Integrating Land Use and Poaching Impacts for Sustainable Wildlife Management in the Atlantic Forest of Misiones, Argentina

Abstract

Misiones, Argentina, holds one of the largest remnants of the Atlantic Forest, with almost 1.4 million hectares of native forest, representing a critical landscape for sustainable biodiversity conservation. However, connectivity across this ecoregion is increasingly threatened by habitat conversion, landscape fragmentation, and poaching pressures that extend beyond protected area boundaries, undermining long-term sustainability of wildlife populations. Using conservation detection dogs, we located, collected, and genetically confirmed 198 scats belonging to four game species: 20 lowland tapir (Tapirus terrestris), 72 white-lipped peccary (Tayassu pecari), 55 collared peccary (Pecari tajacu), and 51 Azara’s agouti (Dasyprocta azarae). Analyses examining species-specific habitat associations emphasized the importance of extending inference beyond point locations to encompass species’ home ranges, with native forest consistently identified as a key component of habitat use. The high prevalence of scats in mosaics of human-modified habitats outside protected areas, especially along their borders, underscores the importance of managing these areas as part of a broader sustainable landscape matrix. While native forest fragments outside of protected areas may serve as important refugia supporting species persistence, their contribution to sustainable management depends on reducing poaching pressure across these landscapes. There is an urgent need to expand antipoaching efforts beyond protected areas and across the Atlantic Forest in the Green Corridor of Misiones while preventing ongoing deforestation and the expansion of monoculture plantations. Achieving sustainable wildlife management in this region will require integrated strategies that promote sustainable land use, conservation planning, and rural development.

1. Introduction

Tropical and subtropical forests contain a disproportionately large share of the world’s biodiversity, yet they are increasingly threatened by habitat loss, fragmentation, and human pressures that undermine the long-term sustainability of ecosystems and biodiversity [1,2,3]. Within this broader context, the Atlantic Forest of South America stands out as a biodiversity hotspot, notable for its high levels of endemism and its long history of extensive deforestation [4,5]. While the province of Misiones, Argentina, represents only 1% of the national territory, it harbors more than half of the country’s known biodiversity, a fact directly linked to its maintaining one of the largest remaining tracts of Atlantic Forest, with approximately 1.4 million hectares, making it a critical region for sustainable conservation planning [6,7,8,9,10,11,12].

The Atlantic Forest ecoregion, with plant communities that include bamboo gallery forests and stands dominated by palm heart (Euterpe edulis), araucaria (Araucaria angustifolia), guatambú (Balfourodendron riedelianum), and palo rosa (Aspidosperma polyneuron), reaches its southernmost distribution in Misiones [4,8,13]. Despite its global conservation importance and high endemism, fragmentation threatens this ecoregion, which harbors an estimated 1–8% of the world’s known species, raising concerns about the sustainability of ecological processes and species persistence [4,5,8,14,15,16,17,18,19]. Half of the remaining forest occurs within protected areas (e.g., national parks, provincial parks, private reserves) of varying size and connectivity, with each embedded in a heterogeneous matrix of human-modified habitats that include monoculture plantations of non-native species (Pinus spp. and Eucalyptus spp.), small-scale agriculture (e.g., tea, yerba mate, tobacco, soy, tung), subsistence farming, pastures for grazing, bare soil, and urban areas, highlighting the need for sustainable land-use strategies across multifunctional landscapes [16,20,21,22,23,24,25].

Beyond habitat loss and degradation, many native vertebrates in Misiones are also threatened by road mortality, infectious diseases transmitted by domestic animals, and high levels of hunting pressure, all of which compromise the sustainability of wildlife populations [16,26,27,28,29,30,31,32,33]. Although wildlife hunting is prohibited in the province, poaching has been shown to severely and negatively affect various mammal populations (e.g., large felids, ungulates, rodents), with declines in population abundance that may lead to local extinction of small, isolated groups, undermining sustainable wildlife management efforts [29,34,35,36]. Among the most affected taxa are four game species: lowland tapir (hereafter tapir; Tapirus terrestris), white-lipped peccary (Tayassu pecari), collared peccary (Pecari tajacu), and Azara’s agouti (hereafter agouti; Dasyprocta azarae), with the white-lipped peccary and agouti considered especially sensitive to hunting pressure [33,37,38,39,40,41]. In Argentina, the white-lipped pecari is classified as Endangered, while the collared peccary and tapir are listed as Vulnerable and the agouti as Near Threatened [27,37,38,42]. These species were selected for their ecological importance, contrasting ecological niches, and differing responses to anthropogenic pressures, making them valuable indicators of ecosystem sustainability and integrity [32,43,44,45].

Functionally, they occupy complementary ecological roles. The tapir, as the largest terrestrial herbivore in the region, is a key long-distance seed disperser [45]. Peccaries act both as seed predators and ecosystem engineers, with their rooting and trampling influencing vegetation structure and regeneration [43,46]. In contrast, the agouti is a scatter-hoarder that plays a critical role in seed dispersal and recruitment at smaller spatial scales [47,48]. Together, these species shape forest composition, structure, and regeneration dynamics, thereby supporting the sustainability of forest ecosystems and their regenerative capacity [44,49,50]. In addition, these species are key in trophic interactions, with each considered important prey to coexisting carnivores [39,51,52,53,54,55,56,57].

