Camera-trap Dataset Research Articles

A landscape-based approach to design flower blocks may reduce mammalian predator activity and protect ground-nesting farmland birds

AbstractHigh predation rates threaten many ground-nesting farmland birds and are difficult to address through conventional measures such as lethal predator control or fencing. Landscape-based approaches for conservation measures promise an alternative by reducing predator - bird encounters, but require detailed knowledge of landscape effects on predation risk. Different habitat elements attractive to predators could have opposing effects on neighbouring nesting habitats, with implications for conservation: Increased predation risk due to higher predator activity (A) or reduced predation risk by distracting predators (B). Here we focus on the placement of conservation measures using flower blocks targeted at Grey Partridges in a Central European Farmland. Based on a three-year camera trap dataset, we investigated effects of landscape structure and composition on mammalian predator activity within flower blocks at two scales (100 m and 500 m radius around the camera) with generalized linear mixed models. Length of linear edge structures, i.e., field block borders, was most important, with a greater availability of linear edge structures leading to a decrease in predator activity at both scales (hypothesis B). Conversely, predator captures at both scales increased with increasing extensive vegetation area (i.e., permanent grassland, flower blocks and fallows) and in proximity to roads, indicating that these may attract predators and increase predator densities (hypothesis A). Our results suggest that a landscape-based approach can mitigate predation risk for ground-nesting birds in flower blocks and analogous conservation measures. Highly structured, small-scale agricultural landscapes seem to be particularly important for reducing mammalian predator activity in flower blocks.

Range Extension and Behavioural Observations of the Flat‐Headed Cusimanse (Crossarchus platycephalus) in Gabon

ABSTRACTThe present work reports on new records of the flat‐headed or Cameroon cusimanse (Crossarchus platycephalus) obtained from camera trap surveys carried out between 2021 and 2023 in central and northwestern Gabon. These records—in the form of high‐definition videos—extend the southern limit of the species' known range by 172 km, as well as provide additional evidence of group living and co‐occurrence with the blue duiker (Philantomba monticola) and moustached guenon (Cercopithecus cephus). These findings suggest that C. platycephalus may be more widespread than previously thought in Gabon, and that a review of existing camera trap datasets could help determine its precise distribution and habitat requirements.

Integrating high-resolution remote sensing and empirical wildlife detection data for climate-resilient corridors across tropical elevational gradients

Corridors are essential tools for promoting biodiversity resilience under climate change. However, corridor design studies are often conducted at spatial scales too coarse to guide implementation by local conservation practitioners. We mapped potential climate-resilient corridors linking lowland to highland protected areas within a highly biodiverse but fragmented landscape of southwestern Costa Rica (6311 km2) using least cost path and circuit theory approaches at high spatial resolution (10 m). We then applied an extensive camera trap dataset of medium-large vertebrates to examine corridor functionality. Although least cost paths (n = 40) were predominantly forested (median = 76 %, range = 57–82 %) and somewhat protected (median = 31 %, range = 3–55 %), they were also highly fragmented. Least cost paths from lowland to highland protected areas traversed medians of 252 forest patches (range = 162–328), 11,186 agriculture patches (range = 822–1,771), and 106 roads (range = 50–252), translating to 2 forest patches, 11 agriculture patches, and 1 road crossed every kilometer. Circuit analyses identified many high-connectivity areas outside of protected areas, including but not limited to least cost paths, but these high-connectivity areas were mostly small forest fragments. Nonetheless, capture rates for medium-to-large mammals at camera traps indicated that many species are currently unlikely to use unprotected, fragmented areas thought to be important for connectivity. In other words, additional conservation and restoration are necessary to establish functional corridors within the landscape. More broadly, this study exemplifies an approach to bridging the gap between regional-scale connectivity analyses and the needs of local practitioners by identifying locations that could be targeted for conservation or restoration within multi-use tropical landscapes.

