Identification errors in camera-trap studies result in systematic population overestimation.
Ontology highlight
ABSTRACT:
SUBMITTER: Johansson O
PROVIDER: S-EPMC7156508 | biostudies-literature | 2020 Apr
REPOSITORIES: biostudies-literature
Publications
Identification errors in camera-trap studies result in systematic population overestimation.
Johansson Örjan Ö Samelius Gustaf G Wikberg Ewa E Chapron Guillaume G Mishra Charudutt C Low Matthew M
Scientific reports 20200414 1
Reliable assessments of animal abundance are key for successful conservation of endangered species. For elusive animals with individually-unique markings, camera-trap surveys are a benchmark standard for estimating local and global population abundance. Central to the reliability of resulting abundance estimates is the assumption that individuals are accurately identified from photographic captures. To quantify the risk of individual misidentification and its impact on population abundance estim ...[more]
Similar Datasets
Project description:In species-habitat association studies, both the type and spatial scale of habitat covariates need to match the ecology of the focal species. We assessed the potential of high-resolution satellite imagery for generating habitat covariates using camera-trapping data from Sabah, Malaysian Borneo, within an occupancy framework. We tested the predictive power of covariates generated from satellite imagery at different resolutions and extents (focal patch sizes, 10-500 m around sample points) on estimates of occupancy patterns of six small to medium sized mammal species/species groups. High-resolution land cover information had considerably more model support for small, patchily distributed habitat features, whereas it had no advantage for large, homogeneous habitat features. A comparison of different focal patch sizes including remote sensing data and an in-situ measure showed that patches with a 50-m radius had most support for the target species. Thus, high-resolution satellite imagery proved to be particularly useful in heterogeneous landscapes, and can be used as a surrogate for certain in-situ measures, reducing field effort in logistically challenging environments. Additionally, remote sensed data provide more flexibility in defining appropriate spatial scales, which we show to impact estimates of wildlife-habitat associations.
| S-EPMC4657010 | biostudies-literature
Project description: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.
| S-EPMC7022311 | biostudies-literature
Project description:Monitoring population trends of threatened species requires standardized techniques that can be applied over broad areas and repeated through time. Sun bears Helarctos malayanus are a forest dependent tropical bear found throughout most of Southeast Asia. Previous estimates of global population trends have relied on expert opinion and cannot be systematically replicated. We combined data from 1,463 camera traps within 31 field sites across sun bear range to model the relationship between photo catch rates of sun bears and tree cover. Sun bears were detected in all levels of tree cover above 20%, and the probability of presence was positively associated with the amount of tree cover within a 6-km2 buffer of the camera traps. We used the relationship between catch rates and tree cover across space to infer temporal trends in sun bear abundance in response to tree cover loss at country and global-scales. Our model-based projections based on this "space for time" substitution suggested that sun bear population declines associated with tree cover loss between 2000-2014 in mainland southeast Asia were ~9%, with declines highest in Cambodia and lowest in Myanmar. During the same period, sun bear populations in insular southeast Asia (Malaysia, Indonesia and Brunei) were projected to have declined at a much higher rate (22%). Cast forward over 30-years, from the year 2000, by assuming a constant rate of change in tree cover, we projected population declines in the insular region that surpassed 50%, meeting the IUCN criteria for endangered if sun bears were listed on the population level. Although this approach requires several assumptions, most notably that trends in abundance across space can be used to infer temporal trends, population projections using remotely sensed tree cover data may serve as a useful alternative (or supplement) to expert opinion. The advantages of this approach is that it is objective, data-driven, repeatable, and it requires that all assumptions be clearly stated.
| S-EPMC5621681 | biostudies-literature
Project description:Camera traps have proven very useful in ecological, conservation and behavioral research. Camera traps non-invasively record presence and behavior of animals in their natural environment. Since the introduction of digital cameras, large amounts of data can be stored. Unfortunately, processing protocols did not evolve as fast as the technical capabilities of the cameras. We used camera traps to record videos of Eurasian beavers (Castor fiber). However, a large number of recordings did not contain the target species, but instead empty recordings or other species (together non-target recordings), making the removal of these recordings unacceptably time consuming. In this paper we propose a method to partially eliminate non-target recordings without having to watch the recordings, in order to reduce workload. Discrimination between recordings of target species and non-target recordings was based on detecting variation (changes in pixel values from frame to frame) in the recordings. Because of the size of the target species, we supposed that recordings with the target species contain on average much more movements than non-target recordings. Two different filter methods were tested and compared. We show that a partial discrimination can be made between target and non-target recordings based on variation in pixel values and that environmental conditions and filter methods influence the amount of non-target recordings that can be identified and discarded. By allowing a loss of 5% to 20% of recordings containing the target species, in ideal circumstances, 53% to 76% of non-target recordings can be identified and discarded. We conclude that adding an extra processing step in the camera trap protocol can result in large time savings. Since we are convinced that the use of camera traps will become increasingly important in the future, this filter method can benefit many researchers, using it in different contexts across the globe, on both videos and photographs.
