Camera trap arrays improve detection probability of wildlife: Investigating study design considerations using an empirical dataset

Temporal niche partitioning among sympatric wild and domestic ungulates between warm and cold seasons

The coexistence of sympatric species with similar ecological niches has been a central issue in ecology. Clarifying the daily activity patterns of sympatric wild ungulates can help understand their temporal niche differentiation and the mechanisms of coexistence, providing information for their conservation. The Baotianman National Nature Reserve in northern China is rich in wild ungulates, but little is known about the daily activity patterns of wild ungulates in the area, making it difficult to develop effective conservation strategies. We studied five representative wild ungulates (i.e. forest musk deer, Chinese goral, Reeve's muntjac, Siberian roe deer, and wild boar) of the region using camera-trapping data, focusing on the seasonal daily activity patterns and effects of seasonal grazing of domestic sheep, to reveal their coexistence based on temporal ecological niche differentiation. Comparative analyses of the seasonal daily activity showed that forest musk deer exhibited a single-peak activity in the warm season. Other ungulates exhibited multipeak activity. All five ungulates differed significantly in daily activity patterns. Notably, wild boar and Reeve's muntjac showed high overlap coefficients between the cold and warm seasons. In both cold and warm seasons, the five wild ungulates and domestic sheep displayed low overlap in their daily activity rhythms potentially indicating temporal ecological niche differentiation. The results suggest that temporal isolation might be a strategy for wild ungulates to avoid domestic sheep and reduce interspecific competition, and that temporal ecological niche differentiation potentially promoted the coexistence among the studied sympatric ungulates. This understanding may provide new insights for the development of targeted conservation strategies.

American marten occupancy and activity patterns at the southern extent of their range in the eastern United States

The relatively recent rediscovery of an American marten (Martes americana) population that was reintroduced over 30 years ago in southern Vermont provides an opportunity to investigate the relative importance of other mesocarnivores, and forest stand (e.g., DBH, downed logs, vertical structure) and habitat variables to their presence on the Green Mountain National Forest. Marten are state-listed as an endangered species in Vermont and occur there at the southern extent of their range in the eastern United States. We collected detection data from camera surveys in 5 km2 units between 2019 and 2021 (December-April; n = 40 units, 238 cameras). We examined activity patterns and applied an occupancy modeling framework to the detection data to assess the relative importance of covariates at unit and camera levels and assess interactions of marten with other mesocarnivores. We did not find any unit-level occupancy models with significant covariates that were better supported than the base model in the single-season unit-level analysis. Distance to the nearest release site was the covariate most supported for detectability at both spatial scales, and marten occupancy at the camera level was positively influenced by the amount of canopy cover. Two species interaction models did not indicate any positive or negative association beyond random with other mesocarnivores and activity patterns among mesocarnivores had substantial overlap. Marten recovery since the time of the reintroduction appears slow, and even 30 years later, the marten distribution is limited and suggests that dispersal is restricted at some level. We recommend a further investigation of the possible impact of other mesocarnivores to juvenile survival or other vital demographic rate (e.g., recruitment) in marten that were not explicitly measured in this study.

What factors influence the rediscovery of lost tetrapod species?

We created a database of lost and rediscovered tetrapod species, identified patterns in their distribution and factors influencing rediscovery. Tetrapod species are being lost at a faster rate than they are being rediscovered, due to slowing rates of rediscovery for amphibians, birds and mammals, and rapid rates of loss for reptiles. Finding lost species and preventing future losses should therefore be a conservation priority. By comparing the taxonomic and spatial distribution of lost and rediscovered tetrapod species, we have identified regions and taxa with many lost species in comparison to those that have been rediscovered-our results may help to prioritise search effort to find them. By identifying factors that influence rediscovery, we have improved our ability to broadly distinguish the types of species that are likely to be found from those that are not (because they are likely to be extinct). Some lost species, particularly those that are small and perceived to be uncharismatic, may have been neglected in terms of conservation effort, and other lost species may be hard to find due to their intrinsic characteristics and the characteristics of the environments they occupy (e.g. nocturnal species, fossorial species and species occupying habitats that are more difficult to survey such as wetlands). These lost species may genuinely await rediscovery. However, other lost species that possess characteristics associated with rediscovery (e.g. large species) and that are also associated with factors that negatively influence rediscovery (e.g. those occupying small islands) are more likely to be extinct. Our results may foster pragmatic search protocols that prioritise lost species likely to still exist.

