Integrating sniffer bees into security and defenses: Advancements in the detection of explosives and illicit substances

Introduction

Insects represent one of the most successful and ecologically diverse groups of eukaryotes on Earth, accounting for approximately half of all known terrestrial species. Their remarkable ecological adaptability is largely attributable to highly evolved olfactory systems capable of detecting and discriminating among a vast repertoire of volatile chemical signals (Schmidt and Benton, 2020) [15] . Among insects, the Western honeybee (Apis mellifera) stands out as an exceptional candidate for applied biosensing. Beyond its well-known roles in honey production and crop pollination, the honeybee has demonstrated profound versatility as a biological sensor across domains ranging from national security and law enforcement to clinical diagnostics and environmental monitoring. Conventional detection technologies for explosives, illicit drugs, and disease biomarkers face significant limitations: metal detectors are ineffective against plastic mines; gas chromatography-mass spectrometry (GC-MS) requires expensive laboratory infrastructure; and trained sniffer dogs while highly effective are costly, require months of training, and carry the physical risk of triggering landmines (Kasban et al., 2010) [7] . In this context, honeybees offer a compelling biological alternative. Their antennae harbor approximately 3,000 sensory organs per antenna, supporting 170 olfactory receptor genes far exceeding the receptor diversity of Drosophila (62 genes) or mosquitoes (79 genes) enabling discrimination of target compounds at concentrations in the parts-per-trillion (ppt) range (Sandoz, 2011) [14] . Research interest in sniffer bee technology has intensified considerably over the past two decades, driven substantially by military funding. The United States Defense Advanced Research Projects Agency (DARPA) has sponsored research programs at multiple institutions, including the University of Montana, Sandia National Laboratories (SNL), and the Los Alamos National Laboratory (LANL), focusing on applications such as Free Flight Landmine Detection, Proboscis Extension Reflex (PER)-based detection devices, and the revolutionary Hybrid Insect Micro-Electro-Mechanical Systems (HI-MEMS) (Hadagali and Suan, 2017) [6] . Beyond security applications, bees have been trained to detect volatile organic compound (VOC) signatures of diseases including lung cancer, tuberculosis, diabetes, and SARS-CoV-2 (Parnas et al., 2023) [12] . The present review synthesizes the current state of knowledge on sniffer bee science, encompassing the neuroanatomy and neurophysiology of the honeybee olfactory system, established and emerging training protocols, a comparative evaluation against existing technologies, and future directions.

Historical Context: Insects in Warfare

The intersection of insects and warfare has deep historical roots, both through vector-borne disease and deliberate weaponization. During the Napoleonic Wars (1803-1815), insect-borne typhoid fever claimed the lives of four soldiers for every one killed in direct combat. Similarly, during the American Civil War (1861-1865), approximately two-thirds of battlefield deaths were attributable to insect-transmitted diseases including malaria, typhoid, yellow fever, and cholera (Lockwood, 2012) [11] . Deliberate deployment of bees as weapons is documented across multiple historical periods. During the Battle of Antietam, a Confederate cannonball struck a row of beehives at Roulette Farm, sowing panic among Union forces (Delaney, 2011) [2] . During World War I, German forces utilized tripwires connected to beehives to impede advancing British troops. In the Italian-Ethiopian conflict (1935-1936), Ethiopian defenders dropped beehive bombs onto Italian tanks. In ancient times, Romans catapulted beehives into enemy fortifications, a tactic also recorded in accounts of the Roman Empire's conquests. Japan's Unit 731 during World War II conducted extensive biological warfare experiments, including the use of plagueinfected fleas (Yersinia pestis) dispersed via bombs to infect civilian and military populations in China one of the most egregious documented cases of entomological warfare (Lockwood, 2012) [11] . During the Cold War, the United States Army conducted "Operation Big Buzz," releasing thousands of uninfected mosquitoes over Georgia to assess insect-vector feasibility for biological warfare. These historical episodes provide important context for the subsequent scientific and military interest in harnessing insect sensory capabilities for defense applications.

