Integrating fungal virulence, formulation, and volatile-mediated attraction for an effective attract-and-kill strategy against Agriotes spp.

The progressive withdrawal of soil insecticides has created an urgent need for sustainable alternatives against Agriotes wireworms. Here, we evaluated an attract-and-kill strategy based on entomopathogenic fungi (EPF) designed to overcome the spatial and behavioral constraints that limit biocontrol efficacy in soil. Our approach integrated in vitro virulence screening, in vivo efficacy assays using starch-enriched alginate beads in sterilized soil microcosms, dual-choice olfactometer bioassays, and volatolomic profiling by thermal desorption gas chromatography coupled with mass spectrometry. Ophiocordyceps sp. MUCL 58550 exhibited the highest in vitro virulence, yet its efficacy declined in soil microcosms. Volatolomic and behavioral analyses suggest that this loss of performance resulted from poor larval attraction, as its profile blend was dominated by ubiquitous alkanes and carbonyl compounds. In contrast, Metarhizium anisopliae MUCL 58549 maintained high efficacy in vivo and elicited positive chemotaxis (65% choice). This response was associated with a distinctive volatile signature enriched in sesquiterpenes (e.g., aristolochene and eremophilene). The non-pathogenic strain Trichoderma harzianum MUCL 29707 induced the strongest attraction (70% choice), consistent with a volatile profile enriched in C8 oxylipins, including 1-octen-3-ol and 3-octanone. However, its inability to infect wireworms highlights the risk of creating pest “safe havens” if attraction is not coupled with lethality. Although the causal role of individual volatile compounds remains to be validated, these results demonstrate that effective biological control relies on a “dual-action” mechanism. Specifically, the integration of volatile-mediated host recruitment with targeted pathogenicity enables formulated EPFs to function as active lethal sinks, offering a viable framework for sustainable wireworm management, pending future field validation.

Wireworms, the polyphagous soil-dwelling larvae of click beetles (Coleoptera: Elateridae), have become increasingly problematic as major pests in temperate agroecosystems, posing significant threats to crop productivity and economic sustainability (Poggi et al., 2021; Paluch, 2024; AHDB, 2023).Their extended subterranean life cycle, which typically lasts three to five years, combined with considerable physiological plasticity that enables survival across a wide range of soil moisture and temperature conditions, makes wireworms particularly difficult to detect and control before crop damage becomes apparent (Traugott et al., 2015; Praprotnik et al., 2023; Szanyi et al., 2023).

In European cropping systems, Agriotes species, particularly A. lineatus, A. obscurus, and A. sputator, are responsible for substantial yield and quality losses in economically important crops, including potato, sugar beet and maize (Veres et al., 2020; Paluch, 2024; Wood et al., 2023). Under high infestation pressure, crop damage can exceed 50% in susceptible crops, posing a particularly serious threat to high-value crops such as potato (Le Cointe et al., 2023; Furlan et al., 2021; Lorenzo et al., 2024). This pressure may be further modulated by shifting climate conditions, which is reshaping the distribution of elaterid pests across Europe. Notably, the thermophilic species A. sordidus, once largely restricted to Southern Europe, has expanded its range northward, with established populations recently reported in Germany, Belgium, and the United Kingdom (Furlan et al., 2021; Szanyi et al., 2023).

The increasing impact of wireworm coincides with major regulatory and agronomic changes in European agriculture. The progressive withdrawal of synthetic soil insecticides, including neonicotinoids and phenylpyrazoles, owing to concerns over their environmental and human health impact, has reduced the availability of effective control options and created a significant protection gap for growers (Möhring et al., 2020; EFSA, 2013; DEFRA, 2024). At the same time, the widespread adoption of conservation agriculture practices, such as reduced tillage and permanent soil cover, may inadvertently favor wireworm populations by creating more stable soil microhabitats that enhance larval survival (Le Cointe et al., 2023). Together, these factors underscore the need for Integrated Pest Management (IPM) strategies that emphasize biologically and behaviorally based approaches for sustainable wireworm control.

Entomopathogenic fungi (EPF) are among the most promising biological control agents for addressing this management gap (Zimmermann, 2007). However, their consistent efficacy remains constrained by the complexity of host-pathogen interactions, which are strongly influenced by both fungal traits and environmental conditions (Kabaluk and Ericsson, 2007; Rogge et al., 2017; Poggi et al., 2021). Virulence is highly dependent on the specific combination of fungal strains and target insect species. For instance, Eckard et al. (2014) reported substantial variation among closely related hosts, with Metarhizium brunneum ART2825 causing more than 80% mortality in A. obscurus but less than 50% mortality in the closely related A. sputator. Similarly, Razinger et al. (2020) showed that certain strains of M. brunneum act as highly virulent pathogens of A. lineatus larvae, while strains of M. robertsii exhibit little direct virulence and instead function as maize growth-promoting endophytes. In addition, fungal morphological and physiological traits, including conidial size and germination rate, can substantially influence infection dynamics and pathogenic success across target species (Seib et al., 2024). Environmental factors further modulate EPF performance. The efficacy of species such as B. bassiana, can vary considerably and is often constrained by soil physicochemical properties (Ladurner et al., 2009). Consequently, the successful integration of EPF into wireworm management programs requires rigorous evaluation of diverse, locally adapted, strains and the development of optimized application strategies, rather than reliance on broadly marketed commercial strains.

Even when a highly virulent EPF strain is identified, its field performance is strongly influenced by the formulation and delivery technology employed. First-generation formulations, consisting primarily of liquid suspensions of unprotected conidia applied inundatively, are highly susceptible to UV degradation, desiccation, and antagonistic interactions with native soil microorganisms, all of which can substantially reduce fungal survival and efficacy (Hernández et al., 2023; Mawcha et al., 2025). Second-generation solid formulations, which utilize substrate such as cereal grains as a carrier, improve fungal persistence by providing a nutrient source for propagules development. However, they remain vulnerable to consumption by non-target organisms and offer only limited protection against environmental stressors (Faria and Wraight, 2007; Lacey et al., 2015). To address these limitations, third-generation biopolymer-based formulations, particularly calcium-alginate encapsulation systems, have emerged as promising delivery platforms for EPF (Lacey et al., 2015; Mawcha et al., 2025). Alginate encapsulation can substantially extend shelf-life and improve the deployment of fungal agents targeting soil-dwelling insect pests (Sarma et al., 2023). In addition, biopolymer matrices provide physical protection against environmental stress, while promoting fungal establishment and persistence in the soil environment (Friuli et al., 2023). For subterranean attract-and-kill (A&K) approaches, encapsulated formulations offer an additional advantage by presenting the inoculum as an edible particle that can be ingested by wireworms, thereby facilitating pathogen acquisition and potentially circumventing some of the barriers associated with infection through the cuticle alone (Brandl et al., 2017). However, despite these physical improvements, third-generation formulations still retain a key vulnerability, as these edible matrices remain susceptible to incidental consumption by non-target soil organisms.

