Abstract
Attractant-based trapping is governed by complex interactions among plume dynamics, insect movement, behavioral thresholds, lure aging, and spatial design. The field is bedeviled by inconsistent terminology and a relative paucity of rigorous mathematical approaches that have the potential to improve general understanding while advancing and facilitating pest management. In this Perspective, we review the current state of the art, with an emphasis on recent developments, and outline initiatives that in our view would benefit the study of insect trapping and its practical applications. Specifically, we propose to: (1) Standardize terminology that unifies concepts such as active space, plume reach, and D50 to enable direct comparison across studies and species. (2) Adopt mathematical frameworks that are aligned with biological reality and can be readily applied by non-specialists. (3) Require operational demonstration: a model is only useful if researchers, growers, and management programs can apply it. (4) Develop optimized experimental protocols and analytical methods and tools that efficiently generate the parameters each model requires. (5) Leverage emerging AI tools to integrate data streams, automate classification, and improve trap interpretation. Benefits of these proposals are exemplified by systems such as USDA’s Slow the Spread Program, which interprets data from ~100000 traps across ~186000 km² each year. A unified, quantitative, and practical framework will allow attractant-based trapping to reach its full potential for early detection and pest management.
1 Introduction
Over the past century, trapping insects using chemical attractants has become a cornerstone of biosecurity detection, delimitation, integrated pest management (IPM) decision support, ecological inference, and population estimation. Global trade, travel, and accelerating climate change have made trapping increasingly important for protecting crops, forests, public health, and biodiversity.
Despite this central importance, the interpretation of trap data remains inconsistent. Datasets developed for one purpose are often re-interpreted, with the same trap data used as evidence of insect presence, seasonal phenology, or absolute abundance. Each of these interpretations hinge on different assumptions about the relationships between trap catch and population size, and the biological processes generating trap catch are rarely scrutinized. Consequently, the meaning of trap catch data is frequently contested, sometimes hotly (e.g., ; ; ).
These challenges are both methodological and conceptual. An insect trap does not sample a population in the same way that a thermometer samples air temperature. Insect traps sample the endpoint of a behavioral process. Trap captures arise from a sequence of processes: insect movement, stimulus response, trap encounter, capture and retention (). Each is affected by environmental conditions, insect physiology, lure composition and trap design. Consequently, disagreements about trap performance characteristics such as “trap efficiency”, “detection probability” or “population density” may reflect differences in assumed measurement models.
We argue that research in this field lacks a shared theoretical framework for interpreting trap data. Agricultural decision support, biosecurity surveillance, and ecological monitoring have developed largely independently, producing parallel terminology and incompatible assumptions about what traps measure. This fragmentation limits data interoperability, prevents accumulation of evidence across programs, and constrains the development of reliable, integrated surveillance systems. Resolving these inconsistencies requires more than improved trap technology; it requires a unified conceptual framework linking trap catch to population processes.
Here we outline the major paradigms used to interpret trap data, identify how inconsistent terminology obscures methodological assumptions, and review a general detection process model. We then propose practical steps to enable trap networks to function as reliable measurement systems and improve interpretation of trap data.
2 Trapping paradigms
Insect traps are used in agricultural pest management, forest pest surveillance, biosecurity detection, ecological monitoring, and population estimation. Although trapping technology is often similar across applications, the questions asked of trap data differ substantially, giving rise to distinct conceptual paradigms.
2.1 Agricultural decision-support
Monitoring agricultural pests with traps aims to determine if the population exceeds a threshold requiring intervention and hinges on a relationship between trap catch and potential crop yield loss (). When yield loss is driven by pest abundance and timing of infestation, these relationships can be relatively robust. Since growers are risk-averse, the intervention thresholds are often set conservatively to avoid economic loss. Interpretation becomes more complex when the insect is a plant pathogen vector. In such cases, trapping shifts toward a biosecurity-type paradigm, where detection may trigger immediate action or intensified sampling.
