile compounds in the VOC blends (66 compounds × 4 relative peak areas [for 3 bacteria and 1 blank, noninoculated medium] in total).To control for multiple comparisons, the P values were adjusted using the Benjamini-Hochberg procedure, for the correlations of each individual species (Benjamini & Hochberg, 1995).This was done by multiplying the ratio between the individual rank of the P value and the number of hypotheses (individual correlations per species = 66), by the false discovery rate.We defined the false discovery rate at 5% (Benjamini & Hochberg, 1995).In all analyses, a significance level of α = 0.05 was used to determine significant effects.All statistical analyses were performed in R 4.0.4(R Core Development Team, 2019).

Olfactory response to bacterial VOC blends
Insect olfactory response to the volatile emissions of the tested bacterial strains and control varied significantly between the different treatments (χ 2 (3) = 10.636,P = 0.014).Olfactory response did not significantly differ between insect trophic levels (χ 2 (2) = 4.688, P = 0.096) or between insect species (χ 2 (6) = 6.774,P = 0.342).However, olfactory response did significantly differ for the interaction between treatment and insect trophic level (χ 2 (11) = 29.686,P = 0.002), as well as for the interaction between treatment and insect species (χ 2 (27) = 57.133,P < 0.001) (Table 1).Amphorophora idaei significantly preferred the VOC blend of the Curtobacterium sp.strain over the blank medium (PI = 0.284, P = 0.012).However, a neutral response (i.e., no preference) was obtained when A. idaei was tested against the other bacterial strains.By contrast, M. persicae var.nicotianae significantly preferred the VOC blend of the B. pumilus strain (PI = 0.291, P = 0.011).A neutral response was found when M. persicae var.nicotianae was tested against the VOCs of the S. saprophyticus strain (PI = 0.012, P = 0.913) and the Curtobacterium sp.strain (PI = 0.215, P = 0.058) (Fig. 1).Out of the primary parasitoids tested, A. colemani showed a significant positive response to B. pumilus (PI = 0.308, P = 0.011) and a significant negative response to the S. saprophyticus strain (PI = −0.284,P = 0.016).Unlike A. colemani, A. matricariae showed a significant preference for the Curtobacterium strain (PI = 0.309, P = 0.006), while a neutral response was obtained for the two other bacterial strains.Similar results were obtained for A. ervi, although effects were less pronounced and not significant (Fig. 1).The hyperparasitoid D. aphidum showed a significant preference for the Curtobacterium (PI = 0.224, P = 0.041) and S. saprophyticus strains (PI = 0.266, P = 0.020) over the blank medium, but preferred the blank medium over the B. pumilus strain (PI = −0.224,P = 0.041).Finally, A. suspensus showed no significant response for any of the bacteria tested compared to the blank medium (Fig. 1).

VOC composition
The composition of the VOC blends was highly similar among the different biological replicates (Fig. 2; Table S1).Furthermore, the bacterial species tested released different VOC blends (perMANOVA: F = 71.08,P = 0.001).This is also displayed in the NMDS ordination plot of the VOC composition which separated all strains from the blank, noninoculated medium to a roughly equal degree along the first NMDS axis, while the strains were separated from each other along the second NMDS axis (Fig. 2).At class level, in comparison with the blank medium, overall bacterial strains produced higher amounts of ketones, organic acids and terpene-derived compounds (Fig. S1; Table S1).The Fig. 1 Olfactory response of the aphid species Amphorophora idaei and Myzus persicae var.nicotianae; their primary parasitoids Aphidius colemani, A. ervi, and A. matricariae; and the hyperparasitoids Asaphes suspensus and Dendrocerus aphidum when given a choice between the odor of a bacterial test strain grown in GYP25 medium and the odor of the blank GYP25 medium in a Y-tube olfactometer.Bacterial strains tested included Bacillus pumilus ST18.16/133,Curtobacterium sp.ST18.16/085, and Staphylococcus saprophyticus ST18.16/160.Insect response is expressed as the preference index, which is calculated by dividing the difference between the number of insects choosing the bacterial odor and the number of insects choosing the control odor, by the total number of responding insects.In total, for each insect species 24 releases of 5 individuals were included in the assay.Nonresponders were excluded from the statistical analysis.Error bars represent standard error of the mean.Overall insect responsiveness was 71.0%, and ranged between 67.7% and 72.3% (indicated in green in the pie charts).Curtobacterium sp.strain produced eight compounds (4-methyl-2-propyl-1-pentanol, nonan-2-ol, 2phenylethanol, 3,5-dimethyl-benzaldehyde, decyl acetate, acetophenone, 3-methyl-2-buten-1-ol, and camphor) in a significantly higher relative concentration than the other strains and the blank medium.The B. pumilus strain produced two compounds (5,5-dimethyl-2,4-hexanedione and acetoin) in a significantly higher relative concentration than the other strains and the blank medium.The S. saprophyticus strain did not produce any compounds in a significantly higher relative concentration than the other strains and the blank medium.Finally, the blank medium emitted two compounds in significantly higher relative amounts compared to the bacterial treatments (isobutyryl chloride and 2-methyl-5-(1-methylethyl)-pyrazine), most probably because these compounds were further metabolized in the bacterial treatments (Table S1).

