Introduction

Soil microbes produce a wide array of volatile organic compounds (VOCs), which have disparate ecological roles such as stimulating plant growth, inhibiting competitors, and influencing the behavior (e.g., attract/repel) of insects and other invertebrates (Davis et al. 2013; Kanchiswamy et al. 2015; Werner et al. 2016; Fincheira and Quiroz 2018). The potential exists for these VOCs to be developed into biopesticides for improved monitoring and control of soil dwelling invertebrate pests impacting plant health, thus improving food security. Entomopathogenic fungi (EPF) such as Beauveria bassiana, Metarhizium anisopliae, and Isaria fumosorosea are constitutively producing an array of VOCs whose ecological role is poorly understood (Bojke et al. 2018). EPF VOCs are responsible for the attractant and repellent behavior exhibited by some insects such as collembola, ants, termites, or mole crickets (Butt et al. 2016; Hamzah et al. 2020; Weithmann et al. 2020; Hummadi et al. 2021). VOCs of Metarhizium brunneum were recently shown to have antimicrobial properties in vitro (Hummadi et al. 2021) with two of the volatiles, 1-octen-3-ol and 3-octanone, attracting nematodes and mollusks at low doses but repelling them at higher concentrations (Khoja et al. 2019, 2021; Hummadi et al. 2021). At even higher doses, these compounds proved lethal to insects, nematodes, and mollusks (Hummadi et al. 2021; Khoja et al. 2019, 2021). The alcohol 1-octen-3-ol is an attractant for several mosquito species, grain beetles, collembola, and tsetse flies (Hall et al. 1984; Kline et al. 2007; Pierce et al. 1991). Similarly, the ketone 3-octanone is also known to influence insect behavior. For example, it is the active component of the alarm pheromone of the leaf cutting ant Acromyrmex (Blum and Brand 1972; Iwabuchi et al. 1987). Determining the semiochemical properties of microbial VOCs could help in the identification of attractant compounds which could be used to improve pest monitoring, mass trapping or attract-and-kill pest control strategies. In the attract-and-kill strategy, attractants are used to lure the pest to a control agent and thereby significantly reducing pesticide inputs. Repellents are equally important as they could protect plants without killing the pest and non-target organism. Both attractants and repellents could be used in push–pull pest control strategies, where repellents drive pests out of the main crop to a trap crop or an attractant (Cook et al. 2007; Khan et al. 2016).

Many major pests of economic significance have a subterranean phase in their life cycle, such as larval stages of click beetles, chafers, rootworms, cutworms, vine weevil, and leatherjackets (Bažok et al. 2021; Hann et al. 2015; Parker 2005). The majority of these pests are difficult to control with conventional chemical pesticides, partly due to the buffering capacity of the soil and the pests cryptic nature. The problem is compounded by the fact that many pesticides have been withdrawn or restricted in use, clearly creating a demand for new products and strategies to monitor and control these pests. Many research projects looked at the development of bioinsecticides using entomopathogenic fungi (EPF) such as Metarhizium; yet, the efficacy of EPF can be limited if the edaphic conditions are limiting their growth (Rath et al. 1995; Kabaluk and Ericsson 2007; Kabaluk et al. 2007; Larroudé and Thibord 2017). On the contrary, VOCs and semiochemicals could be a sustainable active ingredient and used immediately for pest monitoring and control. Carbon dioxide is an attractant for many soil insects (Bernklau and Bjostad 2008; Erb et al. 2013; Barsics et al. 2014; Ambele et al. 2019). Many other attractants have been identified, mostly plant or microbe-derived volatiles (Barsics et al. 2017; Hammack 2001; Hiltpold and Hibbard 2016). Odorant binding proteins (OBPs) located in the maxillary and labial palps play a pivotal role in the detection and subsequent behavioral response to these odors (Zhou et al. 2008; Pelosi et al. 2014; Brito et al. 2016). Insects possess a variable number of OBPs (e.g., 17 for Ceratitis capitata or Locusta migratora to 111 for Aedes aegypti). Each OBP has affinity for several odorant molecules. For example Carpomya vesuviana OBP5 and OBP6 which have the ability to bind a broad spectrum of low molecular weight compounds (Li et al. 2017) or Bactrocera dorsalis OBP56f-2 which can bind methyl eugenol, trans-2-hexenal and 4-carvomenthenol (Chen et al. 2021). Moreover, multiple OBPs can bind to the same compounds with different affinity. In fact, Chen et al. (2021) showed that the B. dorsalis mutant with knock down OBP56f-2 still retained some attraction to methyl eugenol, suggesting other OBPs may also have some affinity for this compound.