Despite these important roles, all four species are subject to intense hunting pressure and habitat alteration, although their vulnerability varies [58,59]. The white-lipped peccary, which depends on large, continuous forest areas and moves in large groups, is particularly sensitive to fragmentation and overhunting, making it a critical species for evaluating landscape-level sustainability. In contrast, the collared peccary and agouti exhibit greater ecological flexibility [33,58]. The tapir, a more solitary species with specific habitat requirements, is especially vulnerable due to its low reproductive rate and large spatial requirements, highlighting challenges for sustainable population persistence [45,60].

In this study, we used conservation detection dogs to investigate the ecology of four game species within a sustainability-oriented monitoring framework. This noninvasive approach offers a powerful alternative to live-capture methods, which are often logistically demanding, invasive, and constrained by low capture probabilities for elusive species [61,62]. In general, some species present challenges for camera-trap monitoring due to the absence of individually distinctive markings, potentially resulting in overestimates of population density, as well as variability in detection across habitats that must be accounted for in management plans [63,64]. Detection dogs help overcome these limitations by enabling the systematic collection of scats across large spatial extents and substantially increasing detection rates for cryptic and low-density species [65,66,67].

This approach minimizes disturbance by eliminating the need for invasive techniques (e.g., physical capture) and avoids biases associated with methods that rely on attracting animals to specific locations (e.g., camera traps, hair snares), aligning with principles of sustainable and ethical wildlife research [20,62,68,69,70,71,72]. Crucially, the recovery of high-quality mitochondrial DNA (mtDNA) from intestinal epithelial cells shed in the mucous during defecation allows for obtaining reliable genetic information of the species origin of each scat, even under conditions of environmental degradation [73,74,75,76,77,78,79]. Leveraging this approach, we evaluated habitat associations and spatial patterns of these species across Misiones, examining their relationships with natural features (e.g., protected areas, water sources) and anthropogenic pressures (e.g., roads, urban areas) as well as the potential influence of poaching on their distribution, with direct implications for sustainable landscape management and conservation planning.

2. Materials and Methods

2.1. Sampling Design

In 2018 and 2022, scats were collected over four months (May–August), which coincided with the season of cooler temperatures [66], with the first survey centered in the northern–central zone of Misiones and the second in the central–southern zone (Figure 1). While sampling typically occurred from sunrise to early afternoon, there was some variation depending on various factors (e.g., temperature, cloud cover, canopy or vegetation cover), with avoidance of late afternoons to allow the dogs sufficient rest and heavy rains that would affect sample quality [66,80,81,82]. Survey effort was standardized across routes by maintaining a consistent walking pace while allowing the dogs to actively search for scats along and adjacent to transects. Environmental conditions known to affect scent detection (e.g., temperature, humidity, precipitation) were considered during survey planning to maximize detection probability and sample quality, as these factors can affect odor production, dispersion, and degradation [82,83,84,85]. A different, but experienced, conservation detection dog, both handled by KED, was used in each survey (2018 and 2022), with both dogs trained using scats from captive and wild animals, to have a positive alert to five game species [tapir, white-lipped peccary, collared peccary, paca (Cuniculus paca), and agouti) and five carnivores [jaguar (Panthera onca), puma (Puma concolor), ocelot (Leopardus pardalis), southern tiger cat (Leopardus guttulus), and bush dog (Speothos venaticus)] [6,86,87]. The latter data are not discussed here.

Daily surveys with the conservation detection dog were divided into unique routes along existing roads (e.g., dirt, logging) and trails (e.g., animal, poaching), with the aim of maximizing coverage of habitat heterogeneity and total area [21]. A systematic, habitat-stratified transect design was not used because detection dogs are more effective when searching along existing trails and routes (e.g., animals trails, roads, poaching paths) than along constructed transects [66]. Furthermore, access across Misiones is limited by landowner permissions and terrain constraints, and strict transect protocols can reduce detection efficiency by restricting natural search behavior and failing to account for environmental effects on scent dispersion or spatially clustered scat distributions [66]. The detection dog’s effective search width was estimated at 15 m on each side of the trail based on handler observations during surveys, resulting in a total search width of approximately 30 m. Each route was georeferenced with a GPS, acknowledging that this was the human track, as the dogs typically moved off the trails, exploring the surrounding area, with an average of 4–6 times this distance [66,84,88]. In 2018, there were 63 unique routes that covered a linear distance of 396.0 km, with the majority (81.6%; 323.3 km) located outside of protected areas (e.g., privately owned, managed by a forestry company) and less than a quarter (18.4%; 72.7 km) inside protected areas (Figure 1). Similar efforts were made in 2022, with 52 unique routes covering a linear distance of 359.7 km; the majority (70.4%; 253.1 km) were located outside protected areas, and less than a third (29.6%; 106.6 km) were inside protected areas (Figure 1).