Overcoming the spatial mismatch between stationary EPF formulations and mobile wireworm larvae requires the strategic manipulation of host behavior through chemical ecology. Wireworm host-finding and foraging behavior is a hierarchical sequence of chemical cues. At long-range, larvae orient toward CO 2 gradients produced by respiring roots (Gfeller et al., 2013; Arce et al., 2021; Nikoukar et al., 2025). As they approach potential hosts, root-emitted Volatile Organic Compounds (VOCs), including hexanal, facilitate host discrimination and localization at intermediate distances (Gfeller et al., 2013; Nikoukar et al., 2025). Furthermore, larvae also rely on aggregation signals, such as 2-pentylfuran, which are produced or stimulated by the presence of conspecifics to locate suitable feeding sites (La Forgia et al., 2023). At close range, feeding stimulants, such as sugars, lipids and proteins promote feeding and prolong larval residence time near the resource (Barsics et al., 2014). Beyond their pathogenic activity, EPF also produce diverse volatile compounds that can influence wireworm behavior, creating opportunities to integrate fungal pathogens and semiochemical-based approaches within A&K management strategies.

EPF-derived volatile compounds can exert diverse and sometimes contrasting effects on insect behavior. Certain EPF volatiles elicit avoidance responses in hosts. For example, B. bassiana produces compounds such as styrene, benzothiazole, and 3-methylbutanol, which have been associated with repellency and altered oviposition behavior in several insect taxa (Crespo et al., 2008; Hung et al., 2015; Bojke et al., 2018; Adame-Garnica et al., 2025). These behavioral effects have been incorporated into “stimulate-and-deter” strategies, in which fungal volatiles discourage larval feeding near vulnerable plant tissues while contributing to improve plant performance (Wood et al., 2023). In contrast, other EPF-produced volatiles can attract insects. For example, some Metarhizium spp. are known to produce sesquiterpene hydrocarbons (Gibson et al., 2014). These compounds are structurally similar to root-derived terpenes that function as host-location cues for subterranean herbivores (Johnson and Nielsen, 2012; Nikoukar et al., 2025), potentially facilitating attraction to fungal inoculum sources. A third category of EPF semiochemicals exhibits dose-dependent effects with behavioral responses shifting from attraction to repellence as emission rates increase. The volatile 1-octen-3-ol produced by M. brunneum illustrates this complexity. In coleopteran insects, low concentrations may attract larvae, while high concentrations induce strong avoidance responses and can reduce feeding damage for up to 18 days (Bourdon et al., 2023).

Evaluating the inherent attractive properties of fungal semiochemicals within encapsulated formulations may offer an alternative approach to traditional A&K strategies that rely on adding external attractants. However, a critical knowledge gap remains: it is unclear which EPF strain combine high virulence against native Agriotes spp. with volatile profile capable of attracting wireworm larvae. Strains isolated from naturally infected wireworm cadavers are particularly promising candidates, because they represent locally adapted genotypes that have persisted under field conditions and may reflect ongoing host-pathogen interactions. Despite growing interest in EPF-based wireworm management, no study has systematically evaluated virulence, efficacy in encapsulated formulation, behavioral attraction, and volatile emission across the same collection of EPF strains. Consequently, the relationship between pathogenicity and semiochemical-mediated host attraction remains poorly understood.

To address this knowledge gap, we developed and evaluated a four-step pre-field framework for the design of a strain-specific A&K system, targeting native Agriotes wireworms. Specifically, our objectives were to: (i) screen a diverse collection of EPF, including newly isolated Metarhizium, and Ophiocordyceps spp., to identify the most virulent strains via direct-contact bioassays; (ii) incorporate selected candidates into calcium-alginate beads and evaluate their persistence and pathogenicity under simulated soil conditions; (iii) quantify the behavioral responses of Agriotes larvae to encapsulated fungal formulations using dual-choice olfactometry; and (iv) characterize the VOC profile of the encapsulated strains using thermal desorption gas chromatography coupled with mass spectrometry (TD-GC-MS). Collectively, these experiments provide a mechanistic basis for the development of a biopolymer-enhanced, behaviorally targeted wireworm management strategy and identify fungal traits associated with successful A&K applications.

Biological materials

Fungal strains

Entomopathogenic fungi (EPFs) were obtained from three distinct sources. First, a collection of reference strains was acquired from the Mycothèque of the Université catholique de Louvain (BCCM/MUCL, Belgium - https://bccm.belspo.be/about-MUCL). These strains were selected following a comprehensive literature review, with priority given to strains recovered from insect cadavers and to phylogenetically related taxa with proven biological control potential. This reference panel included Beauveria bassiana MUCL 39817, Beauveria brongniartii MUCL 131 and MUCL 34853, Aspergillus flavus MUCL 55276, Lecanicillium lecanii MUCL 8115, Paecilomyces fumosoroseus MUCL 15122, Purpureocillium lilacinum MUCL 38701, and Trichoderma harzianum MUCL 29707. Second, two novel strains were isolated from naturally infected wireworm larvae collected in agricultural fields in Gedinne (Belgium) and Sauvagnon (France). Molecular identification, based on the sequencing of the internal transcribed spacer region, beta-tubulin, and elongation factor genes, subsequently confirmed these field-isolated fungi as Metarhizium anisopliae MUCL 58549 and Ophiocordyceps sp. MUCL 58550. Finally, two commercially available EPFs (Metarhizium anisopliae - Commercial strain A and Metarhizium brunneum - Commercial strain B) were included solely as methodological benchmarks to calibrate wireworm larval susceptibility during the initial in vitro screening. All fungal strains were maintained on V8 agar medium (sterilized 121 °C, 15 min) in 90 mm Petri dishes (Sarstedt, Germany) and incubated in the dark at 25 °C until required for the experiments.

Wireworm larvae

Wireworm larvae (Agriotes lineatus and A. obscurus) were obtained from adults collected in spring 2023 in Gembloux, Belgium. Larvae were sorted according to their developmental stage and maintained in plastic containers (11 × 11 × 6 cm) filled with a sterilized (121 °C, 15 min) 1:1 (v:v) mixture of soil and vermiculite (RAJA, France). The substrate moisture was adjusted and maintained at 65% of its water-holding capacity (WHC), and the containers were kept in a climate-controlled chamber at 21 °C. Larvae were fed ad libitum with germinated organic wheat seeds (Walagri, Belgium). Prior to experimentation, L10-L11 instar larvae were individually isolated and starved for seven days to standardize their physiological condition. Larvae were then randomly assigned to the different treatments, with each individual considered a single biological replicate.

Experimental procedures

Biocidal test I: in vitro virulence assay by direct fungal contact

A preliminary assessment of sporulation kinetics was conducted on sterilized V8 agar medium (121 °C, 15 min) in 24-well plates (VWR, USA), before evaluating the virulence of each fungal strain in direct-contact assays. Incubation periods were optimized individually for each strain to ensure comparable conidial production and homogeneous mycelial coverage, targeting a final density of approximately 10 7 conidia per well. For the virulence bioassay, 5 µL of a standardized conidial suspension (1 × 10 3 conidia·mL - ¹) from each strain was inoculated into individual wells containing sterile V8 agar. Plates were incubated in the dark at 25 °C for the strain-specific period determined during the sporulation assessment. Subsequently, a single starved wireworm larvae was introduced into each well and maintained in direct contact with the fungal colony for 5 hours, (to ensure conidial acquisition while minimizing confinement stress) at 22 °C (matching rearing conditions to prevent thermal stress) in darkness. Wells containing uninoculated sterile V8 agar served as untreated control, whereas a commercial lambda-cyhalothrin solution (12 µg·mL - ¹; Karate Garden, Belgium) was used as an insecticide control.