Correlations between insect behavior and VOC composition
To identify potential active compounds, a correlation analysis between the olfactory response of the insects and the relative peak area of the compounds in the VOC blends was performed.Out of the 66 tentatively identified VOCs, the relative peak area of 49 compounds was significantly correlated with the olfactory response of A. idaei, M. persicae var.nicotianae, A. colemani, A. ervi, A. matricariae, and D. aphidum (Pearson's correlation, P < 0.05, r > 0.20 or r < −0.20), while the response of A. suspensus was not correlated with any of the compounds detected (Fig. 3; Tables S2−S7).Among these, 11 compounds were significantly correlated with the olfactory behavior of one species, while 34 and four compounds were significantly correlated with two and three insect species, respectively (Fig. 3; Tables S2−S7).Interestingly, all compounds that were significantly correlated with both the primary parasitoid A. colemani and the hyperparasitoid D. aphidum (11 compounds) showed an opposite correlation, where the behavior of A. colemani was either positively correlated to the compound concentration and D. aphidum negatively, or vice versa.In total, nine compounds were positively correlated with the primary parasitoid, while they were negatively correlated with the hyperparasitoids.Among these, the three compounds with the highest difference in correlation coefficient with a positive correlation for A. colemani and a negative correlation for D. aphidum were n-hexanol (A.colemani: P < 0.001, r = 0.363; D. aphidum: P < 0.001, r = −0.349),3-methyl-pyruvic acid (A.colemani: P < 0.001, r = 0.348; D. aphidum: P < 0.001, r = −0.341),and butyl propanoate (A.colemani: P < 0.001, r = 0.350; D. aphidum: P = 0.001, r = −0.319).The compounds with a significant positive correlation for D. aphidum and a negative correlation for A. colemani were linalool (A.colemani: P = 0.005, r = −0.288;D. aphidum: P < 0.001, r = 0.345) and acetic acid (A.colemani: P = 0.005, r = −0.284;D. aphidum: P = 0.001, r = 0.329) (Fig. 3; Tables S4 and S7).

Discussion
In this study, we assessed the effects of three bacteria on the olfactory behavior of seven insect species from three trophic levels.Results showed that insects Fig. 3 Correlation matrix between insect olfactory response (preference index) and the tentatively identified VOCs in the cell-free media of the three tested bacterial strains and the blank GYP25 medium (n = 4).Investigated insect species included the aphid species Amphorophora idaei and Myzus persicae var.nicotianae; their primary parasitoids Aphidius colemani, A. ervi, and A. matricariae; and their hyperparasitoids Asaphes suspensus and Dendrocerus aphidum.The investigated bacterial strains were Bacillus pumilus ST18.16/133,Curtobacterium sp.ST18.16/085, and Staphylococcus saprophyticus ST18.16/160.Correlations between the calculated preference index of each insect cohort and the relative peak area of each compound in the volatile blends were calculated using Pearson's correlation coefficients (r).Significant correlations (P < 0.05) are indicated with an asterisk.Corrected P values and Pearson's correlation coefficients (r) for significant correlations are given in Tables S2-S7.responded to bacterial volatiles, but responses varied between and within trophic levels.For example, while the VOC blend produced by S. saprophyticus was repellent for the primary parasitoids (especially A. colemani), it was significantly attractive for the hyperparasitoid D. aphidum and to a lesser extent for A. suspensus.Likewise, the B. pumilus strain was significantly attractive for M. persicae var.nicotianae and its primary parasitoid A. colemani, but appeared significantly deterrent for D. aphidum.These findings are in agreement with a preliminary study where A. colemani and D. aphidum were tested against the same bacteria (Goelen et al., 2020a).By contrast, the Curtobacterium sp.strain evoked a neutral to positive response (with the exception of A. suspensus which yielded a slightly negative preference index) across all trophic levels.This result resembles earlier findings showing that the pupal Drosophila parasitoid Trichopria drosophilae is attracted to the same yeast odors that their adult hosts prefer (Ðurović et al., 2021).Drosophila flies are consistently associated with a number of yeasts (Chandler et al., 2012) providing mutual benefits.Volatiles produced by the yeasts are a strong and reliable signal for the flies indicating the presence of an available resource like sugar-rich food.The otherwise immotile yeasts in turn benefit from getting dispersed to another patch of sugar or habitat (Becher et al., 2012;Christiaens et al., 2014;Madden et al., 2018).Whether such scenario also exists for bacteria remains unclear so far.Nevertheless, as parasitoids have great capability