During operational searches, unlike controlled trials, it is difficult to estimate detection probability or assess dog performance because the total number of targets in an area is unknown, and individual scats cannot always be distinguished in the field [89]. However, these limitations can be minimized—supporting data and study replicability—through rigorous dog-handler training [90], as implemented in this study. Although detection probabilities may vary among species due to differences in scat size, deposition rates, and environmental conditions affecting scent persistence, conservation detection dogs have been shown to provide high detection rates across multiple species and habitat types [62,66]. Sampling effort was explicitly incorporated into subsequent analyses to account for variation in detectability and minimize potential bias, with percentages calculated relative to the total sampling effort within each zone and overall (Appendix A). Sampling effort was estimated by converting transects into regularly spaced points (30 m), resulting in minor discrepancies (<1%) relative to GPS-recorded track lengths due to rounding, segmentation of continuous tracks, and rasterization.

2.2. Sample Collection

A visual assessment of the scats located by the detection dog provided a potential species identification and classification of scat quality as fresh (≤24 h), moderately old (between 24 h and 3 days), or old (>3 days) [20], with only those scats classified as fresh or moderately old collected. Each scat was georeferenced with GPS. These data were recorded in a field notebook, along with a unique identifier, date, time, general location, georeferenced position, surrounding habitat (e.g., native forest, monoculture plantation, rural, urban), sample content (e.g., seeds, fruits, herbs, mold), and any other observations, with all data later keyed in for reference in genetic and spatial analyses.

Each scat was swabbed with a cotton-tipped applicator soaked in 1× phosphate-buffered saline solution to capture the intestinal mucus located on the scat’s surface [91,92,93,94]. All samples were collected using sterile gloves and individual swabs to prevent cross-contamination. The tip of each applicator was placed in a 1.7 mL polypropylene tube, labeled, and sealed. Each sample was swabbed in triplicate (two for DNA extraction and one as reference storage), with each taken from different areas of the scat’s exterior. The scat sample was collected and stored in a labeled 18 oz Nasco Whirl-Pak^® bag (Nasco, Fort Atkinson, WI, USA) as a backup for DNA extraction and other analyses (e.g., parasites, viruses, diet). To minimize DNA degradation, samples were protected from direct UV exposure in the field and kept in insulated conditions until the end of each survey day, when both swabs and scats were transferred to a −20 °C freezer until arrival at the laboratory. Given the expected low abundance of tapirs in the central–southern zone of Misiones [27,95], we georeferenced observed tapir tracks on survey routes in 2022. This approach allowed us to maximize the capture of tapir presence, since their tracks are readily distinguishable from coexisting ungulates [96,97,98,99,100].

2.3. Evidence of Poaching

During both surveys (2018 and 2022), all evidence of poaching that was noted during fieldwork was georeferenced, with notes recorded about whether the evidence was direct (presence of hunters, established campsites, elevated hunting platforms or stands, and artificial salt lures) and/or indirect (spent cartridges, corn piles, gunshots, machete-cut trails). Spatial analyses in ArcGIS Pro 3.4.0 (ESRI) examined how poaching locations were associated with habitat type and land management (e.g., protected area, privately owned, forestry company/managed). In addition, whether poaching evidence was in protected areas, within 5 km of a border, or outside of protected areas was determined.

2.4. Genetic Analysis

Initially, DNA extraction was performed from a single swab per scat using a modified protocol (adapted from [101]; Appendix B) incorporating an anionic detergent (TEC-SDS buffer: Tris-EDTA-NaCl-SDS), phenol (saturated in Tris/HCl; pH 7.8), and chloroform:isoamyl alcohol (24:1; saturated in H₂O; pH 7.0). An extraction blank (no DNA added) was included every 20 samples as a check for cross-contamination during processing. For those samples that had low yield, a second DNA extraction was carried out from the second swab using an optimized protocol with a Qiagen DNeasy^® 96 Blood & Tissue Kit (QIAGEN, Hilden, Germany) [102] (Appendix C). The quality of each DNA extraction was verified by horizontal agarose gel (1% w/v) electrophoresis with GelRed Nucleic Acid Gel Stain (Biotium Inc., Fremont, CA, USA). While samples extracted using the manual protocol were diluted 1:50 (v/v) in ultrapure water for subsequent DNA amplification, no dilutions were required for samples extracted using the commercial kit.

To identify species for each sample suspected to be tapir, white-lipped peccary, and collared peccary, a 110 bp region of the mitochondrial cytochrome b (cytb) gene (5′–AAACTGCAGCCCCTCAGAATGATATTTGTCCTCA–3′; 5′–TATTCTTTATCTGCCTATACATRCACG–3′; ref. [103] was amplified, with a modified version of the protocols and reagents published by Farrel et al. [103] and Miotto et al. [104], shown to be effective with these species [39,105]. Amplifications were performed on a Veriti 96-Well Thermal Cycler (Thermo Fisher Scientific, Waltham, MA, USA) in 25 μL volumes containing 2 μL DNA, 1× PCR Green GoTaq Flexi Buffer (Promega Corporation, Madison, WI, USA), 5 mM MgCl₂, 200 μM each dNTP, 0.3 μM each primer, 150 μg/mL BSA (Thermo Fisher Scientific, Waltham, MA, USA), and 0.5 U GoTaq G2 Hot Start Polymerase (Promega Corporation, Madison, WI, USA). To minimize the potential for contamination, all PCR (Polymerase Chain Reaction) setups were carried out in a UV (ultraviolet) PCR Chamber (Ivema C9; Ivema Desarrollos, Valentín Alsina, Buenos Aires, Argentina). A negative control (no DNA added) was included in all PCR runs to test for contamination. The PCR profile consisted of 10 min denaturation at 95 °C, followed by 40 cycles at 95 °C for 30 s, 49 °C for 45 s, and 72 °C for 45 s, with a final extension of 30 min at 72 °C.