To ensure the standardization of the inoculum and the validity of the bioassay, two quality controls were performed. First, to confirm the exact conidial dose available for larval infection during the exposure period, a parallel 24-well plate was prepared under identical conditions; conidia were harvested from representative wells and enumerated using a Thoma counting chamber. Second, to verify the physiological fitness and viability of the inoculum, conidial cultivability was assessed by spreading 10 µL of a 1 × 10 4 conidia·mL - ¹ suspension onto V8 agar plates, followed by monitoring colony formation for one week.

Following the 5 hours exposure, larvae were individually transferred to perforated plastic microcosms containing 20 g of a sterilized substrate composed of vermiculite and potting soil (1:3, v:v). The substrate was sterilized twice at 121 °C for 15 min with a 24 h interval between sterilization cycles. A 1 cm³ potato cube (Solanum tuberosum, cv. Allians) was provided as a food source. Microcosms were maintained in darkness at 22 ± 1 °C. Substrate moisture was adjusted and maintained at 65% of its WHC by gravimetric compensation of evaporation losses. Larval mortality was assessed weekly for 12 weeks. Throughout the monitoring period, substrate moisture and potato cubes were replenished as needed to maintain stable environmental and nutritional conditions.

Biocidal test II: in vivo virulence assay using alginate formulated EPFs

The fungal strains exhibiting the highest virulence in the direct-contact assay were selected for formulation into alginate beads.

Following a strict funnel-based screening approach, only the candidate strains exhibiting the highest intrinsic virulence in Test I were selected for subsequent formulation and soil microcosm trials. The commercially available EPF strains were included in the initial screening solely as methodological benchmarks to calibrate larval susceptibility. As these strains are already commercialized under proprietary formulations, their evaluation in soil microcosms fell outside the scope of this stepwise selection framework.

For each strain, a sterile formulation matrix was prepared by combining 2% sodium alginate (w/v; Sigma-Aldrich, UK) and 30% starch (w/v; Carl Roth, Germany), both sterilized at 121 °C for 15 min. The matrix was subsequently supplemented with a conidial suspension adjusted to a final concentration of 3 × 10 8 conidia·mL - ¹. Alginate beads were produced by dispensing 100-µL droplets of the formulation into a sterile 0.1 M CaCl 2 solution (Sigma-Aldrich, Germany; sterilized at 121 °C for 15 min) under continuous agitation. Beads were allowed to polymerize for 30 min, rinsed with sterile deionized water, and temporarily stored in 90 mm Petri dishes until use. To control for potential non-pathogenic effects associated with bead presence or fungal growth within the formulation, a reference treatment was prepared using T. harzianum MUCL 29707, a non-entomopathogenic biological control fungus.

Forty formulated beads (approximately 2 g fresh weight in total) were introduced into each microcosm. Microcosms consisted of perforated plastic containers filled with 20 g of sterilized substrate composed of vermiculite and potting soil (1:3, v:v), sterilized as described above. A single starved wireworm larva was placed in each microcosm together with a 1 cm³ organic potato cube (S. tuberosum cv. Allians). Larvae remained in continuous contact with the beads throughout the 12-week experimental period. In addition to the T. harzianum MUCL 29707 biological control reference treatment, a second untreated control consisting of sterilized substrate and a potato cube without beads was included. Each treatment comprised 25 biological replicates. Larval mortality was recorded every 10 days. During the monitoring period, potato cubes were replaced as needed, and substrate moisture was maintained at 65% WHC by gravimetric adjustment based on initial microcosm weights.

To confirm EPF infection and fulfill Koch’s postulates, all larvae were subjected to a surface-sterilization procedure prior to incubation. Cadavers were immersed in 2% sodium hypochlorite for 2 min, followed by 70% ethanol for 2 min, and subsequently rinsed thoroughly with sterile deionized water. Surface-sterilized cadavers were then placed on sterile, moistened Whatman filter paper (No. 1001-055, Cytiva, USA) within a 90 mm Petri dishes and incubated in the dark at 25 °C to promote fungal outgrowth and assess infection. The effectiveness of the surface-sterilization procedure was verified by plating an aliquot of the final rinse water onto sterile V8 agar medium. These plates were incubated for 2 weeks at 25 °C and monitored for microbial growth. The absence of colony development confirmed successful removal of external contaminants, ensuring that any fungal outgrowth observed from the cadavers originated from internal infection rather than surface-associated microorganisms.

Wireworm’s orientation bioassay

Behavioral response of wireworms to encapsulated EPFs were evaluated using dual-choice olfactometer adapted from the methodology described by Chacón Hurtado et al. (2023). The assay assessed wireworm orientation in response to volatile emissions from alginate beads inoculated with M. anisopliae MUCL 58549, Ophiocordyceps sp. MUCL 58550, or T. harzianum MUCL 29707 compared with fungus-free alginate beads. To ensure consistency across experiments, all formulated beads used for the behavioral assays and volatile profiling (Section 2.2.4) were produced in parallel. All formulations were inoculated with the same initial conidial concentrations (3 × 10 8 conidia·mL - ¹). Following production, the beads were stored at 4 °C for 24 hours prior to use to standardize the interval between formulation and assay deployment and minimize potential variation in fungal physiological state. The experimental setup consisted of a glass Y-tube olfactometer (20 cm length, 3.2 cm internal diameter) partially filled with moistened vermiculite maintained at 65% WHC (RAJA, France). Each lateral arm was connected via a side-neck (GL45) to a 500-mL borosilicate glass jar (Duran, Belgium) containing humidified vermiculite and either inoculated or blank beads. Prior to each assay, the entire system was allowed to equilibrate for 20 minutes in the dark at 21 °C to ensure adequate volatile diffusion. Individual wireworm larvae, starved as described above, were introduced into the central arm of the olfactometer through a central inlet a glass screw cap (GL14) and allowed to move freely for 60 min. All assays were conducted in complete darkness at a constant temperature of 21 °C. At the end of the observation period, larval position was recorded. Individuals located within the 3 cm of the central inlet were classified as non-responders and assigned to a neutral category. The number of non-responding larvae was recorded at the end of the observation period to determine the overall response rate for each treatment. Responding larvae were scored according to their position in the treatment arm (fungus choice) or the control arm (blank-bead control). To minimize directional bias, the positions of the treatment and control chambers were alternated between replicates. Between assays, the olfactometer was washed with 70% ethanol, rinsed thoroughly with deionized water, and dried at 100 °C. A total of 56 wireworms were tested per fungal treatment.

Volatolomic characterization and biomarkers identification

To characterize the volatile organic compounds (VOCs) emitted by encapsulated fungi, a volatolomic analysis was conducted using the previously selected fungal strains: the EPF M. anisopliae MUCL 58549 and Ophiocordyceps sp. MUCL 58550, together with the biological control reference strain T. harzianum MUCL 29707. A treatment consisting of fungus-free alginate beads was included as a blank-bead control. Fungal encapsulation was performed according to the protocol described for Biocidal Test II. As detailed in Section 2.2.3, VOC collection was conducted in parallel with the behavioral assays using beads of the exact same age (24 hours post-polymerization) and from the same standardized production batches. The inclusion of a fungus-free blank-bead control further allowed us to account for baseline volatile emissions attributed to the formulation matrix.