| S-EPMC4053333 | biostudies-other
Project description:There is an increasing need to monitor activity and population growth of arthropods; however, this is a time-consuming and financially demanding process. Using sensors to detect arthropods in the field can help to follow their dynamics in time. Improving our earlier device, we developed a new camera-supported probe to detect soil microarthropods. An opto-electronic sensor ring detects the caught microarthropod individuals what activates a camera. The camera takes pictures of a specimen when it arrives in the camera chamber. A vacuum device was built into the probe which pumps up the specimen from the probe to a sample container. Here, we describe the construction and operation of the probe. We investigated the precision of the process in a laboratory experiment using living microarthropods and evaluated the accuracy of the probes in a semi-natural investigation when environmental noise was present. Under semi-natural conditions, the percentages of success, i.e., the photographed specimens compared to the caught ones, were between 60% and 70% at the investigated taxa. The automatic camera shooting helped in distinguishing insects from irrelevant detections while collecting the trapped insects allowed species-level determination. This information together serves as a basis for the automatic visual recognition of microarthropod species.
| S-EPMC7240604 | biostudies-literature
Project description:Scientists are increasingly using volunteer efforts of citizen scientists to classify images captured by motion-activated trail cameras. The rising popularity of citizen science reflects its potential to engage the public in conservation science and accelerate processing of the large volume of images generated by trail cameras. While image classification accuracy by citizen scientists can vary across species, the influence of other factors on accuracy is poorly understood. Inaccuracy diminishes the value of citizen science derived data and prompts the need for specific best-practice protocols to decrease error. We compare the accuracy between three programs that use crowdsourced citizen scientists to process images online: Snapshot Serengeti, Wildwatch Kenya, and AmazonCam Tambopata. We hypothesized that habitat type and camera settings would influence accuracy. To evaluate these factors, each photograph was circulated to multiple volunteers. All volunteer classifications were aggregated to a single best answer for each photograph using a plurality algorithm. Subsequently, a subset of these images underwent expert review and were compared to the citizen scientist results. Classification errors were categorized by the nature of the error (e.g., false species or false empty), and reason for the false classification (e.g., misidentification). Our results show that Snapshot Serengeti had the highest accuracy (97.9%), followed by AmazonCam Tambopata (93.5%), then Wildwatch Kenya (83.4%). Error type was influenced by habitat, with false empty images more prevalent in open-grassy habitat (27%) compared to woodlands (10%). For medium to large animal surveys across all habitat types, our results suggest that to significantly improve accuracy in crowdsourced projects, researchers should use a trail camera set up protocol with a burst of three consecutive photographs, a short field of view, and determine camera sensitivity settings based on in situ testing. Accuracy level comparisons such as this study can improve reliability of future citizen science projects, and subsequently encourage the increased use of such data.
| S-EPMC7663993 | biostudies-literature
Project description:In this study, we estimate life history parameters and abundance for a protected jaguar population using camera-trap data from a 14-year monitoring program (2002-2015) in Belize, Central America. We investigated the dynamics of this jaguar population using 3,075 detection events of 105 individual adult jaguars. Using robust design open population models, we estimated apparent survival and temporary emigration and investigated individual heterogeneity in detection rates across years. Survival probability was high and constant among the years for both sexes (φ = 0.78), and the maximum (conservative) age recorded was 14 years. Temporary emigration rate for the population was random, but constant through time at 0.20 per year. Detection probability varied between sexes, and among years and individuals. Heterogeneity in detection took the form of a dichotomy for males: those with consistently high detection rates, and those with low, sporadic detection rates, suggesting a relatively stable population of 'residents' consistently present and a fluctuating layer of 'transients'. Female detection was always low and sporadic. On average, twice as many males than females were detected per survey, and individual detection rates were significantly higher for males. We attribute sex-based differences in detection to biases resulting from social variation in trail-walking behaviour. The number of individual females detected increased when the survey period was extended from 3 months to a full year. Due to the low detection rates of females and the variable 'transient' male subpopulation, annual abundance estimates based on 3-month surveys had low precision. To estimate survival and monitor population changes in elusive, wide-ranging, low-density species, we recommend repeated surveys over multiple years; and suggest that continuous monitoring over multiple years yields even further insight into population dynamics of elusive predator populations.