To identify species in samples suspected to be agouti, a 421 bp region of the cytb gene (5′–TACCATGAGGACAAATATCATTCTG–3′; 5′–CCTCCTAGTTTGTTAGGGATTGATCG–3′; ref. [106]) was amplified. DNA amplifications were performed on a Veriti 96-Well Thermal Cycler in 25 μL volumes containing 2 μL DNA, 1× PCR Green GoTaq Flexi Buffer, 1.5 mM MgCl₂, 100 μM each dNTP, 5 pmol each primer, 150 μg/mL BSA, and 0.5 U GoTaq G2 Hot Start Polymerase. The same PCR setup and negative control were used. The cycling profile consisted of 10 min denaturation at 95 °C, followed by 35 cycles at 95 °C for 45 s, 51 °C for 1 min, and 72 °C for 2 min, with a final extension of 8 min at 72 °C.

Successful PCR amplification was verified by horizontal electrophoresis in 2% (w/v) agarose gels using GelRed Nucleic Acid Gel Stain (Biotium). Amplicons were subsequently purified with a commercial purification kit (GE Healthcare Life Sciences, Chicago, IL, USA) and shipped for bidirectional Sanger sequencing on an ABI 3730xl DNA Analyzer (Macrogen Inc., Seoul, Republic of Korea). Chromas Lite (v2.6.5) was used to edit (e.g., remove primers) the resulting electropherograms, with BioEdit (v7.2.6) used to generate consensus sequences that were compared with the GenBank database using the BLASTn (v2.17.0) algorithm [107].

2.5. Habitat Association Analysis

First, with two land-use grids (30 m × 30 m resolution; MapBiomas Trinational Atlantic Forest Project, Collection 2) for Misiones, one from 2018 and another from 2022, ArcGIS Pro (v3.4.0) was used to extract the corresponding habitat type associated with the georeferenced location of each scat confirmed to be one of the target species [108]. Five land-use categories were then quantified by the proportion of samples found in each habitat type: native forest, monoculture plantations, temporary crops, perennial crops, and non-vegetated areas (urban areas and other areas with no vegetation). To account for differences in habitat availability and sampling effort, we modeled the number of detections per habitat category and protection level using generalized linear mixed models (GLMMs) with a negative binomial error distribution to address overdispersion [109,110]. Sampling effort (total distance surveyed within each habitat category and protection level) was included as an offset term (log-transformed kilometers surveyed), allowing detection rates to be standardized per unit effort [111]. Land use and protection level were included as fixed effects, and transect identity was included as a random intercept to account for spatial dependence among observations [109]. Models were fitted using the glmmTMB package [112] in R (v.4.4.2). Model selection was based on the Akaike Information Criterion (AIC) [113], and model assumptions were assessed using simulation-based residual diagnostics implemented in DHARMa [114]. This framework allows habitat use to be evaluated relative to availability rather than relying solely on observed proportions, providing a more robust assessment of habitat selection [115,116].

Second, for each georeferenced location confirmed to be one of the target species, with this being both scat and tracks for tapir, independent buffer zones were created in ArcGIS Pro (v3.4.0), with the center set to the georeferenced sample location and the buffer zone radius set by a species-specific home range size [20]. The home ranges were based on published literature, with 8.31 km² for the tapir [117], 18.92 km² for the white-lipped peccary [32], 2.16 km² for the collared peccary [118], and 1.895 ha for the agouti [51,119]. While it is recognized that several factors can affect species’ directional movements across the landscape, with circular patterns not always reflecting reality, applying the average home range of the species can provide useful insight into potential habitat associations [20,120]. Specifically, this approach allows habitat associations with each confirmed presence to go beyond a point location and instead be seen relative to the extent of potential habitat a species could be using during its natural movements, allowing for assessment of which habitats the species most frequently associates with compared to those it may potentially avoid [20,121,122]. Using ArcGIS Pro (v3.4.0), the proportion of seven land-use categories (native forest, monoculture plantations, temporary crops, perennial crops, pastures, non-vegetated, and urban areas) was tabulated and summarized within the independent buffer zones for each sample by species.

2.6. Proximity Analysis

The georeferenced locations of each confirmed presence were used to calculate distances to poaching evidence and to protected areas (e.g., provincial parks, private reserves) in ArcGIS Pro (v3.4.0). In addition, distances between these scats and other spatial data [water sources (rivers and streams), roads (dirt and paved), and urban areas] were calculated. To evaluate the influence of landscape features on species detections, we modeled the number of detections per transect, species, and year as a function of mean distance to protected areas, water, roads, and urban areas, allowing for formal testing of relationships between occurrence and spatial predictors [110,123]. Survey effort was standardized as total kilometers surveyed and included as an offset term (log km). Given overdispersed count data with many zeros, we fitted generalized linear models with a negative binomial distribution [111], including species and year as fixed effects and standardized continuous predictors. Models were fitted using glmmTMB [102], comparing a null model (species + year + effort) to a full model including all distance variables. A complementary binomial model (detection presence/absence) was also fitted. Model performance was evaluated using AIC, likelihood ratio tests, and DHARMa residual diagnostics [114]. Poaching was excluded because records were collected along the same transects and were therefore not independent of sampling effort.