2.2 Biosecurity detection
In biosecurity programs, the primary objective is early detection of a species incursion and, if detected, delimitation of its extent (). The approach is typically based on extensive trap grids (). A major limitation is the delay between capture and reporting. This delay can be reduced through automated real-time detection when a species-specific lure is used. When traps capture multiple species, sorting and identification become limiting steps. Emerging tools, including AI-assisted image analysis and environmental DNA (eDNA) may help alleviate these constraints. The principal challenge in this paradigm is the interpretation of zero captures.
2.3 Ecological monitoring
Ecological trapping typically involves sampling from an assemblage of species within an area to characterize diversity, relative abundance, and distribution (; ; ). When traps use general or non-specific attractants (e.g. light or tephritid lures such as Cuelure, methyl eugenol), the implicit assumption is that the relative abundance in traps reflects relative abundance in the environment. However, trap catches may instead reflect relative attractiveness of the lure which is rarely quantified. When traps act as interception devices (e.g. Malaise traps), some of these biases and interpretational issues may be reduced.
2.4 Population abundance
Population monitoring is typically species-specific and aims to track changes in abundance over time. For species of conservation concern, the question is whether the population is stable, increasing, or declining (). In eradication campaigns targeting invasive species, the focus shifts to determining whether the population is decreasing and ultimately whether zero catch can reasonably be interpreted as population absence (). This paradigm often seeks to estimate absolute population size, or at least to establish a consistent relationship between trap catch and population density, to evaluate temporal trends.
3 The fundamental problem
Insect trapping is often treated as a direct measurement of population size or activity. In reality, trap catch represents a complex, inherently stochastic detection process in which insects are captured only after a sequence of biological and physical events. Conceptually, trap catch can be viewed as the outcome of several interacting processes (; ; ; ; ; ; ; ):
Catch=Abundance×Movement×Motivation×Detection×Retention×Sampling geometry.
Each term represents a component of the detection process. Abundance refers to the number of insects present within the area from which a trap can potentially attract individuals. Movement determines whether insects encounter the odor plume. Motivation reflects the behavioral state-governing response to the attractant (e.g., mating status, physiological condition). Detection is the probability that an insect encountering the stimulus responds behaviorally and moves to the trap. Retention reflects the probability that an insect enters the trap and is successfully captured and retained. Finally, sampling geometry describes how trap placement, density, lure longevity, and servicing interval determine the portion of the population exposed to the trap.
Because trap catch depends on all these processes, the same trap catch can arise from many different underlying conditions. A low trap catch may reflect a small population, weak attraction, low insect activity, unfavorable weather, or trap interference. Conversely, a high trap catch may reflect strong attraction or favorable environmental conditions rather than a large population. Importantly, zero catch does not necessarily indicate the absence of insects but may instead reflect low detection probability or unfavorable conditions during the sampling period (; ; ).
These complexities help explain why different research communities have developed distinct interpretations of trap performance. What constitutes evidence of population presence, abundance, or management needs depends on which components of the detection process are assumed to dominate the observed signal. In this sense, disagreements about trap interpretation can reflect different “regimes of perceptibility” (), in which observers implicitly assume different measurement models linking trap catch to underlying population processes.
A unified framework for trapping requires explicit recognition that trap catch reflects a probabilistic detection process rather than a direct population measurement. Once this detection process is made explicit, the parameters describing trap performance – such as attraction distance, sampling area, or capture probability – can be defined and estimated experimentally.
4 Fragmented terminology
Since trap catch reflects a behavioral process, the parameters used to describe trap spatial influence should be explicitly linked to that process. However, the field lacks a shared vocabulary for describing how far and how effectively a trap attracts insects. Over decades of research, multiple terms have been introduced to describe trap efficiency, trap efficacy and spatial influence, often addressing similar biological phenomena but grounded in different conceptual and mathematical frameworks. As a result, biologically comparable studies are often expressed in incompatible terms, producing fragmented literature that hinders adoption by researchers and operational programs.