Being confident in confidence scores: calibration in deep learning models for camera trap image sequences

AbstractIn ecological studies, machine learning models are increasingly being used for the automatic processing of camera trap images. Although this automation facilitates and accelerates the identification step, the results of these models may lack interpretability and their immediate applicability to ecological downstream tasks (e.g. occupancy estimation) remains questionable. In particular, little is known about their calibration, a property that allows confidence scores to be interpreted as probabilities that model's predictions are true. Using a large and diverse European camera trap dataset, we investigate whether deep learning models for species classification in camera trap images are well calibrated. Additionally, as camera traps are often configured to take multiple photos of the same event, we also explore the calibration of predictions aggregated across sequences of images. Finally, we study the effect and the practicality of a post‐hoc calibration method, i.e. temperature scaling, for predictions made at image and sequence levels. Based on five established models and three independent test sets, we show that averaging the logits over the sequence, selecting an appropriate architecture, and optionally using temperature scaling can produce well‐calibrated models. Our findings have clear implication for, for instance, the calculation of error rates or the selection of confidence score thresholds in ecological studies making use of artificial intelligence models.

Feral cat predation of the threatened Pilbara leaf-nosed bat – a key threatening process

Feral cat predation is recognised as a significant threat to Australian wildlife that has contributed to population declines and extinctions. Cat predation of bats worldwide is a growing concern, though there are few studies focusing on Australian species. We consolidated empirical evidence of cat predation on the threatened Pilbara leaf-nosed bat (Rhinonicteris aurantia) in the Pilbara region of Western Australia. Observations in video and camera trap datasets demonstrated repeated predation of R. aurantia by several individual cats at three significant roosts. Between 2020 and 2023, we documented 183–200 bat kills at these sites. We recommend feral cat predation be formally recognised as a key threat to this species. To prevent declines at key colonies that would contribute to a population reduction, ongoing monitoring and proactive control of cats at known R. aurantia roosts should be prioritised. As demonstrated herein, camera trap monitoring is a valid method for quantifying the impact of feral cats at roost sites. Because relatively few individual cats at just a few sites can have a large impact, and most of the larger roost sites of R. aurantia have been discovered, there is potential for amelioration of this threat with multiple established and emerging methods deployed concurrently.

Gentrification drives patterns of alpha and beta diversity in cities

While there is increasing recognition that social processes in cities like gentrification have ecological consequences, we lack nuanced understanding of the ways gentrification affects urban biodiversity. We analyzed a large camera trap dataset of mammals (>500 g) to evaluate how gentrification impacts species richness and community composition across 23 US cities. After controlling for the negative effect of impervious cover, gentrified parts of cities had the highest mammal species richness. Change in community composition was associated with gentrification in a few cities, which were mostly located along the West Coast. At the species level, roughly half (11 of 21 mammals) had higher occupancy in gentrified parts of a city, especially when impervious cover was low. Our results indicate that the impacts of gentrification extend to nonhuman animals, which provides further evidence that some aspects of nature in cities, such as wildlife, are chronically inaccessible to marginalized human populations.

Neotropical mammal responses to megafires in the Brazilian Pantanal.

The increasing frequency and severity of human-caused fires likely have deleterious effects on species distribution and persistence. In 2020, megafires in the Brazilian Pantanal burned 43% of the biome's unburned area and resulted in mass mortality of wildlife. We investigated changes in habitat use or occupancy for an assemblage of eight mammal species in Serra do Amolar, Brazil, following the 2020 fires using a pre- and post-fire camera trap dataset. Additionally, we estimated the density for two naturally marked species, jaguars Panthera onca and ocelots Leopardus pardalis. Of the eight species, six (ocelots, collared peccaries Dicotyles tajacu, giant armadillos Priodontes maximus, Azara's agouti Dasyprocta azarae, red brocket deer Mazama americana, and tapirs Tapirus terrestris) had declining occupancy following fires, and one had stable habitat use (pumas Puma concolor). Giant armadillo experienced the most precipitous decline in occupancy from 0.431 ± 0.171 to 0.077 ± 0.044 after the fires. Jaguars were the only species with increasing habitat use, from 0.393 ± 0.127 to 0.753 ± 0.085. Jaguar density remained stable across years (2.8 ± 1.3, 3.7 ± 1.3, 2.6 ± 0.85/100 km2), while ocelot density increased from 13.9 ± 3.2 to 16.1 ± 5.2/100 km2. However, the low number of both jaguars and ocelots recaptured after the fire period suggests that immigration may have sustained the population. Our results indicate that the megafires will have significant consequences for species occupancy and fitness in fire-affected areas. The scale of megafires may inhibit successful recolonization, thus wider studies are needed to investigate population trends.