3. Results

3.1. Confirmed Species Presence

A total of 259 scats, located by the conservation detection dogs, were visually identified as potentially belonging to one of the targeted game species, with fewer scats in 2018 (n = 63) than in 2022 (n = 196). It was possible to genetically confirm species identity for 198 scats (76.4%), all with BLASTn coverage and identity values at 98–100%, for a total of 20 tapir (11 in 2018 and 9 in 2022), 72 white-lipped peccary (30 in 2018 and 42 in 2022), 55 collared peccary (8 in 2018 and 47 in 2022), and 51 agouti (2022). For the remaining 61 samples, the majority were potentially contaminated by urine (e.g., amplified DNA of coexisting carnivores; [20]) or had poor-quality DNA (e.g., unable to amplify, DNA of dietary items or bacteria; [124]), with some potentially belonging to other coexisting herbivores, but intraspecies contamination (e.g., urine) made this confirmation difficult. In addition, 12 sites with unequivocal tapir tracks were recorded, georeferenced, and included as additional presence records for spatial analyses of this species.

3.2. Evidence of Poaching

Of the 69 sites where poaching was detected, most (66.7%; n = 46) involved indirect evidence, while the remainder consisted of direct evidence (33.3%; n = 23). Although overall levels of poaching evidence were similar between years, the central–southern zone (2022) showed approximately twice as many records within protected areas as the northern–central zone (2018;

Table 1. Summary of 69 locations; poaching evidence was recorded in Misiones, Argentina, in 2018 (n = 31) and 2022 (n = 38), with values reported as the number of sites (#) and percentages (%) for each year (2018 and 2022) and overall (total). In addition to areas inside and outside protected areas (PAs), “associated with PAs” refers to a combined category comprising sites located within protected areas or within ≤5 km of their boundaries. The land-use category corresponds to the habitat at the georeferenced location of each poaching record. The corresponding sampling effort by protection status and land-use categories is reported in Appendix A.

Variable	Category	2018	2022	Total
Protection status	Inside PAs	5 (16.1)	12 (31.6)	17 (24.6)
	Outside PAs	26 (83.9)	26 (68.4)	52 (75.4)
	Associated with PAs	11 (35.5)	23 (60.5)	34 (49.3)
Land-use category	Native forest	29 (93.5)	35 (92.2)	64 (92.8)
	Monoculture plantations	2 (6.5)	3 (7.8)	5 (7.2)

). Within each zone, most records outside protected areas occurred on forestry-managed land in the northern–central zone (61.5%; n = 16) and on privately owned properties in the central–southern zone (92.3%, n = 24). When combining records within protected areas and those within 5 km of their boundaries, approximately one-third of records in 2018 and over half in 2022—and overall across both years—were associated with protected areas or their immediate surroundings (Table 1). An analysis of habitat at georeferenced locations indicates that most poaching sites occurred in native forest, with fewer records in monoculture plantations (Table 1). The corresponding sampling effort by protection status and land-use categories is reported in Appendix A.

3.3. Habitat Association Analysis

The first habitat association analysis showed that most records (83.8%; n = 176) were located in native forest. Both peccary species and agouti had >81.9% of records in native forest, whereas tapir was an exception, with 62.5% of records in native forest and nearly one-third (31.3%) in monoculture plantations (

Table 2. Percentage (%) of each land use, with the corresponding number of georeferenced locations (parentheses) confirmed for each target species—tapir (n = 32; scats and tracks), white-lipped peccary (WLP; n = 72), collared peccary (CP; n = 55), and agouti (n = 51)—as well as for all four game species combined (n = 210). The corresponding sampling effort by protection status and land-use categories is reported in Appendix A.

Land Use	Tapir	WLP	CP	Agouti	ALL
Native forest	62.5 (20)	81.9 (59)	89.1 (49)	94.1 (48)	83.8 (176)
Monoculture plantations	31.3 (10)	15.3 (11)	9.1 (5)	5.9 (3)	13.8 (29)
Temporary crops	3.1 (1)	1.4 (1)	---	---	0.9 (2)
Perennial crops	---	---	1.8 (1)	---	0.5 (1)
Non-vegetated	3.1 (1)	1.4 (1)	---	---	0.9 (2)

). The tapir had the highest proportion of confirmed presence located in modified or altered habitats. Across all species, however, only 13.8% of samples (n = 29) were recorded in monoculture plantations, with the remainder found in agricultural lands (1.4%; n = 3) or non-vegetated areas (<1%; n = 2) (Table 2). The corresponding sampling effort by protection status and land-use categories is reported in Appendix A.

