Scent of a poacher: operational performance of a counter-poaching tracking dog in a Southeast Asian tropical forest

1 Introduction

Poaching remains one of the most pervasive threats to wildlife across Southeast Asia, with snaring and firearm hunting causing extensive mortality among large mammals and contributing to regional biodiversity loss (Benítez-López et al., 2017; Gray et al., 2017). In Peninsular Malaysia, poaching is conducted by both local and transnational teams operating within protected forests, snaring or shooting targeting species such as the critically endangered Malayan tiger (Panthera tigris jacksoni) and its prey (Clements et al., 2010; Kawanishi et al., 2014; Ten et al., 2021). Detecting and intercepting these incursions is challenging due to rugged terrain, dense vegetation, and the limited range and frequency of standard ranger patrols. Across Southeast Asia, enforcement strategies rely heavily on ranger patrols to locate poachers and remove snares. However, detection rates for concealed snares can be low, and patrol based approaches require substantial manpower to achieve sustained reductions in poaching pressure (O’Kelly et al., 2018; Tilker et al., 2024). Strategies prioritising poacher apprehension rather than snare removal alone have shown higher operational impact but depend critically on the ability to track human movement through forest environments (Lam et al., 2023).

Domestic dogs (Canis lupus familiaris) offer a biologically and operationally compelling complement to human visual tracking. With approximately 125–300 million olfactory receptors compared to only about five million in humans (Kowalczyk-Jabło et al., 2026), trained dogs can detect and follow aged human scent trails at concentrations far below the threshold of human perception (Craven et al., 2010). In search and rescue controlled field trials, detection dogs have located targets at significantly faster than human searchers (Greatbatch et al., 2015), particularly in conditions of dense vegetation and small target size.

In counter-wildlife crime contexts, dogs are used to locate wildlife products, carcasses, or snares, and support to law enforcement operations through tracking poachers. These applications have been implemented in Africa, South Asia and parts of Latin America by government agencies and non−governmental organisations (Earthrace Conservation, n.d; TRAFFIC, n.d; Kuiper, 2023; Narayanasamy et al., 2023; Kowalczyk-Jabło et al., 2026). In Kruger National Park, South Africa, canine units have supported operations resulting in 686 arrests between 2017-2021 (Kuiper, 2023). However, despite their growing use, the published evidence base evaluating counter−poaching dogs remains limited. Most available information is derived from case studies, programme reports, or grey literature, with relatively few peer−reviewed studies quantifying performance under operational conditions. This presents a critical gap: while there is clear experiential knowledge suggesting dogs can enhance detection and follow−up capability, there is limited empirical data (Beebe et al., 2016; Grimm-Seyfarth et al., 2021; McKeague et al., 2024) describing when, how, and under what environmental and operational conditions dogs deliver reliable outcomes.

Since 2014, Panthera, a conservation organisation dedicated to the protection of wild cat species, has supported Malayan tiger counter-poaching efforts in Terengganu, Malaysia, initially by providing financial and technical support to Rimba, a Malaysian conservation organisation, from 2014 to 2020, and subsequently assuming direct management of the programme from 2021 onwards. This project supports the Department of Wildlife and National Parks (PERHILITAN) and partners in the Kenyir landscape through the establishment of counter-poaching teams that conduct routine deep-forest patrols and include rangers trained in visual tracking to detect subtle signs of human movement in dense forest environments. Following the detection of fresh signs of intrusion by patrol teams, visual trackers are deployed to efficiently and covertly pursue and locate suspects without alerting them or triggering their escape, thereby facilitating apprehension by law enforcement personnel.

In 2019, Rimba established a canine unit to support counter-poaching tracking efforts in the Kenyir landscape, which was subsequently absorbed into Panthera’s counter-poaching programme in 2021. To the authors’ knowledge, this was the first dog specifically trained and utilized for human scent tracking in an operational forest counter-poaching context in Southeast Asia. The dog was trained to follow human scent and locate targets using both ground-scent and air-scent tracking methods, and to provide a passive proximity alert when close to a target, thereby enhancing the covert nature of tracking operations. The primary objective of the canine unit was to assist visual tracking teams in locating poachers and their associated camps, reducing the time and effort required to follow human tracks, and increasing the likelihood of successfully locating and apprehending targets.

To inform the effective deployment of tracking dogs in counter-poaching operations of this nature, a pilot study was initiated to evaluate performance of the canine unit through three complementary components: (1) controlled tracking trials assessing the dog’s ability to track fresh scent covertly, (2) systematic tracking trials across a range of track ages and environmental conditions, and (3) structured assessment of live reconnaissance and counter poaching deployments. Together, these components provide the first operational performance benchmark for a counter poaching tracking dog in a Southeast Asian tropical forest.

2 Methods

2.1 Study site

The study was conducted in Kenyir State Park, Terengganu, Peninsular Malaysia (300km²), forming a forest corridor linking protected areas across Terengganu, Kelantan, and Pahang. The park experiences a humid equatorial climate with annual rainfall of 2,500–3,500 mm and relative humidity typically 80–90%. Mean temperatures range from approximately 22–33 °C. Vegetation is dominated by secondary and logged over forest, with remaining lowland to hill mixed dipterocarp forest in interior areas. Elevation ranges from approximately 60–600 m above sea level, and former logging roads provide access routes commonly used by poachers.