Using global remote camera data of a solitary species complex to evaluate the drivers of group formation

The social system of animals involves a complex interplay between physiology, natural history, and the environment. Long relied upon discrete categorizations of "social" and "solitary" inhibit our capacity to understand species and their interactions with the world around them. Here, we use a globally distributed camera trapping dataset to test the drivers of aggregating into groups in a species complex (martens and relatives, family Mustelidae, Order Carnivora) assumed to be obligately solitary. We use a simple quantification, the probability of being detected in a group, that was applied across our globally derived camera trap dataset. Using a series of binomial generalized mixed-effects models applied to a dataset of 16,483 independent detections across 17 countries on four continents we test explicit hypotheses about potential drivers of group formation. We observe a wide range of probabilities of being detected in groups within the solitary model system, with the probability of aggregating in groups varying by more than an order of magnitude. We demonstrate that a species' context-dependent proclivity toward aggregating in groups is underpinned by a range of resource-related factors, primarily the distribution of resources, with increasing patchiness of resources facilitating group formation, as well as interactions between environmental conditions (resource constancy/winter severity) and physiology (energy storage capabilities). The wide variation in propensities to aggregate with conspecifics observed here highlights how continued failure to recognize complexities in the social behaviors of apparently solitary species limits our understanding not only of the individual species but also the causes and consequences of group formation.

Improved Wildlife Recognition through Fusing Camera Trap Images and Temporal Metadata

Camera traps play an important role in biodiversity monitoring. An increasing number of studies have been conducted to automatically recognize wildlife in camera trap images through deep learning. However, wildlife recognition by camera trap images alone is often limited by the size and quality of the dataset. To address the above issues, we propose the Temporal-SE-ResNet50 network, which aims to improve wildlife recognition accuracy by exploiting the temporal information attached to camera trap images. First, we constructed the SE-ResNet50 network to extract image features. Second, we obtained temporal metadata from camera trap images, and after cyclical encoding, we used a residual multilayer perceptron (MLP) network to obtain temporal features. Finally, the image features and temporal features were fused in wildlife identification by a dynamic MLP module. The experimental results on the Camdeboo dataset show that the accuracy of wildlife recognition after fusing the image and temporal information is about 93.10%, which is an improvement of 0.53%, 0.94%, 1.35%, 2.93%, and 5.98%, respectively, compared with the ResNet50, VGG19, ShuffleNetV2-2.0x, MobileNetV3-L, and ConvNeXt-B models. Furthermore, we demonstrate the effectiveness of the proposed method on different national park camera trap datasets. Our method provides a new idea for fusing animal domain knowledge to further improve the accuracy of wildlife recognition, which can better serve wildlife conservation and ecological research.

Maximum temperatures determine the habitat affiliations of North American mammals.

Addressing the ongoing biodiversity crisis requires identifying the winners and losers of global change. Species are often categorized based on how they respond to habitat loss; for example, species restricted to natural environments, those that most often occur in anthropogenic habitats, and generalists that do well in both. However, species might switch habitat affiliations across time and space: an organism may venture into human-modified areas in benign regions but retreat into thermally buffered forested habitats in areas with high temperatures. Here, we apply community occupancy models to a large-scale camera trapping dataset with 29 mammal species distributed over 2,485 sites across the continental United States, to ask three questions. First, are species' responses to forest and anthropogenic habitats consistent across continental scales? Second, do macroclimatic conditions explain spatial variation in species responses to land use? Third, can species traits elucidate which taxa are most likely to show climate-dependent habitat associations? We found that all species exhibited significant spatial variation in how they respond to land-use, tending to avoid anthropogenic areas and increasingly use forests in hotter regions. In the hottest regions, species occupancy was 50% higher in forested compared to open habitats, whereas in the coldest regions, the trend reversed. Larger species with larger ranges, herbivores, and primary predators were more likely to change their habitat affiliations than top predators, which consistently affiliated with high forest cover. Our findings suggest that climatic conditions influence species' space-use and that maintaining forest cover can help protect mammals from warming climates.

