Although camera trapping has been shown to be a highly effective non-invasive tool for wildlife monitoring, the technique has not yet been widely applied to studies of arboreal species. Despite the unique challenges that camera trapping in the canopy poses, its versatility and relatively non-invasive nature, combined with recent technological improvements on the cameras themselves, make camera trapping a highly useful tool for arboreal research.
We present data on the methodology and effectiveness of arboreal camera trapping during the first 6 months of a year-long study in the Lower Urubamba Region of Peru investigating animal use of natural crossing points (i.e. branches) over a natural gas pipeline clearing. We placed Reconyx PC800 Hyperfire cameras in 25 crossing points of 13 distinct natural canopy ‘bridges’ at a mean height of 26·8 m.
After 6 months of data collection, we logged 1522 photoevents, recording 20 mammal, 23 bird and four reptile species. An analysis of animal passing events in front of the cameras over time did not suggest any negative response to camera presence. While we found that cameras in the canopy are triggered more frequently by non-target stimuli (e.g. leaves) than cameras on the ground, we demonstrated significantly reduced false triggering following leaf removal within 1·5 m of the camera.
Our results suggest that arboreal camera trapping can provide robust documentation of a diversity of vertebrate species engaged in a variety of activities, and we provide recommendations for other researchers interested in using in this method. This is the most extensive arboreal camera trapping study to date in terms of the length of the study period, the number of cameras being used and the height of the cameras in the trees. Therefore, lessons provided from this experience can be used to improve the design of future arboreal camera trap studies.
Camera trapping has proven to be a cost-effective, non-invasive and relatively low-effort technique for collecting a wide range of ecological data for terrestrial wildlife, yet the technique has not yet been widely applied to the study of arboreal species. Published studies on arboreal mammals generally involve few cameras (one camera: Schipper 2007; Peck 2011; three cameras: Seymour & Batke 2012), low placement in the canopy (2 m: Kierulff et al. 2004; 2–8 m: Olson et al. 2012; 6–14 m: Soanes et al. 2013) and/or relatively low trapping effort (231 trap nights: Ganesh & Devy 2006; 231 nights: Olson et al. 2012). Given the significant challenges associated with observing most arboreal mammals from the ground, arboreal camera trapping has enormous potential for a range of ecological studies. However, to our knowledge, no study has utilized cameras to monitor or survey mammalian wildlife over long periods of time in the high canopy, where many of the least understood mammals dwell.
The potential applications of arboreal camera trapping are similar to those of terrestrial trapping. Particularly in the case of mammals, many species are nocturnal and elusive, making observation challenging. Simply developing a species list for a forest can be a huge challenge, and canopy obligates, particularly smaller species, are more likely to be missed. Like terrestrial camera trapping, arboreal trapping could allow long-term and continuous monitoring of a specific location of interest. For example, use of a flowering or fruiting tree may be monitored through all stages of development (Ganesh & Devy 2006), or use of a particular branch or a crossing point between forest fragments may be observed (Schipper 2007). If a location or travel path is used consistently, there is potential for information on group composition and infant development to be recorded. Finally, cameras provide an opportunity to observe animal condition at close distances, which could reveal evidence of pregnancy or health status, such as the presence of infections or skin parasites (Kelly et al. 2008).
Although camera traps are generally considered to be a non-invasive monitoring tool, there are recognized complications with camera trap use. For example, some cameras, particularly those with a white flash, have been shown to be avoided by animals not only on the ground (Séquin et al. 2003; Wegge, Pokheral & Jnawali 2004) but also in the canopy (Schipper 2007). Cameras on the ground can also be frequently triggered by non-target stimuli (NTS), such as the movement of sun-warmed leaves. Given increased exposure to the wind, and the movement of branches to which cameras are attached, cameras in the canopy may be more susceptible to NTS triggers which can complicate and extend data analysis and management, as well as deplete battery life and memory card space.
In this study, we utilize 6 months of data monitoring natural canopy ‘bridges’ over a pipeline clearing to document the challenges and opportunities of this method and provide focused recommendations for its future use. Our specific objectives were to: (i) document the range of species of all taxonomic groups photographed by the cameras, (ii) investigate potential avoidance of cameras by the most common species photographed, (iii) compare the number of NTS trigger events between arboreal and terrestrial cameras and investigate factors related to NTS events and (iv) document the range of animal behaviours captured with the technique to inform and stimulate future studies. Building on the recent work of Ganesh & Devy (2012), we provide recommendations for maximizing arboreal camera trapping success, and we document specific challenges encountered in the installation and management of cameras in the high canopy including tree climbing, battery and memory management, and wildlife-induced camera damage. This methodological study is part of a larger ongoing study investigating the potential for natural canopy bridges to mitigate fragmentation of arboreal mammal populations by a natural gas pipeline.