2.2 The canine unit

The canine unit was established in 2019. The tracking dog, Phiri, is an intact male Belgian Malinois, a breed selected for its proven performance in military, law enforcement, and conservation operations worldwide, and its documented suitability for demanding field deployments including high working drive, trainability, endurance, and agility. Phiri was acquired from South Africa at two years of age through an anti-poaching canine solutions provider, selected specifically for his endurance and drive for long-distance patrols, ability to remain covert in the field, and a calm temperament suitable for integration within a culturally sensitive ranger team. He was trained in South Africa for three months to track human targets using ground-scenting (following the scent trail and physical disturbances left by a person moving through the environment, with the dog’s nose held close to the ground) and air-scenting (following airborne scent particles carried by air currents, typically with the dog’s nose held up in the air) techniques prior to his arrival in Malaysia, and underwent a two-week acclimatisation period before commencing training and field activities in Malaysia. Phiri was fully vaccinated against canine distemper virus, rabies and other core diseases, and received routine veterinary care throughout the study.

The canine unit comprised of two handlers responsible for the dog’s care, training, and operational deployment. The handlers did not have prior canine handling experience before the establishment of the project and underwent both in-person and remote learning throughout the course of the project. Annual handler and canine training was conducted by a professional canine trainer from the Netherlands to refresh and enhance operational capabilities while introducing additional skills beneficial to the project, including snare detection and passive proximity alerting. Positive reinforcement and marker-based training methods were used throughout the study period, whereby desired behaviours were immediately marked using a consistent verbal cue (“good boy”) from the handler and reinforced with a rubber toy reward to strengthen the association between the behaviour and its positive outcome. General maintenance training was conducted three to four times per week and consisted of structured hide-and-seek exercises in forest environments, with a target team laying tracks ranging from 200 m to 4 km in length and aged between 30 minutes and five hours.

The unit’s development occurred across three phases. An initial training phase (2019–2020) focused on foundational skill development for the dog and handlers (authors A.S. and L.M.), environmental conditioning incorporating eco-socialisation to common forest stimuli, and motivation-building exercises. A pause occurred in 2021–2022 due to COVID-19 restrictions, during which field deployments were substantially reduced, handlers from Phase 1 departed, and a ranger assumed temporary handler responsibilities. Maintenance training was reduced to a minimal level. Two new handlers (authors F.D.G. and B.R.) formally joined in 2023, and from that year onwards, training and operationalisation resumed with a focus on systematic efficacy trials alongside live counter-poaching deployments.

As the canine unit was established as a pilot initiative and little was known about its operational capabilities, limitations, or how it could functionally be integrated into a project of this nature, the canine and patrol units operated independently throughout much of the project. This arrangement allowed patrol teams to continue routine forest patrol activities while the canine unit maintained a regular training schedule to preserve operational readiness for deployment when fresh signs were encountered by patrol teams, while also conducting trials to better understand and evaluate its capabilities for future operational use.

Where possible, the canine unit was included in patrols, as developing long-distance endurance and operational stamina was considered an important component of maintaining field capability.

2.3 Study 1: assessing covert-tracking of fresh tracks

From October 2019 to May 2021, 100 covert-tracking exercises were conducted to assess the dog’s ability to track to a simulated poacher position covertly. In each exercise, a target team of two to three rangers walked a route through forest (200 m–4 km), leaving typical poacher field sign (footprints, machete cuts) at intervals to allow handler recovery should the scent trail be lost. At the endpoint, the target team established a static lay-up position. Low-volume communication was permitted, while excessive noise and movement were prohibited.

After a predetermined track age ranging from approximately 25 minutes to 5 hours, the canine unit, comprising Phiri and one of two handlers, initiated the search. Each trial involved only a single handler working the dog, with handlers rotating across the 100 trials. The handler had no prior knowledge of the route. Both the target and tracking teams recorded their respective movements using GPS units. In addition, the tracking team documented any proximity alerts indicated by the dog, while the target team recorded the time at which they first detected the approaching tracking unit. By synchronising GPS tracks from both teams and comparing detection timestamps, proximity alert distance and the distance at which the tracking team alerted the target team could be determined.

Performance was assessed across five metrics: tracking success (proportion of exercises ending in a successful location of the target team); search independence (proportion of successful location achieved without handler assistance, where assistance consisted of visual tracking to relocate the scent trail within a 15-minute intervention window); covertness (proportion of successful location in which the target team did not detect the tracking team); proximity alert given (proportion of successful location in which Phiri exhibited a behavioural proximity alert); and proximity alert distance (distance in metres from the target team at the time of alert).

2.4 Study 2: assessing track-age and environmental effects on tracking performance

Systematic tracking trials were conducted from August to November 2024, spanning the late dry season and onset of the north-east monsoon. Five track-age categories were tested: 30 minutes, 8 hours, 24 hours, 3 days and 7 days, with up to 6 replicates per category (total n = 23). A blind study design was used: the handler had no prior knowledge of track age or exact route. The same handler worked with Phiri throughout.