For each treatment, five replicates were prepared. Each replicate consisted of an airtight borosilicate glass chambers (20 cm in length, 3.2 cm internal diameter) containing 40 g of sterilized substrate composed of vermiculite and potting soil (1:3 v/v) and 2 g of formulated beads (approximately 40 beads). Chambers were maintained in darkness at 21 °C for 3 h to allow headspace equilibration prior to volatile collection. VOCs were subsequently sampled using a dynamic collection system. Activated-carbon-filtered air was drawn at 21 °C through the chamber at a constant flow rate of 150 mL·min - ¹ using a GilAir™ Plus air sampling pump. Volatiles were trapped on hydrophobic Tenax TA/Carbograph sorbent tubes (Markes International, UK) over a four-hour sampling period. To prevent particulate contamination of the sampling material, a small plug of glass-wool was placed upstream of each collection tube. Immediately after sampling, 25 ng of butylbenzene (Sigma-Aldrich, USA) was added to each sorbent tube as an internal standard. Tubes were then hermetically sealed and stored at 4 °C until analysis. To minimize potential batch effects, at least one replicate per treatment was collected on every sampling day.

VOCs samples were analyzed by thermal desorption gas chromatography coupled with mass spectrometry (TD-GC-MS; Shimadzu, Kyoto, Japan). Prior to desorption, sorbent tubes were subjected to a dry purge to remove residual moisture. Tubes were then desorbed at 280 °C for 8 min using a TD-30R thermal desorption unit (Shimadzu). Released analytes were cryofocused on a Peltier-cooled cold trap at -20 °C for 12 min and subsequently thermally desorbed by rapid heating to 280 °C, Analytes were transferred onto the GC column operating in split mode (1:20). Chromatographic separation was performed on an HP-5 ms capillary column (30 m × 0.25 mm i.d., 0.50 µm film thickness; Agilent Technologies, Santa Clara, CA, USA). Helium was used as the carrier gas at a constant pressure (200 kPa). The GC oven temperature was set as follows: 40 °C hold for 3 min, ramped at 5 °C·min - ¹ to 200 °C, then ramped at 20 °C·min - ¹ to 300 °C, and finally held at 300 °C for 6 min to purge the column.

Mass spectra were acquired by electron ionization in full-scan mode over an m/z range of 35-300. The ion source and interface temperatures were set to 200 °C and 230 °C, respectively. The detector was set at 0.1 kV, following the instrument tuning.

Putative compound identification was achieved by comparing mass spectra with entries in the NIST17 and FFNSC3 spectral libraries, applying a spectral similarity match threshold of > 80%. To confirm these initial identifications, a C7–C30 n -alkane standard mixture (25 ng·µL - ¹; Sigma-Aldrich) was analyzed under identical conditions to calculate Kovats Retention Index (RI). Experimentally determined RI values were subsequently compared with published reference values to confirm compound identification. Due to the absence of authentic standards for all detected compounds, the results are expressed as semi-quantitative relative abundances, calculated as a percentage of the total peak area per replicate.

Statistical analysis

All statistical analyses and data visualizations were performed using R software (version 4.4.2). For biocidal tests I and II, the effect of fungal treatments and chemical reference treatments on wireworm mortality was assessed using time-to-event analyses. Survival analyses were conducted using the survival and survminer packages. To align the time scales of the two assays and allow direct comparison, mortality data from Biocidal Test II, originally recorded at 10-day intervals, were converted into weeks. Furthermore, survival data from Biocidal Test II were right-censored at 84 days (week 12) to ensure strict temporal consistency with Biocidal Test I. Survival probabilities were estimated using the non-parametric Kaplan-Meier method, and overall differences among treatments were assessed using the log-rank test. To quantify treatment-specific effects on mortality risk, univariate Cox proportional hazards models were fitted, and Hazard Ratios (HR) along with their 95% confidence intervals were estimated. The proportional hazards assumption was evaluated using Schoenfeld residuals. When significant overall effects were detected, post-hoc pairwise comparisons were performed using log-rank tests, with p-values adjusted via the Holm procedure.

Behavioral responses from dual-choice olfactometer assays were analyzed using Generalized Linear Models (GLM) with a binomial error distribution and a logit link function, utilizing the base stats package in R. Only responding larvae (N = 106) were included. For each fungal treatment, larval preference was tested against the null hypothesis of random choice (50:50 distribution) using a Wald test on the model intercept. To quantify the magnitude of behavioral preferences, effect sizes were calculated as Odds Ratios (OR) with their corresponding 95% confidence intervals.

GC-MS datasets were aligned using GCAligner 1.0 (Dellicour and Lecocq, 2013). Given the untargeted nature of the volatolomic screening, calibration curves with external standards were not applicable. Consequently, results are expressed as semi-quantitative relative abundances, calculated as a percentage of the total peak area per replicate. The internal standard (butylbenzene) was used strictly to monitor instrument stability and ensure analytical reproducibility across injection batches. Multivariate analyses were conducted using the vegan, mixOmics, FactoMineR, and factoextra packages. To normalize the data distribution and reduce the disproportionate influence of highly abundant compounds without masking biological variation, peak areas were log2-transformed and Pareto-scaled prior to multivariate analysis.

Overall differences in VOC profiles among fungi were assessed using permutational multivariate analysis of variance (perMANOVA) based on Euclidean distances (adonis2 function, 999 permutations), with the homogeneity of multivariate dispersions evaluated via betadisper. While perMANOVA was used to assess global differences among volatile profiles, a Partial Least Squares Discriminant Analysis (PLS-DA) was subsequently performed to identify the specific chemical biomarkers driving the separation between fungal strains. To rigorously validate the model and rule out overfitting, the PLS-DA was subjected to leave-one-out cross-validation (LOOCV) to determine predictive accuracy, and a 1,000-iteration permutation test was conducted to evaluate statistical significance against random chance (p < 0.001). Compounds with Variable Importance in Projection (VIP) scores > 1.0 were considered primary discriminants. Chemical richness was compared using Fisher’s Exact Test with simulated p-values. To further explore profile variations at a broader structural level, differences in the cumulative semi-quantitative abundance of major VOC chemical classes were assessed using a one-way ANOVA followed by Tukey’s HSD post-hoc test. For this univariate analysis of aggregated chemical classes (ANOVA and Tukey HSD), untransformed relative abundances were used instead of log-ratio transformations (e.g., CLR). This approach was chosen because simple pseudo-count imputation creates severe variance artifacts for sparse trace classes (Weiss et al., 2016), while advanced zero-handling CoDA methods (e.g., Bayesian imputation, rclr; Quinn et al., 2019) are optimized for high-dimensional feature sets and are mathematically unstable for low-dimensional, highly aggregated class comparisons. This ensures a transparent and accurate statistical representation of the dominant proportional variations. To evaluate the magnitude of these differences, Eta-squared (η 2) effect sizes were calculated for the ANOVA models.

To visualize the volatile profiles and co-emission patterns, a heatmap was generated using the pheatmap package. Hierarchical clustering was deliberately omitted to manually group biological replicates (columns) by fungal strain and individual VOCs (rows) by their respective chemical classes. Compound names were standardized according to IUPAC nomenclature to ensure chemical rigor. For all analyses, statistical significance was set at α = 0.05.

Biocidal test I: in vitro virulence assay by direct fungal contact

Survival of Agriotes larvae was monitored for 12 weeks following direct exposure to EPF strains or control treatments. Survival analysis revealed a highly significant overall treatment effect on larval mortality (p < 0.0001; Figure 1). The proportional hazards assumptions were met for all Cox regression models (p = 0.8).