Landmine Detection: Traditional Approaches and the Case for Sniffer Bees

Since the mid-19th century, landmines have caused an estimated 15,000-20,000 accidental deaths annually, with children constituting over 30% of victims. Survivors frequently suffer amputations and permanent disabilities. At current clearance rates, it has been estimated that removing all existing landmines could take 450-500 years (Kasban et al., 2010) [7] . Traditional detection involves dividing areas into grids and using metal detectors with prodding sticks a method largely unchanged over 40 years and incompatible with non-metallic or plastic mines. The RAND Corporation has classified emerging detection technologies into five categories, with biological sensors including trained animals and insects representing one promising avenue. Demining dogs are the most established biological detectors, trained to respond to the scent of trinitrotoluene (TNT) and related compounds. However, their performance is susceptible to environmental factors, fatigue impairs accuracy during repeated sweeps, and their weight can trigger anti-personnel mines (Hadagali and Suan, 2017) [6] . Against this backdrop, honeybees present several compelling advantages: they weigh too little to trigger landmines; colonies can be conditioned rapidly and inexpensively; and their foraging behavior naturally distributes hundreds of individuals across a search area simultaneously. Researchers at the University of Montana, contracted by DARPA, demonstrated that honeybees conditioned with trace amounts of TNT (via 2,4dinitrotoluene, a TNT metabolite) can detect explosive vapors at concentrations of 0.01-100 ppt a sensitivity comparable to, and in many contexts exceeding, that of trained dogs (Hadagali and Suan, 2017) [6] . [7] .

Training Methodologies

Three principal paradigms have been developed for training honeybees as detection agents, each with distinct operational characteristics, technical requirements, and deployment profiles.

Free Flight Mechanism

The free flight method, pioneered by researchers at the University of Montana in collaboration with DARPA, Sandia National Laboratories (SNL), and the Air Force Research Laboratory (AFRL), exploits the honeybee's natural foraging behavior. Bees are conditioned over 1-2 days by supplementing their food with trace amounts of TNT (via its degradation product, 2,4-dinitrotoluene, DNT) and caffeine the latter functioning as a memory enhancer that improves retention of learned odor associations (Hadagali and Suan, 2017) [6] .

Field trials conducted at Fort Leonard Wood Army Base, Missouri where active minefields are maintained for testing demonstrated that conditioned bees released into a test area located and congregated above explosive vapor plumes within one minute. Remarkably, bees continued visiting mined sites even in the absence of food rewards. The primary vapor released by buried landmines is TNT, which undergoes partial reduction in soil moisture to form DNT; bees conditioned to either compound responded to both (Delaney, 2011) [2] . Detection rates of 97-99% were achieved for 2,4-DNT vapor at ambient concentrations of 0.01-100 ppt sensitivity that equals or surpasses most man-made sensors and is comparable to trained demining dogs. Tracking of deployed bees was achieved using Light Detection and Ranging (LIDAR) technology the same radar-based mapping approach used by the National Oceanic and Atmospheric Administration to monitor fish populations. LIDAR allowed real-time mapping of bee locations and dwell times, which correlated precisely with mine positions (Delaney, 2011) [2] . Notably, during a three-day trial, bees also congregated above areas with no known mines, where subsequent investigation revealed residual TNT/DNT contamination in the soil demonstrating capability beyond mine localization to residual contamination mapping. Key limitations of the free flight approach include inability to operate at night, in heavy rain or strong winds, or at temperatures exceeding 40 °C; difficulty tracking individual bees during large-area sweeps; and the foraging range of bees (over 2.5 miles), which complicates target localization. Initial tracking methods included fluorescent dust and miniature radio-frequency tags, but LIDAR proved most effective in field conditions (Delaney, 2011) [2] .

Fig 3:

Workflow of honeybee conditioning using explosive-related odour mixtures (DNT, TNT, caffeine) in sugar syrup, followed by field deployment and detection using LIDAR (Delaney, 2011) [2] .

