Research articlesSnap happy: camera traps are an effective sampling tool when compared with alternative methods
Abstract
Camera traps have become a ubiquitous tool in ecology and conservation. They are routinely deployed in wildlife survey and monitoring work, and are being advocated as a tool for planetary-scale biodiversity monitoring. The camera trap's widespread adoption is predicated on the assumption of its effectiveness, but the evidence base for this is lacking. Using 104 past studies, we recorded the qualitative overall recommendations made by study authors (for or against camera traps, or ambiguous), together with quantitative data on the effectiveness of camera traps (e.g. number of species detected or detection probabilities) relative to 22 other methods. Most studies recommended the use of camera traps overall and they were 39% more effective based on the quantitative data. They were significantly more effective compared with live traps (88%) and were otherwise comparable in effectiveness to other methods. Camera traps were significantly more effective than other methods at detecting a large number of species (31% more) and for generating detections of species (91% more). This makes camera traps particularly suitable for broad-spectrum biodiversity surveys. Film camera traps were found to be far less effective than digital models, which has led to an increase in camera trap effectiveness over time. There was also evidence from the authors that the use of attractants with camera traps reduced their effectiveness (counter to their intended effect), while the quantitative data indicated that camera traps were more effective in closed than open habitats. Camera traps are a highly effective wildlife survey tool and their performance will only improve with future technological advances. The images they produce also have a range of other benefits, for example as digital voucher specimens and as visual aids for outreach. The evidence-base supports the increasing use of camera traps and underlines their suitability for meeting the challenges of global-scale biodiversity monitoring.
1. Introduction
Camera traps have come a long way from their beginnings in wildlife photography more than 100 years ago and are now a ubiquitous tool in ecology and conservation, with several hundred scientific studies now published each year using them [1]. The last decade, in particular, has seen the camera trap move from being a niche tool primarily for monitoring big cats (e.g. [2–4]), to taking centre stage in broad-spectrum surveys of whole communities of mammals (e.g. [5–7]). The camera trap has now been adopted by a number of large-scale biodiversity monitoring programmes (e.g. [8,9]) and underpins the Wildlife Picture Index [10], which informs progress towards the Convention on Biological Diversity's Aichi Target 12 (https://www.bipindicators.net/indicators/wildlife-picture-index).
The camera trap's widespread adoption is predicated on an assumption of its effectiveness. Certainly, a number of methodological case studies have found them to compare favourably with other methods for surveying species richness [11–13], species distributions [14–16], relative abundance [17–19] and animal density [20–22]. The key strength of the modern camera trap is its ability to remain in the field for protracted periods of times (on the order of months at a time, if needed), continuously registering detections day and night, with relatively little input from human operators required. This means that even the rarest events, such as nest predation or the passage of an apex predator through an area, can be quantified and studied. Camera traps also facilitate this without significant disturbance to study animals, except for the emission of sound and light [23].
However, a comprehensive assessment of the effectiveness of camera traps compared with alternative methods has never been done. For example, a number of studies have found that camera traps can sometimes be outperformed by other survey methods in certain contexts (e.g. [24–26]). Camera traps are certainly no panacea for surveying wildlife and suffer from a number of limitations in their current form [27], most importantly their relatively high initial cost (typically $200–500 per unit) and the relatively small area that they monitor (notional detection zones of camera traps are typically less than 2 × 10−4 km2). A critical appraisal of camera traps, as has already been done for other technologies (e.g. GPS telemetry: [28]), is now long overdue.
We here draw upon the substantial body of evidence that has accumulated on the effectiveness of camera traps to ask: (i) which sampling methods do camera traps compare most favourably with, and (ii) can we identify the specific study contexts in which camera traps are most effective relative to other methods? In particular, we expected camera trap effectiveness to depend strongly on study objectives, but we also tested specific hypotheses related to the equipment and protocol used, the species under study, and the habitat and location of the study. Ours is the first global assessment of camera trap effectiveness to date, and offers broad insights into when and where this technology can be most effectively deployed.
2. Material and methods
2.1. Literature search
We searched the scientific literature for studies which have compared camera traps with other ecological survey methods, focusing on peer-reviewed journal articles. To do this, on 29 January 2018, we searched the Core Collection of the Web of Science (http://webofknowledge.com) with the query ‘camera trap*’ OR ‘camera-trap*’ OR ‘game camera*’ OR ‘trail camera*’ OR ‘scouting camera*’ OR (‘remote camera*’ AND ‘wildlife’) OR (‘automatic camera*’ AND ‘wildlife’) OR (‘automatic photograph*’ AND ‘wildlife’) for the period 1969–2017. This generated a list of titles and abstracts (n = 1981) which we scanned for suitable studies.
We defined suitable studies as those which presented quantitative data comparing camera traps to one or more other survey methods and which drew explicit conclusions about their relative effectiveness. This included purely methodological studies, as well as ecological studies which also offered explicit comparisons of multiple survey methods as a secondary component of the paper. We only included studies using triggered camera traps, as opposed to time-lapse cameras. Where we were not sure if a study was suitable based on the title and abstract, we downloaded the full text to check the main body of text. We also followed the literature trail; if studies we read in turn cited other studies of potential interest, we assessed their suitability for inclusion as well.
This process yielded 100 peer-reviewed studies, as well as four unpublished technical reports that were cited by other studies, from which we attempted to extract data.
2.2. Data extraction
For each of the 104 studies (figure 1) we identified from the literature search, we attempted to extract both qualitative and quantitative data on the effectiveness of camera traps relative to other survey methods. First, we made an overall qualitative assessment of whether the authors of the study had concluded that camera traps were the superior method for achieving the aims of the study. Author conclusions have previously been used to distil the findings of complex studies involving many variables (e.g. [29]). Although this is a coarse measure of camera trap effectiveness, subject to the attitudes and value systems of the authors of each study, we hypothesized that it would capture information not present in the quantitative data presented in each paper, in particular better capturing the authors' practical experiences of implementing each method.
Figure 1. Locations of primary studies used in this study. The base map displays a measure of overall camera-trapping publishing output (the number of camera trap studies per country published between 2008 and 2014, from [47]).
To qualitatively assess author recommendations, we read each paper, focusing on concluding remarks in the Abstract and Discussion sections. We were able to categorize the author conclusions in most studies (61%) as either ‘In support of camera traps’ or ‘In support of another method’ on this basis. If camera traps were equal in performance to another survey method, then we scored the study as ‘Ambiguous’ (39% of studies). This category was also used if the authors presented heavily caveated conclusions (typically, concluding that one or other survey method was best in specific circumstances, such as particular weather conditions, or for specific species). Given the subjectivity inherent to this scoring, one of us (O.R.W.) did all the scoring in a consistent manner.
In addition to scoring the qualitative author conclusions, we also extracted quantitative data on the effectiveness of camera traps relative to other survey methods, which we fed into a meta-analysis. Suitable quantitative data included comparisons of the following metrics: (i) the number of species detected; (ii) detection rates (per sampling occasion, i.e. per trap night or as defined in the study); (iii) latencies to detection (the sampling effort required to register the first detection); (iv) detection probabilities (species or individual detection probabilities, estimated from occupancy or capture–recapture models, respectively); (v) the number of individuals detected (only possible in cases where individuals could be identified); (vi) the accuracy of state variable estimates (measured in percentage bias from a benchmark, such as a true abundance count); (vii) the precision of state variable estimates (measured using the coefficient of variation); (viii) implementation efforts (measured in person hours of labour), and (ix) costs. For costs, we took account of as many costs as the authors provided information for, typically including equipment, labour and associated field costs, but sometimes also including training, data entry and equipment repair costs. Costs were standardized in different ways across studies, in most cases per survey (e.g. the cost of obtaining a single-session abundance estimate) but in some cases per unit area sampled or per animal detection. If a study presented costs separately for successive sessions of sampling (e.g. costs for years 1–5), then we took the costs for the first session.
Survey methods being compared with camera traps were grouped into broad classes (see electronic supplementary material, table S1.1 in appendix S1 for definitions). Each study used a different experimental design to compare methods, depending on the specific objectives and methods involved, but most studies attempted to spatially and temporally match the deployment of each survey method (e.g. co-deploying methods at the same sampling points). Where this was obviously not the case, we limited comparisons to metrics that we considered would still be valid. For example, if the temporal extents differed, we only considered metrics that are invariant to sampling effort (detection rate, latency to detection, detection probabilities and percentage bias).
We also extracted the sample sizes and variances associated with each data point, where these were presented or could be calculated from available information. However, this was only possible in a minority of cases (39% of data points). Where data were presented in figures, rather than in the main text or tables, we extracted the values using the open source tool ‘Engauge Digitizer’ (http://markummitchell.github.io/engauge-digitizer).
