Explore the vast range of titles available.

Camera‐traps are valuable tools for estimating wildlife population density, and recently developed models enable density estimation without the need for individual recognition. Still, processing and analysis of camera‐trap data are extremely time‐consuming. While algorithms for automated species classification are becoming more common, they have only served as supporting tools, limiting their true potential in being implemented in ecological analyses without human supervision. Here, we assessed the capability of two camera‐trap based models to provide robust density estimates when image classification is carried out by machine learning algorithms. We simulated density estimation with Camera‐Trap Distance Sampling (CT‐DS) and Random Encounter Model (REM) under different scenarios of automated image classification. We then applied the two models to obtain density estimates of three focal species (roe deer Capreolus capreolus, red fox Vulpes vulpes and Eurasian badger Meles meles) in a reserve in central Italy. Species detection and classification was carried out both by the user and machine learning algorithms (respectively, MegaDetector and Wildlife Insights), and all outputs were used to estimate density and ultimately compared. Simulation results suggested that the CT‐DS model could provide robust density estimates even at poor algorithm performances (down to 50% of correctly classified images), while the REM model is more unpredictable and depends on multiple factors. Density estimates obtained from the MegaDetector output were highly consistent for both models with the manually labelled images. While Wildlife Insights' performance differed greatly between species (recall: badger = 0.15; roe deer = 0.56; fox = 0.75), CT‐DS estimates did not vary significantly; on the contrary, REM systematically overestimated density, with little overlap in standard errors. We conclude that CT‐DS and REM models can be robust to the loss of images when machine learning algorithms are used to identify animals, with the CT‐DS being an ideal candidate for applications in a fully unsupervised framework. We propose guidelines to evaluate when and how to integrate machine learning in the analysis of camera‐trap data for density estimation, further strengthening the applicability of camera‐traps as a cost‐effective method for density estimation in (spatially and temporally) extensive multi‐species monitoring programmes.

Camera traps are used by scientists and natural resource managers to acquire ecological data, and the rapidly increasing camera trapping literature highlights how popular this technique has become. Nevertheless, the methodological information reported in camera trap publications can vary widely, making replication of the study difficult. Here we propose a series of guiding principles for reporting methods and results obtained using camera traps. Attributes of camera trapping we cover include: (i) specifying the model(s) of camera traps(s) used, (ii) mode of deployment, (iii) camera settings, and (iv) study design. In addition to suggestions regarding best practice data coding and analysis, we present minimum principles for standardizing information that we believe should be reported in all peer-reviewed papers. Standardised reporting enables more robust comparisons among studies, facilitates national and global reviews, enables greater ease of study replication, and leads to improved wildlife research and management outcomes.

Camera traps, combined with AI, have emerged to achieve automated, scalable biodiversity monitoring. However, passive infrared (PIR) sensors that typically trigger camera traps are poorly suited for detecting small, fast‐moving ectotherms such as insects. Insects comprise over half of all animal species and are key components of ecosystems and agriculture. The need for an appropriate and scalable insect camera trap is critical in the wake of concerning reports of declines in insect populations. This study proposes an alternative to the PIR trigger: ultra‐lightweight convolutional neural networks running on low‐powered hardware to detect insects in a continuous stream of captured images. We train such models to distinguish insect images from backgrounds. Our design achieves zero latency between trigger and image capture. Our models are rigorously tested and achieve high accuracy ranging from 91.8% to 96.4% AUROC on test data and 58.8% to 87.2% AUROC on field data from distributions unseen during training. The high specificity of our models ensures minimal saving of false positive images, maximising deployment storage efficiency. High recall scores indicate a minimal false negative rate, maximising insect detection. Analysis using saliency maps shows the learned representation of our models to be robust, with low reliance on spurious background features. Our method is also shown to operate deployed on off‐the‐shelf, low‐powered microcontroller units, consuming a maximum power draw of less than 300 mW. This paves the way for scalable systems with longer deployment times. Overall, we fully define the properties of a successful trigger for camera traps and show how lightweight AI models, made bespoke for efficient hardware, can be realised with a specific focus on insect ectotherms. We provide these models to the community alongside a complete codebase for future modifications, and we demonstrate how they can be deployed on an example ESP32‐S3 microcontroller platform. This step potentiates a major advancement for ectotherm camera traps and insect monitoring.