Contained Detection Devices/Proboscis Extension Reflex (PER)

Following the success of free flight studies, DARPA sponsored Los Alamos National Laboratory (LANL) to develop the Stealthy Insect Sensor Project, establishing contained detection devices based on the Proboscis Extension Reflex (PER). The PER is a well-characterized Pavlovian conditioned response: when honeybees associate a specific odor with a sucrose reward, they subsequently extend their proboscis upon detecting that odor alone, in a manner functionally analogous to Pavlov's salivation response in dogs (Rehman et al., 2017) [13] . In the LANL system, restrained bees were conditioned by simultaneous presentation of explosive vapors (TNT, DNT, C4) and a sucrose syrup reward. Once conditioned, three bees were placed in a "sniffer box" equipped with a camera monitoring each bee's head. When ambient air containing explosive vapors was drawn through the device, conditioned bees exhibited PER, which was captured by the camera and displayed on a laptop monitor. In a 2005 demonstration for DARPA, the US Southern Command, and US Marines, the sniffer box successfully discriminated contaminated from uncontaminated crates and vehicles containing C4 and TNT (Hadagali and Suan, 2017) [6] . A significant advancement was the development of VASOR 136 (Volatile Analysis by Specific Olfactory Recognition), a handheld device accommodating 36 bees secured in reusable harnesses distributed across 6 cassettes. The device presented unfiltered ambient air to bees maintained at reduced temperature (to slow movement without harm). A red screen indicated detection (PER triggered by two or more of three monitored bees); green indicated no detection. VASOR 136 eliminated human-dependent monitoring via automated image analysis of bee movement, pattern, and orientation displayed on a PDA. Critically, the device detected not only TNT, DNT, and C4, but also Howitzer propellant grains used in improvised explosive devices, gunpowder, Semtex, PE4, DMNR, and TATP a compound notoriously difficult to detect even with trained dogs.

Retraining for novel explosive compounds requires less time and cost than initial conditioning (Hadagali and Suan, 2017) [6] . [6] .

Hybrid Insect Micro-Electro-Mechanical Systems (HI-MEMS)

The most technologically advanced bee-based detection paradigm is the Hybrid Insect Micro-Electro-Mechanical Systems (HI-MEMS) program, funded by DARPA with the goal of developing cybernetic insect organisms ("cyborgs"). MEMS chips containing sensors, probes, and wireless transmitters are implanted into developing insects during the late pupal stage, when metamorphosis maximizes tissue renewal around the implant, establishing a stable tissuemachine interface (Bozkurt et al., 2008) [1] . The chips harvest energy from the insect's own movement and power onboard sensors including microphones, cameras, or gas sensors. Remote control of cyborg insect locomotion is achieved via GPS, radio frequency, or ultrasound. Crucially, HI-MEMS allows operators to access the insect's native visual system directly rather than attaching external cameras, enabling covert intelligence collection, target tracking, and sample retrieval while maintaining the insect's natural appearance (Siljak et al., 2022) [17] . Research at Cornell University successfully demonstrated MEMS implantation in tobacco hornworms (Manduca sexta), establishing muscular electrical control via this technology (Bozkurt et al., 2008) [1] .

Preconcentration Techniques

Remote Explosive Trace Sampling (REST) collects air samples from suspected mine areas for subsequent laboratory analysis using detection animals or instruments.

While useful, REST exposes detection animals to contaminated zones. To mitigate this, passive preconcentrator systems have been developed using Fluoropolymer Aflas (FPA) a copolymer of tetrafluoroethylene, vinylidene fluoride, and propylene that captures nitroaromatic vapors (TNT, 2,4-DNT) through hydrogen-bonding surface absorption, analogous to solidphase extraction in HPLC (Gillanders et al., 2019) [5] . In practical implementation, FPA cartridges are placed at hive entrances. Returning forager bees pass through the cartridges, depositing residues collected electrostatically from their body hairs. Thermal desorption of the cartridge releases accumulated compounds, which are quantified via luminescence quenching using luminescent organic semiconductor polymer films. This preconcentration approach amplifies analyte signals by orders of magnitude before bee-based or instrument-based detection, substantially improving sensitivity for deeply buried or lowvapor-pressure mines (Gillanders et al., 2019) [5] .