Comparing wildlife habitat suitability models based on expert opinion with camera trap detections

Expert knowledge is used in the development of wildlife habitat suitability models (HSMs) for management and conservation decisions. However, the consistency of such models has been questioned. Focusing on 1 method for elicitation, the analytic hierarchy process, we generated expert-based HSMs for 4 felid species: 2 forest specialists (ocelot [Leopardus pardalis] and margay [Leopardus wiedii]) and 2 habitat generalist species (Pampas cat [Leopardus colocola] and puma [Puma concolor]). Using these HSMs, species detections from camera-trap surveys, and generalized linear models, we assessed the effect of study species and expert attributes on the correspondence between expert models and camera-trap detections. We also examined whether aggregation of participant responses and iterative feedback improved model performance. We ran 160 HSMs and found that models for specialist species showed higher correspondence with camera-trap detections (AUC [area under the receiver operating characteristic curve] >0.7) than those for generalists (AUC < 0.7). Model correspondence increased as participant years of experience in the study area increased, but only for the understudied generalist species, Pampas cat (β = 0.024 [SE 0.007]). No other participant attribute was associated with model correspondence. Feedback and revision of models improved model correspondence, and aggregating judgments across multiple participants improved correspondence only for specialist species. The average correspondence of aggregated judgments increased as group size increased but leveled off after 5 experts for all species. Our results suggest that correspondence between expert models and empirical surveys increases as habitat specialization increases. We encourage inclusion of participants knowledgeable of the study area and model validation for expert-based modeling of understudied and generalist species.

Camera traps reveal that terrestrial predators are pervasive at riverscape cold-water thermal refuges

Perceived predation risks by terrestrial predators are major ecological forces in aquatic systems, particularly for aggregating fish. Riverscape thermal refuges are discrete, localized cold-water patches where fish temporarily aggregate to buffer against heat events. Predation pressures by terrestrial predators at thermal refuges may decrease the thermoregulatory benefits of refuge use, but quantifying such effects can be challenging and controversial when sampling can impose additional stress on fish. We passively monitored terrestrial predator visitation patterns and predation at four thermal refuges in the Housatonic River, Connecticut, USA, between May 18th and September 29th, 2022, with camera traps, a common wildlife monitoring method. Specifically, we (1) assessed diel visitation patterns by different categories of terrestrial predators at thermal refuges and determined if patterns varied among predator categories or with prevailing environmental conditions, and (2) estimated the probability of predation by hour of the day combined across all predator categories, quantifying general predation pressures at refuges. We detected at least one terrestrial predator at a thermal refuge each day, and mean hourly visitation rates (count/h) were highly variable across predator categories and sampling dates. The most supported generalized additive mixed model indicated that terrestrial predator visitation rates (count/h/day) varied with mean daily river discharge and water temperature differential, and relationships differed across categories of terrestrial predators. We observed 22 separate predation attempts on thermoregulating salmonids and predicted that the probability of predation by any terrestrial predator increased from 0.002 to 0.017 throughout a 24 h day (p = .004). Camera traps provided novel evidence that terrestrial predators are pervasive at riverine thermal refuges, which is relevant for refuge conservation and management globally. We recommend the implementation of a coordinated monitoring network across riverine thermal refuges using camera traps, further enriching our ecological understanding of cumulative predator effects in refuges across complex riverscapes.

N-mixture model-based estimate of relative abundance of sloth bear (Melursus ursinus) in response to biotic and abiotic factors in a human-dominated landscape of central India