Much of the terminology used today originates from early works on odor plume structure and insect response (; ; ; ), which defined active space as the region in which odor concentration exceeds the threshold necessary to elicit a behavioral response. Although foundational, the concept of active space was not intended as a trapping metric. To provide a measurable basis for comparing attractiveness using trap data, and introduced the effective attraction radius (EAR), defined as the radius of a hypothetical passive trap that would capture the same number of insects as a baited trap. Although EAR is not a real biological distance, it correlates positively with attraction distance ().
introduced the effective sampling area (ESA), which integrates distance-dependent capture probability over space to enable inference about population density. Building on earlier work on plume structure and probabilistic views of trapping, formalized distance-dependent capture through the concept of (instantaneous) plume reach, defined as the distance to which behaviorally active pheromone filaments extend from a point source under field conditions. Attempts to apply plume reach as a distance parameter governing trap catch in spongy moth proved inadequate, motivating the development of a probability-based distance metric, D50, derived from first principles of capture probability and shown to apply across multiple species and attractant types (). D50 is defined as the distance at which capture probability declines to 50% of its near-source value, explicitly linking insect behavior, trap function, and distance.
Although active space, EAR, ESA, plume reach, and D50 all attempt to quantify the spatial consequences of distance-dependent capture, they arise from different assumptions and are not interchangeable (Table 1). The field would benefit from clear definitions, explicit articulation of equivalence and non-equivalence among metrics, and convergence toward a minimal set of parameters sufficient to describe attractant-based trap performance. Because EAR, ESA, and D50 are all based on the same underlying distance-dependent capture idea, quantitative relationships among them may exist, but further work is needed to determine when such relationships hold and how they depend on insect behavior, trap design, and environmental conditions. Establishing these connections would allow parameters measured in one framework to be translated into others, facilitating comparison across studies, improving integration into decision algorithms, and enabling historical datasets to be reinterpreted within a common framework. Clarifying terminology would also allow the parameters governing this detection process to be more clearly defined and more readily quantified.
Table 1
| Term | Definition | Established and possible relationship to other metrics |
|---|---|---|
| Active space (AS) | Region in which odor concentration exceeds the threshold required to elicit a behavioral response | With additional assumptions about response success within that region, it can be related to maximum, average, or 50% attraction distance. In simplified models, its overall effect can be represented by EAR (). |
| Effective attraction radius (EAR) | Radius of a hypothetical passive trap that would catch the same number of insects as an attractive trap, but not a real biological distance | May be positively related to maximum, mean, or 50% attraction distance (). In some systems it may also be monotonic with D50, though not identical to it. |
| Plume reach (PR) | Spatial extent of the behaviorally active plume under field conditions | Explicitly not equivalent to D50 (). May help set the effective outer extent over which the capture process operates. |
| Effective sampling area (ESA) | Area-based coefficient that converts trap catch to absolute population density; obtained by integrating distance-dependent capture probability over space. | Can be derived from the same distance-dependent capture function that D50 summarizes (). It likely depends on both the shape of decay of trap catch with distance and the effective outer sampling domain. Distinct from EAR, which is an equivalence metric rather than a catch-to-density integral. |
| D50 | Distance at which capture probability declines to 50% of its near-source value | At least in some species, it corresponds closely to observed behavioral or physiological response distance () and thus may relate to the effective extent of AS. May be related to ESA and EAR as an alternative summary of the same underlying capture process but not an equivalence metric. |
Comparison of conceptual metrics used in attractant-based insect trapping literature.
5 Operationalizing the trapping model
Applying a trapping model to an insect-trap system requires estimating model parameters. These parameters are not abstract statistical quantities; they represent biological or operational processes, including movement, encounter rates, attraction, and activity duration. Accordingly, every trapping model must have a corresponding experimental protocol to generate data for parameterization.
Mark-release-recapture experiments remain the primary method for quantitatively evaluating attractant performance (; ; ; ). Recapture rates are often highly variable, and reducing statistical uncertainty makes these experiments time-, labor-, and resource-intensive, underscoring the need for efficient protocols.