Neuroanatomy and Neurophysiology of the Honeybee Olfactory System 5.1 Peripheral Detection: Antennal Sensilla

Olfactory detection in A. mellifera begins at the antennae, which carry sensilla placodea thin, oval-shaped olfactory sensilla (9 μm × 6 μm) perforated by numerous pores and housing 5-35 olfactory receptor neurons (ORNs) each. Odor molecules traverse pores, dissolve in sensillum lymph which contains odorant-binding proteins (OBPs) facilitating molecular transport and bind to olfactory receptor (OR) proteins on ORN dendrites. The honeybee OR complex differs fundamentally from vertebrate ORs, functioning as a heteromeric ion channel composed of a ligand-specific OR and a co-receptor (AmOr2, orthologous to Drosophila Or83b) (Schmidt and Benton, 2020) [15] . The honeybee genome encodes 170 OR genes (including 7 pseudogenes), a repertoire substantially larger than Drosophila (62) or mosquito (79), reflecting the ecological demands of plant volatile discrimination and colony pheromone communication (Sandoz, 2011) [14] .

Primary Processing: The Antennal Lobe

ORN axons project via the antennal nerve to the antennal lobe (AL), the primary olfactory processing center. The honeybee AL contains 165 glomeruli spherical neuropil units where ORNs synapse onto local neurons (LNs) and projection neurons (PNs). Each glomerulus receives convergent input from ORNs expressing the same receptor type, encoding a specific olfactory channel. ORNs release acetylcholine (ACh) to activate both uniglomerular projection neurons (uPNs) and local interneurons (Sandoz, 2011) [14] . LNs are predominantly inhibitory, utilizing GABA, glutamate, or histamine as neurotransmitters. They are classified as homogeneous (homo-LNs, connecting all glomeruli for global gain control) or heterogeneous (hetero-LNs, connecting a subset of glomeruli for contrast enhancement). The AL performs two critical computations: gain control normalizing PN activity across a wide range of stimulus intensities and contrast enhancement sharpening glomerular activity patterns to increase odor discriminability (Sandoz, 2011) [14] . Two major output tracts relay processed olfactory information to higher brain centers: the lateral antenno-protocerebral tract (l-APT), which innervates the lateral horn (LH) before the mushroom bodies (MB); and the medial antenno-protocerebral tract (m-APT), which innervates the MB before the LH. These tracts draw from non-overlapping sets of glomeruli, forming two largely independent olfactory subsystems (Sandoz, 2011) [14] .

Higher Processing: Mushroom Bodies and Lateral Horn

The mushroom bodies (MB) are bilobed neuropil structures central to olfactory learning and memory. Projection neuron axons terminate in MB calyces, where they form microglomerular contacts with ~170,000 Kenyon cells (KCs). KCs typically require convergent input from many PNs to fire a threshold mechanism generating sparse coding meaning only a small fraction of KCs respond to any given odor. This sparse representation is theoretically optimal for high-capacity associative memory storage: the critical basis for Pavlovian odor-reward conditioning exploited in bee training paradigms (Sandoz, 2011) [14] . The Pe-1 neuron, a major MB output neuron, integrates multimodal inputs (olfactory, visual, mechanosensory) and modulates behavioral responses. The VUM-mx1 neuron a singular octopaminergic neuron projects to the AL, MB, and LH simultaneously, acting as a reinforcement signal during reward learning by encoding sucrose unconditioned stimulus (US) value. This neuron is the neurophysiological substrate of the PER conditioning protocol. The lateral horn (LH) receives inputs from both uPNs and multiglomerular PNs (mPNs) and likely functions as a pre-motor center integrating innate olfactory responses (pheromones) with learned associations (Sandoz, 2011) [14] .

Molecular Mechanisms of Olfactory Transduction

The molecular basis of honeybee olfactory detection centers on the OR/co-receptor complex. Ligand binding triggers conformational changes in the heteromeric ion channel, causing cation influx and ORN depolarization generating action potentials relayed to the AL. Odorant-binding proteins (OBPs) in the sensillum lymph facilitate solubilization and transport of hydrophobic odorant molecules to the OR complex. Odorant-degrading enzymes (ODEs) within the sensillum lymph terminate signaling by metabolizing odorants, enabling temporal resolution of rapidly changing olfactory stimuli (Schmidt and Benton, 2020) [15] . The expanded OR repertoire of honeybees reflects evolutionary pressures for discrimination among thousands of floral VOCs, colony pheromones, and environmental cues. Comparative genomic analyses suggest significant positive selection on bee OR genes, particularly in ligandbinding domains, consistent with ecological specialization (Schmidt and Benton, 2020) [15] . Plasticity in olfactory receptor expression regulated by experience and colony social context may contribute to the facility with which honeybees acquire new odor-reward associations in training protocols.