2.3. Quantifying relative camera trap effectiveness
For each pair of data points—one for camera-trapping and one for the other survey method—we calculated an effect size to capture the magnitude of the difference in performance between methods. We used the log response ratio [30,31] as our measure of effect size, which is appropriate when combining information across values measured on very different scales. It also has the practical benefit in our case of not requiring sample sizes or variances, which were unavailable in most cases.
We calculated the log response ratio as: ln ( Metriccamera trap / Metricother sampling method), using the values we extracted in one of the nine categories of metric outlined above. For metrics in which lower values indicate better performance (latency to detection, percentage bias, coefficient of variation, implementation effort and cost), we instead took Metriccamera trap as the denominator. Larger effect size values in this paper therefore always indicate that camera traps outperformed the survey method they were compared with. Log response ratios cannot be calculated when either the numerator or denominator is zero. This issue was mostly confined to the comparison of detection rates, and in practice means that the comparison is of detection rates when both methods detected the given species at least once. In total, we obtained 662 quantitative comparisons across 97 of the 104 studies in our sample.
To obtain the median effect sizes for each type of metric and for each survey method comparison, including estimates of uncertainty, we used a stratified bootstrapping approach (e.g. [32]). This approach deals effectively with the lack of independence between multiple comparisons provided by a single study, because in each bootstrap, only a single comparison per study (selected at random) is chosen. Bootstrap sampling with replacement (n = 10 000) was done separately for each metric and each method with more than 15 comparisons available in the data. We were also a priori interested in effect sizes according to three other factors—the type of camera trap used (film or digital), whether baits or lures were used or not with the camera traps (attractant or no attractant), and the openness of the habitat (closed forested habitats or open shrubland, grassland and desert habitats)—and calculated bootstrapped median effect sizes in the same way as for the metrics and survey methods. We calculated 95% confidence intervals for the bootstrapped effect sizes from the quantiles of the distribution of bootstraps and judged effect sizes to be significant if the intervals did not overlap zero. For comparing across effect size estimates, we used post hoc pairwise Wald tests [33] and judged significant differences using an error rate (α) of 0.05.
We could not weight the effect size for each comparison by its precision because the required data for this were not presented in most studies. This means that the precision of our overall effect size estimates may be reduced, but does not necessarily mean that the estimates are biased [34].
2.4. Hypothesis testing
We modelled the author recommendations and the effect sizes separately but tested a similar set of hypotheses. In particular, we hypothesized that studies which used digital instead of film camera traps, used an attractant with the camera traps and were done in closed instead of open habitats would be more effective. Habitat openness was thought to be important on the basis of anecdotal information that the most common type of camera trap sensor—the passive infrared (PIR) sensor—suffers from misfires and missed detections in open habitat, due to denser ground vegetation and higher temperatures, respectively [1]. We also hypothesized that studies which included small-sized animals (i.e. animals with a weaker infrared signal) would be less effective and modelled this using the minimum of the body weights of the focal species in a given study. The absolute latitude of the study area was also included to test if studies done in the tropics were less effective than those at higher latitudes, due to higher temperatures and possibly greater rates of equipment failure [1].
For the models of effect sizes, we also included covariates for the metric used and the comparison survey method used, but we could not include these covariates in the models of the author recommendations, because each study typically involved multiple metrics and/or methods. We tested for collinearity among the covariates using variance inflation factors, which provided no cause for concern (all factors less than 3; [35]).
We treated author recommendations as an ordinal response and used proportional odd models implemented in the ordinal package in R [36]. We fit the model Recommendation ∼ Camera type + Use of attractants + Habitat openness + Latitude + Minimum weight, as well as all nested versions of this model (for a total of 32 models), using the MuMIn package in R [37]. The continuous covariates, Latitude and Minimum weight, were centred at their mean and scaled by twice their standard deviation [38]. We took the natural logarithm of the minimum body weights (before standardization), owing to the strong positive skew in these data. We did not consider interactions among variables, owing to a lack of data.
Important variables were those that appeared in the set of ‘best’ models based on information-theoretic criteria (i.e. models for which ΔAICc < 2 [39]). The relative importance of each variable was assessed using the standardized parameter estimates [40] and by summing the Akaike weights of the models in which each variable appeared [39,41]. We calculated model-averaged parameter estimates (using ‘natural averaging’, to avoid shrinkage of estimates towards zero) and 95% confidence intervals, which incorporate model uncertainty as well as parameter uncertainty ([39,42], but see [43,44]).
We modelled effect sizes, i.e. log response ratios, using linear mixed-effects models in the lme4 package in R [45]. The saturated model in this case was ln(Response ratio) ∼ Camera type + Use of attractants + Habitat openness + Latitude + Minimum weight + Metric + Comparison survey method, as well as a random intercept for Study to account for the lack of independence among multiple measures made from the same study. As for the author recommendations, we fit all nested versions of this model (for a total of 128 models), assessed the relative importance of variables using the standardized parameter estimates and sums of Akaike weights, and calculated model-averaged parameter estimates and 95% confidence intervals.
We made all calculations and fit the models in R v. 3.4.1. [46].
3. Results
Studies took place in 32 countries, on all continents except Antarctica, with most studies occurring in the USA (n = 33), Brazil (n = 9) and Australia (n = 8). This reflects broader geographical trends in camera-trapping publishing output ([47]; figure 1). Studies spanned the period 1990–2017, with fieldwork conducted from the late 1980s to 2016. Across this period, film camera traps were gradually replaced by digital camera traps, with the latter becoming more commonly used from 2007 onwards (figure 2). Camera-trapping was compared with a wide range of other survey methods (n = 22 methods), including methods as disparate as acoustic recording, eDNA, radio-tracking and local ecological knowledge (a full list of the methods is provided in electronic supplementary material, appendix S2).
Figure 2. Author recommendations over time and the proportion of studies using digital camera traps over the same period. The proportion of studies using digital camera traps (dashed line) was predicted using a generalized linear model with a binomial response. This was based on the year fieldwork began, rather than the publication date, owing to the substantial lag between the two in some studies.
3.1. Author recommendations
Most studies recommended the use of camera traps overall (n = 46), though many studies also found that other survey methods were equal in performance or presented caveated support for camera traps (i.e. ‘Ambiguous’; n = 41). A minority of studies supported the use of other survey methods overall (n = 17). Three models were identified as the ‘best’ at explaining the variation in author recommendations (table 1). These contained the variables Camera type (importance = 0.99; model-averaged estimate of the difference between factor levels, with digital as the reference level = −1.42; 95% CI: −2.24 to −0.61), Use of attractants (importance = 0.84; estimate = −0.94; 95% CI: −1.74 to −0.14), Habitat openness (importance = 0.34; estimate = −0.40; 95% CI: −1.21 to 0.40) and Latitude (importance = 0.27; estimate = 0.20; 95% CI of the slope: −0.67 to 1.07). Digital camera traps were 4.1 times more likely to result in an affirmative author recommendation than film camera traps, in support of our hypothesis (figure 3). Studies which did not use attractants during camera-trapping were 2.5 times more likely to result in an affirmative author recommendation than those that did use attractants (figure 3), which is counter to our hypothesis that attractants would make camera traps more effective. This effect appears to have been driven, at least in part, by a cluster of studies on North American forest carnivores (e.g. [24,25,48]), which routinely baited their camera traps and found them to be less effective than other survey methods.
Figure 3. Author recommendations according to the type of camera used (film or digital) and whether attractants (i.e. baits or lures) were used or not. The percentage of studies has been calculated separately within each factor level combination.
| model | parameters | log-likelihood | ΔAICc | Akaike weight |
|---|---|---|---|---|
| author recommendations—proportional odds model | ||||
| Camera type + Use of attractants | 4 | −98.42 | 0 | 0.30 |
| Camera type + Use of attractants + Habitat openness | 5 | −97.93 | 1.23 | 0.16 |
| Camera type + Use of attractants + Latitude | 5 | −98.31 | 1.99 | 0.11 |
| effect sizes—linear mixed-effects model | ||||
| Camera type + Habitat openness + Metric + Comparison survey method | 17 | −1022.24 | 0 | 0.27 |
| Camera type + Metric + Comparison survey method | 16 | −1023.95 | 1.29 | 0.14 |
3.2. Relative camera trap effectiveness
The bootstrapped effect size over the studies in the sample was 0.33, which corresponds to a 39% (95% CI: −2% to 100%) higher effectiveness of camera traps relative to the methods they were compared with. Digital camera traps, which have now completely superseded film-based models (figure 2), were 65% (95% CI: 2–169%) more effective than other methods. The effect sizes per metric and method showed that in no case were camera traps significantly inferior to other survey methods (figure 4; electronic supplementary material, tables S2.1 and S2.2 in appendix S2). On the contrary, camera traps were significantly more effective than other methods at detecting a higher number of species (31% higher effectiveness for species richness; 95% CI: 11–60%) and at quickly generating a large number of detections for individual species (91% higher effectiveness for detection rate; 95% CI: 29–192%). Similarly, camera traps were significantly more effective than live traps (88%; 95% CI: 1–239%).