Several factors are critical in defining an effective experimental protocol. (1) Trap configuration is an integral component of protocol design and should minimize directional bias and interference while preserving distance-specific information. This can be achieved by arranging multiple traps symmetrically around a single release point to average over wind direction and plume variability. At very short distances, where trap interference becomes likely (), multiple release points arranged around a single trap can preserve the focal distance while reducing plume overlap. Because plume structure and insect movement can be influenced by landscape features such as vegetation structure, edges, and topography, the spatial context surrounding traps should be considered when interpreting results ().
(2) An initial release should be used to estimate converged catch: the time (days) until cumulative trap catch stabilizes (; ). Subsequent releases should be conducted for durations sufficient to reach converged catch.
(3) Replication across independent releases at each distance is necessary to account for variability, and release numbers should be scaled with distance to ensure measurable capture probabilities at larger distances.
Designing protocols that maximize parameter accuracy with limited resources (e.g., limited insect availability) presents a challenge. In models that estimate a characteristic attraction distance (e.g. D50), intuition suggests that releases concentrated near the true D50 would be most informative. However, because D50 is unknown a priori, this intuition cannot be applied directly without refinement. To address this limitation, we propose a protocol that includes release distances spanning the full response range, including 0 m (near the trap), a distance beyond which capture is negligible, and at least two intermediate distances to resolve the distance–response curve (https://arcg.is/0HDaC8). Further model-specific analysis may yield protocols that improve parameter estimation with the same or fewer releases.
Adoption of analytical methods increasingly depends on how easily they can be applied in practice. In many scientific fields, successful approaches are distributed not only as publications but also as software packages or decision-support tools that allow users to apply methods directly rather than having to derive them independently from underlying theory. For insect trapping, progress will depend not only on improved models and experiments, but also on frameworks that translate those models into user-friendly operational tools.
6 Machine learning and AI offer new opportunities
Many of the processes that determine whether an insect is captured in a trap are poorly observed or difficult to parameterize explicitly. Given this complexity, artificial intelligence (AI), and machine learning (ML) in particular, may offer significant potential as was recently demonstrated in other areas where difficult biological problems resisted effective solutions for decades (; ). However, ML approaches are not yet being utilized for interpreting trap data and supporting management decisions.
Most current efforts focus on automated capture of trap data, particularly computer vision methods used to detect, classify, and count insects from trap images (; ; ; ). These methods replace manual counting with scalable, high-frequency data streams. However, integration of these monitoring data streams into predictive models and operational decision algorithms remains limited.
At the same time, reliance on modern AI methods, particularly deep learning, introduces a critical and often underappreciated risk. Model accuracy is fundamentally constrained by the size and quality of the training data, and performance outside that domain may degrade sharply or fail catastrophically ().
A promising solution explored in other fields is hybrid modeling, in which a mechanistic, non-AI solver is combined with an AI component that learns the residual error between the combined model estimates and observations (). In this framework, a science-based model provides a stable, interpretable baseline, while AI corrects systematic biases where data are available. Such methods have been shown to outperform both standalone mechanistic and purely data-driven models, while retaining accuracy outside the AI training domain (; ). This approach can be especially beneficial when training data are limited, which is often the case in field biology.
We propose that this hybrid strategy is particularly well suited to insect trapping and trap-based decision algorithms. For example, the decision algorithm used in the USDA’s National Slow the Spread (STS) Program of the spongy moth provides a well-developed example of how trap catches may be translated into management actions using a biologically motivated, rule-based framework that has proven effective at large spatial and temporal scales (; ; ). However, such frameworks necessarily omit important sources of interannual variability, including weather-mediated pathogen dynamics. For example, wet spring conditions can promote epizootics of Entomophaga maimaiga, leading to elevated larval mortality and subsequent decline in adult populations (; ). These biologically driven fluctuations may substantially alter trap catch and the perceived need for management interventions. Rather than replacing biologically grounded frameworks, we suggest augmenting them with AI components that learn deviations between model estimates and observed trap outcomes, thereby preserving interpretability while improving model accuracy.