Reliable estimation of abundance is a prerequisite for a species' conservation planning in human-dominated landscapes, especially if the species is elusive and involved in conflicts. As a means of population estimation, the importance of camera traps has been recognized globally, although estimating the abundance of unmarked, cryptic species has always been a challenge to conservation biologists. This study explores the use of the N-mixture model with three probability distributions, i.e., Poisson, negative binomial (NB) and zero-inflated Poisson (ZIP), to estimate the relative abundance of sloth bears (Melursus ursinus) based on a camera trapping exercise in Sanjay Tiger Reserve, Madhya Pradesh from December 2016 to April 2017. We used environmental and anthropogenic covariates to model the variation in the abundance of sloth bears. We also compared null model estimates (mean site abundance) obtained from the N-mixture model to those of the Royle-Nichols abundance-induced heterogeneity model (RN model) to assess the application of similar site-structured models. Models with Poisson distributions produced ecologically realistic and more precise estimates of mean site abundance (λ = 2.60 ± 0.64) compared with other distributions, despite the relatively high Akaike Information Criterion value. Area of mixed and sal forest, the photographic capture rate of humans and distance to the nearest village predicted a higher relative abundance of sloth bears. Mean site abundance estimates of sloth bears obtained from the N-mixture model (Poisson distribution) and the RN model were comparable, indicating the overall utility of these models in this field. However, density estimates of sloth bears based on spatially explicit methods are essential for evaluating the efficacy of the relatively more cost-effective N-mixture model. Compared to commonly used index/encounter-based methods, the N-mixture model equipped with knowledge on governing biotic and abiotic factors provides better relative abundance estimates for a species like the sloth bear. In the absence of absolute abundance estimates, the present study could be insightful for the long-term conservation and management of sloth bears.

Understanding External Influences on Target Detection and Classification Using Camera Trap Images and Machine Learning

Using machine learning (ML) to automate camera trap (CT) image processing is advantageous for time-sensitive applications. However, little is currently known about the factors influencing such processing. Here, we evaluate the influence of occlusion, distance, vegetation type, size class, height, subject orientation towards the CT, species, time-of-day, colour, and analyst performance on wildlife/human detection and classification in CT images from western Tanzania. Additionally, we compared the detection and classification performance of analyst and ML approaches. We obtained wildlife data through pre-existing CT images and human data using voluntary participants for CT experiments. We evaluated the analyst and ML approaches at the detection and classification level. Factors such as distance and occlusion, coupled with increased vegetation density, present the most significant effect on DP and CC. Overall, the results indicate a significantly higher detection probability (DP), 81.1%, and correct classification (CC) of 76.6% for the analyst approach when compared to ML which detected 41.1% and classified 47.5% of wildlife within CT images. However, both methods presented similar probabilities for daylight CT images, 69.4% (ML) and 71.8% (analysts), and dusk CT images, 17.6% (ML) and 16.2% (analysts), when detecting humans. Given that users carefully follow provided recommendations, we expect DP and CC to increase. In turn, the ML approach to CT image processing would be an excellent provision to support time-sensitive threat monitoring for biodiversity conservation.

Human land‐use effects on mammalian mesopredator occupancy of a northeastern Connecticut landscape

Mammalian mesopredators—mid‐sized carnivores—are ecologically, economically, and socially important. With their adaptability to a variety of habitats and diets, loss of apex predators, and forest regrowth, many of these species are increasing in number throughout the northeastern United States. However, currently the region is seeing extensive landscape alterations, with an increase in residential and industrial development, especially at the expense of existing forest and small‐scale farmland. We sought to understand how important an existing mosaic of working lands (timberland and farmland) in a forested landscape is to mesopredator species. We did this by studying mesopredator occupancy across three land uses (or habitat types): forest reserve (protected), timber harvest (shelterwood cuts), and field (both crop yielding and fallow) in and around a 3200‐ha forest in northeastern Connecticut. We examined coyote (Canis latrans), bobcat (Lynx rufus), fisher (Pekania pennanti), and raccoon (Procyon lotor) occupancy using paired camera traps across juxtaposed reserve, shelterwood, and field units from April 2018 to March 2019. We created a priori habitat variable models for each species and season, as well as analyzed the impact of habitat types on each species. Throughout the year bobcats were positively associated with foliage height diversity and had the highest use in shelterwoods and lowest use in fields. Land use utilization varied seasonally for coyotes and raccoons, with higher use of fields than reserves and shelterwoods for half the year and no difference between land uses and the other half. Both species were not strongly associated with any particular habitat variables. Reserve forest was moderate to highly used by all species for at least half the year, and highly use year‐round by fishers. Our findings reveal that a mosaic of intact forest and working lands, timber harvest, and agriculture can support mesopredator diversity.