Materials and methods
Our study area, Pagoreni, is located within a natural gas concession block (no. 56) in the Lower Urubamba Region of the Peruvian Amazon (11°42′S, 72°48′W, Fig. 1). Pagoreni is in topographically variable terra firme primary forest dominated by Iriartea deltoidea (Araceae) and Pentagonia parvifolia (Rubiaceae). Comiskey et al. (2001) found that the forest was similar to terra firme sites elsewhere in the Amazon basin, with high alpha-diversity, high species richness and a large proportion of tall trees (5·7% of trees in a 1 ha plot were >30 m tall). The study area is characterized by a wet season from May through September and a dry season from October to April, with annual rainfall of 3000–3500 mm (Alonso et al. 2001). The average annual temperature is 27 °C, with 80% average annual humidity (Alonso et al. 2001).
In July 2012, clearing of the right-of-way (RoW) for a natural gas pipeline began. The RoW is between 10 and 25 m wide, and natural bridges are composed of pairs of branches from trees on either side that connect over the top of the RoW. Thirteen natural canopy bridges were deliberately left along a five-kilometre portion of the RoW, and bridges were an average of 410 m apart along the RoW (range = 80–810 m). Most of the trees that made up these natural bridges were some of the largest in the surrounding forest (Fig. 2).
Data Collection and Analysis
The natural canopy bridges were exposed in July by the clearing of the natural gas pipeline RoW, and the full suite of camera traps was installed in September 2012. The 13 natural canopy bridges were monitored from September 2012 to March 2013 with Reconyx PC800 Hyperfire™ Professional (Reconyx Inc., Holmen, WI, USA) camera traps with infrared flash. The bridges generally consisted of pairs of trees on opposite sides of the RoW (Fig. 2), although five of the bridges were composed of connections between three or more trees. In addition to the camera traps in the canopy, in October 2012, we placed pairs of cameras on the ground below canopy bridges to monitor animal crossings. These cameras were angled diagonally down the pipeline RoW in opposite directions. Because the RoW was used for intermittent transport of personnel and machinery during the study period, there was relatively little vegetation in front of the cameras.
Camera trap photographs were processed by a team of three people. Two people developed spreadsheets of trigger events, coding the events by category (e.g. NTS, mammal, bird, reptile), and a third person (FC) reviewed the vertebrate photographs and identified the species in each. Coding events by category was achieved at a rate of approximately 1200 photographs per hour and allowed vertebrate identifications to be performed without the distraction of NTS photographs, likely reducing errors. Equivocal vertebrate identifications were confirmed by experts (mammals: L. Emmons and R. Voss; birds: D. Lane and G. Servat; reptiles: JD and G. Chavez).
Arboreal camera placement
Camera traps were placed in the natural bridges at all possible crossing points (1–4 cameras per bridge, N = 25) to capture all animal crossings. Cameras were set to take three pictures per trigger, with less than one second between photographs, and no ‘quiet period’ between triggers; the photograph size was 3·1 MP. We placed the cameras facing along the target branch (example of camera orientation: Fig. 3 and photographs of subjects: Fig. 4a), rather than perpendicular to the branch, thereby maximizing time spent by the animals in the detection zone. We also placed the cameras on the largest branches and as close to the trunk as possible to reduce movement of the cameras due to wind.
The cameras were placed at an average height of 26·8 m (range = 13·5–33·7 m). Each camera was equipped with 12 AA lithium batteries and 4 GB memory cards until early December 2012, when 16 GB cards were installed in all of the cameras. The cameras were checked and photographs were downloaded in late October, early December and early March 2013.
We used a Sidewinder slingshot to set a line in the target tree, and TG and FCR then climbed the trees using the single rope technique (SRT, Fig. 2), navigated the canopy using the double rope technique (DRT), and finally placed the camera/s in the location with the best view of the crossing point/s (SRT and DRT described in Maher 2010). A paracord guide line was left in each tree after climbing to facilitate future access. Placing the cameras was relatively time-consuming (5–10 h per bridge, with more time spent in bridges with more cameras), and they were placed in an average of one bridge (up to four cameras) per day per climber. During maintenance periods, two to three bridges could be checked per climber per day.