Most fungal strains did not significantly affect larval survival relative to the untreated control, which exhibited a final survival rate of approximately 92%. These strains included P. lilacinum MUCL 38701, B. bassiana MUCL 39817, L. lecanii MUCL 8115, P. fumosoroseus MUCL 15122, B. brongniartii MUCL 131 and 34853, and A. flavus MUCL 55276 (p ≥ 0.431). Similarly, the biological control reference T. harzianum MUCL 29707 did not significantly alter larval survival compared with the untreated control (p = 0.431), with a final survival rate of approximately 94% (see Figure 1 under label “Strains with non-significant mortality”).

In contrast, the chemical reference treatment (lambda-cyhalothrin) caused rapid and complete larval mortality, resulting in 0% survival within the first week of exposure. Survival was significantly lower than in the untreated control (p = 2.6 × 10 -6), confirming the sensitivity of the experimental system and the ability of the bioassay to detect treatment-induced mortality.

Among the pathogenic candidates, Ophiocordyceps sp. MUCL 58550 was the most virulent strain, significantly reducing larval survival relative to the untreated control (p = 0.030 – Figure 1). Exposure to this strain significantly increased the mortality risk (HR of 11.7, 95% CI: 5.16-26.49), resulting in the lowest survival rate among the fungal treatments (approximately 43%). The commercial strain M. brunneum strain B also significantly increased larval mortality risk (HR = 9.2, 95% CI: 3.72-22.58; p = 0.046), yielding a final survival rate of approximately 50%.

Although neither the commercial strain M. anisopliae strain A nor M. anisopliae MUCL 58549 significantly reduced larval survival (p = 0.145 and p = 0.173, respectively), both exhibited substantial biological activity, as indicated by elevated HR (HR = 4.6, 95% CI: 1.43–15.07 and HR = 4.1, 95% CI: 1.25–13.65, respectively). The most active of these treatments resulted in a final survival rate of approximately 64% (Figure 1).

Overall, Ophiocordyceps sp. MUCL 58550 was identified as the most promising candidate for further evaluation in soil microcosm assays. Despite the absence of a statistically significant effect, M. anisopliae MUCL 58549 was selected for subsequent soil assays based on the magnitude of the observed biological effect, as reflected by a Hazard Ratio of 4.1.

Biocidal test II: in vivo virulence of alginate-formulated EPFs in soil microcosms

The survival of wireworm larvae was monitored in soil microcosms over a 12-week period to assess fungal virulence under soil conditions. Survival analysis revealed a significant effect of fungal treatment on larval survival (p = 0.031; Figure 2). Among the strains tested, the encapsulated M. anisopliae MUCL 58549 significantly reduced larval survival relative to the untreated control (p = 0.013), resulting in a final survival rate of approximately 38%. Exposure to this treatment increased the risk of mortality by approximately threefold (HR = 3.08, 95% CI: 1.26-7.49).

In contrast, Ophiocordyceps sp. MUCL 58550 exhibited a more moderate effect on larval survival (Figure 2). Although exposure to this strain was associated with an increased mortality risk (HR = 2.23, 95% CI: 0.88-5.67), the effect was not statistically significant relative to the untreated control (p = 0.091), and the final survival rate remained approximately 46%.

Similarly, the biological control reference T. harzianum MUCL 29707 did not significantly affect larval survival relative to the untreated control (p = 0.659; HR = 1.26, 95% CI: 0.46-3.47). Final survival was approximately 67% in this treatment with 72% in untreated control. The absence of a significant effect of T. harzianum supports the conclusion that the increased mortality in the pathogen-treated groups was attributed to fungal pathogenicity rather than handling procedures or experimental stress.

All surface-sterilized cadavers (100%) from the M. anisopliae MUCL 58549 and Ophiocordyceps sp. MUCL 58550 treatments, yielded outgrowth of the respective inoculated fungi. The identity of the re-isolated strains was confirmed microscopically based on conidial morphology and sporulation structures. These observations demonstrate that mortality in both treatments resulted from successful EPF infection.

Wireworm’s orientation bioassay

Behavioral responses of wireworm to fungal formulation were assessed using dual-choice olfactometer. Overall responsiveness was high (63.1% on average), and the specific response rates for each treatment (T. harzianum MUCL 29707: 66.1%, Ophiocordyceps sp. MUCL 58550: 64.3%, M. anisopliae MUCL 58549: 58.9%) are detailed in Figure 3.

Larvae were significantly attracted to beads containing the biological control reference T. harzianum MUCL 29707, with 70% of responding individuals selecting the fungal treatment arm over the blank-bead control (Z = 2.683, p = 0.007, OR = 2.33 [95% CI: 1.25-4.35]). A similar pattern was observed for the EPF M. anisopliae MUCL 58549 beads, with 65% of responding larvae choosing the EPF treatment over the blank-bead control (Z = 2.331, p = 0.021, OR = 1.85 [95% CI: 1.10-3.12]). In contrast, beads containing the EPF Ophiocordyceps sp. MUCL 58550 did not induce a significant orientation response. Responding larvae were distributed equally between the fungal and the blank-bead control (Z = -0.665, p = 0.506, OR = 0.88 [95% CI: 0.50-1.55]).

Volatolomic characterization and biomarker identification

Overall VOC diversity and multivariate differentiation among fungal genera

Dynamic headspace sampling of the three encapsulated fungal formulations, analyzed by TD-GC-MS, detected a total of 80 volatile organic compounds (VOCs) spanning six chemical classes: alcohols, ketones, aldehydes, terpenes, alkanes, and aromatic compounds. A comprehensive, annotated database detailing the occurrence, chromatographic parameters, and relative abundance of all detected compounds is provided in Supplementary Table S1. Although the three fungal strains shared a broadly similar qualitative chemical composition, marked differences were observed in the relative abundance of individual compounds and chemical classes.

To visually assess the variation in volatile profiles, a heatmap of the VOC emissions was generated (Supplementary Figure S1). The heatmap grouped all replicates according to fungal strain with 100% accuracy, revealing distinct, block-like co-emission patterns that characterized each fungus. Specifically, a sesquiterpene cluster was uniquely associated with M. anisopliae MUCL 58549, a dense alcohol and ketone cluster with T. harzianum MUCL 29707, and a low-intensity, alkane-dominant cluster with Ophiocordyceps sp. MUCL 58550.

To statistically validate the strain-specific clustering patterns observed in the HCA, a PerMANOVA analysis was performed. Prior to interpretation, the assumption of homogeneity of multivariate dispersions among groups was confirmed using the betadisper test (F 2, 12 = 1.15, p = 0.349). The analysis revealed a significant effect of fungal strain on overall VOC composition (F 2, 12 = 7.32, R² = 0.768, p = 0.001), indicating that volatile profiles differed significantly among strains. Pairwise comparisons further confirmed significant differences between all fungal strains (p < 0.05).

To visualize these differences in multivariate chemical space, an unsupervised Principal Component Analysis (PCA) was conducted. Biological replicates (n = 5 per strain) formed tight, well-defined clusters, highlighting the high reproducibility of the VOC emission profiles (Figure 4). The first two PCA components explained 67.7% of the total variance and provided a clear separation among M. anisopliae MUCL 58549, Ophiocordyceps sp. MUCL 58550, and T. harzianum MUCL 29707.