The camera trap is a powerful research tool that has a wide range of ecological applications and facilitates monitoring over large spatial and temporal scales. To improve the reliability of camera trap studies and provide more knowledge on camera performance, we evaluated three aspects of camera traps that researchers should consider – camera height, blank images and missed detections. We deployed 20 camera stations, each consisting of one low camera (0.6 m) and two adjacent high cameras (3 m). We tested for differences in detection rates and blank images between camera heights. We calculated missed detections using the two high cameras and used a subset of cameras (n = 14) to examine whether missed detections were caused by late triggers or failed triggers. We found that placing cameras high to minimize theft and damage did not influence detection rates. There were, however, more blank images which can increase the time required for analysis. These blank images increased as temperature increased. Missed detections were primarily the result of failed triggers and increased as species size decreased. Failed detections are particularly significant for distribution surveys of low‐density species. Detection at the camera is imperfect, even when working with larger species. We evaluated three aspects of camera traps that researchers should consider – camera height, blank images and missed detections. We deployed 20 camera stations, each consisting of one low camera (0.6 m) and two adjacent high cameras (3 m). We found that placing cameras high did not influence detection rates and high cameras generated more blank images which increased as temperature increased. Missed detections increased as species size decreased and were primarily the result of failed triggers as opposed to late triggers. We recommend that cameras used for documenting species presence be placed at 2.5 m and above and urge researchers to avoid assumptions of certain detectability inside the detection zone, even when working with larger species.

We present ClassifyMe a software tool for the automated identification of animal species from camera trap images. ClassifyMe is intended to be used by ecologists both in the field and in the office. Users can download a pre-trained model specific to their location of interest and then upload the images from a camera trap to a laptop or workstation. ClassifyMe will identify animals and other objects (e.g., vehicles) in images, provide a report file with the most likely species detections, and automatically sort the images into sub-folders corresponding to these species categories. False Triggers (no visible object present) will also be filtered and sorted. Importantly, the ClassifyMe software operates on the user’s local machine (own laptop or workstation)—not via internet connection. This allows users access to state-of-the-art camera trap computer vision software in situ, rather than only in the office. The software also incurs minimal cost on the end-user as there is no need for expensive data uploads to cloud services. Furthermore, processing the images locally on the users’ end-device allows them data control and resolves privacy issues surrounding transfer and third-party access to users’ datasets.

1. The abundance and distribution of a species are affected by processes which operate at multiple scales. Large-scale dynamics are increasingly recognized in conservation responses such as metapopulation management, transfrontier protected areas and softening the agricultural matrix. Landscape-scale monitoring is needed both to inform and judge their efficacy. In this Special Profile we address some of the challenges presented by monitoring at the landscape scale, how models of species distribution can be used to inform policy, and we discuss how monitoring at the global-scale could be approached. 2. Collecting data over a large area is inherently costly, so methods which can provide robust information at low-cost are particularly valuable. We present two papers which test low-cost approaches against more data-hungry methods (indices of abundance vs. direct density estimates, and species distribution models built from presence-only vs. presence/absence data). 3. Occupancy modelling is a useful approach for landscape-scale monitoring due to the relatively low-cost of collecting detection/non-detection data. We discuss challenges, such as non-random sampling locations and periodical unavailability for detection, in using detection/non-detection data for monitoring species distribution. Such data can also provide estimates of abundance and we show how existing models have been modified to allow the abundance of multiple species to be estimated simultaneously. 4. Models of species distribution can be used to project likely future scenarios and thus inform conservation planning where distributions are likely to change because of climate change or changing disturbance patterns. We also discuss how an optimization framework can be used to make efficient management decisions for invasive species management in the light of imperfect information. 5. Synthesis and applications. Monitoring is needed for many purposes including auditing past management decisions and informing future choices. Much monitoring data are collected at the site scale, although management authorities increasingly recognize landscape-scale dynamics. Recent global targets for conservation require monitoring which can report trends at the global-scale. Integrating data collected at a variety of scales to draw robust inference at the scale required is a challenge which deserves more attention from applied ecologists.