A method for automatic identification and separation of wildlife images using ensemble learning

Camera traps have revolutionized wildlife resource surveys by enabling the acquisition of comprehensive ecosystem information. Camera traps usually produce massive images. Snapshot datasets captured in nature reserves proximate to human habitation often comprise a significant volume of human activity images containing humans or livestock. Manually identifying and labeling wildlife images from large-scale datasets is a labor-intensive task that necessitates a substantial number of professionals and incurs expensive personnel costs. By harnessing the capabilities of deep learning technology to automatically differentiate between wildlife and human activity images, ecologists can solely focus on manually labeling wildlife images that account for a minor fraction of the dataset. This strategic shift can significantly reduce personnel expenses and enhance work efficiency. Existing research usually treats human activity images as ordinary categories and utilizes species recognition methods to automatically identify and filter them. However, when human activity images overwhelmingly predominate the dataset, established species recognition methods are susceptible to misclassifying a substantial proportion of wildlife images as human activity images. This misclassification can potentially result in ecologists missing opportunities to discover or observe wildlife. To tackle this challenge, we proposed an ensemble learning method based on a conservative strategy and current mainstream deep learning frameworks to automatically identify wildlife and human activity images. We validated our method on a camera trap dataset from Lasha Mountain (LSM) in Yunnan, China. The experimental results demonstrated that our method automatically identified wildlife images from the dataset with accuracy, recall, and precision of 95.75%, 94.07%, and 83.89%, respectively. This led to an approximately 80% reduction in personnel costs.

Mammalian resistance to megafire in western U.S. woodland savannas

AbstractIncreasingly frequent megafires are dramatically altering landscapes and critical habitats around the world. Across the western United States, megafires have become an almost annual occurrence, but the implication of these fires for the conservation of native wildlife remains relatively unknown. Woodland savannas are among the world's most biodiverse ecosystems and provide important food and structural resources to a variety of wildlife, but they are threatened by megafires. Despite this, the great majority of fire impact studies have only been conducted in coniferous forests. Understanding the resistance and resilience of wildlife assemblages following these extreme perturbations can help inform future management interventions that limit biodiversity loss due to megafire. We assessed the resistance of a woodland savanna mammal community to the short‐term impacts of megafire using camera trap data collected before, during, and after the fire. Specifically, we utilized a 5‐year camera trap data set (2016–2020) from the Hopland Research and Extension Center to examine the impacts of the 2018 Mendocino Complex Fire, California's largest recorded wildfire at the time, on the distributions of eight observed mammal species. We used a multispecies occupancy model to quantify the effects of megafire on species' space use, to assess the impact on species size and diet groups, and to create robust estimates of fire's impacts on species diversity across space and time. Megafire had a negative effect on the detection of certain mammal species, but overall, most species showed high resistance to the disturbance and returned to detection and site use levels comparable to unburned sites by the end of the study period. Following megafire, species richness was higher in burned areas that retained higher canopy cover relative to unburned and burned sites with low canopy cover. Fire management that prevents large‐scale canopy loss is critical to providing refugia for vulnerable species immediately following fire in oak woodlands, and likely other mixed‐forest landscapes.