The aim of this study was to analyze the influence of 1-octen-3-ol and 3-octanone on the behavior of three major crop pests larvae, namely wireworm (larvae of click beetles), western corn rootworm and garden chafer (Ruther and Mayer 2005; Benjamin et al. 2018). These Coleoptera were chosen as they have different life cycles and behaviors. Indeed, wireworms are a generalist pest that spend 2 to 5 years in the soil before pupating, while corn rootworm, a specialist pest, and garden chafer, a generalist pest, complete their life cycle in one year. In vivo, trials were performed on the three species to test the attractant and repellent properties of 1-octen-3-ol and 3-octanone; moreover, as corn rootworm is fully sequenced, we choose this species as a model to perform in silico studies to identify corn rootworm OBPs and establish if they had affinity for the VOCs 1-octen-3-ol and 3-octanone.

Materials and methods

Insect source and maintenance

Late instar wireworm (Agriotes lineatus) collected in Britany (France) in September 2020, were kept in 1L pots filled with medium loam soil and fed with potato slices. Western corn rootworm (Diabrotica v. virgifera) eggs were collected by the Austrian Agency for Health and Food Safety Ltd. in Spring 2020. The eggs were incubated at 25 °C in a soil-sand mixture until hatching. Emergent larvae were provided corn seedlings as a food source. Only first and second instar larvae were used. These were gently transferred to the experimental containers using a fine paint brush. Garden chafer (Phyllopertha horticola) larvae, collected in August–September 2020 from pesticide free sites in Tyrol (Austria), were quarantined for a week and healthy third instar individuals were used in the trials.

Behavioral response choice trial

Preliminary lab studies showed that direct exposure to liquid 1-octen-3-ol and 3-octanone were toxic to wireworm, western corn rootworm, and garden chafer at doses of 0.5 µl per gram of soil or less (Bourdon et al. 2022). High mortality (> 90%) was observed in less than 4 days.

The behavioral responses of test insects to the VOCs to these same compounds was investigated in choice studies conducted in 5 l plastic (HDPE) buckets (30 × 13 × 13 cm) filled with 4 l medium loam mixed with 10% silver sand (jardineries Truffaut). The test arena was evenly allocated into three zones: treated, central, and untreated (Fig. 1). The treated and untreated zone were either placed on the right or left side of the buckets to account for external attraction cues. Three five-day old corn seedlings were planted at opposite ends of the test area, approximately 5 cm from the edge of the container and 3 cm from each other. Two sublethal doses (100 µl and 2 × 100 µl, which correspond to 0.05 µl/g of soil and 0.1 µl/g of soil) were tested for 1-octen-3-ol and for 3-octanone (Sigma-Aldrich, France). Each VOC was injected (100 µl) into 7 mm dia Sharrow cellulose filter tips (Wilsons & Co Ltd) which were placed either directly behind the central 5-day old maize seedlings or between the seedlings for the 2 × 100 µl treatment (Fig. 1). Filters were covered with 1 cm of soil immediately after they had been loaded with the VOC. Four larvae were released in the center of each bucket on the soil surface. There were ten replicates for each treatment and dose. Untreated buckets were prepared by injecting water in sharrow filters placed in the “treated” side. Five days post treatment, each zone was collected individually and checked manually to recover live and dead larvae. Due to the very delicate nature of corn rootworm, the recovery method had to be modified to avoid as much handling as possible. Briefly, the soil was separated in two parts (treated and untreated, see Fig. 1). Each half of the soil was mixed with water, the living larvae could be recovered floating on the water, while the dead ones could not be found. Larvae from all treatments were incubated in clean, fresh soil for 5 days to assess the final mortality.

Fig. 1
figure 1

Behavioral response choice trial design. In the treated area, either one filter was placed in position A, or two filters were placed in position B1 and B2. In each filter 100 µl of 1-octen-3-ol or 3-octanone was injected. The two black arrows indicate how the soil was split for corn rootworm position assessment. The assessment was made by carefully washing the soil to recover the floating living larvae

Full size image

The response index (RI) was calculated according to Usseglio et al. (2017), using the equation:

$${\text{RI}} = \, \left[ {\left( {T - C} \right)/{\text{Total}}} \right] \, \times { 1}00,$$

where T and C represent the number of insects found in treated and untreated control zones, respectively. Total is the number of insects which responded and excludes the unresponsive larvae in the central zone and those never found. Positive RI values indicate attraction while negative values indicate repellency. The trials were performed in greenhouses (temperature ranging between 20 and 25 °C, RH = 75 ± 16% for A. lineatus trial, RH = 46 ± 4% for D. v. virgifera and P. horticola).