Remote monitoring of communal marking sites, or latrines, provides a unique opportunity to observe undisturbed scent marking behaviour of African wild dogs (Lycaon pictus). We used remote camera trap observations in a natural experiment to test behavioural scent mark responses to rivals (either familiar neighbours or unfamiliar strangers), to determine whether wild dogs exhibit the “dear enemy” or “nasty neighbour” response. Given that larger groups of wild dogs represent a greater threat to smaller groups, including for established residents, we predicted that the overarching categories “dear enemy” vs. “nasty neighbour” may be confounded by varying social statuses that exists between individual dyads interacting. Using the number of overmarks as a metric, results revealed an interaction between sender and receiver group size irrespective of familiarity consistent with this prediction: in general, individuals from large resident packs overmarked large groups more than they overmarked smaller groups, whereas individuals from smaller packs avoided overmarking larger groups, possibly to avoid detection. Monitoring a natural system highlights variables such as pack size that may be either overlooked or controlled during scent presentation experiments, influencing our ability to gain insights into the factors determining territorial responses to rivals. Using camera traps to generate behavioural data to act as a “natural experiment” to determine African wild dog responses to rivals (unfamiliar strangers vs. familiar neighbours). Authors found that group size impacts behavioural responses, and highlights shortcomings of scent presentation experiments in determining territorial responses to rivals.

The forest‐like characteristics of agroforestry systems create a unique opportunity to combine agricultural production with biodiversity conservation in human‐modified tropical landscapes. The cacao‐growing region in southern Bahia, Brazil, encompasses Atlantic forest remnants and large extensions of agroforests, locally known as cabrucas, and harbors several endemic large mammals. Based on the differences between cabrucas and forests, we hypothesized that: (1) non‐native and non‐arboreal mammals are more frequent, whereas exclusively arboreal and hunted mammals are less frequent in cabrucas than forests; (2) the two systems differ in mammal assemblage structure, but not in species richness; and (3) mammal assemblage structure is more variable among cabrucas than forests. We used camera‐traps to sample mammals in nine pairs of cabruca‐forest sites. The high conservation value of agroforests was supported by the presence of species of conservation concern in cabrucas, and similar species richness and composition between forests and cabrucas. Arboreal species were less frequently recorded, however, and a non‐native and a terrestrial species adapted to open environments (Cerdocyon thous) were more frequently recorded in cabrucas. Factors that may overestimate the conservation value of cabrucas are: the high proportion of total forest cover in the study landscape, the impoverishment of large mammal fauna in forest, and uncertainty about the long‐term maintenance of agroforestry systems. Our results highlight the importance of agroforests and forest remnants for providing connectivity in human‐modified tropical forest landscapes, and the importance of controlling hunting and dogs to increase the value of agroforestry mosaics. Resumo Por suas similaridades à florestas nativas, sistemas agroflorestais representam uma opotunidade única de conciliar produção agrícola e conservação de biodiversidade em paisagens tropicais modificadas pelo homem. A região cacaueira do sul da Bahia, Brasil, possui remanescentes de floresta Atlântica e extensas áreas de agroflorestas, localmente conhecidas como cabrucas, e abriga diversos mamíferos de maior porte em risco de extinção. Com base em diferenças entre cabrucas e florestas, hipotetizamos que: (1) mamíferos não nativos e não arborícolas são mais frequentes, ao passo que mamíferos exclusivamente arborícolas e caçados são menos frequentes em cabrucas do que em florestas; (2) os dois sistemas diferem quanto à estrutura das assembleias de mamíferos, mas não quanto à riqueza de espécies; e (3) a estrutura das assembleias de mamíferos é mais variável entre cabrucas do que entre florestas. Utilizamos armadilhas fotográficas para amostrar mamíferos de maior porte em nove pares (cabruca‐floresta) de sítios. O alto valor de conservação das agroflorestas foi apoiado pela presença de espécies de interesse para a conservação nas cabrucas e pela similaridade na riqueza e composição de espécies entre florestas e cabrucas. No entanto, espécies arborícolas foram registradas com menor frequência e uma espécies não nativa e uma terrestre adaptada a ambientes abertos (Cerdocyon thous) foram registradas com maior frequência nas cabrucas. Fatores que podem superestimar o valor de conservação das cabrucas são: a alta proporção de cobertura florestal na paisagem de estudo, o empobrecimento da fauna de mamíferos de maior porte nas florestas e a incerteza sobre a manutenção dos sistemas agroflorestais no longo prazo. Nossos resultados destacam a importância das agroflorestas e remanescentes florestais para promover conectividade em paisagens florestais tropicais dominadas pelo homem e a importância do controle da caça e de populações de cães domésticos para aumentar o valor de conservação de mosaicos agroflorestais.

