The study, led by researchers from Panthera Malaysia, Rimba, LEAD Conservation, Vrije Universiteit Amsterdam, the Zoological Society of London and Panthera Wildlife Trust, addresses a persistent weakness in ranger-based wildlife protection. The authors note that counter-poaching patrols are a primary tool for protecting wildlife, yet detection rates of active poaching incursions are typically low, and patrol planning frameworks that have been both operationalised and performance-assessed remain rare. While ranger data collection systems such as SMART and EarthRanger provide clarity on where patrols went and what they found, the authors argue that advancing from this record to planning where patrols should go next has lagged behind. Over-dependence on ranger experience alone, they write, has left gaps and has not delivered intended conservation outcomes. The stakes are high: more than 150 rangers lose their lives each year, making well-planned patrols a matter of safety as well as effectiveness.

The framework draws on two established analytical traditions. The first is search theory, a discipline first developed during the Second World War to improve the efficiency of locating enemy vessels and missing allied ships at sea, and later applied to search and rescue operations for missing persons. The parallels to counter-poaching are direct: both involve optimising limited team resources while balancing thoroughness of search against speed in locating the quarry. The second is actor-based crime scripting, an approach from criminology that shifts attention from where crimes have occurred to where and how they are enabled, identifying the micro-locations — access points, movement corridors, rest areas — critical at each stage of an offence. The authors also adapted the military F3EAD cycle — Find, Fix, Finish, Exploit, Analyse, Disseminate — which structures the transition from detecting a quarry’s general presence through pinpointing its precise location to interdiction and learning from each operation. Almost all published counter-poaching patrol planning frameworks, the authors note, end at the point of initial detection.

The first component of the new framework is the Operational Patrol Space, or OPS: a connected network of prioritised patrol routes derived from how poachers actually use the terrain. Building the network began with a poaching risk map produced using the species distribution modelling tool Maxent, run separately for three categories of human activity observations — camps, snares and other human signs such as litter, parang markings on trees, footprints and trail cuts — against five environmental variables chosen to reflect the factors influencing a poacher’s path and camp placement: elevation, slope, distance to lake, distance to ridge and distance to river. The model performances, measured by Area Under the Curve scores, were 0.71 plus or minus 0.05 for camps, 0.89 plus or minus 0.08 for snares, and 0.77 plus or minus 0.03 for other human signs. The authors caution that the camp model’s modest score reflects the limited number of confirmed camp locations and an overfitting risk, and that the combined output should be interpreted as a structured synthesis of prior spatial knowledge rather than a validated predictive surface.

That map was then fused with ranger knowledge in participatory consultation sessions involving the protection lead, chief operations officer and analyst. Working over a large topographical map with a transparent overlay, alongside the risk map and plotted patrol GPS tracks, the analyst traced lines over key landscape features corresponding to important poacher space use, while experienced rangers ensured the lines were logical for patrols lasting five to eight days. The resulting network was deliberately selective, prioritising ridgelines that function as poacher access highways into the reserve, ridgelines with documented repeat victimisation across multiple years, and ridgelines with a high tiger-value profile and strong snaring potential. Hill flanks between ridgelines were excluded on ecological grounds: wildlife movement concentrates on ridgelines as least-cost travel paths in montane terrain, concentrating snaring risk there. By condensing the patrollable area from 1,200 square kilometres to 334 square kilometres of high-value ridgelines and movement corridors, the framework made the task achievable and purposeful.

The network’s geometry was grounded in measured detection capabilities. The authors estimated, from operational experience, how far rangers on foot can reliably detect different sign types in mountainous, dense tropical rainforest, and set a pragmatic Effective Search Width of 400 metres — a weighted composite across sign types — used to buffer the search polylines, which were then cut into segments two kilometres long. Ground sign such as footprints and discards was reliably detectable only within two to three metres of a ranger, while navigational markers and parang cuts on vegetation could be visible up to 20 metres. Camouflaged snares were almost undetectable off the main trail beyond two metres, though they were almost exclusively set on ridge crests where topography created natural pinch points. Sounds of tree felling and agarwood chopping could carry up to 500 metres from a ridgeline, and the smell of cooking or a campfire up to 200 metres. Detection distances were markedly reduced in denser valley vegetation compared to open ridgelines.