Identification of corn rootworm OBPs

In silico analyses were performed on western corn rootworm. This species was used as a model, as it is fully sequenced with its genome obtained from NCBI (National Center of Biotechnology Information) (GCF_003013835.1). Previously annotated OBPs from mosquitoes (Aedes aegypti and Aedes albopictus) were obtained from the paper by Zhou et al. (2008). Additional classic, Plus-C, Minus-C and atypical OBPs were obtained from A. aldopictus (FJ040863.1, FJ040862.1, FJ040861.1) (Armbruster et al. 2009), Locusta migratora (FJ959365, JN247410, JN129989) (Yu et al. 2009), Tribolum castaneum (KQ971354.1, KQ971352.1) (Richards et al. 2008). All these sequences were blasted using local BLASTp module in BioEdit Tool, using a blast cutoff of 10 to identify all corn rootworm OBPs (Hall et al. 2011). All identified OBPs were tested on motif search website (https://www.genome.jp/tools/motif/) to verify that the identified sequences were OBPs. Corn rootworm OBPs were classified according to their number of conserved cysteine residues. Phylogenetic analysis were performed on MEGA11 using the maximum likelihood method and the Jones–Taylor–Thornton method as an amino acid-based substitution model (Fatma et al. 2021). The phylogenetic tree was generated with bootstrap values of 1000 replicates using OBPs from D. v. virgifera, D. melanogaster, and Cylas formicarius (https://www.ncbi.nlm.nih.gov/nuccore/?term=cylas+formicarius+odorant).

Binding affinity of the VOCs to OBPs

To predict the binding affinities of the VOCs with the OBPs, we followed the protocol described by Fatma et al. (2021). Briefly, the thread-based structure model of the multiple classic, plus-C and minus-C OBPs was predicted using RoseTTAFold module of Robetta (https://robetta.bakerlab.org/submit.php). Selected models were verified by ERRAT and procheck (https://saves.mbi.ucla.edu/) to analyze the patterns of nonbonded atomic interaction and obtain information on the overall structural geometry of the proteins (Colovos and Yeates 1993; Laskowski et al. 1996).

Binding site prediction was then performed using CASTp webserver (http://sts.bioe.uic.edu/castp/calculation.html). The 3D structure of 1-octen-3-ol, 3-octanone, as well as the corn rootworm attractants: (E)-β-caryophyllene, carbon dioxide, indole, (+)-linalool and 4-methoxycinnamaldehyde (Hammack 2001; Hiltpold and Hibbard 2016; Walsh et al. 2020) were obtained from PubChem. Finally, the molecular docking was performed using Autodock Vina (Trott and Olson 2010). The protein–ligand interactions were determined using Protein–Ligand Interaction Profiler (https://plip-tool.biotec.tu-dresden.de/plip-web/plip/index) (Adasme et al. 2021).

Statistical analysis

For the attractant/repellent trial, the difference among treatment RIs was determined using an ANOVA followed by a Tukey post hoc analysis pairwise T test, to determine the difference between treatment pairs. Moreover, the number of larvae not found (considered as dead) was compared using a Kruskal Wallis test. All statistical analysis were performed using R version 4.1.2 (R Core Team 2021).

Results

Behavioral response choice trial

The VOCs elicited different responses in the test insects. They had no effect on garden chafer which were spread evenly between the treated and untreated area (Fig. 2, Table 1). On the contrary, corn rootworms were slightly attracted toward 1-octen-3-ol, while wireworms were slightly repelled by this VOC as well as with the high dose of 3-octanone (Fig. 2); yet, no significant differences were found between the treatment and control. The absence of significant differences can be explained by the high dispersion of the insects in the untreated buckets (Fig. 2). Moreover, many wireworms did not appear to have moved since 46% to 84% were found in the central zone (Table 1). The presence of wireworms in the water-treated side of the untreated bucket should not be due to external cues as the water-treated side was placed either right or left of the buckets. Their presence on the water-treated side appears to be due to random movement toward the plants.