Applications as Detection Agents 6.1 Lung Cancer Biomarker Detection

Exhaled breath contains over 3,500 known volatile organic compounds (VOCs), whose relative composition reflects systemic metabolic states. Lung cancer the leading cause of cancer-related mortality globally is associated with a characteristic VOC profile including decane, undecane, 2methylheptane, 2,2,4,6,6-pentamethylheptane (PMH), pentanal, hexanal, nonanal, propylbenzene, and trichloroethylene. Current gold-standard analysis (GC-MS) offers high sensitivity but requires specialized instrumentation and expertise. Electronic noses (e-noses) are portable alternatives but exhibit lower sensitivity and susceptibility to environmental drift (Parnas et al., 2023) [12] . Parnas et al. (2023) [12] proposed exploiting the broadband sensitivity of the honeybee antennal lobe neural circuit as a biologically inspired gas sensor. in vivo electrophysiological recordings from antennal lobe projection neurons demonstrated distinct, reproducible population-level activity patterns for each of nine lung cancer biomarkers, as well as for synthetic lung cancer breath mixtures versus healthy breath mixtures. Critically, the system reliably classified all nine biomarkers and their mixtures without requiring prior identification of specific VOC targets a key advantage over e-nose systems that must be designed around known analytes. Experimental bees were restrained in dental wax harnesses, antennal lobe tissue was surgically exposed, multielectrode arrays recorded neural responses during olfactometer-delivered odor presentations. Computational analysis of population response patterns using biological neural algorithms achieved reliable multi-class classification, validating the proof-of-concept for a bee olfactory circuit-based lung cancer breath test (Parnas et al., 2023) [12] .

Detection of Tuberculosis

Suckling and Sagar (2011) [18] demonstrated that Apis mellifera workers can be conditioned to detect Mycobacterium tuberculosis signature volatile compounds specifically methyl phenylacetate, methyl p-anisate, and methyl nicotinate using the PER paradigm. Forager bees were collected and immobilized, then harnessed in 1 mL Eppendorf tubes to allow head movement. Following sucrose satiation and overnight starvation at 28 °C, bees underwent five conditioning trials at 10-minute intervals, each pairing 5-second odor presentation with 5-second sucrose reward. Bees that failed to show PER to sucrose or showed PER at baseline were excluded (Suckling and Sagar, 2011) [18] . The sensitivity achieved in laboratory conditions supports potential application in low-resource settings where conventional TB diagnostics (sputum culture, PCR, chest Xray) are unavailable. The scalability of bee conditioning, combined with the low cost of maintaining colonies, positions this approach as a complement to existing diagnostics in endemic regions.

Detection of Diabetes

A collaborative project between the Joslin Diabetes Center (Dr. Allison Goldfine and colleagues Tobias Horstmann and Juliet Phillips) and Los Alamos National Laboratory has trained honeybees to detect elevated acetone concentrations in exhaled breath a biomarker of ketoacidosis associated with uncontrolled diabetes. A purpose-built device houses conditioned bees and allows patients to breathe into the apparatus; bees conditioned to acetone extend their proboscis (or aggregate toward the breath source) when the patient's breath contains elevated acetone.

The primary motivation is portability and non-invasiveness: the device requires no blood sampling and could be deployed as a rapid screening tool in low-resource or rural settings in developing countries, where laboratory infrastructure is absent. This application exemplifies the potential of bee-based biosensors to democratize diagnostic access.