After accounting for sampling effort and spatial structure, detection rates differed among surveyed zones, with significantly higher rates in the central–southern zone than in the northern–central zone (β = 1.72 ± 0.40 SE, p < 0.001). Because survey areas differed between sampling periods, the zone effect was interpreted as a spatial rather than a temporal pattern. Land-use effects were generally weak, although there was a marginal trend toward lower detection rates in perennial crops relative to native forest (β = −2.02 ± 1.08 SE, p = 0.062). Detection rates did not differ between protected and non-protected areas (p = 0.99). Model diagnostics indicated a good fit, with no evidence of overdispersion (p = 0.28) or zero inflation (p = 0.98). Overall, these results provide a quantitative assessment of habitat associations while accounting for variation in sampling effort across land-use categories.

The habitat analysis based on home range size showed that all four species had more than 79.8% of their ranges in native forests, with a value of 82.3% across the four species (

Table 3. Summary of percentages (%) of each land-use type within species-specific buffer zones, approximating home range size for each species—tapir (n = 32; scats and tracks), white-lipped peccary (WLP; n = 72), collared peccary (CP; n = 55), and agouti (n = 51)—as well as for all species combined (ALL).

Land-Use Category	Tapir	WLP	CP	Agouti	ALL
Native forest	80.1	83.0	79.8	92.0	82.3
Monoculture plantations	15.0	12.2	18.4	6.0	13.1
Perennial crops	2.2	1.8	1.3	0.1	1.8
Temporary crops	1.1	1.9	0.1	0.4	1.6
Pastures	0.8	0.6	0.3	1.3	0.6
Non-vegetated	0.6	0.4	0.1	0.2	0.4
Urban areas	0.2	0.1	---	---	0.1

). The agouti, the smallest of the four species, had the highest proportion (92.0%), and the collared peccary had the lowest (79.8%), with similar values between the white-lipped peccary (83.0%) and tapir (80.1%) (Table 3). Of the remaining land-use categories, monoculture plantations accounted for the largest proportion (13.1%), followed by perennial crops (1.8%) and temporary crops (1.6%), with all others being ≤1.0% (Table 3).

3.4. Proximity Analysis

Of the 210 presence records (198 scats and 12 tracks), more than half occurred within protected areas or within 5 km of their boundaries, with the remainder within 15.9 km (

Table 4. Summary of distances for the 210 georeferenced presence locations (198 scats and 12 tracks) relative to protected areas, water sources, urban areas, roads, and poaching evidence. For each species—tapir (n = 32; scats and tracks), white-lipped peccary (WLP; n = 72), collared peccary (CP; n = 55), and agouti (n = 51)—as well as all species combined (ALL), values are as the number of sites (n), with the corresponding percentage (%) of each distance (parentheses).

Protected Areas
Distance	Tapir	WLP	CP	Agouti	ALL
0–5	24 (75.0)	40 (55.6)	33 (60.0)	36 (70.6)	133 (63.3)
5–10	4 (12.5)	6 (8.3)	18 (32.7)	6 (11.7)	34 (16.2)
10–15	3 (9.4)	25 (34.7)	4 (7.3)	2 (3.9)	34 (16.2)
15–20	1 (3.1)	1 (1.4)	---	7 (13.7)	9 (4.3)
Poaching evidence
Distance	Tapir	WLP	CP	Agouti	ALL
0–5	23 (71.9)	64 (88.9)	37 (67.3)	45 (88.2)	169 (80.5)
5–10	1 (3.1)	---	8 (14.5)	---	9 (4.3)
10–15	7 (21.9)	5 (6.9)	---	1 (2.0)	13 (6.2)
15–20	1 (3.1)	3 (4.2)	10 (18.2)	5 (9.8)	19 (9.0)
Water Sources
Distance	Tapir	WLP	CP	Agouti	ALL
0–5	32 (100)	66 (91.7)	55 (100)	49 (96.1)	202 (96.2)
5–10	---	6 (8.3)	---	2 (3.9)	8 (3.8)
Roads
Distance	Tapir	WLP	CP	Agouti	ALL
0–5	31 (96.9)	57 (79.2)	37 (67.3)	29 (56.9)	154 (73.3)
5–10	1 (3.1)	15 (20.8)	18 (32.7)	22 (43.1)	56 (26.7)
Towns
Distance	Tapir	WLP	CP	Agouti	ALL
0–5	1 (3.1)	1 (1.4)	---	3 (5.9)	5 (2.4)
5–10	2 (6.2)	9 (12.5)	12 (21.8)	16 (31.4)	39 (18.6)
10–15	7 (21.9)	12 (16.6)	14 (25.5)	15 (29.4)	48 (22.8)
15–20	14 (43.8)	21 (29.2)	18 (32.7)	2 (3.9)	55 (26.2)
20–35	8 (25.0)	29 (40.3)	11 (20.0)	15 (29.4)	63 (30.0)

). An even greater proportion occurred within 5 km of poaching evidence, with the remainder within 19 km (Table 4). All detections were within 5.2 km of a water source (Table 4). Approximately three-quarters of records were within 4.9 km of a road, with the rest between 5.1 and 7.6 km (Table 4). Most records occurred 3.5–32.2 km from the nearest town, with fewer than 3% within 2.7–3.5 km (Table 4). The model, including landscape–distance predictors, improved fit relative to the null model (AIC = 582.1 vs. 586.4; χ² = 12.25, df = 4, p = 0.0156). Detection counts increased significantly with distance to protected areas (β = 0.397 ± 0.173, p = 0.022), whereas distances to water and roads were not significant predictors. Distance to urban areas showed a negative, marginal effect (β = −0.310 ± 0.178, p = 0.082). Detection rates were higher in 2022 than in 2018 (β = 0.898 ± 0.369, p = 0.015). The presence/absence model showed consistent temporal effects (β = 1.142 ± 0.354, p = 0.001) and indicated lower detection probability with increasing distance from roads (β = −0.428 ± 0.172, p = 0.013) and urban areas (β = −0.465 ± 0.182, p = 0.011). Residual diagnostics indicated no evidence of overdispersion, zero inflation, or lack of fit.