To identify the specific biomarkers driving the separation already naturally observed in the unsupervised PCA, a supervised PLS-DA was performed. To rigorously rule out overfitting, the model was subjected to leave-one-out cross-validation (LOOCV) and permutation testing. The LOOCV demonstrated a robust predictive capability with a cross-validated accuracy of 100%. Furthermore, a 1,000-iteration permutation test confirmed that the model’s classification performance was significantly better than chance (p < 0.001). Consequently, the model explained 62% of the total variance in the dataset, indicating that the chemical signatures of the three strains were highly distinct (Supplementary Figure S2). Collectively, these results demonstrate that fungal strains can be accurately discriminated based solely on their VOC emission profiles.

Qualitative and quantitative chemical fingerprinting

Qualitative comparison of VOC chemical families revealed no significant differences in overall composition among fungal strains (Fisher’s Exact Test, p = 1; Supplementary Figure S3). All three strains emitted compounds belonging to each of the six identified chemical families, indicating a shared qualitative volatile framework. Thus, differences among isolates were driven by the relative abundance of chemical classes rather than by their presence or absence (Figure 5).

Analyses of relative abundances revealed marked isolate-specific differences in VOC emissions. The volatile profile of M. anisopliae MUCL 58549 was characterized by a significantly higher relative proportion of terpenes compared to both other strains (p < 0.001). In contrast, Ophiocordyceps sp. MUCL 58550 was distinguished by emitting significantly greater relative amounts of alkanes and esters compared to T. harzianum (p < 0.05), although these levels did not differ significantly from those of M. anisopliae. Finally, T. harzianum MUCL 29707 was uniquely characterized by significantly higher proportions of alcohols (p < 0.001). Regarding ketones, both Ophiocordyceps and T. harzianum emitted significantly higher proportions than M. anisopliae (p < 0.05). The relative production of aldehydes did not differ significantly among any of the fungal strains (p > 0.05).

Discriminating biomarkers

To identify the VOCs driving separation among fungal strains, Variable Importance in Projection (VIP) scores were extracted from the supervised PLS-DA model. Twenty compounds exceeded the significance threshold (VIP > 1.0) and were therefore considered the principal discriminant biomarkers underlying the observed clustering pattern (Figure 6). Among the most influential compounds were putatively identified bromoacetone, 2,2,4,4-tetramethyloctane, and 1-methoxydecane, which showed consistently high influence power across the model.

An analysis of the semi-quantitative relative abundances of the top ten most emitted VOCs per strain further highlighted their distinct volatile signatures (Figure 7). In M. anisopliae MUCL 58549, the relative volatile profile was dominated by sesquiterpene hydrocarbons, led by putatively identified aristolochene (20.1%) and eremophilene (19.0%). These compounds represented the largest proportion of its volatilome, making them the most characteristic VOCs associated with this strain and strongly associated with the significantly higher total terpene emission previously described.

The volatilome of Ophiocordyceps sp. MUCL 58550 contrasted markedly, being characterized by a predominant contribution of short-chain ketones and alkanes. This profile was primarily dominated by putatively identified 2-pentanone (10.0%), heptane (7.6%), and butanone (6.5%).

In contrast, the volatile output of T. harzianum MUCL 29707 was heavily dominated by characteristic C8 compounds, particularly putatively identified 1-octen-3-ol (18.6%) and 3-octanone (17.6%). The prominence of these specific alcohols and ketones distinguished T. harzianum from the two EPF and accounted for a substantial proportion of its total relative abundance.

This study was undertaken to address the growing need for sustainable alternatives to synthetic insecticides, following the restriction of neonicotinoids use, which has increased the vulnerability of economically important crops such as potatoes, maize and sugar beet to Agriotes wireworm damage. To support Integrated Pest Management (IPM) frameworks, our objective was to develop a mechanistic understanding of a subterranean attract-and-kill (A&K) strategy using entomopathogenic fungi (EPF). By transitioning from high-throughput in vitro screenings to soil-based microcosm experiments, we evaluated the performance of EPFs formulated in a starch-alginate matrix under conditions that better approximate the belowground environment. Coupling these bioassays with detailed volatile analysis allowed us to disentangle fungal pathogenicity from volatile-mediated behavioral effects, thereby providing a comprehensive assessment of the factors governing A&K efficacy.

Our results revealed a clear discrepancy between the intrinsic pathogenicity of fungal strains and their biocontrol performance under soil conditions. Although Ophiocordyceps sp. MUCL 58550 induced the highest mortality in direct-contact microplate assays, its efficacy declined significantly when formulated into alginate beads and evaluated in soil microcosms. In contrast, M. anisopliae MUCL 58549 maintained consistent virulence across both experimental scales and additionally induced significant attraction of wireworm larvae. Volatile profiling by TD-GC-MS indicated that these differences were associated with distinct strain-specific VOC signatures emitted by the formulated EPF.

Fungal pathogenicity: from in vitro screening to soil microcosms

In the initial in vitro screening (Test I), we aimed to identify fungal candidates capable of overcoming the physical and physiological defenses of Agriotes larvae. As expected, the insecticide control, lambda-cyhalothrin, induced rapid and complete mortality, whereas the efficacy of the biological treatments varied considerably. No significant mortality was observed for the generalist entomopathogens P. lilacinum and B. bassiana. This lack of mortality may, at least in part, reflect challenges associated with penetrating the relatively thick integument of Agriotes larvae (Traugott et al., 2015). However, the initial screening was primarily designed to identify acute and relatively fast-acting pathogenic phenotypes. Therefore, the absence of significant mortality under these short-term assays conditions does not necessarily preclude biological activity under different soil conditions or following longer exposure periods, as previously reported for B. bassiana (Ladurner et al., 2009). Within the conceptual framework proposed by St. Leger et al. (1991), successful infection depends on the ability of an EPF to recognize, colonize and breach host-specific chitinous barriers. Our results support this view, as only the strains recovered from naturally mycosed wireworm cadavers (M. anisopliae MUCL 58549 and Ophiocordyceps sp. MUCL 58550) caused substantial mortality under direct-contact conditions.

The virulence of EPFs relies on a sequential enzymatic degradation of the host integument (Ansari et al., 2009; Gibson et al., 2014). Following conidial adhesion, the secretion of Pr1 subtilisin-like proteases initiates protein digestion of the host cuticle (St. Leger et al., 1991). This process is supported by mechanical pressure and the secretion of chitinases, lipases, and esterases that further degrade the polymer matrix and epicuticular wax layers (Ortiz-Urquiza and Keyhani, 2013; Gao et al., 2011). Following penetration, the pathogen faces the insect’s immune response. In Agriotes larvae, these defenses primarily involve cuticular microbiome homeostasis, hemocyte encapsulation, and phenoloxidase-mediated melanization, particularly around the spiracles (Ladurner et al., 2009; Poggi et al., 2021). The high virulence of M. anisopliae observed in our assays is consistent with the well-documented pathogenic arsenal of this species. In addition to its efficient cuticle-degrading enzyme system, M. anisopliae produces secondary metabolites such as destruxins, that interfere with host cellular processes and suppress immune responses, thereby facilitating fungal establishment and dissemination within the host (Schrank and Vainstein, 2010). These mechanisms may explain its ability to maintain high infectivity against Agriotes larvae, despite the substantial defensive barriers characteristic of wireworms.