| S-EPMC5489181 | biostudies-literature
Project description:Density estimates for large carnivores derived from camera surveys often have wide confidence intervals due to low detection rates. Such estimates are of limited value to authorities, which require precise population estimates to inform conservation strategies. Using lures can potentially increase detection, improving the precision of estimates. However, by altering the spatio-temporal patterning of individuals across the camera array, lures may violate closure, a fundamental assumption of capture-recapture. Here, we test the effect of scent lures on the precision and veracity of density estimates derived from camera-trap surveys of a protected African leopard population. We undertook two surveys (a 'control' and 'treatment' survey) on Phinda Game Reserve, South Africa. Survey design remained consistent except a scent lure was applied at camera-trap stations during the treatment survey. Lures did not affect the maximum movement distances (p = 0.96) or temporal activity of female (p = 0.12) or male leopards (p = 0.79), and the assumption of geographic closure was met for both surveys (p >0.05). The numbers of photographic captures were also similar for control and treatment surveys (p = 0.90). Accordingly, density estimates were comparable between surveys (although estimates derived using non-spatial methods (7.28-9.28 leopards/100km2) were considerably higher than estimates from spatially-explicit methods (3.40-3.65 leopards/100km2). The precision of estimates from the control and treatment surveys, were also comparable and this applied to both non-spatial and spatial methods of estimation. Our findings suggest that at least in the context of leopard research in productive habitats, the use of lures is not warranted.
| S-EPMC4822812 | biostudies-literature
Project description:Information from camera traps is used for inferences on species presence, richness, abundance, demography, and activity. Camera trap placement design is likely to influence these parameter estimates. Herein we simultaneously generate and compare estimates obtained from camera traps (a) placed to optimize large carnivore captures and (b) random placement, to infer accuracy and biases for parameter estimates. Both setups recorded 25 species when same number of trail and random cameras (n = 31) were compared. However, species accumulation rate was faster with trail cameras. Relative abundance indices (RAI) from random cameras surrogated abundance estimated from capture-mark-recapture and distance sampling, while RAI were biased higher for carnivores from trail cameras. Group size of wild-ungulates obtained from both camera setups were comparable. Random cameras detected nocturnal activities of wild ungulates in contrast to mostly diurnal activities observed from trail cameras. Our results show that trail and random camera setup give similar estimates of species richness and group size, but differ for estimates of relative abundance and activity patterns. Therefore, inferences made from each of these camera trap designs on the above parameters need to be viewed within this context.
| S-EPMC8630032 | biostudies-literature
Project description:Wild animals are essential for ecosystem structuring and stability, and thus they are important for ecological research. Since most wild animals have high athletic or concealable abilities or both, it is used to be relatively difficult to acquire evidence of animal appearances before applications of camera traps in ecological researches. However, a single camera trap may produce thousands of animal images in a short period of time and inevitably ends up with millions of images requiring classification. Although there have been many methods developed for classifying camera trap images, almost all of them follow the pattern of a very deep convolutional neural network processing all camera trap images. Consequently, the corresponding surveillance area may need to be delicately controlled to match the network capability, and it may be difficult to expand the area in the future. In this study, we consider a scenario in which camera traps are grouped into independent clusters, and images produced by a cluster are processed by an edge device installed with a customized network. Accordingly, edge devices in this scenario may be highly heterogeneous due to cluster scales. Resultantly, networks popular in the classification of camera trap images may not be deployable for edge devices without modifications requiring the expertise which may be hard to obtain. This motivates us to automatize network design via neural architecture search for edge devices. However, the search may be costly due to the evaluations of candidate networks, and its results may be infeasible without considering the resource limits of edge devices. Accordingly, we propose a search method using regression trees to evaluate candidate networks to lower search costs, and candidate networks are built based on a meta-architecture automatically adjusted regarding to the resource limits. In experiments, the search consumes 6.5 hours to find a network applicable to the edge device Jetson X2. The found network is then trained on camera trap images through a workstation and tested on Jetson X2. The network achieves competitive accuracies compared with the automatically and the manually designed networks.
| S-EPMC8843777 | biostudies-literature