4. Discussion

This study reinforces the effectiveness of conservation detection dogs for locating noninvasive samples across multiple species and heterogeneous environments while providing reliable genetic confirmation for downstream spatial analyses within a sustainable wildlife monitoring framework. We successfully identified four target species (tapir, white-lipped peccary, collared peccary, and agouti), detected variation in detection rates between two spatially distinct regions, and found strong associations among species occurrence, protected area status, and poaching activity. This study also provides the first systematic scientific data on the occurrence and distribution of game species in the southern–central zone of Misiones, Argentina, a region historically underrepresented in ecological research.

In Misiones, native forest remains a key habitat despite extensive loss and fragmentation of the Atlantic Forest; however, many detections occur in human-modified landscapes, particularly near protected area boundaries, underscoring the need for broader-scale analyses beyond point locations and supporting a sustainable landscape perspective. These patterns suggest that forest fragments outside protected areas may function as refugia only if poaching pressure is reduced, highlighting the importance of expanding antipoaching efforts beyond protected areas and strengthening sustainable land-use planning to limit deforestation and the expansion of monoculture.

4.1. Spatial Variation in Species Detection

Despite comparable survey effort in both years and the exclusion of samples from an additional target species surveyed in 2022 (agouti), which accounted for approximately one-quarter of detections, more scats were located in 2022 than in 2018. Because different regions were surveyed in each year (northern–central in 2018 and central–southern in 2022), this difference should be interpreted as a spatial rather than a temporal effect and is unlikely to reflect methodological differences, including the detection dog used. This pattern is notable given the greater habitat fragmentation and lower proportion of legally protected area in the central–southern zone [16,23,87,108], raising important considerations for regional sustainability and conservation prioritization. Even excluding UNESCO’s Reserva de Biosfera Yabotí (approximately 230,000 ha), with Parque Provincial Esmeralda (approximately 30,000 ha) at its core, the northern–central zone still contains approximately 7.5 times as much protected area as the central–southern zone. These results underscore the importance of not overlooking the central–southern zone in conservation and management planning by the Ministerio de Ecología y Recursos Naturales Renovables (MEyRNR), as it remains a key area for maintaining sustainable biodiversity and ecosystem function.

Despite known challenges associated with herbivore scat (e.g., high concentrations of bile salts, polysaccharides, proteins, glycolipids, urea, polyphenols) [91,125,126,127,128], most samples were successfully identified genetically, confirming the presence of four target species: tapir, white-lipped peccary, collared peccary, and agouti. Although the number of tapir scat detections was similar between the two surveys, both were substantially lower than reported by Delgado et al. [39] (n = 28). However, differences in survey design —particularly the proportion of sampling conducted within protected areas—limit direct comparison. The lower number of detections in this study likely reflects the fact that most tapir scats reported by Delgado et al. [39] were located within protected areas, whereas sampling here was conducted primarily outside them.

4.2. Habitat Associations

Our results suggest that tapirs use monoculture plantations more frequently than the other species studied; however, this pattern should be interpreted cautiously given the limited sample size (n = 20). This finding highlights the species’ ecological flexibility in Misiones, where it tolerates a range of habitats [28,129,130,131,132] but remains closely associated with habitats linked to protected areas, emphasizing the importance of maintaining sustainable habitat mosaics [6,27,133,134,135,136,137]. White-lipped peccaries were detected across both regions, indicating persistence in the landscape. Although the decline in detections relative to earlier studies could reflect natural population fluctuations or movement dynamics, it is more plausibly explained by disease impacts and, in particular, poaching, given documented increases in hunting pressure in the region [6,39,40]. Collared peccary detections were low in earlier surveys but higher in the central–southern zone in 2022, suggesting potential spatial variation in distribution and/or habitat use.

The shift in tapir habitat associations from point-based analyses to home-range scale highlights the importance of considering spatial context beyond the location of scat deposition. Across all four species, native forest predominated within estimated home ranges, consistent with studies showing that mammals in this region depend on structurally complex forest environments that provide forage, thermal regulation, and protection from poaching [38,137,138,139,140]. For example, agoutis are highly sensitive to habitat degradation and rely on continuous canopy cover for movement and access to resources [16,38,53,141,142,143]. Similarly, tapirs and peccaries exhibit higher occupancy and activity in structurally complex forests characterized by dense understory and stable food availability [130,132,138,140,144,145], consistent with ecological niche modeling predictions for Misiones [6]. This reinforces the central role of native forest in supporting long-term ecological sustainability and resilience. In contrast, the limited use of plantations and croplands aligns with findings from the Atlantic Forest in Brazil, where commercial plantations and agricultural mosaics generally support lower mammal richness and reduced activity of forest specialists [6,87,129,130,146,147,148], underscoring trade-offs between production and biodiversity sustainability.