to learn to associate chemical cues with rewards (Petitt et al., 1992;Vet, et al., 1995;Olson et al., 2003), it is very likely that they could have evolved a specific response to volatiles from microbes that are associated with their hosts (Lewis & Lizé, 2015), which may lead to similar responses over trophic levels.Although Curtobacterium strains have been found in many hemipteran insects (Gai et al., 2011;Azevedo et al., 2016), including aphids (Goelen et al., 2020a;He et al., 2021), it is not yet clear whether such strong relationships exist between the Curtobacterium sp. and insect species investigated in this study.Our results also show that insect species from the same trophic level, and even from the same genus, responded differently to bacterial VOCs.This can be illustrated for the B. pumilus strain, evoking a positive or neutral response within the tested aphids and Aphidius parasitoids.Likewise, the VOC blend was highly repellent for D. aphidum, while attractive (albeit not significantly; P = 0.068) to A. suspensus.Similar trends were observed for S. saprophyticus.
Although VOC composition differed significantly between bacterial strains, the tested strains emitted the same set of VOCs, but often in different relative amounts.The Curtobacterium sp.strain produced eight compounds in a significantly higher amount compared to the other strains and blank, noninoculated medium, whereas the B. pumilus strain produced two compounds in a significantly higher amount than the other strains and blank medium.By contrast, the S. saprophyticus strain produced no compounds in a significantly higher amount than the other strains.This suggests that the bacterial VOCs may elicit a different response in insects depending on the concentration of the VOCs and the composition of the blend, most probably determined by the presence of particular compounds at high concentrations or Fig. 4 Olfactory response of the aphid Myzus persicae var.nicotianae, its primary parasitoid Aphidius colemani, and the hyperparasitoid Dendrocerus aphidum when given a choice between an individual compound dissolved in diethyl ether and diethyl ether as a control in a Y-tube olfactometer.Individual compounds included acetic acid, linalool, and n-hexanol and were tested in three doses (100 ng, 1 μg, and 10 μg).Insect response is expressed as the preference index, which is calculated by dividing the difference between the number of insects choosing the individual compounds offered and the number of insects choosing the control odor, by the total number of responding insects.In total, for each insect species 12 releases of 5 individuals were included in the assay.Nonresponders were excluded from the statistical analysis.P values marked in bold indicate a significant olfactory response (P < 0.05) when compared to a 50 : 50 distribution.Error bars represent standard error of the mean.Overall insect responsiveness was 73.9%, and ranged between 63.3% and 86.7% (indicated in green in the pie charts).specific ratios between compounds (Bruce et al., 2005;Webster et al., 2010;McCormick et al., 2014;Takemoto & Takabayashi, 2015;Liu et al., 2019).When zooming in at the VOC composition of the tested strains, 49 compounds were significantly linked to insect behavior, among which 38 could be linked with the behav-ior of more than one insect species.Interestingly, all compounds that correlated with both A. colemani and D. aphidum behavior (11 compounds) showed an opposite pattern, where the behavior of A. colemani was either positively correlated to the compound concentration while it was negatively correlated to the behavior of D. aphidum, or vice versa.This contrasting correlation was confirmed for a selection of compounds in Y-tube olfactometer experiments, but results were dependent on the dose of compounds tested.A similar dose-dependent effect of individual compounds has been found in a number of other studies (Shiojiri et al., 2010;Takemoto & Takabayashi, 2015;Goelen et al., 2021).Interestingly, the olfactory response of the insects toward the investigated compounds was stronger than that to the bacterial VOC blends.This suggests that the bacterial VOC blends may contain certain compounds that have an inhibitory or masking effect on the active compounds in the blend, or that they occurred in suboptimal concentrations (Cha et al., 2013;Verschut et al., 2019).Two compounds present in the bacterial VOC blends, linalool and acetic acid, were significantly positively correlated with the olfactory response of D. aphidum, while negatively correlated with the olfactory response of A. colemani and resulted in a similar response when used individually.Linalool is a volatile monoterpenoid alcohol that mediates interactions between plants and pollinators, herbivores, carnivores, and microbes (Raguso, 2016) and has been shown to be attractive