Trials were conducted from six access points, with entry into the forest via former logging trails representative of operational patrol conditions. Within the forest, conditions varied across terrain including rivers, valleys and ridgelines. To minimise contamination from non-target human scent, access points with low visitor frequency were selected, and locations were shifted if fresh non-target human sign was detected during track-laying.

Two rangers acting as poachers walked a route of 500 m into the forest using a GPS unit to record the track. At the 500 m mark they established a mock camp (a worn towel and a pot on a stick stand), remained for five minutes to create a scent pool, then exited via a separate route at least 200 m from the inbound track to avoid contamination. An access point was not used again to lay tracks for subsequent trials for at least two weeks to prevent incidences of overlapping tracks. Environmental variables (temperature, relative humidity) were recorded at the start of each trial using a Kestrel 500 handheld weather station. Rainfall events between track-laying and deployment were recorded using bioacoustic recorders (Rugged SwiftOne) placed at each access point. The terrain of the tracks was also noted down (Figure 1).

The tracking unit deployed from the start point once the designated track age was reached with the dog working on a 6 m leash. The handler recorded behavioural indicators of scent acquisition (increased pace, leash tension, sniffing intensity, head-pop) and scent loss (tail-down posture, circling, backward gazing). If Phiri lost the track, the handler could intervene for up to 15 minutes of visual tracking, conducting a 360° search from the last known point, presenting visible field signs. A maximum of three track-loss events were permitted per trial before the trial was classified as unsuccessful. Time to locate the mock camp was recorded. Track-age success thresholds were determined by identifying the point at which success rate began to decline across increasing age categories. Given the small and unequal sample sizes per category, formal statistical comparison of success rates across track-age categories was not performed; results were interpreted as descriptive.

The effect of heat stress on tracking performance was examined using the heat index, which combines ambient air temperature and relative humidity to estimate apparent temperature and provides a more physiologically relevant measure of thermal load than ambient temperature alone. Statistical significance was assessed using Fisher’s Exact Test across three heat index bands (27–30 °C, 31–34 °C, 35–39 °C) defined during analysis to span the observed range; given the small sample sizes per band, this analysis should be considered exploratory. Only seven trials were associated with rainfall events, precluding formal analysis of rainfall effects.

2.5 Study 3: assessing operational performance in live deployments

Between 2020 and 2025, the canine unit participated in routine patrols and counter-poaching operations. Routine patrols involved multi-day deep forest deployments conducted under standard patrol conditions, with opportunistic tracking when poaching signs were encountered. Counter-poaching operations were purpose-driven responses to confirmed or suspected poaching incursions, where the canine unit was typically deployed 1–2 days after initial detection of sign, with the primary objective of locating and following human scent associated with poachers. Counter-poaching operations were divided into three stages; reconnaissance, main operation and post-operation. Reconnaissance involved scouting for poaching sign following an alert received from a community contact or real-time cameras, main operations involved following up on confirmed sign to locate the exact position of a poaching team to facilitate apprehension by a law enforcement team, and post-operations involved locating snares and additional poacher sign after a successful or failed operation.

Performance during deployments was assessed through handler observation and structured operational debriefs, examining: (1) the dog’s contribution to locating targets or navigating towards poacher sign; (2) coordination and complementarity between the canine and human tracking units; and (3) operational challenges encountered and the responses adopted.

2.6 Transferability

To enhance transferability of the programme, we break down the components and financial costs involved in setting up and running the dog unit using the TEPIDOIL-F framework (Kerr et al., 2008) and use this to identify a recommended minimally viable operational capability. TEPIDOIL-F is an acronym for ‘Training, Equipment, Personnel, Information, Doctrine, Organisation, Infrastructure, Logistics and Finance’, a framework developed by military planners that provides a systematic assessment of the organisational, logistical, and resource requirements needed to deliver and sustain a capability. While developed in a defence context, TEPIDOIL-F is directly applicable to conservation settings as it identifies the conditions required for the effective and sustained deployment of a specific capability within a broader counter-poaching system.

Current practices were documented using operational records, deployment observations, and training records. Costs incurred are reported and scaled to provide cost estimates for a minimally viable operational capability, with estimates for elements not present derived conservatively from local market rates as indicative planning values. Based on the strengths, limitations, and operational bottlenecks identified, a tiered capability model (Basic, Recommended Minimum, and Optimised) was developed to provide practical guidance for conservation programmes considering canine unit implementation.

3 Results

3.1 Study 1: assessing covert-tracking of poachers following fresh tracks

Across 100 covert−tracking exercises on tracks aged between 25 minutes and five hours, the dog successfully fixed the target in 97% of trials (Table 1). The dog worked independently without handler intervention in 71% of the successful exercises,; in the remaining cases, the handler intervened an average of two times (range: 0-4) to recover the track using visual sign. Covertness was maintained in 85% of the successful trials and clear proximity alert was given in 72% of the successful trials at distances ranging from 3 -372m with a mean distance of 98 m from the target.