Figure 4. Bootstrapped effect sizes for each metric and survey method, with the distribution of bootstrap estimates shown in each case with a violin plot. Asterisks after the labels indicate cases in which camera traps were significantly more effective than the methods they were compared with. Points and error bars indicate medians and standard errors. Sample sizes are the number of effect sizes extracted from the primary literature, with non-independence among effect sizes from the same study accounted for with stratification during each bootstrap. The red vertical dashed line at zero indicates no difference in effectiveness.
In contrast with the author recommendations, the bootstrapped effect sizes calculated from the quantitative data in the same studies did not show a difference in camera trap effectiveness according to whether attractants were used (z = 0.05; p = 0.96; figure 5). Effect sizes were larger for studies using digital rather than film camera traps, and for studies in closed rather than open habitats, but the differences were not significant, given the large sampling variances (Camera type: z = 1.45; p = 0.15; Habitat openness: z = 0.88; p = 0.38; figure 5).
Figure 5. Bootstrapped effect size according to camera trap type, whether attractants were used and habitat openness. Asterisks after the labels indicate cases in which camera traps were significantly more effective than the methods they were compared with. Points and error bars indicate medians and standard errors. Sample sizes are the number of effect sizes extracted from the primary literature, with non-independence among effect sizes from the same study accounted for with stratification during each bootstrap. The red vertical dashed line at zero indicates no difference in effectiveness.
Mixed-effects modelling identified two models that were ‘best’ at explaining the variation in effect sizes (table 1). These models contained the variables Metric (importance = 0.930; see electronic supplementary material, table S2.3 in appendix S2 for estimates and 95% CIs), Camera type (importance = 0.928; model-averaged estimate of the difference between factor levels = −0.55; 95% CI: −0.91 to −0.18), Comparison survey method (importance = 0.89; see electronic supplementary material, table S2.4 in appendix S2 for estimates and 95% CIs) and Habitat openness (importance = 0.66; estimate = −0.41; 95% CI: −0.77 to −0.05). There was no support for the importance of Use of attractants, Latitude or Minimum weight. The lack of support for any effect of attractants again contrasts with the author recommendations.
4. Discussion
We found that studies comparing camera traps to other survey methods, in general, support the notion that camera traps are a highly effective wildlife survey tool. This was confirmed by the recommendations made by study authors as well as the quantitative data underlying the studies. The quantitative data showed that camera traps were effective across a range of different metrics, but were especially effective at detecting a wide range of species and at recording a large number of detections of focal species. This makes camera traps ideally suited for broad-spectrum biodiversity surveys of the kind often required for research or management purposes. Camera traps performed significantly better than live traps and were otherwise comparable in performance to other survey methods. Confirming a widely held belief, the transition to digital camera traps over the last 15 years has substantially increased the effectiveness of the method. There was also evidence that camera trap studies done in open habitats are less effective. Other characteristics of studies—the use of baits and lures, the body size of focal taxa and latitude—were not found to have a strong overall bearing on the effectiveness of camera traps, although this does not discount the fact they are likely to be important considerations in specific cases.
4.1. How do camera traps compare with other survey methods?
There was substantial variation in camera trap effectiveness depending on the method they were compared with, as evidenced by the bootstrapped effect sizes (figure 4) and the mixed-effects model results (table 1). Camera traps were most effective in comparison to live traps and hair traps, and least effective in comparison to detector dogs, although the difference was only significant in the case of live-trapping (see electronic supplementary material, appendix S3 for discussion of hair traps and detector dogs). Live traps performed poorly compared with camera traps in part because they are typically ‘single-catch’ traps, i.e. they require manual resetting after each capture, while camera traps are ‘multi-catch’ traps (e.g. [15,49,50]). Live traps also require an animal to be attracted by a bait or lure, interact with a foreign object in the environment (i.e. the trap) and be physically caught successfully. This sequence of events, which may fail for a wide range of reasons, contrasts with camera-trapping, which only requires an animal to enter the camera detection zone and trigger the sensor. As a result, live-trapping methods are typically optimized for a relatively narrow set of species (e.g. [50,51]). A major benefit of live-trapping, however, is that animals are physically caught, meaning that samples can be taken from animals, they can be marked for the purposes of studying population dynamics and they can be GPS- or radio-tagged.
4.2. Camera trap effectiveness depends strongly on study objectives
The most important predictor identified in the mixed-effects models was Metric (table 1), indicating that camera trap effectiveness varied widely depending on the type of measurement made (figure 4). Camera traps were especially effective at detecting a large number of species, owing to the relatively low specificity of most camera trap sensors. The PIR sensors typical of commercial camera traps are known to be effective for any endothermic species larger than 100 g [1], and it is possible to detect even smaller species (including ectothermic species) at distances less than 1 m using specialized protocols [52–54]. Camera traps were also effective at making a large number of detections of focal species, owing to their capacity to record multiple detections without requiring maintenance. While this was less true of film camera traps, which were limited to a single roll of film, modern digital camera traps are now capable of recording more than 10 000 images on a single set of batteries. Camera traps were less effective than other methods at detecting species quickly (for example, in a rapid assessment survey) and cheaply, though in neither case was this difference significant (figure 4; see electronic supplementary material, appendix S3 for additional discussion).
4.3. Differences in camera trap effectiveness according to equipment and habitat
The type of camera trap used—film or digital—had the most important bearing on camera trap effectiveness (table 1), with film camera traps being much less effective (figures 3 and 5). In fact, after removing film camera traps from the sample, camera traps were overall significantly more effective than other methods. As well as the limited number of images that film camera traps could store, they were often paired with active infrared sensors, which have much smaller detection zones (they require animals to cross a beam of infrared) compared with PIR sensors, further limiting their effectiveness. Improvements in camera trap performance must be accounted for when making inferences about wildlife population changes over long time-frames from camera trap data (e.g. by accounting for detection probabilities). In addition, it suggests that next-generation camera trap technology (e.g. replacing PIR sensors with pixel-based change detection algorithms) might lead to similarly dramatic improvements in camera trap effectiveness [27].
Mixed-effects modelling of the effect sizes identified habitat openness as an important variable (table 1), with camera traps being more effective in closed habitats. Authors were not, however, more likely to recommend using camera traps in closed habitats. Most camera trap studies are done in closed habitats [27,55], perhaps because of their greater effectiveness in this habitat. We hypothesized that this is because PIR sensors are more effective in closed habitats, but it might also be because researchers have a more limited set of survey methods available to them in dense forest habitat (for example, aerial surveys of terrestrial animals are not possible) or because certain survey methods are less effective (e.g. line transects). Either way, it underlines the importance of applying best-practice methods to help address the potentially large number of problems that open environments can pose [1].
4.4. Limitations: experimental design and representativeness
During this review, we identified a number of shortcomings across studies. In particular, some studies suffered from poor spatial or temporal matching across the sampling methods being compared. We limited the impacts of this by comparing metrics that would be sensitive to this only across similar spatial or temporal extents. However, we did include studies which sampled over the same overlapping area, albeit not at exactly the same spatial point locations (this was necessary to compare sampling methods which are not point-based, such as line transects and plot surveys), and we also included studies which used the methods at different times (for example, in different months or different years). We anticipate that this will introduce additional noise into our data, but not undermine our main results.
We also included studies that did not standardize financial costs across the methods being compared (and instead we measured any cost differences as an effect size). This was because financial costs are one of several possible resources that might be a limiting factor for any given research team carrying out a camera trap survey, for example the time taken to complete a survey in the case of a rapid survey, or field labour in the case of a team which has limited personnel. Although many studies explicitly acknowledged the resource that they had matched across sampling methods (e.g. [12,49]), others did not, and in these cases the resources allocated to each sampling method were probably determined by accepted ‘best-practice’ (e.g. [19,22]), or by logistical factors external to the study. We only made comparisons across methods where it was sensible to do so, for example using a metric that was insensitive to resources (e.g. detection rate). However, it remains the case that our overall results are reflective of the resources allocated to each sampling method across the studies in our sample.