4.3. Poaching Patterns

Poaching in Misiones is concentrated in native forest habitats and is strongly associated with protected areas rather than with more heavily modified landscapes, directly threatening the sustainability of wildlife populations. The predominance of indirect evidence suggests sustained hunting pressure, while direct evidence confirms that poaching remains active across the landscape. The high frequency of records within native forests and within or near (≤5 km) protected areas indicates that poaching pressure extends beyond reserve boundaries into adjacent forest patches.

More than half of the genetically confirmed scats were located within or along the borders of protected areas, despite most survey effort occurring outside them, underscoring the role of these areas as critical refugia [134,149,150,151]. At the same time, the occurrence of many records outside protected areas highlights the importance of managing human-modified landscapes to support sustainable species persistence and connectivity [152,153,154,155]. Across the Atlantic Forest of Argentina and Brazil, many medium- to large-sized mammals are increasingly occupying unprotected or altered habitats due to extensive habitat loss and fragmentation [152,156,157], indicating that protected areas alone are insufficient to meet their spatial requirements [158,159,160]. Nevertheless, degraded landscapes outside protected areas can still contribute to connectivity when they retain native forest fragments and are appropriately managed under sustainable land-use practices [161,162,163].

The proportion of illegal activity within protected areas was higher in the central–southern zone than in the northern–central zone, despite the latter containing a substantially greater extent of protected area. When surrounding buffer zones (<5 km) are included, the apparent impact of poaching on protected areas increases further [164]. Overall, nearly half of all poaching records occurred either within them or in their immediate surroundings, and most species detections were located within 5 km of poaching evidence, indicating a concerning spatial overlap [39,105,165]. This pattern suggests that poaching pressure is concentrated along protected area boundaries and in adjacent buffer zones, likely facilitated by greater human accessibility and animal movement across reserve edges. This pattern also underscores the need to expand antipoaching efforts beyond protected area boundaries. In particular, the Grupo de Operaciones en Selva (GOS), under the MEyRNR, would benefit from extending patrols into these peripheral areas, where enforcement is currently limited. Strengthening enforcement—especially in the central–southern zone, where evidence of prey was strongest—to maintain biodiversity and prevent further wildlife declines, along with managing buffer zones and high-access areas, is essential to achieving sustainable wildlife management outcomes across the region.

4.4. Limitations and Methodological Considerations

While our approach provides a robust framework, several limitations warrant consideration. Occurrence records—regardless of the methods used—may underrepresent certain species due to detectability biases associated with low and uneven population densities across heterogeneous landscapes. In Misiones, Argentina, these biases are exacerbated when using camera traps in remote or human-dominated areas, where frequent theft or damage reduces their effectiveness and limits their suitability for long-term monitoring. In contrast, detection dogs are not constrained by these limitations, allowing for unbiased sampling across habitat types, including areas outside protected zones and irrespective of human presence, thereby providing a scalable and sustainable approach to monitoring elusive species. However, detection success can still be influenced by environmental conditions and sample quality. Although sample sizes varied among species—particularly for tapir—the dataset provides valuable baseline information for typically elusive species. Because sampling was conducted in different regions in each year, the result should be interpreted as reflecting spatial rather than temporal variation in species occurrence.

4.5. Conservation Implications

Risk to species outside protected areas is amplified by the combined effects of poaching and road access [129,133,135,166]. Our results show substantial spatial overlap between high numbers of genetically confirmed records of game species and poaching evidence located outside and along the borders of protected areas. These findings underscore the need to protect all remaining Atlantic Forest in Misiones, under the Green Corridor framework (Corredor Verde; Provincial Law XVI No 60), regardless of land tenure, as part of a broader strategy for sustainable landscape conservation [6,87,167,168]. Achieving this will require extending antipoaching efforts, including GOS patrols, beyond protected area boundaries. Without such expansion, continued overhunting in unprotected areas is likely to increase pressure on protected areas, reducing their effectiveness and undermining long-term conservation sustainability [29,36,169,170,171,172,173].

Responsibility for addressing these challenges should extend beyond the MEyRNR to include additional stakeholders, particularly in the northern–central zone, where extensive monoculture plantations and forest reserves occur [16,24]. Forestry companies managing large areas of native and production forests, especially those with Forest Stewardship Council certification, could play a more active role by maintaining adequately resourced monitoring and enforcement capacity in support of sustainable forest management and conservation outcomes [174]. Practical actions include partnering with the MEyRNR to support additional GOS patrols, implementing long-term biodiversity monitoring programs, and establishing formal systems for reporting illegal activities on their properties, thereby strengthening sustainable governance of forest landscapes. Although some of these measures have been implemented previously, their decline has coincided with increasing poaching pressure in the region. Overall, the observed distribution of species and poaching records indicates that without expanded, coordinated antipoaching efforts supported by sufficient personnel and resources, current conservation strategies will be insufficient to achieve long-term sustainability of wildlife populations and ecosystem integrity.