Table 1. Tracking performance across 100 covert-tracking exercises on scent trails under five-hours old in Study 1.

3.2 Study 2: assessing tracking performance in rainforest conditions

Tracking success declined with increasing track age (Figure 2). Success rates were 100% for 30−minute tracks (n = 5), 75% for 8−hour tracks (n = 4), 67% for 24−hour tracks (n = 6), 100% for 3−day tracks (n = 2), and 33% for 7−day tracks (n = 6). The 3-day category achieved only two replicates due to time and logistical constraints during the monsoon onset period. The mean time to locate the target increased with track age: approximately 20–25 minutes for the 30 minutes and 8 hours track age, approximately 40 minutes at 24 hours, approximately 85 minutes at 3 days (elevated by one outlier trial; remaining replicates were completed more rapidly), and approximately 60 minutes at 7 days (Table 2).

Table 2. Tracking performance across five track-age categories in Study 2.

Tracking success varied across heat index bands (27-30 °C, 31–34 °C, 35-39 °C) with success rates of 70%, 78%, and 50%, respectively (Figure 3). Pairwise Fisher’s Exact Tests showed no statistically significant differences between bands with p values from 0.53-1.00.

There were seven trials associated with rainfall events of between 2 to 26 hours of rain between the track laying and target search. Phiri achieved a 57% success rate (4 out of 7) with successful tracks at 24 hours (n=2) with 14 hours of rain in both cases, and 3 days (n=2) with 2 and 26 hours of rain (Table 2). The dog did require two interventions by the handler in all the successful rain events and exhibited frequent pauses and gaze back to the handler, indicating uncertainty and temporary losses of track.

Across the trials, several recurring tracking patterns and challenges were observed under field conditions (Figure 4). These included temporary loss of track associated with non-target wildlife scent, shifts between ground and air scenting, and reduced tracking efficiency in areas with older or overlapping tracks. In multiple instances, handler intervention was required to return the dog to the last confirmed track, although the dog was also capable of independently reacquiring scent. The mapped trajectories further illustrate how track age, environmental complexity, and track density influenced tracking behaviour and outcomes.

3.3 Study 3: assessing operational performance in live deployments

Between 2020 and 2025, the dog unit participated in five out of 649 routine patrols (Supplementary Table 1) and four of 14 counter−poaching operations (Supplementary Table 2).

The canine-integrated routine patrols were monitoring deployments of pre-planned routes involving the canine unit and members of the patrol team designed to fulfil multiple objectives, including routine forest patrolling as well as environmental socialisation and endurance training for the tracking dog. In addition, they provided opportunities for cross-learning between patrol and canine unit personnel, as well as familiarisation of the patrol team members with the dog to support future operational integration.

Routine patrols involved minimal reliance on active tracking. Phiri maintained strong energy levels and motivation throughout, consistently keeping pace with average patrol distances of approximately 8 km per day and reaching up to 13 km per day over patrol durations of up to seven days. However, in some instances, additional support was required, including being carried over difficult terrain or given encouragement to continue forward.

The canine unit contributed to all four of the counter-poaching operations it was involved in through detection of abandoned poacher rest areas (7–12 days old), active snares, agarwood-processing sites and confirmation of non-poacher camps. In one case, the canine unit played an instrumental role in a reconnaissance patrol while in the remaining instances, the canine unit played assisting roles. Five of the 14 counter-poaching operations conducted during the study period resulted in an apprehension. However the canine unit was absent from all five primarily due to the constraints detailed in Supplementary Table 2, including recurring episodes of severe atopic dermatitis affecting Phiri, periods of handler unavailability, terrain presenting high injury risk, incursion signs discovered deep in the forest beyond practical rapid-deployment range, and sign age exceeding the assessed operational threshold of approximately two weeks.

The canine unit was typically deployed 1–2 days following the initial alert of fresh poaching sign detected by the ranger team. Tracking distances ranged from 2–4 km per search, with extended operations covering cumulative distances of up to 16-20km over a 12-day period. During the first two days of operations the dog generally maintained high energy levels, strong motivation, and a clear drive to work. In extended deployments, he showed a noticeable decline in speed and eagerness by the third to fourth day of operations. However, the dog’s willingness to track remained consistent when called to search. Motivational challenges were managed with structured breaks and verbal praise. In instances where Phiri exhibited signs of fatigue or heat distress during an ongoing operation, the tracking dog search was terminated and responsibility for the search was transferred to the ranger team. The canine unit would then undergo rest and cooling in accordance with field protocols before being either redeployed to support ranger teams or withdrawn from the field.

On all deployments the canine unit functioned as a complementary asset to the ranger visual-tracking team; when visible signs were lost or ambiguous, Phiri took the lead, tracking ahead until the next interpretable visual cue was located by rangers. Given this inherently collaborative search mechanism, it was not possible in this study to isolate the tracking dog’s independent contribution to detections or to determine whether these finds would have been achieved in the absence of the canine unit.