Hare today, gone tomorrow: the role of interspecific competition in shaping riverine rabbit occurrence

Effective conservation, particularly of threatened species, requires an understanding of both abiotic and biotic drivers of distribution. In the case of one of Africa’s most endangered mammals, the riverine rabbit Bunolagus monticularis, only environmental covariates of presence have been used to provide coarse predictions of their distribution. Two potential competitors, namely scrub hare Lepus saxatilis and cape hare L. capensis, have significant (>90%) range overlap with the riverine rabbit, yet little is known about how these species interact. We used multi-species occupancy models, which model co-occurrence as a function of environmental variables, to assess the spatial response of riverine rabbits to both species of hare in Sanbona Wildlife Reserve, South Africa. We also examined temporal overlap between riverine rabbits and hares. Camera trapping data were collected from 150 camera traps distributed in clusters of 5 cameras at 30 independent sites, covering 223.24 km2. Contrary to prior studies, we found that riverine rabbits were not restricted to riparian habitat, and that their occurrence was conditional on hare absence and was negatively affected by terrain ruggedness. Whilst hare occurrence was independent of terrain ruggedness, it was negatively affected by rabbit presence. Activity patterns revealed high temporal overlap between hares and rabbits (Δ = 0.828, CI = 0.745-0.940); however, neither species co-occurred at any given site. Our results suggest that conservation management has greatly underestimated the importance of competition with other lagomorphs in understanding riverine rabbit occurrence.

The relationship between landscape features and domestic species on the occupancy of native mammals in urban forests

Landscape features such as roads, hiking trails, and forest edge are prevalent features of urban forests. These features can negatively influence biodiversity through fragmentation and facilitate the penetration of invasive species, particularly predators such as domestic dogs and cats. This study examined how these landscape features affect the distribution of native and domestic mammals and how domestic species affect the occupancy and activity patterns of native mammals. To this end, we conducted camera trap surveys in forested urban and suburban parks in Indiana and Kentucky, USA, to record the presence of native species (deer, raccoons, coyote, and opossums) and non-native predators (domesticated cats and dogs) in relation to distance to nearest roads, hiking trails, and forest edge. We found negative correlations between proximity to roads, hiking trails, and forest edge and the occupancy of the native species with this effect extending up to 300 m. We also found evidence that the presence of dog and cat was negatively correlated with the occupancy for all native species. We recommend park management consider the impact of roads, hiking trails, and forest edge when designing parks and the enforcement of dog leash regulations for the conservation of large mammals in urban parks.

Intraguild dynamics of understudied carnivores in a human-altered landscape

Mesocarnivores constitute a diverse and often abundant group of species, which are increasingly occupying hweigher trophic levels within multi-use landscapes. Yet, we know relatively little about their interactions with each other, especially in human-altered areas. Using camera trap data collected in a forestry concession in the Greater Gorongosa ecosystem of central Mozambique, we examined the spatiotemporal relationships and potential for intraguild competition among three understudied African carnivores: African civets (Civettictis civetta), bushy-tailed mongooses (Bdeogale crassicauda), and large-spotted genets (Genetta maculata). After accounting for habitat preferences and tolerance to anthropogenic factors, we found that African civets and bushy-tailed mongooses avoid each other spatially and temporally. Additionally, civets and mongooses were also both more likely to use sites farther away from human settlements, possibly decreasing the total available habitat for each species if competition is driving this spatial partitioning. In contrast, we did not find evidence for spatial or temporal partitioning between large-spotted genets and African civets, but bushy-tailed mongooses altered their activity patterns where they co-occurred with genets. Our study contributes to scant ecological knowledge of these mesocarnivores and adds to our understanding of community dynamics in human-altered ecosystems.

Component processes of detection probability in camera-trap studies: understanding the occurrence of false-negatives

Camera-trap studies in the wild record true-positive data, but data loss from false-negatives (i.e. an animal is present but not recorded) is likely to vary and widely impact data quality. Detection probability is defined as the probability of recording an animal if present in the study area. We propose a framework of sequential processes within detection – a pass, trigger, image registration, and images being of sufficient quality. Using closed-circuit television (CCTV) combined with camera-trap arrays we quantified variation in, and drivers of, these processes for three medium-sized mammal species. We also compared trigger success of wet and dry otter Lutra lutra, as an example of a semiaquatic species. Data loss from failed trigger, failed registration and poor capture quality varied between species, camera-trap model and settings, and were affected by different environmental and animal variables. Distance had a negative effect on trigger probability and a positive effect on registration probability. Faster animals had both reduced trigger and registration probabilities. Close passes (1 m) frequently did not generate triggers, resulting in over 20% data loss for all species. Our results, linked to the framework describing processes, can inform study design to minimize or account for data loss during analysis and interpretation.