Fig. 2
figure 2

Response index (RI) of the three insect species. Positive RI values indicate attractancy, while negative values indicate repellency. Corn rootworms were slightly attracted by 1-octen-3-ol but were repelled by the high dose of 3-octanone. Garden chafers were spread nearly evenly between the treatment and control. Wireworms were repelled by 1-octen-3-ol and the high dose of 3-octanone

Full size image
Table 1 Average proportion of larvae found in each area across each pot 5 days after the beginning of the behavioral response trials
Full size table

During the assessment, 7.5% to 17.5% wireworm could not be located and were assumed dead (Table 2) as these quickly turn black and therefore are difficult to find in the soil. The number of dead larvae was not significantly different between the treatments (p = 0.8932, df = 4), indicating that the mortality was not induced by the treatment. Corn rootworm larvae mortality ranged between 45 and 77.5% (Table 2); however, no significant differences were found between the treatments and control (p = 0.08965, df = 4), confirming deaths were not induced by the VOCs. Most garden chafer were recovered, except in the 1-octen-3-ol treatment where 2.5% of the larvae could not be found (Table 2).

Table 2 Mortality during the behavioral trial
Full size table

Identification of corn rootworm OBPs

86 OBPs were identified in the genome of corn rootworm. 19 OBPs were classic OBPs with six conserved cystine residues, 48 were minus-C OBPs with 4 or 5 cystine residues, 17 were Plus-C OBPs with 7 or 8 cystine residues and an additional conserved proline residue, and 2 were atypical OBPs with 9 or 10 conserved cystine residues.

The phylogenetic tree (Fig. 3) showed that the majority of D. v. virgifera OBPs were not closely related and belong to different subgroups. OBPs from D. v. virgifera and C. formicarius were closely related as compared to D. melanogaster OBPs.

Fig. 3
figure 3

Phylogenetic tree of Diabrotica v. virgifera (in yellow), Drosophila melanogaster (in blue) and Cylas formicarius (in red) OBPs. The red box highlights the sequences used for molecular docking. The phylogenetic tree shows that OBPs are well conserved between even between different order, some D. v. virgifera OBPs are closely related to D. melanogaster OBPs

Full size image

Binding affinity of the VOCs to OBPs

Four classic OBPs, one Plus-C and one Minus-C OBP were selected for evaluation of the ERRAT score, Ramachandran plots of the predicted models and Q-mean. The best models score for each selected OBP is presented in Table 3. The binding affinity of 1-octen-3-ol and 3-octanone was compared with known attractants which included linalool and 4-methoxycinnamaldehyde (Hammack 2001; Hiltpold and Hibbard 2016; Walsh et al. 2020).

Table 3 Evaluation of four classic OBPs models (DvirOBP10, DvirOBP21, DvirOBP22, DvirOBP80), a plus-C OBP model (DvirOBP12), and a minus-C OBP model (DvriOBP28)
Full size table

Carbon dioxide had the least binding affinity with the OBPs (Fig. 4, Table 4), while (E)-β-caryophyllene and 3-octanone had the strongest binding affinity followed by indole and 1-octen-3-ol (Fig. 4, Table 4).

Fig. 4
figure 4

Binding affinity of four classic corn rootworm OBPs (DvirOBP10, DvirOBP21, DvirOBP22, DvirOBP80), one Minus-C (DvriOBP28) OBP, and one Plus-C OBP (DvirOBP12) to seven potential attractants. Lowest score indicate better binding affinity. (E)-β-caryophyllene and 3-octanone had the best binding affinity with the OBPs, while carbon dioxide had the least binding affinity

Full size image
Table 4 Binding affinity of four classic OBPs models (DvirOBP10, DvirOBP21, DvirOBP22, DvirOBP80), a plus-C OBP model (DvirOBP12), and a minus-C OBP model (DvriOBP28) to the VOCs 1-octen-3-ol and 3-octanone and plant derived compounds known to be potential corn earworm attractants
Full size table

Carbon dioxide only had hydrogen bonds with the OBPs, while the other ligands had a majority of hydrophobic bonds (Fig. 5). Hydrogen bonds are usually responsible for the binding of small proteins such as carbon dioxide, while larger proteins recognition is more dependent on hydrophobic bonds (Hubbard and Haider 2010). The ligands with more hydrophobic bonds were (+)-linalool, 1-octen-3-ol, 3-octanone and (E)-β-caryophyllene, while carbon dioxide, 4-methoxycinnamaldehyde and indole had the lowest number of binding interactions (Fig. 5).