Coordination challenges arose during two of the four counter-poaching operations. On one operation, at a junction where two paths split, with the left path ascending uphill and the right path descending downhill. Visual cues were present along the left path, while Phiri persistently indicated movement towards the right. As rangers prioritised observable signs and time was limited, the team proceeded along the left path. The right path which descended downhill, was not supported by visible ground sign, however it is possible that scent from the uphill path travelled downslope and accumulated along the lower path, influencing Phiri’s indication towards the right. Given time constraints, only one of the two possible routes could be followed and the alternative route was not subsequently verified. As the target individuals were not located, it was not possible to determine whether the dog’s indication represented the correct direction of travel, although deprioritising Phiri’s input may have resulted in a missed opportunity if the scent trail continued along the unsearched route.

In both routine patrols and live counter-poaching operations, Phiri was exposed to a range of environmental obstacles that directly challenged both movement and performance. The terrain characteristics increased the risk of physical injury such as sprains or chipped nails. Leech infestation was persistent and at times severe; up to 15–20 leeches were observed attached to Phiri’s body, including on sensitive areas such as the nostrils and gums, where their presence appeared to temporarily impair scent-work. Stinging or biting insects such as lice and ants further contributed to Phiri’s discomfort. These were managed through routine in-field monitoring and brief pauses for targeted removal, with particular attention to sensitive areas such as the nostrils and oral region when irritation or sensory interference was observed. Terrain hazards such as thorny vegetation, fallen spiny material, and rocky stream crossings occasionally resulted in minor paw injuries (cuts and bleeding) and was treated immediately in the field using saline cleansing and antiseptic application, with continued tracking permitted or modified based on severity of the injury. Phiri was occasionally distracted by wildlife scent or animal scat, though refocusing was typically rapid following handler intervention.

3.4 Study 3: transferability

Application of the TEPIDOIL-F framework (Table 3) highlights current practices and recommendations for improved program design for strengthening the effectiveness, integration, and long-term sustainability of canine-assisted counter-poaching operations. While the programme demonstrates the potential utility of integrating a tracking dog into tropical forest counter-poaching operations, recurring constraints related to handler availability, canine health, logistics, and deployment doctrine affected operational readiness. The tiered assessment indicates that a one-dog, two-handler configuration (Tier 2) represents the minimum recommended capability for programmes operating under conditions similar to Kenyir, as it addresses the most frequently observed operational limitations while remaining substantially more cost-effective than maintaining multiple dogs.

Table 3. TEPIDOIL-F assessment of canine-assisted counter-poaching operations: current practice at Kenyir and tiered capability recommendations.

Costs for current practice and Tier 2 reflect actual and projected Kenyir expenditure. Tier 1 and Tier 3 figures are indicative estimates derived from local market rates. ★ = tier most directly supported by the Kenyir data.

Tier 1 (Basic): 1 dog, 1 handler. Lowest cost; highest operational risk — any handler or canine health episode eliminates deployment capacity entirely. Suitable only as a feasibility starting point or for programmes with very infrequent expected deployments.

Tier 2 (Recommended minimum, ★): 1 dog, 2+ handlers. Addresses the primary observed deployment gap in Kenyir (handler unavailability, n = 3 of 5 gaps) at moderate cost. Most directly supported by this study’s evidence.

Tier 3 (Optimised): 2+ dogs, 3+ handlers. Maximises operational readiness and canine welfare cover. Justified only where deployment demand is substantially higher than observed in Kenyir; not currently cost-justifiable at Kenyir’s deployment frequency.

4 Discussion

This study provides the first operational benchmark of counter−poaching tracking dog performance in a humid tropical forest in Southeast Asia. Across controlled trials and live deployments, the dog demonstrated high effectiveness on fresh tracks and retained operational value on tracks several days old, while also highlighting environmental and logistical constraints that shape deployment feasibility.

4.1 Collaborative tracking rather than autonomous search

The combination of high overall tracking success for fresh tracks (97%) alongside lower success rate for search independence (71%) underscores that optimal performance in dense tropical forest environments is achieved through collaborative tracking, where canine olfaction is integrated with handler interpretation and, where applicable, ranger visual tracking. The collaborative search mode reflects well-established principles in working dog deployment, where handler interpretation significantly influences success rates in complex environments (Lit et al., 2011; Beebe et al., 2016). As such, handler intervention should not be considered a failure mode but rather a functional component of operational tracking which allows handlers to integrate canine behavioural cues with visible field signs to make informed, real-time decisions. Consequently, handler competence in visual tracking is also important for maximising the effectiveness of the canine unit (Lit et al., 2011; Beebe et al., 2016; Sandor et al., 2025).

Phiri’s ability to maintain covertness in 85% of successful approaches, while also providing a proximity alert in 72% of cases, significantly enhances its tactical value within the counter-poaching team. Prior to this study, it was uncertain whether a dog would be able to operate covertly at all; therefore, these findings are particularly encouraging as this capability is an essential aspect of the Kenyir counter-poaching operational tactics.