However, the transition from direct-contact in vitro bioassays (Test I) to soil microcosms (Test II) highlighted the critical role of host behavior and spatial ecology in determining infection success, revealing what can be described as a “virulence reversal” for specific strains. Despite the high in vitro mortality rate of Ophiocordyceps sp. MUCL 58550 (HR = 11.69), likely supported by the specialized enzymatic toolkit characteristic of this genus (Zheng et al., 2013; Wang et al., 2020), the strain’s efficacy declined markedly when larvae were not confined with the inoculum. This discrepancy suggests that intrinsic pathogenicity alone is insufficient to ensure successful biological control under ecologically more realistic conditions. In a three-dimensional soil matrix, the lack of an attractive cue results in only transient or accidental contact, creating a “race against time” between pathogen penetration and host evasion. As originally conceptualized by Vey and Fargues (1977), limited exposure may allow the host to physically “outpace” the pathogen via grooming (Yanagawa et al., 2008) or molting before the germ tube can successfully initiate penetration. This hypothesis is supported by recent findings from Wolfgang et al. (2025), who demonstrated that wireworms are not passive targets; they can actively reduce infection risks through behavioral defenses, including increased molting frequency to physically shed conidia. Together, these findings indicate that infection success in soil depends not only on fungal virulence traits but also on the capacity of the pathogen to maintain sufficient host contact to overcome behavioral avoidance and defensive responses.

In contrast, the efficacy of M. anisopliae MUCL 58549 in soil microcosms (HR = 3.08) supports the suitability of starch-alginate beads as a ‘nutritional oasis’ for EPF (Shah et al., 2023). The biopolymer matrix not only regulates moisture for conidial survival but also facilitates a ‘millet effect’ (Brandl et al., 2017). This micro-fermentation process may increase the local abundance of infective propagules increasing the conidial pressure in the soil (Paluch et al., 2024), and thereby enhance the probability of host encounter and infection. The incorporation of 30% starch may have further contributed to efficacy by stimulating feeding activity in wireworms (Antwi et al., 2018), potentially increasing both the frequency and duration of contact with fungal propagules. While this high starch content may accelerate the biodegradation of the physical matrix in natural soils, potentially reducing the bait’s long-term attractiveness, this breakdown process can conversely facilitate the release and establishment of the fungus into the surrounding soil, extending its presence as a secondary inoculum. Under these conditions, infection may occur not only through prolonged cuticular exposure but also through ingestion-associated pathways, thereby reducing the importance of the cuticle as the sole barrier to infection. Together, these mechanisms may explain why M. anisopliae MUCL 58549 maintained high efficacy in soil despite the behavioral and physiological defenses of wireworms, including grooming and molting. However, it is important to note that these soil experiments were conducted under sterilized microcosm conditions. While this controlled setup provides valuable mechanistic insights, it does not capture the influence of native microbial communities, fluctuating abiotic conditions, and heterogeneous soil structures that characterize natural agroecosystems. Consequently, further validation under more realistic field conditions is required to determine the efficacy and robustness of these formulations under agricultural conditions.

Volatolomic signatures and behavioral responses of wireworms

An effective A&K strategy must take into consideration the behavioral responses of the target pest. Kabaluk and Ericsson (2007) demonstrated that the ability of wireworms to detect and avoid lethal fungal conidia can substantially reduce the field efficacy of EPF. To circumvent this avoidance behavior, formulated beads should exploit the sensory mechanisms underlying wireworm host-seeking. In this context, Nikoukar et al. (2025) recently established a sensory hierarchy in Agriotes larvae showing that although CO 2 mediates long-range orientation, root-emitted VOCs are indispensable for final host discrimination. Consistent with this framework, our results confirm that M. anisopliae MUCL 58549 (65% attraction) and the biological control reference T. harzianum MUCL 29707 (70% attraction) produce volatile cues capable of overriding this hierarchy and eliciting significant larval attractions, whereas Ophiocordyceps sp. MUCL 58550 produced no directional response. Furthermore, it is crucial to recognize that behavioral attraction alone does not necessarily translate into epidemiological transmission. While VOCs may guide the pest to the bead, successful infection ultimately depends on prolonged contact, potential ingestion, and the subsequent overcoming of host defenses.

Our semi-qualitative volatile profiling revealed high chemical similarity among the tested fungal species (Fisher’s Exact Test, p = 1.00). This aligns with the “volatolomic toolkit” concept (Davis et al., 2013), suggesting that biochemical pathways producing common fungal volatiles (e.g., 1-octen-3-ol, 3-octanone, and 2-pentanone) are broadly conserved across diverse taxa. Because these ubiquitous microbial metabolites indicate resource-rich habitats, soil-dwelling insects likely evolved to exploit them. However, rather than the simple presence or absence of individual compounds, our results indicate that wireworm behavioral responses are shaped by quantitative variation in volatile production.

Although all strains produced largely overlapping VOCs, their differing attractiveness was associated with marked shifts in relative abundance. For M. anisopliae MUCL 58549, the profile featured elevated emissions of the putatively identified sesquiterpenes aristolochene and eremophilene (20.1% and 19.0% of total VOCs, respectively; VIP > 2.0; p < 0.01), compounds associated with highly virulent Metarhizium strains (Ansari et al., 2009). By emitting these compounds, the fungus may mimic an active root system, increasing its attractiveness to foraging wireworms and supporting its utility in A&K strategies (Brandl et al., 2017). This process is likely driven by an active farnesyl pyrophosphate (FPP) biosynthetic pathway. Given the role of volatile cues in short-range host discrimination (Nikoukar et al., 2025), quantitative modulation of FPP-derived metabolites may represent a key mechanism underlying attractiveness. However, it must be emphasized that our study currently demonstrates a strong correlation between these specific VOC profiles and larval attraction. The definitive causal role of individual biomarkers, such as aristolochene, or specific volatile blends remains to be strictly established through future dose-response bioassays using synthetic pure compounds.

While wireworms rely on specific root-emitted VOC blends for host location (Barsics et al., 2017; Nikoukar et al., 2025), our results suggest that M. anisopliae MUCL 58549 may exploit these same olfactory pathways. Furthermore, although its profile differs from strains producing aggregation-associated compounds like 2-pentylfuran (La Forgia et al., 2023), it nevertheless elicited strong attraction, indicating that distinct fungal volatile blends may converge functionally by activating wireworm sensory mechanisms to identify suitable feeding sites.

Beyond attraction, fungal VOCs may contribute to pathogenicity through biofumigant effects operating over distance within the soil matrix (Bourdon et al., 2022). By diffusing through soil pore spaces, these volatiles may interfere with insect immune defenses, such as inhibiting phenoloxidase activity and the melanization response responsible for pathogen encapsulation, prior to direct contact with propagules (Butt et al., 2016; Hummadi et al., 2021). This complements the attractive function, creating a multifaceted strategy where VOCs both facilitate host location and may enhance susceptibility to infection.

In contrast, the lack of a behavioral response to Ophiocordyceps sp. MUCL 58550 may result from a deficit of specialized attractants. While the metabolic “toolkit” for sesquiterpene production is present, their reduced relative abundance likely limits their perceptibility within the chemically complex soil environment (Ansari et al., 2009). Instead, its volatilome is dominated by ubiquitous ketones and alkanes, such as 2-pentanone (10.0%), heptane (7.6%), and butanone (6.5%), which are widespread products of general microbial metabolism in soil ecosystems (Davis et al., 2013). Because wireworm attraction is mediated by a restricted subset of biologically active VOCs rather than general background soil volatiles (Barsics et al., 2017; Antwi et al., 2018), our findings suggest that the predominance of these ubiquitous compounds may explain the strain’s chemical neutrality and lack of attraction.