The set of studies in our sample reflects broader biases in where camera-trapping is done and what it is used for [27,47,55]. Therefore, studies from the USA, and on mammalian carnivores, are more common in our dataset than other types of study. If we are to make more general conclusions about the effectiveness of camera-trapping, irrespective of current geographical and taxonomic biases, then we would ideally have access to a much more balanced dataset. This will only be possible if steps are taken to address the broader geographical biases in conservation research [56].
4.5. An evidence-base for more effective wildlife surveys
More than a decade ago, Rowcliffe & Carbone [57] asked whether camera traps might have a bright future in ecological monitoring. We have established the evidence-base to show that they do. Across a range of metrics, camera traps are either comparable to or outperform existing alternative methods, and their performance is only likely to improve with future advances in camera trap technology (outlined in [27]). In addition, we have not considered here the considerable value that camera trap images and videos have as digital museum voucher specimens, and as powerful visual aids for outreach, science communication, lobbying and community engagement. The raw data recorded by other methods are typically much less rich in information (with the exception of DNA sequences) and usually do not provide information on group size, behaviour or other natural history details.
Camera traps have recently been advocated as a key component of a proposed global biodiversity monitoring network [58]. Camera traps are prime candidates because they allow for highly standardized data collection, they minimally disturb wildlife, and they generate data that can be up-scaled to regional or global scales while accounting for imperfect observation. Adding to this, we have here shown that they are also highly effective relative to other methods. We anticipate that the coming decade will be even brighter for camera-trapping than the last.
Data accessibility
All of the raw data used in this publication, including the author recommendations and calculated effect sizes, are available online in the Zenodo repository [59].
Authors' contributions
O.R.W. conceived and designed the study, analysed the data and led the manuscript writing. Both authors collected the data, edited the manuscript and gave final approval for publication.
Competing interests
We declare that we have no competing interests.
Funding
Primary funding for this study was provided by WWF-UK. O.R.W. was also supported by an AXA Research Fellowship during this work.
Acknowledgements
We thank all of the authors of the primary research we drew upon. We have included in the reference list those studies not cited in the main text [60–140]. We also thank Joanna Burgar and an anonymous reviewer for their constructive comments on the manuscript.
References
- 1.
Wearn OR, Glover-Kapfer P . 2017 Camera-trapping for conservation: a guide to best-practices. WWF Conservation Technology Series 1(1). Woking, UK: WWF-UK, 180 p. Google Scholar - 2.
Karanth KU, Nichols JD . 1998 Estimation of tiger densities in India using photographic captures and recaptures. Ecology 79, 2852-2862. (doi:10.1890/0012-9658(1998)079[2852:EOTDII]2.0.CO;2) Crossref, Web of Science, Google Scholar - 3.
Silver SC, Ostro LET, Marsh LK, Maffei L, Noss AJ, Kelly MJ, Wallace RB, Gómez H, Ayala G . 2004 The use of camera traps for estimating jaguar Panthera onca abundance and density using capture/recapture analysis. Oryx 38, 148-154. (doi:10.1017/s0030605304000286) Crossref, Web of Science, Google Scholar - 4.
Jackson RM, Roe JD, Wangchuk R, Hunter D . 2006 Estimating snow leopard population abundance using photography and capture–recapture techniques. Wildl. Soc. Bull. 34, 772-781. (doi:10.2193/0091-7648(2006)34[772:ESLPAU]2.0.CO;2) Crossref, Web of Science, Google Scholar - 5.
Tobler MW, Hartley AZ, Carrillo-Percastegui SE, Powell GVN . 2015 Spatiotemporal hierarchical modelling of species richness and occupancy using camera trap data. J. Appl. Ecol. 52, 413-421. (doi:10.1111/1365-2664.12399) Crossref, Web of Science, Google Scholar - 6.
Wearn OR, Rowcliffe JM, Carbone C, Pfeifer M, Bernard H, Ewers RM . 2017 Mammalian species abundance across a gradient of tropical land-use intensity: a hierarchical multi-species modelling approach. Biol. Conserv. 212, 162-171. (doi:10.1016/j.biocon.2017.05.007) Crossref, Web of Science, Google Scholar - 7.
Rich LN, Miller DAW, Robinson HS, McNutt JW, Kelly MJ . 2016 Using camera trapping and hierarchical occupancy modelling to evaluate the spatial ecology of an African mammal community. J. Appl. Ecol. 53, 1225-1235. (doi:10.1111/1365-2664.12650) Crossref, Web of Science, Google Scholar - 8.
Beaudrot L 2016 Standardized assessment of biodiversity trends in tropical forest protected areas: the end is not in sight. PLoS Biol. 14, e1002357. (doi:10.1371/journal.pbio.1002357) Crossref, PubMed, Web of Science, Google Scholar - 9.
McShea WJ, Forrester T, Costello R, He Z, Kays R . 2016 Volunteer-run cameras as distributed sensors for macrosystem mammal research. Landsc. Ecol. 31, 55-66. (doi:10.1007/s10980-015-0262-9) Crossref, Web of Science, Google Scholar - 10.
O'Brien TG, Baillie JEM, Krueger L, Cuke M . 2010 The Wildlife Picture Index: monitoring top trophic levels. Anim. Conserv. 13, 335-343. (doi:10.1111/j.1469-1795.2010.00357.x) Crossref, Web of Science, Google Scholar - 11.
Silveira L, Jácomo ATA, Diniz-Filho JAF . 2003 Camera trap, line transect census and track surveys: a comparative evaluation. Biol. Conserv. 114, 351-355. (doi:10.1016/S0006-3207(03)00063-6) Crossref, Web of Science, Google Scholar - 12.
Paull DJ, Claridge AW, Cunningham RB . 2012 Effective detection methods for medium-sized ground-dwelling mammals: a comparison between infrared digital cameras and hair tunnels. Wildl. Res. 39, 546-553. (doi:10.1071/WR12034) Crossref, Web of Science, Google Scholar - 13.
Espartosa KD, Pinotti BT, Pardini R . 2011 Performance of camera trapping and track counts for surveying large mammals in rainforest remnants. Biodivers. Conserv. 20, 2815-2829. (doi:10.1007/s10531-011-0110-4) Crossref, Web of Science, Google Scholar - 14.
Clare J, McKinney ST, Depue JE, Loftin CS . 2017 Pairing field methods to improve inference in wildlife surveys while accommodating detection covariance. Ecol. Appl. 27, 2031-2047. (doi:10.1002/eap.1587) Crossref, PubMed, Web of Science, Google Scholar - 15.
De Bondi N, White JG, Stevens M, Cooke R . 2010 A comparison of the effectiveness of camera trapping and live trapping for sampling terrestrial small-mammal communities. Wildl. Res. 37, 456-465. (doi:10.1071/WR10046) Crossref, Web of Science, Google Scholar - 16.
Larrucea ES, Brussard PF . 2008 Efficiency of various methods used to detect presence of pygmy rabbits in summer. West North Am. Nat. 68, 303-310. (doi:10.3398/1527-0904(2008)68[303:EOVMUT]2.0.CO;2) Crossref, Web of Science, Google Scholar - 17.
Ford AT, Clevenger AP, Bennett A . 2009 Comparison of methods of monitoring wildlife crossing-structures on highways. J. Wildl. Manage. 73, 1213-1222. (doi:10.2193/2008-387) Crossref, Web of Science, Google Scholar - 18.
Güthlin D, Storch I, Küchenhoff H . 2014 Toward reliable estimates of abundance: comparing index methods to assess the abundance of a mammalian predator. PLoS ONE 9, e94537. (doi:10.1371/journal.pone.0094537) Crossref, PubMed, Web of Science, Google Scholar - 19.
Villette P, Krebs CJ, Jung TS, Boonstra R . 2016 Can camera trapping provide accurate estimates of small mammal (Myodes rutilus and Peromyscus maniculatus) density in the boreal forest? J. Mammal. 97, 32-40. (doi:10.1093/jmammal/gyv150) Crossref, Web of Science, Google Scholar - 20.
Balme GA, Hunter LTB, Slotow R . 2009 Evaluating methods for counting cryptic carnivores. J. Wildl. Manage. 73, 433-441. (doi:10.2193/2007-368) Crossref, Web of Science, Google Scholar - 21.
Anile S, Ragni B, Randi E, Mattucci F, Rovero F . 2014 Wildcat population density on the Etna volcano, Italy: a comparison of density estimation methods. J. Zool. 293, 252-261. (doi:10.1111/jzo.12141) Crossref, Web of Science, Google Scholar - 22.
Spehar SN, Loken B, Rayadin Y, Royle JA . 2015 Comparing spatial capture-recapture modeling and nest count methods to estimate orangutan densities in the Wehea Forest, East Kalimantan, Indonesia. Biol. Conserv. 191, 185-193. (doi:10.1016/j.biocon.2015.06.013) Crossref, Web of Science, Google Scholar - 23.