In this study, Phiri demonstrated a passive proximity alert by ‘freezing’ with its head oriented towards the target. This alert type was deliberately selected over active alerts such as barking in order to maintain operational covertness. Alert distances ranged from 3 m to 372 m, with an average of 98 m. By comparison, other studies have reported average detection distances of 13.91m (range: 0.50 m and 62.82m) for tortoise targets (Cablk et al., 2008) and up to 150m for rodents in forest environments (Gsell et al., 2010). In Kenyir, visual tracking teams typically establish observation positions at approximately 40–50 m from a suspected poacher camp. With an average proximity alert distance of 98 m, Phiri operates well within the tactical envelope of the visual tracking team, providing an earlier indication of target presence. With consistent training, it is possible for these alert distances to be increased and made more predictable, giving ranger teams a clearer expectation of how far ahead a target may be when the dog signals proximity (Reed et al., 2011).

4.2 Track age and operational thresholds

The ability of a dog to follow days-old tracks is vital in counter-poaching operations in South East Asia. In Kruger National Park operations, 12-hour-old tracks were considered good performance (Kuiper, 2023) but in Southeast Asia, where poaching teams establish long-term camps deep inside the forest, counter-poaching operation success relies on being able to follow poaching sign up to two weeks old, hence the ability of Phiri to track signs above two-days old become operationally important. Although success rates declined beyond three days, the capacity to achieve successful tracks extending up to seven days was notable, though this should be treated as indicative as this was only based on two replicates.

Fresher tracks (<24 h) were associated with more confident and decisive tracking behaviour: consistent orientation on the target scent, steady pace, and minimal handler intervention. This supports a deployment approach in which the handler allows greater autonomy, focuses on reading the dog’s behaviour, and prioritises efficient forward movement rather than frequent confirmation of direction (Hall et al., 2021). As track age increased (>24 h, particularly ≥3 days), tracking became more variable and effort intensive: slower pace, less consistent orientation, and increasing reliance on handler input. Although scent seemed to be still detectable, it was less continuous and required more deliberate searching. This marked a transition from an autonomous tracking mode to a collaborative search mode, where handler interpretation, decision-making, and integration with ranger observations become critical to maintaining progress.

Rainfall washes away scent molecules from the air and surfaces (McKeague et al., 2024). In this study however, successful tracking was achieved despite prolonged periods of rainfall. This outcome was unexpected and may be influenced by the structural characteristics of the tropical forest having a dense canopy. Dense canopy cover could have reduced the direct impact of rain on the ground, preserving scent. Tracking dogs operating in tropical forest may retain effectiveness even under rainy conditions, potentially to a greater extent than previously thought. However, this observation is based on only seven trials with rain. Further replication is required to better assess the influence of rainfall on tracking performance.

4.3 Heat stress as a primary limiting factor

The observed decline in work rate and endurance at higher heat index measures (35-39 °C) was not statistically significant, due to the limited sample size and the narrow range of temperatures observed, but is consistent with established physiological trade−offs between thermoregulation and olfactory work. Dogs cannot sniff and pant simultaneously: as thermal load increases, panting rate rises and olfactory efficiency declines as airflow is diverted from the nasal passage to the mouth (Gazit and Terkel, 2003; McKeague et al., 2024). The effectiveness of panting as a cooling mechanism is itself constrained by humidity, since evaporative heat loss requires the surrounding air to be unsaturated. At the 80–90% relative humidity typical of Kenyir, panting provides substantially less cooling than in temperate or arid settings, while at 100% evaporative cooling ceases entirely (Slotta-Bachmayr et al., 2024).

Here, canopy cover emerged as a critical operational variable. Dense canopy can create cooler and more stable microclimatic conditions, with air temperatures up to 3 °C lower and ground surface temperatures up to 8 °C lower under shaded conditions than in areas exposed to direct sunlight (Tukiran et al., 2016). Under closed canopy, heat accumulation is slow enough that extended search periods were feasible. Transitions onto exposed features such as logging roads or ridgelines impose rapid thermal stress that can terminate deployments even when scent remains detectable. On one deployment involving a three-day-old track with a target approximately 1 km away, Phiri worked steadily under dense canopy before emerging onto an exposed logging road. Despite continued scent engagement and a brief rest period, the deployment was terminated due to visible heat stress. Subsequent verification confirmed the target camp was approximately 100 m beyond the termination point. This finding has direct operational relevance; deployment feasibility and route planning should account for canopy continuity and shade availability.

Active management of Phiri’s welfare was essential to minimise heat stress. Heat related illness has been identified as a cause of mortality in 10% of detection dogs in Iraq (O’Brien et al., 2020). Handlers monitored behavioural indicators of fatigue and overheating, including reduced tracking drive, self-initiated stopping, excessive panting, tongue curling, and changes in ear posture. Where heat stress was suspected, Phiri was immediately moved to shade, cooled gradually using water or cooling packs, and rehydrated incrementally. Where rivers or streams were available along deployment routes, Phiri was also allowed to partially immerse in water to facilitate body cooling. These practices are consistent with studies demonstrating that immersive and evaporative cooling methods as effective heat stress management strategies (Parnes et al., 2023). Handlers always carried dedicated water supplies and food for Phiri during all patrols and operations, regardless of water availability along routes. Work periods were interspersed with as many breaks needed depending on terrain, elevation, and canopy cover. Rest periods typically lasted 10–15 minutes but were extended, when necessary, based on the Phiri’s condition.