Elucidating Patterns in the Occurrence of Threatened Ground-Dwelling Marsupials Using Camera-Traps

Simple Summary

Being able to effectively monitor the continued plight of highly vulnerable animals against management efforts over time is critical for their conservation. In south-eastern New South Wales, Australia, we used a camera trapping array to collect baseline information about patterns of occurrence of three threatened native ground-dwelling marsupials of conservation interest: the long-nosed bandicoot (Perameles nasuta), long-nosed potoroo (Potorous tridactylus) and southern brown bandicoot (Isoodon obesulus). Over a four-year period, detections of the two bandicoots were more erratic and less predictable than that of the potoroo, resulting in higher uncertainty about occupancy estimates and adequacy of sampling effort. The detection probability of each bandicoot species and that of the potoroo differed variously with structural complexity of vegetation. Detection probability of the southern brown bandicoot was highest where ground cover was most dense and shrub cover most open. The reverse pattern was found for the long-nosed bandicoot. Finally, the detection probability of the long-nosed potoroo was highest where ground and shrub cover was densest. Future camera trapping monitoring efforts need to take better account of these nuances and be flexible to including additional sampling for at least the two bandicoots. In short, when it comes to monitoring approach, one size doesn’t fit all.

Abstract

Establishing trends in endangered fauna against management efforts is a key but often challenging enterprise. Camera-traps offer a new and literal window into monitoring many different mammalian species. Getting it right demands seeking baseline information about how often target species interact with these devices, prior to setting a long-term monitoring strategy. We used a camera-trap array to collect detection data on three species of threatened ground-dwelling marsupials in south-eastern mainland Australia. Over a four-year period, occupancy estimates for two species of bandicoot (southern brown bandicoot Isoodon obesulus and long-nosed bandicoot Perameles nasuta) and a single species of rat-kangaroo (long-nosed potoroo Potorous tridatylus) were generated. These estimates were variously robust depending on visitation history, but nevertheless indicated persistence of these rare and otherwise under threat species. Detection probability for each species differed between study areas, type of management and with complexity of ground and shrub vegetation cover. The relationship between detection and vegetation structure dictated that survey effort was only robust where conditions were optimal for a given species. Outside of that further survey effort would be required to have confidence in survey outcome. In the future this would demand a different sampling strategy, be that through lengthening survey time or adding additional camera units at sites.

Assessing arrays of multiple trail cameras to detect North American mammals

Motion triggered camera traps are an increasingly popular tool for wildlife research and can be used to survey for multiple species simultaneously. As with all survey techniques, it is crucial to conduct camera trapping research following study designs that include adequate spatial and temporal replication, and sufficient probability of detecting species presence. The use and configuration of multiple camera traps within a single survey site are understudied considerations that could have a substantial impact on detection probability. Our objective was to test the role that camera number (one, two or three units), and spacing along a linear transect (100 m or 150 m), have on the probability of detecting a species given it is present. From January to March, 2017 we collected data on six mammal species in Maine, USA: coyote (Canis latrans), fisher (Pekania pennanti), American marten (Martes americana), short-tailed weasel (Mustela erminea), snowshoe hare (Lepus americanus), and American red squirrel (Tamiasciurus hudsonicus). We used multi-scale occupancy modelling to compare pooled detection histories of different configuration of five cameras deployed at the same survey site (n = 32), and how the configuration would influence the probability of detecting a species given it was available at the site. Across all six species, we found substantial increases in probability of detection as the number of cameras increased from one to two (22 to 400 percent increase), regardless of the spacing between cameras. For most species the magnitude of the increase was less substantial when adding a third camera (4 to 85 percent increase), with coyote and snowshoe hare showing a pronounced effect. The influence of survey station features also varied by species. We suggest that using pooled data from two or three cameras at a survey site is a cost effective approach to increase detection success over a single camera.