Fig. 5
figure 5

Number of hydrogen and hydrophobic bond between seven different ligands ((+)-linalool, (E)-beta-caryophyllene, 1-octen-3-ol, 3-octanone, 4-methoxycinnamaldehyde, carbon dioxide and indole) to four classic corn rootworm OBPs (DvirOBP10, DvirOBP21, DvirOBP22, DvirOBP80), one Minus-C (DvriOBP28) OBP, and one Plus-C OBP (DvirOBP12)

Full size image

Discussion

This study shows that the Metarhizium-derived VOCs 1-octen-3-ol and 3-octanone had no significant effect on soil insects; yet, more wireworms were found in the control zone while corn rootworm were mostly found in the treated zone. In silico experiments presented herein validated that corn rootworm had odor binding proteins with strong affinity for these compounds. The relatively conserved OBPs (Fig. 3) suggest that wireworms and garden chafer should be able to bind the VOCs similarly. In silico studies are a good way to screen potential new attractants or repellents; yet they are limited to insect species which have been fully sequenced and cannot be used on their own to determine if a compound has attractant or repellent properties. The different behavioral responses observed in this study indicate diverse preferentiality for specific rhizosphere VOCs in isolation, supporting prior works (Fäldt et al. 1999; Sawahata et al. 2008; Badri and Vivanco 2009; Erb et al. 2013; Holighaus et al. 2014; Bont et al. 2017). It is probable that in nature, plants and fungi emit a bouquet of VOCs, and the overall blend might have a greater influence on insect’s behavior than single compounds (Hammack 2001; Renou and Anton 2020).

The behavioral response of invertebrate to volatiles cues is also known to be dose dependent. Cook et al. (2007) showed that the attraction levels of 1-octen-3-ol and R-1-octen-3-ol differed between the mosquitoes Aedes aegypti and Culex quinquefasciatus. Moreover, it was shown that low doses of VOCs could attract mosquitoes, mollusks, and insects, while higher doses were repellent (Xu et al. 2015; Khoja et al. 2019, 2021; Hummadi et al. 2021).

The absence of response of garden chafer to 1-octen-3-ol and 3-octanone could indicate that this species does not have specific receptors for these compounds. Alternatively, it could be that garden chafers do not centrally process these in terms of chemotaxis, and use other orientation cues. In fact, these two VOCs are commonly emitted by numerous fungi (Insam and Seewald 2010) as well as trees such as Quercus robur or Q. petraea (Weissteiner et al. 2012), indicating that they could be commonly abundant in the soil and therefore of little utility to the insect. Measuring the release rates of the VOCs, their absorption, and their diffusion in the soil was out of the scope of this study, but we can expect the VOCs to have dissipated quickly as has been seen for similar compounds (Cui et al. 2021). Rapid dispersal would lead to a short window for potential action, should one insect be less mobile or have a less rapid response to any given stimuli then the behavioral effect would be likely to be much less evident. Moreover, the life cycle of wireworms and corn rootworms could also explain the limited efficacy of the VOCs to attract or repel these species. Wireworms only feed actively during 29% of their life cycle (Furlan 2004), and only on a limited number of seeds and tubers (Chaton et al. 2008). Therefore, wireworms which are not in their feeding phase may avoid unnecessary movement to conserve energy (Charnov 1974). This would explain the high number of wireworms found in the central area (Table 1). Due to the limited number of wireworm available for our study, we could not pre-select larvae in feeding stage for the trials. First and second instar corn rootworm were used for these trials because they are the most damaging stages; however, the natural mortality of first instar can reach 95%, while second instar mortality can reach 10% (Toepfer and Kuhlmann 2006), which explains the high number of mortality found in our study.

Many of these issues have the potential for mitigation through the development of appropriate formulation technologies. Further trials will be done to encapsulate the VOCs into absorbent materials to significantly reduce the volatility of the compounds and extend their period of activity. The VOCs will be tested for attract-and-kill or push–pull strategies to increase their efficacy and drag the pests out of the crops. Push–pull strategies using the VOCs and a trap crop such as wheat or peas could be effective against wireworm (Adhikari 2017). The repellent effect of the VOCs could also explain why Kabaluk and Ericsson (2007) observed wireworms avoiding soil treated with Metarhizium spores, moreover, they found that the avoidance increased with the increase of concentration of conidia even from distance, suggesting that VOCs may play a role in insect avoidance of entomopathogens. Push–pull strategies for western corn rootworm could be effectively achieved through the combined deployment of attractive VOCs in a trap crop and a simultaneous repellent within the main crop. Such repellents could include methyl anthranilate (Bernklau et al. 2019).

The VOCs used in this paper had no significant influence the behavior of soil insects. However, the high mortality and absence of movement of the larvae impacted negatively the results from this study. The slight behavioral change observed were encouraging, thus, we will do further trials to confirm the effect of encapsulated VOCs on soil insects.