Detection of SARS-CoV-2

SARS-CoV-2 infection induces characteristic changes in host VOC emission profiles detectable in exhaled breath and biological samples. Kontos et al. (2022) [10] demonstrated that honeybees can be conditioned using Pavlovian protocols to discriminate between VOC signatures of SARS-CoV-2-infected and healthy samples (derived from throat swabs of American mink [Neovison vison] experimentally infected with SARS-CoV-2). In these experiments, 150 bees were distributed in cylindrical containers and harnessed after cold immobilization. Sample VOCs from SARS-CoV-2-soaked filter paper were introduced via air pumps, with simultaneous presentation of control (non-infectious) and target (infectious) VOC streams. Sugar reward was paired with infectious sample exposure. After 7-10 conditioning trials, bees reliably extended their proboscis in response to infectious samples without sucrose reward, confirming acquired conditioned discrimination (Kontos et al., 2022) [10] . The approach offers advantages in speed, cost, and portability over PCR-based assays in resource-limited settings. A key challenge is controlling for confounding VOCs from sample matrices, which requires standardized sample preparation protocols.

Detection of Illicit Drug

The legalization and decriminalization of cannabis in multiple jurisdictions has undermined the legal standing of trained sniffer dog alerts as probable cause for searches. Retraining dogs to selectively ignore cannabis while retaining sensitivity to other drugs is operationally challenging and costly. Trained insects offer a legally and technically flexible alternative (Schott et al., 2015) [16] . Schott et al. (2015) [16] systematically evaluated three insect species the European grapevine moth (Lobesia botrana), the Madagascar hissing cockroach (Gromphadorhina portentosa), and the Western honeybee (A. mellifera) for their antennal electrophysiological responses to volatiles of heroin, cocaine, amphetamine, and cannabis using a portable electroantennography (EAG) device. Honeybees demonstrated physiological antennal responses to all tested drugs, including street-grade adulterated samples (heroin at 0.2% and 47.3% purity; cocaine at 20% and 70% purity; amphetamine at 16% purity; cannabis flower). Behavioral conditioning studies used automated acrylic measurement chambers (148 × 20 × 6 mm) instrumented with 26 infrared sensor pairs tracking bee position at 16 Hz. Conditioned stimulus (target drug odor, 8 seconds) was paired with a mild electric shock (10 V, 1.2 Hz) as unconditioned stimulus; a second odor (unpunished) served as control. Bees successfully learned to avoid the punished drug odor, demonstrating acquisition of aversive olfactory conditioning. Post-conditioning EAG analysis confirmed specific binding of drug compounds to antennal receptor sites (Schott et al., 2015) [16] .

The Sniffer Wasp: An Alternative Biosensor

Beyond honeybees, the braconid parasitic wasp Microplitis croceipes has attracted significant research interest for its olfactory capabilities, particularly in food safety monitoring. M. croceipes exhibits nearly 100-fold greater sensitivity to specific compounds (notably myrcene and 3-octanone) compared to the Cyranose 320 electronic nose. This wasp demonstrates context-dependent conditioned behavioral responses: when conditioned with an odorant while feeding on sucrose, subsequent exposure elicits foraging behavior; when conditioned during host stinging (Helicoverpa zea), subsequent exposure elicits host-seeking behavior (Gaidos, 2008) [4] .

Researchers at the University of Georgia developed the "Wasp Hound" a cartridge-based detection device in which live M. croceipes are exposed to air samples drawn through the device. A webcam monitors wasp crowding behavior near the air inlet (indicating target odor detection), analyzed in real-time by Visual Cortex software. Applications have included detection of aflatoxin contamination in food stores and location of animal carcasses. The Wasp Hound represents a complementary platform to bee-based devices, potentially deployable for food safety screening (Gaidos, 2008) [4] . [12] . Principal limitations of bee-based detection include weather and temperature dependency (bees cannot fly at night, in rain, or above 40 °C); limited operational lifespan of conditioned bees; challenges in tracking individual bees during free-flight deployment; and the ethical considerations of exposing insects to aversive conditioning stimuli. Continued advances in LIDAR tracking, automated PER monitoring (VASOR 136), and HI-MEMS integration are progressively addressing the operational constraints.

Comparative Assessment: Sniffer Bees vs. Conventional Detectors

Effects on Foraging and Pollinating Behaviour

An important consideration in any bee-based detection application is the potential impact of conditioning on natural foraging and pollination behavior. Detection using the Sting Extension Reflex (SER) an innate defensive response to noxious stimuli used in some explosive detection protocols suppresses foraging due to the negative emotional state associated with aversive conditioning. To restore foraging activity post-deployment, extinction training (repeated odor presentation without shock) or counterconditioning (pairing the previously aversive odor with sucrose reward) is required (Filipi et al., 2022) [3] . PER-based conditioning using positive reinforcement (sucrose reward) is less disruptive to foraging behavior and is preferred for applications requiring bees to also fulfill their ecological pollination role. Quantitative assessments of foraging efficiency, colony weight gain, and brood development in conditioned versus control colonies remain an important area for future investigation, particularly for applications involving long-term or repeated deployment of the same colonies.