7 Roadmap for improvement
7.1 Short-term
Immediate progress can be made through clarification and standardization. Firstly, the field would benefit from explicit definitions of commonly used metrics (e.g., active space, EAR, ESA, D50) and clear articulation of the assumptions underlying each metric. Publishing crosswalk analyses that distinguish equivalence from non-equivalence among these metrics would reduce ambiguity and improve comparability across studies. Secondly, reporting standards for trap-based experiments should be strengthened to ensure consistent documentation of lure characteristics, trap design, servicing intervals, environmental conditions, and sampling geometry. Finally, experimental protocols should be explicitly aligned with the parameters required by detection-process models, ensuring that studies are designed to estimate biologically meaningful quantities rather than relative indices.
7.2 Medium-term
Beyond clarification, coordinated research efforts are needed to validate and generalize detection-process models across taxa, landscapes, and environmental regimes. Multi-site studies that quantify distance-dependent capture, movement processes, and environmental modulation would allow model structures to be tested under diverse conditions. In parallel, hybrid modelling methods that integrate mechanistic frameworks with data-driven components should be developed and rigorously evaluated. Such efforts would allow structured environmental variability (e.g., weather-mediated pathogen dynamics) to be incorporated without sacrificing interpretability. Machine learning components could be used to learn systematic deviations between mechanistic estimates and observed trap outcomes, improving system performance while retaining biological interpretability. Reanalysis of historical trapping datasets within unified measurement frameworks could further accelerate knowledge integration.
7.3 Long-term
Sustained progress will ultimately require institutional alignment. Surveillance networks should evolve toward interoperable measurement systems in which parameters, models, and decision rules are transparently documented and transferable across programs. The STS program provides a practical example of a largely integrated system that combines extensive publicly accessible trapping datasets, a formalized decision algorithm linking trap catch to management actions, and a suite of web-based tools supporting operational decision making (https://www.slowthespread.org/). Together, these elements illustrate how trapping networks can evolve from simple monitoring systems into integrated measurement and decision-support platforms. Regulatory guidelines and operational frameworks should adopt standardized detection-process terminology and calibrated metrics, enabling consistent interpretation of trap data across agricultural, biosecurity, and ecological applications. Over time, this shift would allow trapping systems to move from heuristic relative indices toward explicitly parameterized measurement tools that support robust, data-informed decision-making.
8 Conclusions
Attractant-based insect trapping is an important tool for ecological research, invasive species surveillance, and pest management, yet the conceptual and analytical frameworks used to interpret trap data remain fragmented, limiting progress. We have highlighted the critical issues and possible directions for moving forward. Trap catch should be understood not as a direct measure of population size but as the outcome of a probabilistic detection process shaped by insect behavior, environmental conditions, and sampling geometry.
Traps themselves are not flawed instruments; the problem lies in how the data they capture are interpreted. By treating traps as sensors of a complex detection process rather than simple population counters, the field can move toward more consistent interpretation, stronger quantitative models, and more reliable decision support. This reframing provides a foundation for a unified framework enabling ecological research, invasive species surveillance, and pest management to share analytical approaches, models, and datasets rather than repeatedly reinventing methods. Without this intellectual shift, inconsistencies in interpretation will persist; with it, trapping can function as a reliable and transferable measurement system across applications.
Statements
Author contributions
KO: Investigation, Project administration, Funding acquisition, Writing – original draft, Writing – review & editing. DK: Writing – review & editing. MZ: Writing – review & editing. AO: Conceptualization, Writing – original draft, Writing – review & editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. APC support was provided by the Slow The Spread Foundation, Inc through a subaward under a USDA Forest Service federal grant (Grant No. 24-DG-11083150-103 Mod. 4).
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The author AO declared that they were an editorial board member of Frontiers at the time of submission. This had no impact on the peer review process and the final decision.