Meek PD, Ballard GA, Fleming PJS, Schaefer M, Williams W, Falzon G . 2014 Camera traps can be heard and seen by animals. PLoS ONE 9, e110832. (doi:10.1371/journal.pone.0110832) Crossref, PubMed, Web of Science, Google Scholar - 24.
Long RA, Donovan TM, Mackay P, Zielinski WJ, Buzas JS . 2007 Comparing scat detection dogs, cameras, and hair snares for surveying carnivores. J. Wildl. Manage. 71, 2018-2025. (doi:10.2193/2006-292) Crossref, Web of Science, Google Scholar - 25.
Hackett HM, Lesmeister DB, Desanty-Combes J, Montague WG, Millspaugh JJ, Gompper ME . 2007 Detection rates of eastern spotted skunks (Spilogale putorius) in Missouri and Arkansas using live-capture and non-invasive techniques. Am. Midl. Nat. 158, 123-131. (doi:10.1674/0003-0031(2007)158[123:DROESS]2.0.CO;2) Crossref, Web of Science, Google Scholar - 26.
Garrote G, de Ayala RP, Tellería JL . 2014 A comparison of scat counts and camera-trapping as means of assessing Iberian lynx abundance. Eur. J. Wildl. Res. 60, 885-889. (doi:10.1007/s10344-014-0855-7) Crossref, Web of Science, Google Scholar - 27.
Glover-Kapfer P, Soto-Navarro CA, Wearn OR . 2019 Camera-trapping version 3.0: current constraints and future priorities for development. Remote Sens. Ecol. Conserv. 1-15. (doi:10.1002/rse2.106) Web of Science, Google Scholar - 28.
Hebblewhite M, Haydon DT . 2010 Distinguishing technology from biology: a critical review of the use of GPS telemetry data in ecology. Phil. Trans. R. Soc. B 365, 2303-2312. (doi:10.1098/rstb.2010.0087) Link, Web of Science, Google Scholar - 29.
Luskin MS, Lee JSH, Edwards DP, Gibson L, Potts MD . 2018 Study context shapes recommendations of land-sparing and sharing; a quantitative review. Glob. Food Sec. 16, 29-35. (doi:10.1016/j.gfs.2017.08.002) Crossref, Web of Science, Google Scholar - 30.
Hedges L V, Gurevitch J, Curtis PS . 1999 The meta-analysis of response ratios in experimental ecology. Ecology 80, 1150-1156. (doi:10.1890/0012-9658(1999)080[1150:TMAORR]2.0.CO;2) Crossref, Web of Science, Google Scholar - 31.
Nakagawa S, Santos ESA . 2012 Methodological issues and advances in biological meta-analysis. Evol. Ecol. 26, 1253-1274. (doi:10.1007/s10682-012-9555-5) Crossref, Web of Science, Google Scholar - 32.
Crouzeilles R, Curran M, Ferreira MS, Lindenmayer DB, Grelle CEV, Benayas JM . 2016 A global meta-analysis on the ecological drivers of forest restoration success. Nat. Commun. 7, 11666. (doi:10.1038/ncomms11666) Crossref, PubMed, Web of Science, Google Scholar - 33.
Guillera-Arroita G, Lahoz-Monfort JJ . 2012 Designing studies to detect differences in species occupancy: power analysis under imperfect detection. Methods Ecol. Evol. 3, 860-869. (doi:10.1111/j.2041-210X.2012.00225.x) Crossref, Web of Science, Google Scholar - 34.
Gurevitch J, Hedges LV . 1999 Statistical issues in ecological meta-analysis. Ecology 80, 1142-1149. (doi:10.1890/0012-9658(1999)080[1142:SIIEMA]2.0.CO;2) Crossref, Web of Science, Google Scholar - 35.
Zuur AF, Ieno EN, Elphick CS . 2010 A protocol for data exploration to avoid common statistical problems. Methods Ecol. Evol. 1, 3-14. (doi:10.1111/j.2041-210X.2009.00001.x) Crossref, Web of Science, Google Scholar - 36.
Christensen RHB . 2015 ordinal: Regression models for ordinal data. R package version 2015.6-28. See http://cran.r-project.org/package=ordinal. Google Scholar - 37.
Bartoń K . 2018 MuMIn: Multi-model inference. R package version 1.40.4. See http://cran.r-project.org/package=cooccur. Google Scholar - 38.
Gelman A . 2008 Scaling regression inputs by dividing by two standard deviations. Stat. Med. 27, 2865-2873. (doi:10.1002/sim.3107) Crossref, PubMed, Web of Science, Google Scholar - 39.
Burnham KP, Anderson DR . 2002 Model selection and multimodel inference: a practical information-theoretical approach, 2nd edn. New York, NY: Springer. Google Scholar - 40.
Galipaud M, Gillingham MAF, Dechaume-Moncharmont FX . 2017 A farewell to the sum of Akaike weights: the benefits of alternative metrics for variable importance estimations in model selection. Methods Ecol. Evol. 8, 1668-1678. (doi:10.1111/2041-210X.12835) Crossref, Web of Science, Google Scholar - 41.
Giam X, Olden JD . 2016 Quantifying variable importance in a multimodel inference framework. Methods Ecol. Evol. 7, 388-397. (doi:10.1111/2041-210X.12492) Crossref, Web of Science, Google Scholar - 42.
Grueber CE, Nakagawa S, Laws RJ, Jamieson IG . 2011 Multimodel inference in ecology and evolution: challenges and solutions. J. Evol. Biol. 24, 699-711. (doi:10.1111/j.1420-9101.2010.02210.x) Crossref, PubMed, Web of Science, Google Scholar - 43.
Cade BS . 2015 Model averaging and muddled multimodel inferences. Ecology 96, 2370-2382. (doi:10.1890/14-1639.1) Crossref, PubMed, Web of Science, Google Scholar - 44.
Banner KM, Higgs MD . 2017 Considerations for assessing model averaging of regression coefficients. Ecol. Appl. 27, 78-93. (doi:10.1002/eap.1419) Crossref, PubMed, Web of Science, Google Scholar - 45.
Bates D, Maechler M, Bolker B, Walker S . 2015 Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1-48. (doi:10.18637/jss.v067.i01) Crossref, Web of Science, Google Scholar - 46. R Development Core Team. 2017 R: a language and environment for statistical computing. Google Scholar
- 47.
Burton AC, Neilson E, Moreira D, Ladle A, Steenweg R, Fisher JT, Bayne E, Boutin S . 2015 Wildlife camera trapping: a review and recommendations for linking surveys to ecological processes. J. Appl. Ecol. 52, 675-685. (doi:10.1111/1365-2664.12432) Crossref, Web of Science, Google Scholar - 48.
Bull EL, Holthausen RS, Bright LR . 1992 Comparison of 3 techniques to monitor marten. Wildl. Soc. Bull. 20, 406-410. Web of Science, Google Scholar - 49.
Driessen MM, Jarman PJ . 2014 Comparison of camera trapping and live trapping of mammals in Tasmanian coastal woodland and heathland. In Camera trapping: wildlife management and research (edsMeek PD, Fleming PJS, Ballard GA, Banks PB, Claridge AW, Sanderson J, Swann D ), pp. 253-262. Melbourne, Australia: CSIRO Publishing. Google Scholar - 50.
Rockhill AP, Sollman R, Powell RA, DePerno CS . 2016 A comparison of survey techniques for medium- to large-sized mammals in forested wetlands. Southeast Nat. 15, 175-187. (doi:10.1656/058.015.0112) Crossref, Web of Science, Google Scholar - 51.
Swan M, Di Stefano J, Christie F, Steel E, York A . 2014 Detecting mammals in heterogeneous landscapes: implications for biodiversity monitoring and management. Biodivers. Conserv. 23, 343-355. (doi:10.1007/s10531-013-0604-3) Crossref, Web of Science, Google Scholar - 52.
Jumeau J, Petrod L, Handrich Y . 2017 A comparison of camera trap and permanent recording video camera efficiency in wildlife underpasses. Ecol. Evol. 7, 7399-7407. (doi:10.1002/ece3.3149) Crossref, PubMed, Web of Science, Google Scholar - 53.
Molyneux J, Pavey CR, James AI, Carthew SM . 2017 The efficacy of monitoring techniques for detecting small mammals and reptiles in arid environments. Wildl. Res. 44, 534-545. (doi:10.1071/WR17017) Crossref, Web of Science, Google Scholar - 54.
Welbourne DJ, MacGregor C, Paull D, Lindenmayer DB . 2015 The effectiveness and cost of camera traps for surveying small reptiles and critical weight range mammals: a comparison with labour-intensive complementary methods. Wildl. Res. 42, 414-425. (doi:10.1071/WR15054) Crossref, Web of Science, Google Scholar - 55.