Post-apprehension interviews with offenders — seven interviews conducted by Malaysia’s Department of Wildlife and National Parks as part of standard post-arrest procedure — and ranger observations were formalised into crime scripts revealing a consistent activity pattern. A typical four-man Vietnamese or Thai snaring and agarwood poaching team spent little time traversing the main ridgelines, yet those routes concentrated the most detectable sign, making ridgelines the most reliable terrain for detecting an active incursion despite low poacher dwell time there. During the snare-setting phase, all four team members spent daylight hours establishing lines; during snare maintenance, one or two checked and reset snares each morning while the others searched hill flanks for agarwood. By late afternoon all returned to a basecamp beside a stream in a gulley. Over a typical two-week incursion, approximately 65 percent of poacher time was spent at camp, 24 percent on the flanks and 10 percent at the snare line. The camp was therefore the highest-value target for apprehension, and after-action reviews showed the apprehension rate was highest when the team was at camp and the apprehension occurred before dawn.

The second component, vigilance, tracks time elapsed since each OPS segment was last patrolled. Awareness is high when a patrol enters an area and decays linearly with each passing day until the next patrol restores it. Four thresholds were set to align with the multi-month incursion durations of the three priority poaching groups — Vietnamese, Cambodian and Thai teams conducting multi-month incursions targeting agarwood and carnivores with arrays of steel snares: High for under three months since last patrol, Medium for three to six months, Low for six to nine months, and Very Low for over nine months. For a segment to count as patrolled, a team needed to cover more than 500 metres — 25 percent — inside it. Sectors were classed as adequately patrolled when the proportion of High-vigilance segments exceeded 30 percent, an adjustable threshold balancing coverage ambition against the four patrol teams available. High-priority sectors were patrolled at least monthly, Medium every two months and Low every three months, with route selection varied to avoid repetitive patterns that poachers could adapt to.

The results, drawn from operations between 2018 and 2025 and assessed with Fisher’s exact tests, show detection rate rising from 15.6 percent (95 percent confidence interval 6.9 to 31.8 percent) to 50.0 percent (95 percent confidence interval 33.6 to 66.4 percent), a statistically significant improvement with p equal to 0.007. Camp-fixing success rose from 20.0 percent to 56.2 percent, an improvement that did not reach statistical significance (p equal to 0.311), which the authors attribute to the small number of detected incursions in the pre-implementation period — five — rather than an absence of effect. Snare outcomes also improved: before the framework, 89 percent of the 148 snares detected had already caught wildlife or weathered beyond use; after implementation, missed snares dropped to 59 percent of 714 detected, and 41 percent were disarmed or seized before being deployed. Sixty-one percent of snares removed lay within OPS; the remaining 39 percent were found by patrols tracking off-OPS as per protocol, and after two revisions of the network this off-OPS proportion fell to 24 percent.

The authors also report effects on ranger morale: shrinking the patrollable area and the traffic-light vigilance display made the task more achievable and communicated priorities clearly, with rangers reporting that precise spatial targets gave patrols a concrete sense of purpose that open-ended area searches did not. The authors caution that other factors, including ranger tracking training and increased poaching alerts from other sources, contributed to the performance enhancement, and that the framework was developed for a specific threat profile of multi-month, camp-based incursions by foreign syndicates. It has since been adopted at two additional Malaysian protected areas using the same parameters and is under trial in an African savannah landscape with a different threat type, local bushmeat poaching, though formal comparative evaluation across sites remains a priority for future work. The authors present the framework as modular and replicable, adaptable to other ecosystems by redefining patrol geometry around the spatial constraints and detection processes of the focal threat.

Subject of Research: A structured search framework for counter-poaching patrol planning that improved detection of active poaching incursions in a Malaysian tiger landscape

Article Title: Catching needles in haystacks: a structured search framework to optimise counter-poaching patrols

Article References: Lam, W. Y., Phung, C.-C., Lemieux, A. M., Clements, G. R., Booton, M. D., & Pickles, R. S. A. (2026). Catching needles in haystacks: a structured search framework to optimise counter-poaching patrols. Frontiers in Conservation Science, 7, 1942233. https://doi.org/10.3389/fcosc.2026.1942233

Image Credits: AI Generated

DOI: 10.3389/fcosc.2026.1942233

Keywords: counter-poaching, patrol planning, search theory, vigilance, Operational Patrol Space, Malayan tiger, Kenyir, rangers, SMART, crime scripting, protected area management, poaching detection