or repellent to many insect species (McCormick et al., 2012;Takemoto & Takabayashi, 2015).Acetic acid is a volatile carboxylic acid and is a common bacterial fermentation product from the oxidation of ethanol (Sievers & Swings, 2005).It has been found to be an important attractant for Drosophila spp.and many other insect species when used in conjunction with other volatile compounds (El Sayed et al., 2005;Mansourian & Stensmyr, 2015).By contrast, nine compounds were significantly positively correlated with the olfactory response of A. colemani, while they were negatively associated with D. aphidum.Among these, n-hexanol, an organic alcohol, was the compound with the largest difference in correlation coefficient between both insect species, and also appeared to evoke a contrasting response in both insect species when tested individually.n-hexanol is a common plant and microbial volatile compound (Knudsen et al., 2006;Weisskopf et al., 2021) and has been shown to affect the olfactory behavior of insect species in the orders Hemiptera, Hymenoptera, and Lepidoptera (Wager & Breed, 2000;von Arx et al., 2011;Koczor et al., 2021).In general, the volatile compounds identified in this study are common compounds found in the volatile blends of many microorganisms.However, as the volatiles produced by bacteria can be influenced by both abiotic factors such as growth medium composition (Blom et al., 2011;Garbeva et al., 2014) and biotic factors such as growth phase and cell physiology (Kai et al., 2010), it has to be noted that the volatiles identified in our study may be different com-pared to the volatiles that the bacteria produce in their natural habitats.
Volatile compounds evoking a differential response in parasitoids and hyperparasitoids could be particularly interesting for improving present-day aphid biocontrol.Hyperparasitoids can disrupt biological aphid control by suppressing the populations of their parasitoid hosts, especially in confined environments such as greenhouses where augmentative biological control is commonly used (Acheampong et al., 2012;Yang et al., 2017).So far, there is no effective sustainable strategy that can be used to control hyperparasitoids.Volatile compounds that induce an attractive or repellent response in primary parasitoids and the reverse in hyperparasitoids offer a promising perspective in developing a so-called "push-pull" strategy.In such a strategy, hyperparasitoids could be "pushed" away from areas with their parasitoid hosts and potentially "pulled" into traps.Primary parasitoids, on the other hand could be "pulled" into the crop and "pushed" away from the hyperparasitoid traps, leading to enhanced biocontrol (Cusumano et al., 2020).A previous study has shown that synthetic VOCs (pure compounds), especially when blended together in specific ratios, may be better candidates for push-pull systems than the microbial volatile blends in which they were identified (Goelen et al., 2021).Future studies are needed to further investigate the efficacy of synthetic VOCs as either "push" or "pull" components in field or greenhouse conditions.
Altogether, our results have shown that bacterial volatiles can elicit olfactory responses in insects from different trophic levels, and that responses vary between and within trophic levels, and even between species within the same genus.For a number of VOCs we found contrasting correlations between VOC concentration in the volatile blend and the behavior of primary and secondary parasitoids.These VOCs evoked similar olfactory responses when used individually.This may lead to novel semiochemical-based strategies to manage hyperparasitoids, and improve the biological control of aphids.

Table 1
Effects † of VOC blend, trophic level, insect species, and their interactions on the olfactory response of seven selected model insect species ‡ toward the volatile blends of three bacteria § .All generalized linear mixed models used the binary choice for treatment or control side of the Y-tube olfactometer tests as response variable and individual cohort of insects as random variable.All models were made using a binomial distribution with a logit link function.

Table 2
Effects † of compound, dose, insect species, and their interactions on the olfactory response of three selected insect species ‡ toward three selected individual compounds § .All generalized linear mixed models used the binary choice for treatment or control side of the Y-tube olfactometer tests as response variable and individual cohort of insects as random variable.All models were made using a binomial distribution with a logit link function.‡Tested insect species included the aphid Myzus persicae var.nicotianae, the primary parasitoid Aphidius colemani and the hyperparasitoid species Dendrocerus aphidum.§ Tested compounds were acetic acid, linalool, and n-hexanol.