McCallum J . 2013 Changing use of camera traps in mammalian field research: habitats, taxa and study types. Mamm. Rev. 43, 196-206. (doi:10.1111/j.1365-2907.2012.00216.x) Crossref, Web of Science, Google Scholar - 56.
Di Marco M 2017 Changing trends and persisting biases in three decades of conservation science. Glob. Ecol. Conserv. 10, 32-42. (doi:10.1016/j.gecco.2017.01.008) Crossref, Web of Science, Google Scholar - 57.
Rowcliffe JM, Carbone C . 2008 Surveys using camera traps: are we looking to a brighter future? Anim. Conserv. 11, 185-186. (doi:10.1111/j.1469-1795.2008.00180.x) Crossref, Web of Science, Google Scholar - 58.
Steenweg R 2017 Scaling-up camera traps: monitoring the planet's biodiversity with networks of remote sensors. Front. Ecol. Environ. 15, 26-34. (doi:10.1002/fee.1448) Crossref, Web of Science, Google Scholar - 59.
Wearn OR, Glover-Kapfer P . 2018Supporting data for ‘Snap happy: camera traps are an effective sampling tool when compared to alternative methods’. Zenodo (http://doi.org/10.5281/zenodo.1324998) Google Scholar - 60.
Jones LLC, Raphael MG . 1990 Ecology and management of marten in fragmented habitats of the Pacific Northwest—Progress Report FY90. 60 p. Google Scholar - 61.
Fowler CH, Golightly RT . 1993 Fisher and marten survey techniques on the Tahoe National Forest. Unpublished report. Google Scholar - 62.
Laymon SA, Holterman MD, Barrett RH . 1993 Distribution of marten, fisher, wolverine, and Sierra Nevada red fox in portions of the Sierra National Forest. Unpublished report. Google Scholar - 63.
Zielinski WJ, Kucera TE . 1995 American marten, fisher, lynx, and wolverine: survey methods for their detection. Collingdale, PA: DIANE Publishing. 163 p. Google Scholar - 64.
Koerth BH, Mckown CD, Kroll JC . 1997 Infrared-triggered camera versus helicopter counts of white-tailed deer. Wildl. Soc. Bull. 25, 557-562. Web of Science, Google Scholar - 65.
Foresman KR, Pearson DE . 1998 Comparison of proposed survey procedures for detection of forest carnivores. J. Wildl. Manage. 62, 1217-1226. (doi:10.2307/3801985) Crossref, Web of Science, Google Scholar - 66.
Noyce KV, Garshelis DL, Coy PL . 2001 Differential vulnerability of black bears to trap and camera sampling and resulting biases in mark-recapture estimates. Ursus 12, 211-225. Google Scholar - 67.
Harrison RL, Barr DJ, Dragoo JW . 2002 A comparison of population survey techniques for swift foxes (Vulpes velox) in New Mexico. Am. Midl. Nat. 148, 320-337. (doi:10.1674/0003-0031(2002)148[0320:ACOPST]2.0.CO;2) Crossref, Web of Science, Google Scholar - 68.
Jeganathan P, Green RE, Bowden CGR, Norris K, Pain D, Rahmani A . 2002 Use of tracking strips and automatic cameras for detecting critically endangered Jerdon's coursers Rhinoptilus bitorquatus in scrub jungle in Andhra Pradesh, India. Oryx 36, 182-188. (doi:10.1017/S003060530200025X) Crossref, Web of Science, Google Scholar - 69.
Srbek-Araujo AC, Chiarello AG . 2005 Is camera-trapping an efficient method for surveying mammals in Neotropical forests? A case study in south-eastern Brazil. J. Trop. Ecol. 21, 121-125. (doi:10.1017/S0266467404001956) Crossref, Web of Science, Google Scholar - 70.
Gompper ME, Kays RW, Ray JC, Lapoint SD, Bogan DA, Cryan JR . 2006 A comparison of noninvasive techniques to survey carnivore communities in northeastern North America. Wildl. Soc. Bull. 34, 1142-1151. (doi:10.2193/0091-7648(2006)34[1142:ACONTT]2.0.CO;2) Crossref, Web of Science, Google Scholar - 71.
Harrison RL . 2006 A comparison of survey methods for detecting bobcats. Wildl. Soc. Bull. 34, 548-552. (doi:10.2193/0091-7648(2006)34[548:ACOSMF]2.0.CO;2) Crossref, Web of Science, Google Scholar - 72.
Roberts CW, Pierce BL, Braden AW, Lopez RR, Silvy NJ, Frank PA, Ransom D . 2006 Comparison of camera and road survey estimates for white-tailed deer. J. Wildl. Manage. 70, 263-267. (doi:10.2193/0022-541X(2006)70[263:COCARS]2.0.CO;2) Crossref, Web of Science, Google Scholar - 73.
Barea-Azcón JM, Virgós E, Ballesteros-Duperón E, Moleón M, Chirosa M . 2007 Surveying carnivores at large spatial scales: a comparison of four broad-applied methods. Biodivers. Conserv. 16, 1213-1230. (doi:10.1007/s10531-006-9114-x) Crossref, Web of Science, Google Scholar - 74.
King CM, McDonald RM, Martin RD, Tempero GW, Holmes SJ . 2007 Long-term automated monitoring of the distribution of small carnivores. Wildl. Res. 34, 140-148. (doi:10.1071/WR05091) Crossref, Web of Science, Google Scholar - 75.
Dillon A, Kelly MJ . 2008 Ocelot home range, overlap and density: comparing radio telemetry with camera trapping. J. Zool. 275, 391-398. (doi:10.1111/j.1469-7998.2008.00452.x) Crossref, Web of Science, Google Scholar - 76.
Lyra-Jorge MC, Ciocheti G, Pivello VR, Meirelles ST . 2008 Comparing methods for sampling large- and medium-sized mammals: camera traps and track plots. Eur. J. Wildl. Res. 54, 739-744. (doi:10.1007/s10344-008-0205-8) Crossref, Web of Science, Google Scholar - 77.
Trolle M, Noss AJ, Passos Cordeiro JL, Oliveira LF . 2008 Brazilian tapir density in the Pantanal: a comparison of systematic camera-trapping and line-transect surveys. Biotropica 40, 211-217. (doi:10.1111/j.1744-7429.2007.00350.x) Crossref, Web of Science, Google Scholar - 78.
Can ÖE, Togan I . 2009 Camera trapping of large mammals in Yenice Forest, Turkey: local information versus camera traps. Oryx 43, 427-430. (doi:10.1017/S0030605308000628) Crossref, Web of Science, Google Scholar - 79.
Lucherini M, Reppucci J, Vidal E 2009 A comparison of three methods to estimate variations in the relative abundance of moutain vizcachas (Lagidium viscacia) in the High Andes ecosystems. Mastozool. Neotrop. 16, 223-228. Google Scholar - 80.
Rovero F, Marshall AR . 2009 Camera trapping photographic rate as an index of density in forest ungulates. J. Appl. Ecol. 46, 1011-1017. (doi:10.1111/j.1365-2664.2009.01705.x) Crossref, Web of Science, Google Scholar - 81.
Vine SJ, Crowther MS, Lapidge SJ, Dickman CR, Mooney N, Piggott MP, English AW . 2009 Comparison of methods to detect rare and cryptic species: a case study using the red fox (Vulpes vulpes). Wildl. Res. 36, 436. (doi:10.1071/WR08069) Crossref, Web of Science, Google Scholar - 82.
Perry TW, Newman T, Thibault KM . 2010 Evaluation of methods used to estimate size of a population of desert bighorn sheep (Ovis canadensis mexicana) in New Mexico. Southwest Nat. 55, 517-524. (doi:10.1894/SGM-07.1) Crossref, Web of Science, Google Scholar - 83.
Comer CE, Symmank ME, Kroll JC . 2011 Bobcats do not exhibit rub response despite presence at hair collection stations. Wildl. Biol. Pract. 7, 116-122. (doi:10.2461/wbp.2011.7.9) Crossref, Google Scholar - 84.
Janečka JE, Munkhtsog B, Jackson RM, Naranbaatar G, Mallon DP, Murphy WJ . 2011 Comparison of noninvasive genetic and camera-trapping techniques for surveying snow leopards. J. Mammal. 92, 771-783. (doi:10.1644/10-MAMM-A-036.1) Crossref, Web of Science, Google Scholar - 85.
Morgan TW, Elliott CL . 2011 Comparison of remotely-triggered cameras vs. howling surveys for estimating coyote (Canis latrans) abundance in central Kentucky. J. Ky. Acad. Sci. 72, 84-90. Google Scholar - 86.