Maximising camera trap data: Using attractants to improve detection of elusive species in multi-species surveys

Camera traps are a key tool in ecological studies, and are increasingly being used to understand entire communities. However, robust inferences continue to be hampered by low detection of rare and elusive species. Attractants can be used to increase detection rates, but may also alter behaviour, and little research has evaluated short-term, localized response to the presence of attractants. We conducted three camera trap surveys in Kibale National Park, Uganda, using food baits and scent lures (“attractants”) at each camera station to entice small carnivores to pass in front of camera stations. To examine the interrelationship between scavenging and response to attractants, we also placed camera traps at five food refuse pits. We modelled the effect of attractant and duration of trap placement on the detection probability of small carnivores and selected African golden cat Caracal aurata prey items. We examine transient site response of each species, by comparing our observed likelihood of detection in each 24 h period from 1–7 d following refreshing of attractants to randomly generated capture histories. African civet Civettictis civetta, rusty-spotted genet Genetta maculata, African palm civet Nandinia binotata, and marsh mongoose Atilax paludinosus detection probabilities were highest and Weyns’s red duiker Cephalophus wenysi detection probability was lowest immediately after attractants were placed. Within 24 h after attractant was placed, rusty-spotted genet and African palm civet were more likely to be detected and African golden cat, red duiker, and blue duiker Philantomba monticola were less likely to be detected. Our results suggest that attractants can increase detection of small-bodied species and include some arboreal species in terrestrial camera trap sampling. However, attractants may also alter short-term visitation rates of some species, with potentially cascading effects on others. Community level and intraguild interaction studies should control for the potentially confounding effects of attractants on spatial activity patterns.

Framing pictures: A conceptual framework to identify and correct for biases in detection probability of camera traps enabling multi-species comparison

Obtaining reliable species observations is of great importance in animal ecology and wildlife conservation. An increasing number of studies use camera traps (CTs) to study wildlife communities, and an increasing effort is made to make better use and reuse of the large amounts of data that are produced. It is in these circumstances that it becomes paramount to correct for the species- and study-specific variation in imperfect detection within CTs. We reviewed the literature and used our own experience to compile a list of factors that affect CT detection of animals. We did this within a conceptual framework of six distinct scales separating out the influences of (a) animal characteristics, (b) CT specifications, (c) CT set-up protocols, and (d) environmental variables. We identified 40 factors that can potentially influence the detection of animals by CTs at these six scales. Many of these factors were related to only a few overarching parameters. Most of the animal characteristics scale with body mass and diet type, and most environmental characteristics differ with season or latitude such that remote sensing products like NDVI could be used as a proxy index to capture this variation. Factors that influence detection at the microsite and camera scales are probably the most important in determining CT detection of animals. The type of study and specific research question will determine which factors should be corrected. Corrections can be done by directly adjusting the CT metric of interest or by using covariates in a statistical framework. Our conceptual framework can be used to design better CT studies and help when analyzing CT data. Furthermore, it provides an overview of which factors should be reported in CT studies to make them repeatable, comparable, and their data reusable. This should greatly improve the possibilities for global scale analyses of (reused) CT data.

What are we missing? Advantages of more than one viewpoint to estimate fish assemblages using baited video

Counting errors can bias assessments of species abundance and richness, which can affect assessments of stock structure, population structure and monitoring programmes. Many methods for studying ecology use fixed viewpoints (e.g. camera traps, underwater video), but there is little known about how this biases the data obtained. In the marine realm, most studies using baited underwater video, a common method for monitoring fish and nekton, have previously only assessed fishes using a single bait-facing viewpoint. To investigate the biases stemming from using fixed viewpoints, we added cameras to cover 360° views around the units. We found similar species richness for all observed viewpoints but the bait-facing viewpoint recorded the highest fish abundance. Sightings of infrequently seen and shy species increased with the additional cameras and the extra viewpoints allowed the abundance estimates of highly abundant schooling species to be up to 60% higher. We specifically recommend the use of additional cameras for studies focusing on shyer species or those particularly interested in increasing the sensitivity of the method by avoiding saturation in highly abundant species. Studies may also benefit from using additional cameras to focus observation on the downstream viewpoint.