Generative AI statement
The author(s) declared that generative AI was used in the creation of this manuscript. The authors declare that generative AI was used in the creation of this manuscript for organizing and articulating ideas. All intellectual contributions and conceptual developments are entirely those of the authors, who remain solely responsible for the accuracy and integrity of the final content.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
AbramsonJ.AdlerJ.DungerJ.EvansR.GreenT.PritzelA.et al. (2024). Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature630, 493–500. doi: 10.1038/s41586-024-07487-w
2
AdamsC. G.SchenkerJ. H.McgheeP. S.GutL. J.BrunnerJ. F.MillerJ. R. (2017). Maximizing information yield from pheromone-baited monitoring traps: Estimating plume reach, trapping radius, and absolute density of Cydia pomonella (Lepidoptera: Tortricidae) in Michigan apple. J. Econ. Entomol.110, 305–318. doi: 10.1093/jee/tow258
3
AhmedS.MarwatS. N. K.BrahimG. B.KhanW. U.KhanS.Al-FuqahaA.et al. (2024). IoT based intelligent pest management system for precision agriculture. Sci. Rep.14, 31917. doi: 10.1038/s41598-024-83012-3
4
AndreadisT. G.WeselohR. M. (1990). Discovery of Entomophaga-Maimaiga in North-American gypsy-moth, lymantria-dispar. PNAS87, 2461–2465. doi: 10.1073/pnas.87.7.2461
5
BakerT.RoelofsW. (1981). Initiation and termination of oriental fruit moth male response to pheromone concentrations in the field. Environ. Entomol.10, 211–218. doi: 10.1093/ee/10.2.211
6
BassL.ElderL. H.FolescuD. E.ForouzeshN.TolokhI. S.KarpatneA.et al. (2024). Improving the accuracy of physics-based hydration-free energy predictions by machine learning the remaining error relative to the experiment. J. Chem. Theory Comput.20, 396–410. doi: 10.1021/acs.jctc.3c00981
7
BerecL.KeanJ. M.Epanchin-NiellR.LiebholdA. M.HaightR. G. (2015). Designing efficient surveys: Spatial arrangement of sample points for detection of invasive species. Biol. Invasions17, 445–459. doi: 10.1007/s10530-014-0742-x
8
ByersJ. (2008). Active Space of Pheromone Plume and its Relationship to Effective Attraction Radius in Applied Models. J. Chem. Ecol.34, 1134–1145. doi: 10.1007/s10886-008-9509-0
9
ByersJ. A. (2009). Modeling distributions of flying insects: Effective attraction radius of pheromone in two and three dimensions. J. Theor. Biol.256, 81–89. doi: 10.1016/j.jtbi.2008.09.002
10
ByersJ.AnderbrantO.LöqvistJ. (1989). Effective attraction radius. J. Chem. Ecol.15, 749–765. doi: 10.1007/BF01014716
11
CardéR. T.ElkintonJ. (1984). “ Field trapping with attractants: Methods and interpretation,” in Techniques in pheromone research (New York, NY: Springer). doi: 10.1007/978-1-4612-5220-7_4
12
CareyJ. R.PapadopoulosN.PlantR. (2017). The 30‐year debate on a multi‐billion‐dollar threat: Tephritid fruit fly establishment in California. Am. Entomol.63, 100–113. doi: 10.1093/ae/tmx036
13
CarnioV.PretiM.FavaroR.LeciniO.GiannottaG.AngeliS. (2026). Design, training, and field validation of a YOLOv8-based automated trap for remote detection of codling moth, Cydia pomonella. Entomol. Exp. Appl.174, 212–224. doi: 10.1111/eea.70050
14
ColemanT. W.LiebholdA. M. (2023). Slow the Spread: a 20-year reflection of the national Lymantria dispar integrated pest management program (Madison, WI: U.S. Department of Agriculture, Forest Service, Northern Research Station). doi: 10.2737/NRS-GTR-212
15
DrewR.ZaluckiM. P.HooperG. H. S. (1984). The ecology of Australian Tephritidae in their endemic habitat (1) Temporal variation in abundance. Oecologia64, 267–272. doi: 10.1007/BF00376881