Munari DP, Keller C, Venticinque EM . 2011 An evaluation of field techniques for monitoring terrestrial mammal populations in Amazonia. Mamm. Biol. 76, 401-408. (doi:10.1016/j.mambio.2011.02.007) Crossref, Web of Science, Google Scholar - 87.
Núñez-Pérez R . 2011 Estimating jaguar population density using camera-traps: a comparison with radio-telemetry estimates. J. Zool. 285, 39-45. Web of Science, Google Scholar - 88.
Roberts NJ . 2011 Investigation into survey techniques of large mammals: surveyor competence and camera-trapping vs. transect-sampling. Biosci. Horizons 4, 40-49. Crossref, Google Scholar - 89.
Bartolommei P, Manzo E, Cozzolino R . 2012 Evaluation of three indirect methods for surveying European pine marten in a forested area of central Italy. Hystrix 23, 90-92. Web of Science, Google Scholar - 90.
DeSa MA, Zweig CL, Percival HF, Kitchens WM, Kasbohm JW . 2012 Comparison of small-mammal sampling techniques in tidal salt marshes of the central Gulf Coast of Florida. Southeast Nat. 11, 89-100. (doi:10.1656/058.011.0109) Crossref, Web of Science, Google Scholar - 91.
Li S, McShea WJ, Wang D, Huang J, Shao L . 2012 A direct comparison of camera-trapping and sign transects for monitoring wildlife in the Wanglang National Nature Reserve, China. Wildl. Soc. Bull. 36, 538-545. (doi:10.1002/wsb.161) Crossref, Web of Science, Google Scholar - 92.
Di Cerbo AR, Biancardi CM . 2013 Monitoring small and arboreal mammals by camera traps: effectiveness and applications. Acta Theriol. (Warsz) 58, 279-283. (doi:10.1007/s13364-012-0122-9) Crossref, Google Scholar - 93.
Head JS, Boesch C, Robbins MM, Rabanal LI, Makaga L, Kühl HS . 2013 Effective sociodemographic population assessment of elusive species in ecology and conservation management. Ecol. Evol. 3, 2903-2916. (doi:10.1002/ece3.670) Crossref, PubMed, Web of Science, Google Scholar - 94.
Kluever BM, Gese EM, Dempsey SJ, Knight RN . 2013 A comparison of methods for monitoring kit foxes at den sites. Wildl. Soc. Bull. 37, 439-443. (doi:10.1002/wsb.261) Crossref, Web of Science, Google Scholar - 95.
Ribeiro P, de Melo FR . 2013 Mamíferos de médio e grande porte de uma área agricultável em Terezópolis (GO) com notas sobre métodos de amostragem. Neotrop. Biol. Conserv. 8, 68-78. (doi:10.4013/nbc.2013.82.02) Crossref, Google Scholar - 96.
Zero VH, Sundaresan SR, O'Brien TG, Kinnaird MF . 2013 Monitoring an endangered savannah ungulate, Grevy's zebra Equus grevyi: choosing a method for estimating population densities. Oryx 47, 410-419. (doi:10.1017/S0030605312000324) Crossref, Web of Science, Google Scholar - 97.
Ariefiandy A, Purwandana D, Seno A, Chrismiawati M, Jessop TS . 2014 Evaluation of three field monitoring-density estimation protocols and their relevance to Komodo dragon conservation. Biodivers. Conserv. 23, 2473-2490. (doi:10.1007/s10531-014-0733-3) Crossref, Web of Science, Google Scholar - 98.
Castleberry SB, Mengak MT, Menken TE . 2014 Comparison of trapping and camera survey methods for determining presence of Allegheny woodrats. Wildl. Soc. Bull. 38, 414-418. (doi:10.1002/wsb.383) Crossref, Web of Science, Google Scholar - 99.
Duquette JF, Belant JL, Svoboda NJ, Beyer DE, Albright CA . 2014 Comparison of occupancy modeling and radiotelemetry to estimate ungulate population dynamics. Popul. Ecol. 56, 481-492. (doi:10.1007/s10144-014-0432-7) Crossref, Web of Science, Google Scholar - 100.
Fisher JT, Bradbury S . 2014 A multi-method hierarchical modeling approach to quantifying bias in occupancy from noninvasive genetic tagging studies. J. Wildl. Manage. 78, 1087-1095. (doi:10.1002/jwmg.750) Crossref, Web of Science, Google Scholar - 101.
Glen A, Warburton B, Cruz J, Coleman M . 2014 Comparison of camera traps and kill traps for detecting mammalian predators: a field trial. New Zeal. J. Zool. 41, 155-160. (doi:10.1080/03014223.2014.898667) Crossref, Web of Science, Google Scholar - 102.
Gužvica G, Bošnjak I, Bielen A, Babić D, Radanović-Gužvica B, Šver L . 2014 Comparative analysis of three different methods for monitoring the use of green bridges by wildlife. PLoS ONE 9, e106194. (doi:10.1371/journal.pone.0106194) Crossref, PubMed, Web of Science, Google Scholar - 103.
Monterroso P, Rich LN, Serronha A, Ferreras P, Alves PC . 2014 Efficiency of hair snares and camera traps to survey mesocarnivore populations. Eur. J. Wildl. Res. 60, 279-289. (doi:10.1007/s10344-013-0780-1) Crossref, Web of Science, Google Scholar - 104.
Rodgers TW, Giacalone J, Heske EJ, Janečka JE, Phillips CA, Schooley RL . 2014 Comparison of noninvasive genetics and camera trapping for estimating population density of ocelots (Leopardus pardalis) on Barro Colorado Island, Panama. Trop. Conserv. Sci. 777, 690-705. (doi:10.1177/194008291400700408) Crossref, Web of Science, Google Scholar - 105.
Clare JDJ, Anderson EM, MacFarland DM, Sloss BL . 2015 Comparing the costs and detectability of bobcat using scat-detecting dog and remote camera surveys in central Wisconsin. Wildl. Soc. Bull. 39, 210-217. (doi:10.1002/wsb.502) Crossref, Web of Science, Google Scholar - 106.
Harrison RL . 2015 A comparison of sign searches, live-trapping, and camera-trapping for detection of American badgers (Taxidea taxus) in the Chihuahuan Desert. West North Am. Nat. 75, 387-395. (doi:10.3398/064.075.0409) Crossref, Web of Science, Google Scholar - 107.
Jordan MJ, Lobb-Rabe M . 2015 An evaluation of methods to attract urban mesocarnivores to track plates and camera traps. Northwest Sci. 89, 383-392. (doi:10.3955/046.089.0406) Crossref, Web of Science, Google Scholar - 108.
Rydell J, Russo D . 2015 Photography as a low-impact method to survey bats. Mamm. Biol. 80, 182-184. (doi:10.1016/j.mambio.2014.11.003) Crossref, Web of Science, Google Scholar - 109.
Velli E, Bologna MA, Silvia C, Ragni B, Randi E . 2015 Non-invasive monitoring of the European wildcat (Felis silvestris silvestris Schreber, 1777): comparative analysis of three different monitoring techniques and evaluation of their integration. Eur. J. Wildl. Res. 61, 657-668. (doi:10.1007/s10344-015-0936-2) Crossref, Web of Science, Google Scholar - 110.
Bowler MT, Tobler MW, Endress BA, Gilmore MP, Anderson MJ . 2017 Estimating mammalian species richness and occupancy in tropical forest canopies with arboreal camera traps. Remote Sens. Ecol. Conserv. 3, 146-157. (doi:10.1002/rse2.35) Crossref, Web of Science, Google Scholar - 111.
Carvalho WD, Rosalino LM, Adania CH, Esbérard CEL . 2016 Mammal inventories in seasonal Neotropical forests: traditional approaches still compensate drawbacks of modern technologies. Iheringia Série Zool. 106, e2016005. (doi:10.1590/1678-4766e2016005) Web of Science, Google Scholar - 112.
Day CC, Westover MD, Hall LK, Larsen RT, McMillan BR . 2016 Comparing direct and indirect methods to estimate detection rates and site use of a cryptic semi-aquatic carnivore. Ecol. Indic. 66, 230-234. (doi:10.1016/j.ecolind.2016.01.039) Crossref, Web of Science, Google Scholar - 113.
Diggins CA, Gilley LM, Kelly CA, Ford WM . 2016 Comparison of survey techniques on detection of northern flying squirrels. Wildl. Soc. Bull. 40, 654-662. (doi:10.1002/wsb.715) Crossref, Web of Science, Google Scholar - 114.