This chemical neutrality likely explains the “virulence reversal” observed between the in vitro and soil assays. Although Ophiocordyceps was highly virulent in vitro (HR = 11.69), its efficacy declined significantly in the complex 3D soil matrix (Brandl et al., 2017; Bourdon et al., 2022). Formulation in the starch-rich matrix did not induce an attractive signaling profile, indicating that VOC emission is an intrinsic metabolic trait (Wang et al., 2023) rather than a response to substrate availability (Butt et al., 2016). Unable to bridge spatial gaps in heterogeneous soils through chemical signaling, this strain struggles to establish necessary host-pathogen contact, potentially reducing its suitability for granular A&K systems.

The inclusion of T. harzianum MUCL 29707 provided an informative non-pathogenic reference, confirming that physical contact with fungal biomass alone was insufficient to induce mortality. Despite its lack of pathogenicity, this strain elicited the highest recruitment rate (70%), significantly exceeding that of the other strains in dual-choice assays. This strong behavioral response was possibly associated with an elevated relative abundance of C8 oxylipins, particularly 1-octen-3-ol (18.6%) and 3-octanone (17.6%), compounds widely recognized as biologically active fungal volatiles involved in mediating interactions with wireworms (Bennett and Moore, 2025; Hummadi et al., 2021).

While Trichoderma species are recognized as effective plant biostimulants (González-Pérez et al., 2022), the performance of T. harzianum MUCL 29707 highlights an important constraint of “A&K” systems, strong attractivity must be coupled with sufficient intrinsic virulence to ensure that behavioral recruitment translates into effective pest mortality. This distinction between behavioral recruitment and pathogenic efficacy underscore the importance of combining these two functions when developing effective A&K systems, as exemplified by the virulent M. anisopliae MUCL 58549 strain.

Formulation strategy, larval defenses, and agronomic perspectives

Our findings suggest that Agriotes larvae function as complex holobionts, in which resistance emerges from multiple interacting layers of symbiotic defense. For instance, Wolfgang et al. (2025) examined interactions between larvae and the soil microbiota and reported that exposure to fungal pathogens induced a form of “immune priming”. This response accelerates molting (ecdysis), enabling larvae to physically shed spores before they can breach the cuticle. Similarly, Tang et al. (2024) showed that variation in gut-associated microbial communities influences host susceptibility, demonstrating that the symbiont Acinetobacter baumannii can enhance the expression of host immune genes, and thereby inhibit the establishment of M. anisopliae.

In this context, we developed a starch-alginate formulation to bypass these barriers by promoting oral acquisition of fungal propagules. As demonstrated by Thorpe et al. (1947), carbohydrate-rich substrates elicit a strong biting response in wireworms; accordingly, our 30% starch matrix was incorporated as a phagostimulant to promote direct ingestion of the beads. We hypothesize that this strategy delivers a high conidial dose directly into the midgut, with the aim of overwhelming microbiota-mediated defenses and reducing the protective effects of symbiotic microorganisms previously described by Kabaluk and Ericsson (2007). However, it should be noted that larval biting rates on the formulated beads were not directly measured in this study; whether the starch content elicits a biting response of sufficient frequency to contribute to the observed mortality remains to be confirmed in vivo.

From an agronomic perspective, these volatolomic insights provide support for a “stimulate-and-deter” (push-pull) strategy. Bourdon et al. (2022) and Wood et al. (2023) investigated the dose-dependent responses of wireworms to fungal oxylipins and reported that, while low concentrations may be attractive, high concentrations of 1-octen-3-ol and 3-octanone function as biofumigants or repellents. Accordingly, high-concentration oxylipin formulations could be deployed to repel larvae from susceptible crops “push”, while our attractive starch-enriched beads could attract “pull” and eliminate the pest in the inter-row, creating a lethal sink in the soil matrix. Together, these complementary mechanisms may concentrate pest populations in targeted areas where infection pressure is maximized, thereby enhancing overall control efficacy.

Translating this volatile-mediated A&K strategy to field conditions will require consideration of several ecological variables. In natural soils, the starch-alginate formulation will inevitably interact microbial communities, including saprotrophic microorganisms. Although microbial degradation may reduce the physical longevity of the bait, we hypothesize that this process could transiently enhance volatile and CO 2 emissions. Matrix degradation may also facilitate the release and dissemination of fungal conidia into the surrounding soil, potentially extending the functional lifetime of the formulation beyond that of the original bait matrix. Such persistence could be particularly relevant for regulating overlapping pest generations, as our current bioassays focused specifically on the attractiveness of late-instar larvae. Whether this formulation is equally effective against younger instars, eggs, or adults remains an important question for future research. Beyond life-stage-specific attractiveness, the behavioral consequence of sublethal exposure may also influence field efficacy. If a larva encounters or consumes a sublethal dose, it remains unclear whether it would maintain attraction to the bait or subsequently develop behavioral avoidance. Conversely, if sublethal exposure induces disease-associated anorexia or reduce feeding activity, such effect could nevertheless contribute to crop protection by limiting pest damage. Moreover, the dual-infection capability of EPFs (oral ingestion and cuticular penetration) could increase the likelihood that even brief or non-sustained encounters with the formulation result in infection. Ultimately, the field performance of this encapsulation strategy will depend not only on its efficacy against the target pest but also on its ecological specificity. Although the precise host range and non-target effects of our M. anisopliae strain remain to be determined, the formulation may reduce exposure of non-target organisms by relying on olfactory cues associated with root-foraging behavior. By emitting VOCs that mimic or resemble root-associated chemical signals, the bait could preferentially attract root-feeding pests, while limiting attraction of soil fauna that do not feed on roots. This hypothesis, however, requires experimental validation. Ongoing studies in our laboratory are therefore evaluating the precise impact of the formulation on soil microbial communities and non-target arthropods, providing an essential step toward assessing its ecological safety and field applicability.

This study demonstrates that intrinsic virulence and volatile-mediated host recruitment are independent, yet equally necessary, components of an effective Attract-and-Kill strategy. The behavioral neutrality of the highly virulent Ophiocordyceps sp. MUCL 58550 illustrates how a lack of chemical attraction limits host-pathogen encounters in the soil, explaining the frequent disconnection between laboratory bioassays and field performance. Conversely, our findings with the biological control reference T. harzianum highlight that attraction without lethality risks creating “safe havens” for pests. Bridging this gap, the dual-action mechanism of M. anisopliae MUCL 58549 successfully hijacks wireworm foraging behavior through root-mimicking sesquiterpenes and starch phagostimulation, transforming a passive formulation into an active “lethal sink”. While our study identifies strong correlations between specific volatile profiles and behavioral responses, the causal role of individual compounds or blends remains to be confirmed through future experiments using synthetic lures. Ultimately, by integrating principles of chemical ecology with formulation engineering, such as prioritizing strains with attractive volatile profiles or incorporating synthetic lures. Future research should focus on bridging the gap between controlled laboratory microcosms and complex field environments, thereby supporting the development of robust and effective belowground pest management strategies.