4.4 Operational integration and decision−making

A recurring operational challenge was the integration of the canine unit with experienced ranger teams who had established visual-tracking practice. Kenyir rangers possess strong fieldcraft and justified confidence in their tracking skills, therefore, the introduction of a dog raised questions about necessity and operational value. With established counter-poaching tracking protocols developed over years based primarily on accumulated experience and knowledge of visual tracking, the introduction of a scent-tracking dog without prior operational guidance on how it would integrate within these existing frameworks led to differences in how search operations were conducted or how information was interpreted and prioritised.

For canine olfaction-based tracking to be effective, the dog must lead the search to minimise scent contamination and ensure unobstructed access to the target scent plume. However, during joint operations, visual trackers often found it challenging to remain in a follow-only role, particularly when visual signs suggested alternative directions of movement. These differences occasionally led to friction and in some cases, search efforts were unintentionally redirected away from the dog’s line of work due to stronger confidence in visual evidence.

Building trust in the canine unit therefore requires repeated demonstration of reliability through cross-learning opportunities between the rangers and canine unit alongside the establishment of clear decision-making protocols that formalise the role of canine indication within the operational hierarchy. Such frameworks should specify from the outset how operations are conducted, who leads at each stage, when to transition between tracking approaches, and how to resolve situations where canine indications and visual-tracking cues diverge. Without agreed protocols, search efforts risk becoming disorganised, leading to confusion and reduced operational success. In practice, the more familiar evidence type – visible sign is often prioritised, potentially at the expense of scent-derived intelligence that may be accurate but less immediately interpretable to the team.

Trust in the dog’s indication is a critical component of effective scent-tracking operations (Beebe et al., 2016). Without such trust, canine input risks being undervalued or disregarded, limiting the effectiveness of the unit. Observations from training showed that Phiri’s persistence in a particular direction frequently reflected accurate scent interpretation, even when the terrain appeared counterintuitive from a human perspective. In such situations, premature redirection based on handler or ranger assumptions can interrupt the dog’s tracking process. Repeated interruptions may lead to confusion and reduced motivation, as the dog is prevented from fully committing to and resolving the scent trail.

Cultural context is an important consideration for conservation dog programmes in Peninsular Malaysia. Within predominantly Muslim ranger teams, sensitivities surrounding contact with dogs influenced operational comfort and willingness to work in close proximity to the canine unit. These concerns are rooted in beliefs that dogs are ritually unclean and it is wrong to touch or keep a dog (Sarwar et al., 2023). Instances where rangers shared concerns about patrolling in the same team or sharing the same vehicle as Phiri were not uncommon. Over time, however, acceptance increased through repeated joint deployments, opportunities to observe the dog’s operational value, and the establishment of clear physical boundaries that respected individual preferences, including separate vehicles and designated distances between the dog and ranger sleeping or eating areas in the field. Positive relationships were further strengthened through patrols conducted with handlers without Phiri, facilitating trust-building and knowledge exchange between handlers and ranger teams. While these challenges are generally manageable through early dialogue, clear protocols, and respect for individual boundaries, they require explicit attention given to team composition, deployment planning, and programme design rather than treated as incidental.

4.5 Multi−day tropical deployments

The literature on tracking dogs is predominantly derived from search-and-rescue, scat-dogs, endangered species research and conservation-detection context (Wasser et al., 2004; Greatbatch et al., 2015; Sandor et al., 2025) involving single operational periods of hours. The Kenyir deployments represent a qualitatively different model: multi-day operations with repeated scent-reacquisition cycles, sustained environmental exposure and significant logistical demands. Dog temperament and physical resilience are central to this. Phiri’s calm, methodical disposition and capacity to remain quiet and covert during tracking supported sustained performance in operationally sensitive environments. In contrast, high-drive dogs which are often better suited to short, intense search tasks may be less suited to extended deployments (McKeague et al., 2024; Kowalczyk-Jabło et al., 2026).

The logistical burden associated with deploying the canine unit was considerable, as carrying sufficient kibble and wet food for multi-day patrols added substantial weight alongside standard camping and operational equipment. Combined with the additional welfare considerations and accommodations by the rangers for the dog during extended field deployments, these constraints reduced the practicality and utilisation of the canine unit for routine reconnaissance patrols, which accounted for only 0.8% of deployments over the six-year study period. In addition, at 36 kg, Phiri’s size limited the handlers’ ability to assist him across difficult terrain. Smaller dogs may offer greater manoeuvrability and reduced logistical burden but possibly, at the cost of endurance.

A structural deployment gap also emerged; when incursions were detected several days into a patrol, deep within the forest, the canine unit could not reach the patrol team in time to assist at the point of detection, and when it did arrive it was already fatigued from the approach journey. Addressing this gap required either pre-positioning the dog unit alongside patrol teams as a standard element of extended operations or establishing a rapid-insertion capability; neither of which is logistically simple given the terrain and size of the unit. Alternatively, deployment for operations could be limited to incursions detected closer to the forest edge or access points, where response times and accessibility are more manageable.