The jaguarundi (Puma yagouaroundi) is a small felid with a historical range from central Argentina through southern Texas. Information on the current distribution of this reclusive species is needed to inform recovery strategies in the United States where its last record was in 1986 in Texas. From 2003 to 2021, we conducted camera‐trap surveys across southern Texas and northern Tamaulipas, México to survey for medium‐sized wild cats (i.e., ocelots [Leopardus pardalis], bobcats [Lynx rufus], and jaguarundi). After 350,366 trap nights at 685 camera sites, we did not detect jaguarundis at 16 properties or along 2 highways (1050 km2) in Texas. However, we recorded 126 jaguarundi photographic detections in 15,784 trap nights on 2 properties (125.3 km2) in the northern Sierra of Tamaulipas, Tamaulipas, México. On these properties, latency to detection was 72 trap nights, with a 0.05 probability of detection per day and 0.73 photographic event rate every 100 trap nights. Due to a lack of confirmed class I sightings (e.g., specimen, photograph) in the 18 years of this study, and no other class I observations since 1986 in the United States, we conclude that the jaguarundi is likely extirpated from the United States. Based on survey effort and results from México, we would have expected to detect jaguarundis over the course of the study if still extant in Texas. We recommend that state and federal agencies consider jaguarundis as extirpated from the United States and initiate recovery actions as mandated in the federal jaguarundi recovery plan. These recovery actions include identification of suitable habitat in Texas, identification of robust populations in México, and re‐introduction of the jaguarundi to Texas. From 2003 to 2021, we conducted camera‐trap surveys across southern Texas and northern Tamaulipas, México to survey for wild cats (i.e., jaguarundi, ocelots, bobcats). After 350,366 trap nights at 685 camera sites, we failed to detect jaguarundis at 16 properties and along 2 highways (1050 km2) in Texas, but had 126 detections in Mexico. Due to a lack of confirmed class I sightings (e.g., specimen, photograph) in the 18 years of this study, and no other class I observations or roadkill since 1986, we conclude that the jaguarundi is likely extirpated from the United States.

We present evidence of scent marking in the large‐antlered muntjac (Muntiacus vuquangensis). Given the importance of scent marking in individual recognition among ungulates, this behavior may serve to communicate the fitness cost of antagonistic interactions among rival males and could serve as a mechanism for mate assessment among females. We document the first recorded evidence of scent marking in the large‐antlered muntjac (Muntiacus vuquangensis), a little‐known and threatened deer endemic to the Annamites ecoregion of Indochina. We discuss the function of this behavior within a wider behavioral and evolutionary ecology framework for mammals living in tropical forests.