Future Prospects and Emerging Applications

The following domains represent priority areas for future development of sniffer bee technology:

Conclusion

Sniffer bee technology represents a compelling convergence of evolutionary biology, behavioral neuroscience, and engineering. The honeybee olfactory system characterized

References

1. Bozkurt A, Gilmour R, Stern D, Lal A. MEMS based bioelectronic neuromuscular interfaces for insect cyborg flight control. In: Proceedings of the 21st IEEE International Conference on Micro Electro Mechanical Systems; 2008 Jan 13-17; Tucson, AZ, USA. New York: IEEE; 2008. p. 160-163.
2. Delaney L, Marine Corps Command and Staff College. Military applications of apiculture: The (other) nature of war. In: Bees and Battles, Loudoun Beekeepers Association Newsletter. 2011;2(6).
3. Filipi J, Stojnić V, Muštra M, Gillanders RN, Jovanović V, Gajić S, et al. Honeybee-based biohybrid system for landmine detection. Sci Total Environ. 2022;803:150041.
4. Gaidos S. Sting operation: Scientists use bees and wasps to sniff out the illicit and the dangerous. Sci News. 2008;174(7):16-19.
5. Gillanders RN, Glackin JM, Filipi J, Kezic N, Samuel ID, Turnbull GA. Preconcentration techniques for trace explosive sensing. Sci Total Environ. 2019;658:650658.
6. Hadagali MD, Suan CL. Advancement of sensitive sniffer bee technology. Trends Anal Chem. 2017;97:153-158.
7. Kasban H, Zahran O, Elaraby SM, El-Kordy M, Abd El-Samie FE. A comparative study of landmine detection techniques. Sensing Imaging. 2010;11(2):89112.
8. Kerk WC, Chua LS. Sniffer bees as a good alternative for the current sniffing technology. Biointerface Res Appl Chem. 2016;6(4):1507-1513.
9. Kerk WC, Chua LS, Sarmidi MR, Aziz R. Sniffer bees as a reliable tool for Andrographis paniculata detection. Sensing Imaging. 2018;19:1-16.
10. Kontos E, Samimi A, Hakze-Van der Honing RW, Priem J, Avarguès-Weber A, Haverkamp A, et al. Bees can be trained to identify SARS-CoV-2 infected samples. Biol Open. 2022;11(4):bio059111.
11. Lockwood JA. Insects as weapons of war, terror, and torture. Annu Rev Entomol. 2012;57(1):205-227.
12. Parnas M, Cox E, Sanchez S, Farnum A, Lefevre N, Miller S, et al. Employing a honeybee olfactory neural circuit as a novel gas sensor for the detection of human lung cancer biomarkers. bioRxiv [Preprint]. 2023 [cited 2026 Jun 8]. Available from: https://doi.org/10.1101/2023.10.
13. Rehman I, Mahabadi N, Sanvictores T, Rehman CI. Classical conditioning. Treasure Island (FL): StatPearls Publishing; 2017.
14. Sandoz JC. Behavioral and neurophysiological study of olfactory perception and learning in honeybees. Front Syst Neurosci. 2011;5:98.
15. Schmidt HR, Benton R. Molecular mechanisms of olfactory detection in insects: Beyond receptors. Open Biol. 2020;10(10):200252.
16. Schott M, Klein B, Vilcinskas A. Detection of illicit drugs by trained honeybees (Apis mellifera). PLoS One. 2015;10(6):e0128528.
17. Siljak H, Nardelli PH, Moioli RC. Cyborg insects: Bug or a feature. IEEE Access. 2022;10:49398-49411.
18. Suckling DM, Sagar RL. Honeybees Apis mellifera can detect the scent of Mycobacterium tuberculosis. Tuberculosis (Edinb). 2011;91(4):327-328.