4.6 Transferability

A central question this study could not fully resolve is whether the canine unit improved counter-poaching outcomes beyond what the ranger team would have achieved alone, and whether the investment was cost-effective. Interpreting the operational record requires an important caveat: the unit reached full operational readiness in 2023, immediately following a two-year period in which the canine unit was dormant and poaching incursions were virtually absent due to COVID-19 travel restrictions, with incursion frequency remaining low in subsequent years (Lam et al., 2023). This severely compressed the window of opportunity for live counter-poaching deployments, and the four operations in which the unit participated represented a limited sample from which to draw firm conclusions. Logistical, terrain, handler availability, and cultural constraints further reduced deployment frequency within that already narrow window. That the unit contributed to detections in all four operations it attended is therefore more informative than the raw participation rate of 28% suggests.

Start-up costs for the canine unit were approximately USD 39,000, with annual operating costs of USD 78,300 (Table 3). These expenditures represented approximately 22% of the counter-poaching programme’s total annual operating budget of USD 360,000. Costs were incurred not only in the acquisition and maintenance of the dog itself, but also in supporting a dedicated team of three to four personnel, including two primary handlers and additional staff to lay tracks during routine training to prevent skill fade. For this investment to be justifiable on conservation grounds, the canine unit would need to generate a proportionate reduction in poaching threat beyond what the ranger team achieved independently, a threshold that current data cannot confirm. However, Lam et al. (2023) demonstrated that apprehending experienced poaching team leaders generated disruption effects extending to subsequent incursion attempts within the same poaching community, and that a 40% decline in incursion frequency was associated with the stabilisation of a tiger population estimated at fewer than 200 individuals nationally (Ten et al., 2021). At this population size, the persistence of a single breeding female carries measurable demographic consequence. If canine-assisted detection contributed to even one additional apprehension of a team leader who would otherwise have evaded the ranger team, the downstream deterrence effect could plausibly justify a substantial proportion of the unit’s annual operating cost.

The Kenyir context may not represent the most favourable conditions for demonstrating canine unit value. The poaching threats addressed here involving transnational teams travelling long distances from their communities of origin and operating deep inside the forest for extended periods represented an unusually demanding deployment environment in which the canine deterrence effect was structurally constrained. Deterrence through awareness of a canine unit requires that potential offenders know of its existence and factor that into their decision to attempt an incursion. As Lam et al., 2023 showed, Vietnamese poachers in particular had limited awareness of conditions at Kenyir prior to deployment and limited ability to change target once financially committed, making general deterrence through any mechanism difficult to achieve.

In landscapes where poaching is conducted primarily by local communities conducting bushmeat hunting on forest fringes, common example across Southeast Asia, the conditions for canine-mediated deterrence are more favourable. Such poachers typically operate shorter incursions closer to the forest edge, reducing logistical barriers to rapid deployment, and knowledge of a local canine unit would be more likely to spread rapidly within the community. The unpredictability inherent in a dog’s detection capability can generate elevated risk perceptions that exceed the dog’s actual performance, a well-documented effect in crime prevention contexts. Research in urban settings has shown that the visible presence of dogs was associated with reductions in property crime, attributed to their salience as signals of heightened detection risk (Pinchak et al., 2023). This is relevant to a potential deterrence pathway among forest-fringe poachers with awareness of a local canine unit, and suggests that for protected area managers facing a primarily local poaching problem, the deterrence value of a canine unit may represent a return on investment beyond what operational detection metrics alone would indicate. However this constitutes a further evidence gap requiring research.

5 Conclusions

A trained tracking dog can perform effectively in the demanding thermal, humidity, and terrain conditions of a Southeast Asian tropical forest, providing a valuable complement to visual tracking in counter−poaching operations. Performance was highest on tracks up to three days old, declined under high heat stress, and was strongly influenced by canopy cover and handler integration.

For protected−area managers considering canine unit integration, tracking dogs represent a potentially powerful force multiplier when supported by clear SOPs, skilled handlers, cultural integration, and logistics suited to multi−day tropical deployments. The deterrence value of a canine unit is likely to be greatest in landscapes where poaching is conducted by local or forest-fringe communities, and this dimension should be factored into cost-effectiveness assessments alongside direct detection contributions. As enforcement agencies across Southeast Asia confront increasingly adaptive poaching networks, canine−assisted tracking offers a practical, evidence−supported enhancement to ranger capacity - but realising its full potential requires investment not only in the dog and handler, but in the organisational doctrine, integration protocols, and performance monitoring systems that connect canine deployments to measurable conservation outcomes. The three-component evaluation framework applied here; controlled covert-fixing trials, systematic environmental trials, and structured live deployment assessment provides a replicable model for other programmes seeking to benchmark and improve their canine unit performance.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI tools were used for language editing and clarity improvement. All scientific content, analyses, and conclusions are the work of the authors. Any references generated or suggested were independently checked, and all sources included in the manuscript were verified by the authors.

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.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcosc.2026.1858501/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Jesús A. Rivas, New Mexico Highlands University, United States

Reviewed by: Shoshana Rapley, Australian National University, Australia

Holly Elizabeth Marshall, Royal Commission of Alula, Saudi Arabia