Forced sympatry? Spatiotemporal interactions of ursids, the Himalayan brown bear and the Asiatic black bear, along a gradient of anthropic disturbances in Western Himalaya

AbstractThe coexistence of species with similar ecological niches implies that species must segregate along one or more niche axes in order to avoid competition. Partitioning of time, space, or resources is an essential niche dimension along which competitor species tend to coexist. The intensity of interspecific competition may vary under varying anthropogenic pressures, but such knowledge is scarce. We investigated the co‐occurrence patterns of two bear species, Asiatic Black Bear Ursus thibetanus and Himalayan Brown Bear Ursus arctos isabellinus, at two sites in the Western Himalayan region of India with similar ecological conditions and varying anthropogenic pressures. Camera trap datasets from the Great Himalayan National Park (GHNP) in Himachal Pradesh and the Bhagirathi basin in Uttarakhand were used to assess the species interactions. We used two species interaction models and kernel density estimates to understand the co‐occurrence pattern of the bear species and their interactions with humans. Our results provide insights into the spatiotemporal behavior of the two bear species and reveal their sympatric and allopatric relationships in two different anthropogenic environments. We found (1) a prominent association between the two bear species (98%) indicating spatial overlap in GHNP, but found spatial separation among the two species (93%) in Bhagirathi basin (2). In GHNP, the spatially co‐occurring bear species were found to be temporarily segregated, whereas in Bhagirathi, they showed spatial avoidance and similar activity patterns (3). In GHNP, human activities in the alpine habitat had a significant negative impact on brown bear presence. The comparative study revealed that the brown bear uses the forested habitats of GHNP due to high anthropogenic disturbances in the alpine areas, whereas no spatial association was observed in the Bhagirathi basin with comparatively low anthropogenic disturbances. The results highlight that, in addition to biological factors, anthropogenic pressures can also influence interspecific competition.

Mid-sized felids threatened by habitat degradation in Southeast Asia

Deforestation and poaching in Southeast Asia have driven a stark decline in the region's apex predators, including large felids like tigers and leopards. Meanwhile, some small felids thrive in the region's human-modified landscapes. The extent to which medium-sized felids cope with anthropogenic disturbances remains poorly understood, but this information is crucial for the conservation of threatened felids and key trophic interactions that maintain high-diversity food webs. Here, we use the largest camera-trap dataset from Southeast Asia to conduct a multi-scale synthesis of the habitat associations of two cryptic felids, the Near-Threatened Asiatic golden cat (Catopuma temminckii) and the Endangered bay cat (Catopuma badia). Unlike many mesopredators, both species exhibited poor tolerance to habitat degradation (i.e. selective logging, edges or fragmentation). The golden cat was positively associated with forest patch size and elevation, and negatively associated with degraded forests, and the bay cat was negatively associated with human population density. Our habitat suitability model suggests that ongoing forest fragmentation and degradation have critically reduced suitable habitat for the golden cat, giving reason to suspect a population decline that calls for a revision of the species' IUCN Red List status to Vulnerable. There is also evidence that the bay cat may be more widely distributed in Borneo than previously thought, including in areas currently threatened by deforestation. Our results indicate both species face a high risk of becoming extirpated from many of the region's remaining forests. In areas where apex predators have been extirpated, these charismatic mid-sized felids can become umbrella species to protect forests with high biodiversity value.

‘The human shield effect’: Human-wildlife co-occurrence patterns in the coffee forests of southwestern Ethiopia

Understanding species (co)-occurrence patterns and how these are affected and mediated by humans is essential for the development of management plans to guide coexistence between humans and wildlife. Here, we evaluated two opposing hypotheses regarding the effects of humans on species occurrence: “humans as super-predators” and “humans as shield”, using an existing camera-trap dataset of mammal species occurrence collected in the coffee forests of southwestern of Ethiopia. We applied a multispecies occupancy modelling framework to explicitly examine co-occurrence patterns between humans, top-predators, prey, and crop-raiders, along a gradient of forest integrity (characterized by forest cover and fragmentation). We examined co-occurrence patterns during both coffee and non-coffee harvest seasons. Our results show partial support for the “humans as shield” hypothesis. We found (i) signs of co-occurrence between humans and prey in areas of low forest integrity during both survey seasons, and between humans and raiders during the coffee-season, (ii) signs of co-occurrence between prey and raiders during both seasons, (iii) no signs of negative or positive co-occurrence between humans and top-predators. Our findings indicate that a possible “shield effect” between humans and prey within a predator space, might be undergoing at the edges of coffee forests. Our findings suggest that incorporating humans as one more species in the ecological system can contribute to shed light into the effects of humans on species occurrence and ultimately contribute to inform management for coexistence.