16
ElderL. H.OnufrievA. V. (2025). Using deep graph neural networks improves physics-based hydration free energy predictions even for molecules outside of the training set distribution. J. Phys. Chem. B129, 7483–7498. doi: 10.1021/acs.jpcb.5c02263
17
ElkintonJ. S.CardeR. T. (1988). Effects of intertrap distance and wind direction on the interaction of gypsy-moth (Lepidoptera, Lymantriidae) pheromone-baited traps. Environ. Entomol.17, 764–769. doi: 10.1093/ee/17.5.764
18
HajekA. E.ButlerL.WalshS. R. A.SilverJ. C.HainF. P.HastingsF. L.et al. (1996). Host range of the gypsy moth (Lepidoptera: Lymantriidae) pathogen Entomophaga maimaiga (Zygomycetes: Entomophthorales) in the field versus laboratory. Environ. Entomol.25, 709–721. doi: 10.1093/ee/25.4.709
19
HongS.-J.KimS.-Y.KimE.LeeC.-H.LeeJ.-S.LeeD.-S.et al. (2020). Moth detection from pheromone trap images using deep learning object detectors. Agriculture10, 170. doi: 10.3390/agriculture10050170
20
JumperJ.EvansR.PritzelA.GreenT.FigurnovM.RonnebergerO.et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature596, 583–589. doi: 10.1038/s41586-021-03819-2
21
KeanJ. M.SucklingD. M. (2005). Estimating the probability of eradication of painted apple moth from Auckland. NZ Plant Prot.58 (0), 7–11. doi: 10.30843/nzpp.2005.58.4246
22
LarssonM. C. (2016). Pheromones and other semiochemicals for monitoring rare and endangered species. J. Chem. Ecol.42, 853–868. doi: 10.1007/s10886-016-0753-4
23
LiebholdA. M.TobinP. C. (2008). Population ecology of insect invasions and their management. Annu. Rev. Entomol.53, 387–408. doi: 10.1146/annurev.ento.52.110405.091401
24
MacKenzieD. I.NicholsJ. D.RoyleJ. A.PollockK. H.BaileyL.HinesJ. E. (2017). Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence (Netherlands, Academic Presss: Elsevier).
25
MillerJ. R.AdamsC. G.WestonP. A.SchenkerJ. H. (2015). Trapping of small organisms moving randomly: principles and applications to pest monitoring and management (Germany: Springer International).
26
MurlisJ.ElkintonJ. S.CardeR. T. (1992). Odor plumes and how insects use them. Annu. Rev. Entomol.37, 505–532. doi: 10.1146/annurev.ento.37.1.505
27
NakamuraK.KawasakiK. (1977). The active space of the Spodoptera litura (F.) sex pheromone and the pheromone component determining this space. Appl. Entomol. Zool.12, 162–177. doi: 10.1303/aez.12.162
28
OnufrievaK. S.OnufrievA. V. (2021). How to count bugs: A method to estimate the most probable absolute population density and its statistical bounds from a single trap catch. Insects12, 932. doi: 10.3390/insects12100932
29
OnufrievaK. S.OnufrievA. V.HickmanA. D.MillerJ. R. (2020). Bounds on absolute gypsy moth (Lymantria dispar dispar)(Lepidoptera: Erebidae) population density as derived from counts in single milk carton traps. Insects11, 673. doi: 10.3390/insects11100673
30
ÖstrandF.AnderbrantO.JönssonP. (2000). Behaviour of male pine sawflies, Neodiprion sertifer, released downwind from pheromone sources. Entomol. Exp. Appl.95, 119–128. doi: 10.1046/j.1570-7458.2000.00649.x
31
RamakrishnanR.DralP. O.RuppM.Von LilienfeldO. A. (2015). Big data meets quantum chemistry approximations: The Δ-machine learning approach. J. Chem. Theory Comput.11, 2087–2096. doi: 10.1021/acs.jctc.5b00099
32
RizviS. A. H.GeorgeJ.ReddyG. V. P.ZengX.GuerreroA. (2021). Latest developments in insect sex pheromone research and its application in agricultural pest management. Insects12, 484. doi: 10.3390/insects12060484
33