Dupuis-Désormeaux M, Davidson Z, Mwololo M, Kisio E, MacDonald SE . 2016 Comparing motion capture cameras versus human observer monitoring of mammal movement through fence gaps: a case study from Kenya. Afr. J. Ecol. 54, 154-161. (doi:10.1111/aje.12277) Crossref, Web of Science, Google Scholar - 115.
Glen AS, Anderson D, Veltman CJ, Garvey PM, Nichols M . 2016 Wildlife detector dogs and camera traps: a comparison of techniques for detecting feral cats. New Zeal. J. Zool. 43, 127-137. (doi:10.1080/03014223.2015.1103761) Crossref, Web of Science, Google Scholar - 116.
Greene DU, McCleery RA, Wagner LM, Garrison EP . 2016 A comparison of four survey methods for detecting fox squirrels in the southeastern United States. J. Fish Wildl. Manag. 7, 99-106. (doi:10.3996/082015-JFWM-080) Crossref, Web of Science, Google Scholar - 117.
Hausser Y, Tagand R, Vimercati E, Mermod S, Fischer C . 2017 Comparing survey methods to assess the conservation value of a community-managed protected area in western Tanzania. Afr. J. Ecol. 55, 1-11. (doi:10.1111/aje.12301) Crossref, Web of Science, Google Scholar - 118.
Lee P-S, Gan HM, Clements GR, Wilson J-J, Adamowicz S . 2016 Field calibration of blowfly-derived DNA against traditional methods for assessing mammal diversity in tropical forests. Genome 59, 1008-1022. (doi:10.1139/gen-2015-0193) Crossref, PubMed, Web of Science, Google Scholar - 119.
Mills CA, Godley BJ, Hodgson DJ . 2016 Take only photographs, leave only footprints: novel applications of non-invasive survey methods for rapid detection of small, arboreal animals. PLoS ONE 11, e0146142. (doi:10.1371/journal.pone.0146142) Crossref, PubMed, Web of Science, Google Scholar - 120.
Pirie TJ, Thomas RL, Fellowes MDE . 2016 Limitations to recording larger mammalian predators in savannah using camera traps and spoor. Wildl. Biol. 22, 13-21. (doi:10.2981/wlb.00129) Crossref, Web of Science, Google Scholar - 121.
Rahman DA, Gonzalez G, Aulagnier S . 2016 Benefit of camera trapping for surveying the critically endangered Bawean deer Axis kuhlii (Temminck, 1836). Trop. Zool. 29, 155-172. (doi:10.1080/03946975.2016.1199763) Crossref, Web of Science, Google Scholar - 122.
Wearn OR, Carbone C, Rowcliffe JM, Bernard H, Ewers RM . 2016 Grain-dependent responses of mammalian diversity to land-use and the implications for conservation set-aside. Ecol. Appl. 26, 1409-1420. (doi:10.1890/15-1363) Crossref, PubMed, Web of Science, Google Scholar - 123.
Whitworth A, Dominie Braunholtz L, Huarcaya RP, Macleod R, Beirne C . 2016 Out on a limb: arboreal camera traps as an emerging methodology for inventorying elusive rainforest mammals. Trop. Conserv. Sci. 9, 675-698. (doi:10.1177/194008291600900208) Crossref, Web of Science, Google Scholar - 124.
Caruso N, Luengos Vidal E, Guerisoli M, Lucherini M . 2017 Carnivore occurrence: do interview-based surveys produce unreliable results? Oryx 51, 240-245. (doi:10.1017/S0030605315001192) Crossref, Web of Science, Google Scholar - 125.
Chitwood MC 2017 Are camera surveys useful for assessing recruitment in white-tailed deer? Wildl. Biol. 1, wlb.00178. (doi:10.2981/wlb.00178) Crossref, Web of Science, Google Scholar - 126.
Després-Einspenner ML, Howe EJ, Drapeau P, Kühl HS . 2017 An empirical evaluation of camera trapping and spatially explicit capture-recapture models for estimating chimpanzee density. Am. J. Primatol. 79, 1-12. (doi:10.1002/ajp.22647) Crossref, Web of Science, Google Scholar - 127.
Ellis KS, Larsen RT, Whiting JC, Wilson TL, McMillan BR . 2017 Assessing indirect measures of abundance and distribution with remote cameras: simplifying indices of activity at pygmy rabbit burrows. Ecol. Indic. 77, 23-30. (doi:10.1016/j.ecolind.2017.01.041) Crossref, Web of Science, Google Scholar - 128.
Humphrey JE, Robert KA, Leonard SWJ . 2017 Elliott traps found to be ineffective for the survey of swamp skink (Lissolepis coventryi): a cautionary tale of outdated survey guidelines. Wildl. Res. 44, 514-522. (doi:10.1071/WR17012) Crossref, Web of Science, Google Scholar - 129.
Ishige T, Miya M, Ushio M, Sado T, Ushioda M, Maebashi K, Yonechi R, Lagan P, Matsubayashi H . 2017 Tropical-forest mammals as detected by environmental DNA at natural saltlicks in Borneo. Biol. Conserv. 210, 281-285. (doi:10.1016/j.biocon.2017.04.023) Crossref, Web of Science, Google Scholar - 130.
Mazzolli M, Haag T, Lippert BG, Eizirik E, Hammer ML, Al Hikmani K . 2017 Multiple methods increase detection of large and medium-sized mammals: working with volunteers in south-eastern Oman. Oryx 51, 290-297. (doi:10.1017/S0030605315001003) Crossref, Web of Science, Google Scholar - 131.
Pepper MA, Herrmann V, Hines JE, Nichols JD, Kendrot SR . 2017 Evaluation of nutria (Myocastor coypus) detection methods in Maryland, USA. Biol. Invasions 19, 831-841. (doi:10.1007/s10530-016-1312-1) Crossref, Web of Science, Google Scholar - 132.
Peres PHF, Polverini MS, Oliveira ML, Duarte JMB . 2017 Accessing camera trap survey feasibility for estimating Blastocerus dichotomus (Cetartiodactyla, Cervidae) demographic parameters. Iheringia Sér. Zool. 107, e2017041. (doi:10.1590/1678-4766e2017041) Crossref, Web of Science, Google Scholar - 133.
Pfeffer SE, Spitzer R, Allen AM, Hofmeester TR, Ericsson G, Widemo F, Singh NJ, Cromsigt JP . 2018 Pictures or pellets? Comparing camera trapping and dung counts as methods for estimating population densities of ungulates. Remote Sens. Ecol. Conserv. 4, 173-183. (doi:10.1002/rse2.67) Crossref, Web of Science, Google Scholar - 134.
Riley M, Soutyrina S, Miquelle D, Hayward G, Goodrich J, Buskirk S . 2017 Comparison of methods for estimating Amur tiger abundance. Wildl. Biol. 1, wlb.00253. (doi:10.2981/wlb.00253) Crossref, Web of Science, Google Scholar - 135.
Rodgers TW 2017 Carrion fly-derived DNA metabarcoding is an effective tool for mammal surveys: evidence from a known tropical mammal community. Mol. Ecol. Resour. 17, e133-e145. (doi:10.1111/1755-0998.12701) Crossref, PubMed, Web of Science, Google Scholar - 136.
Smith DHV, Weston KA . 2017 Capturing the cryptic: a comparison of detection methods for stoats (Mustela erminea) in alpine habitats. Wildl. Res. 44, 418-426. (doi:10.1071/WR16159) Crossref, Web of Science, Google Scholar - 137.
Soofi M . 2017 Precision and reliability of indirect population assessments for the Caspian red deer Cervus elaphus maral. Wildlife Biol. 2017, wlb.00230. (doi:10.2981/wlb.00230) Crossref, Web of Science, Google Scholar - 138.
Torrents-Ticó M, Rich L, McNutt JW, Nthomiwa M, Mothala M, Motsamai G, Jordan NR . 2017 On the right track? Comparing concurrent spoor and camera-trap surveys in Botswana. Afr. J. Wildl. Res. 47, 128-137. (doi:10.3957/056.047.0128) Crossref, Web of Science, Google Scholar - 139.
Villette P, Krebs CJ, Jung TS . 2017 Evaluating camera traps as an alternative to live trapping for estimating the density of snowshoe hares (Lepus americanus) and red squirrels (Tamiasciurus hudsonicus). Eur. J. Wildl. Res. 63, 1-9. (doi:10.1007/s10344-016-1064-3) Crossref, Web of Science, Google Scholar - 140.
Zak AA, Riley EP . 2017 Comparing the use of camera traps and farmer reports to study crop feeding behavior of moor macaques (Macaca maura). Int. J. Primatol. 38, 224-242. (doi:10.1007/s10764-016-9945-6) Crossref, Web of Science, Google Scholar