A Dynamic Occupancy Model for Interacting Species with Two Spatial Scales

AbstractOccupancy models have been extended to account for either multiple spatial scales or species interactions in a dynamic setting. However, as interacting species (e.g., predators and prey) often operate at different spatial scales, including nested spatial structure might be especially relevant to models of interacting species. Here we bridge these two model frameworks by developing a multi-scale, two-species occupancy model. The model is dynamic, i.e. it estimates initial occupancy, colonization and extinction probabilities—including probabilities conditional to the other species’ presence. With a simulation study, we demonstrate that the model is able to estimate most parameters without marked bias under low, medium and high average occupancy probabilities, as well as low, medium and high detection probabilities, with only a small bias for some parameters in low-detection scenarios. We further evaluate the model’s ability to deal with sparse field data by applying it to a multi-scale camera trapping dataset on a mustelid-rodent predator–prey system. Most parameters are estimated with low uncertainty (i.e. narrow posterior distributions). More broadly, our model framework creates opportunities to explicitly account for the spatial structure found in many spatially nested study designs, and to study interacting species that have contrasting movement ranges with camera traps.Supplementary materials accompanying this paper appear online.

Bag of tricks for long-tail visual recognition of animal species in camera-trap images

Camera traps are a method for monitoring wildlife and they collect a large number of pictures. The number of images collected of each species usually follows a long-tail distribution, i.e., a few classes have a large number of instances, while a lot of species have just a small percentage. Although in most cases these rare species are the ones of interest to ecologists, they are often neglected when using deep-learning models because these models require a large number of images for the training. In this work, a simple and effective framework called Square-Root Sampling Branch (SSB) is proposed, which combines two classification branches that are trained using square-root sampling and instance sampling to improve long-tail visual recognition, and this is compared to state-of-the-art methods for handling this task: square-root sampling, class-balanced focal loss, and balanced group softmax. To achieve a more general conclusion, the methods for handling long-tail visual recognition were systematically evaluated in four families of computer vision models (ResNet, MobileNetV3, EfficientNetV2, and Swin Transformer) and four camera-trap datasets with different characteristics. Initially, a robust baseline with the most recent training tricks was prepared and, then, the methods for improving long-tail recognition were applied. Our experiments show that square-root sampling was the method that most improved the performance for minority classes by around 15%; however, this was at the cost of reducing the majority classes' accuracy by at least 3%. Our proposed framework (SSB) demonstrated itself to be competitive with the other methods and achieved the best or the second-best results for most of the cases for the tail classes; but, unlike the square-root sampling, the loss in the performance of the head classes was minimal, thus achieving the best trade-off among all the evaluated methods. Our experiments also show that Swin Transformer can achieve high performance for rare classes without applying any additional method for handling imbalance, and attains an overall accuracy of 88.76% for the WCS dataset and 94.97% for Snapshot Serengeti using a location-based training/test partition. Despite the improvement in the tail classes' performance, our experiments highlight the need for better methods for handling long-tail visual recognition in camera-trap images, since state-of-the-art approaches achieve poor performance, especially in classes with just a few training instances.

Biological invasions disrupt activity patterns of native wildlife: An example from wild pigs

Activity patterns are an important behavioral trait linked to animal interactions within food webs. Biological invasions can have impacts on the behavior of native species through predation and competition, but few studies have determined if invasions lead to shifts in activity patterns of native species. Wild pigs (Sus scrofa) are a generalist, invasive vertebrate in North America that may affect activity patterns of native species through competition and predation. We hypothesized native species' activity patterns would differ from their expected activity pattern (i.e., typical hours of daily activity) and be more variable across sites occupied by wild pigs. We used a U.S. nationwide and a regional camera trapping dataset to determine if wild pig presence affected activity patterns of a nocturnal, diurnal, and crepuscular species: northern raccoon (Procyon lotor), eastern gray squirrel (Sciurus carolinensis), and white-tailed deer (Odocoileus virginianus), respectively. We simulated expected activity patterns of each species and overlapped them with observed activity patterns of respective species in populations where wild pigs were present and where they were not detected. Overlap coefficients were used in generalized linear mixed models to determine if wild pig presence explained deviations from the expected activity pattern. When wild pigs were detected, raccoons remained nocturnal, but their activity patterns were less variable, squirrel activity on the ground was less diurnal and their activity patterns were more variable, and deer activity patterns remained similar in both respects. These results highlight how biological invasions may differentially influence native species activity patterns, potentially altering food web-level interactions.