RobinetC.LanceD. R.ThorpeK. W.OnufrievaK. S.TobinP. C.LiebholdA. M. (2008). Dispersion in time and space affect mating success and Allee effects in invading gypsy moth populations. J. Anim. Ecol.77, 966–973. doi: 10.1111/j.1365-2656.2008.01417.x
34
RoquesA.RenL.RassatiD.ShiJ.AkulovE.AudsleyN.et al. (2023). Worldwide tests of generic attractants, a promising tool for early detection of non-native cerambycid species. NeoBiota84, 169–209. doi: 10.3897/neobiota.84.91096
35
SedellJ. K. (2021). No fly zone? Spatializing regimes of perceptibility, uncertainty, and the ontological fight over quarantine pests in California. Geoforum123, 162–172. doi: 10.1016/j.geoforum.2019.04.008
36
SharovA. A.LeonardD.LiebholdA. M.RobertsE. A.DickersonW. (2002). Slow the Spread": A national program to contain the gypsy moth. J. For.100, 30–35. doi: 10.1093/jof/100.5.30
37
ShellyT. E.LanceD. R.TanK. H.SucklingD. M.BloemK.EnkerlinW.et al. (2017). To repeat: Can polyphagous invasive tephritid pest populations remain undetected for years under favorable climatic and host conditions? Am. Entomologist (Lanham Md.)63, 224–231. doi: 10.1093/ae/tmx075
38
ThompsonL. M.GraysonK. L.JohnsonD. M. (2016). Forest edges enhance mate‐finding in the invasive European gypsy moth, Lymantria dispar. Entomol. Exp. Appl.158, 295–303. doi: 10.1111/eea.12402
39
TobinP. C.BlackburnL. M. (2007). “ Slow the spread: a national program to manage the gypsy moth,” in General Technical Report NRS-6 ( USDA Forest Service, Newtown Square, PA). doi: 10.2737/NRS-GTR-6
40
TobinP. C.ZhangA.OnufrievaK.LeonardD. S. (2011). Field evaluation of effect of temperature on release of disparlure from a pheromone-baited trapping system used to monitor gypsy moth (Lepidoptera: Lymantriidae). J. Econ. Entomol.104, 1265–1271. doi: 10.1603/ec11063
41
TurchinP.OdendaalF. J. (1996). Measuring the effective sampling area of a pheromone trap for monitoring population density of southern pine beetle (Coleoptera: Scolytidae). Environ. Entomol.25, 582–588. doi: 10.1093/ee/25.3.582
42
WorkT. T.BuddleC. M.KorinusL. M.SpenceJ. R. (2002). Pitfall trap size and capture of three taxa of litter-dwelling arthropods: Implications for biodiversity studies. Environ. Entomol.31, 438–448. doi: 10.1603/0046-225x-31.3.438
43
ZaluckiM. P.DrewR.HooperG. H. S. (1984). The ecology of Australian Tephritidae in their endemic habitat (2) Spatial variation in abundance. Oecologia64, 273–280. doi: 10.1007/BF00376882
44
ZarboubiM.BelloutA.ChabaaS.DliouA. (2026). Enhancing integrated pest management with IoT and YOLO-Evo: A smart, low-cost monitoring system for sustainable apple farming. Results Eng.29, 108850. doi: 10.1016/j.rineng.2025.108850
Summary
Keywords
attractant-based insect trapping, detection probability, modelling, pheromone, population density estimation, sampling framework, trap catch
Citation
Onufrieva KS, Kriticos DJ, Zalucki MP and Onufriev AV (2026) A perspective on chemical attractant-based insect trapping: from disparate paradigms to an interoperable scientific toolbox. Front. Ecol. Evol. 14:1839762. doi: 10.3389/fevo.2026.1839762
Received
26 March 2026
Revised
25 April 2026
Accepted
04 May 2026
Published
10 June 2026
Volume
14 - 2026
Edited by
Sergei Petrovskii, University of Leicester, United Kingdom
Reviewed by
Rosalvo Oliveira Neto, Federal University of São Francisco Valley, Brazil
Updates
Copyright
© 2026 Onufrieva, Kriticos, Zalucki and Onufriev.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Ksenia S. Onufrieva, ksenia@vt.edu
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.