Divergent effects of lure on multi-species camera-trap detections and quality of photos

The use of attractants to increase detection of target species, such as carnivores, in camera trap studies must be tested for its effectiveness and be carefully planned, as it can lead to misleading comparisons among species. We analyzed a five-year multi-species camera trap dataset of lured and control stations in a protected area in the Amazon rainforest. We aimed to identify the lure effect on a wider range of species and assess whether its use is an efficient strategy to increase the number and the quality of carnivore records. From the 14 vertebrate species analyzed, we found that the use of lures had a negligible effect on nine species, and did not improve the number of records or the detection probability of the carnivores. On the other hand, lured stations attracted omnivores and scavengers (common opossum, black-and-white tegu, and turkey vulture) while had the opposite effect on potential prey species (Black-capped capuchin and Northern Amazon squirrel). We detected a stronger effect of the lure when considering the number of records (relative abundance models) than the probability of detection (occupancy models). The lure increased the proportion of high-quality photos, suitable for individual recognition of jaguars, but only for the first weeks, when the lure was fresh. Therefore, we suggest that the sardine and egg-based lures should be refreshed every-two weeks to ensure greater effectiveness in the quality of photos for jaguar individualization. However, it is important to consider that lure renewing will imply a significant increase in field-related costs and it’s likely to bias other species studies. Thus, we advise that lures should only be used if researchers are certain that the focus is only to increase carnivore data at the expense of using non-target species. Camera traps survey design must be carefully planned a priori and the cost-benefit of lure use and refreshment should be weighed in the study context.

Alpine Musk Deer (Moschus chrysogaster) Adjusts to a Human-Dominated Semi-Arid Mountain Ecosystem.

Comprehension of whether human and livestock presence affects wildlife activity is a prerequisite for the planning and management of humans and livestock in protected areas. Xinglong Mountain Nature Reserve (XMNNR) in northwest China, as a green island in a semi-arid mountain ecosystem, is one of the scattered and isolated areas for Alpine musk deer (AMD), an endangered species. AMD cohabits their latent habitat area with foraging livestock and humans. Hence, habitat management within and outside the distribution areas is crucial for the effective conservation of AMD. We applied camera traps to a dataset of 2 years (September 2018-August 2020) to explore seasonal activity patterns and habitat use and assess the impacts of AMD habits in XMNNR. We investigated AMD responses to livestock grazing and human activities and provided effective strategies for AMD conservation. We applied MaxENT modeling to predict the distribution size under current conditions. The activity patterns of the AMD vary among seasons. The optimum habitat average distance to cultivated land ranges of AMD (150~3300 m during grass period/100~3200 m during withered grass period), distances to the residential area ranges (500~5700 m during the grass period/1000~5300 m during the withered grass period), elevation ranges (2350~3400 m during the grass period/2360~3170 m during the withered grass period), aspect ranges (0~50° and 270~360°), normalized vegetation index ranges (0.64~0.72 during the grass period/0.14~0.60 during the withered grass period), and land cover types (forest, shrub, and grassland). Results present that the predicted distributions of AMD were not confined to the areas reported but also covered other potential areas. The results provide evidence of strong spatial-temporal avoidance of AMD in livestock, but gradually adjusting to human activities. These camera trap datasets may open new opportunities for species conservation in much wider tracts, such as human-dominated landscapes, and may offer guidance and mitigate impacts from livestock, as well as increase artificial forest planting and strengthen the investigation of the potential population resources of AMD.