The cameras were affixed to their respective branches with either a RAM® mount with a single ball joint or a home-made mount with two ball joints (Fig. 5a). The double ball joint allowed for necessary increased versatility in the positioning of the camera (Fig. 5b). The base of the mount is made of 4 inch (102 mm) SCH40 PVC split into quarters, cut 10 cm long. The base has a milled 2·4 cm slot for the webbing on either side and a hole drilled in the centre for ball joint no. 1 (http://www.rammount.com, part number RAP-B-366U, Fig. 5a, lower left). Ball joint no. 1 is affixed to the PVC base with a wing nut and attached to the arm (part number RAM-B-201U-A, Fig. 5a, lower right). The arm connects to ball joint no. 2, which is screwed into the back of the camera. One-inch (2·4 cm) nylon webbing was passed through the slots in the PVC base and around the tree. We also used a piece of coated copper wire to further secure the camera to the tree trunk as a backup in case of damage to the webbing.
Camera trap avoidance
To detect camera trap avoidance, we investigated the rates at which animals passed in front of the cameras over time. Because cameras were placed in the bridges approximately 2 months after the bridges were exposed, we expected animals to be relatively habituated to the bridges and that any changes in passing rates would be unrelated to bridge acclimation. We calculated the number of times mammals passed in front of the cameras (hereafter ‘passing events’) per camera per week for each species by dividing the total number of events per week by the number of cameras that were functioning. A passing event was defined as an individual or social group being photographed, regardless of the number of photographs taken while that individual/group was in the field of view. Passing events were considered to be independent if they were separated by more than 1 min of no photographs. We chose this conservative 1-min threshold because our data set showed numerous instances of distinct solitary individuals passing the same camera just outside 1 min of one another (e.g. Potos falvus, individuals recognizable by scars), each crossing event was critical to document, and all species photographed were either solitary or photographed in small, cohesive, fast-moving groups.
For all mammal species with more than 100 passing events over the 23 weeks, we evaluated the total number of cameras functioning per week (mean = 18 ± 2·4 cameras functioning, range = 13–23 cameras) and calculated the number of passing events for those cameras alone, removing the weeks of events for the malfunctioning cameras from the analysis. We then performed regression analyses and tested the null hypothesis that the number of events per week per camera would remain constant over time (i.e. the slope of the regression line would approach zero).
Triggers by non-target stimuli
Both arboreal and terrestrial cameras were triggered by non-target stimuli (NTS), both during the day and night. An NTS event was defined as a set of three photographs (the camera was triggered to take three photographs per stimulus) that did not contain vertebrates. In contrast to the method used to define a target stimulus (i.e. vertebrate) trapping event above, we did not group events because it was not possible to determine whether the same stimulus was activating the camera during consecutive events.
The first arboreal camera check at 6 weeks revealed that some camera memory cards had filled with NTS photographs of what appeared to be sun-warmed leaves close to the camera. Therefore, we removed all of the leaves within reach (approximately 1·5 m) from in front of the cameras. Removing the leaves required some movement around the canopy with additional climbing lines and the use of garden pruners and was not possible for all cameras. For each arboreal camera that received the leaf removal treatment, we compared the daily rate of NTS events before and after leaf removal for the maximum equal number of camera nights available with a one-way paired t-test.
To investigate whether NTS events were more common in the canopy than on the ground, each arboreal camera (after leaf removal) was paired with one of the two ground cameras at the same location, and the daily rate of NTS events was compared for the period during which both were functioning in a one-sided paired t-test. For canopy bridges that contained more than two arboreal cameras (N = 4), we randomly selected two of the cameras for the analysis.
Of all photoevents, 1522 (8239 photographs) were of vertebrates, with a total of 47 species representing 23 families and 14 orders (mammals: 20 spp., nine families, five orders; birds: 23 spp., 12 families, eight orders; reptiles: four spp., two families, one order; Table 1, photographed examples in Fig. 4). The smallest-bodied species of mammal, bird and reptile detected by the cameras were Rhipidomys sp. (25–170 g), Tolmomyias assimilis (16·5 g) and Anolis sp. (1·5 g), respectively (Museum of Vertebrate Zoology 1977; Emmons & Feer 1997; Duellman 2005). The majority of the photographs were of solitary individuals, but there were multiple individuals in some of the photographs of the following species: Aotus nigriceps, three adults and one infant; Saguinus imperator, one adult and two infants (Fig. 4c); Saguinus fuscicollis, one adult and one infant; Bassaricyon alleni, two adults; Coendou cf. ichillus, two adults; and Choloepus cf. hoffmanni, one adult and one infant. There were also two events of A. nigriceps and P. flavus in a single frame together.
Table 1. Number of passing events and event rate for species documented by camera traps while crossing 13 natural canopy bridges in primary rain forest of the Lower Urubamba Region, Peru over 3608 trap nights
Notable animal behaviours recorded included: carrying infant (on back or in mouth), social grooming, auto grooming, face threatening, feeding and potential mating. Various pathologies on the animals' bodies were also recorded in the photographs including a broken limb, a missing eye, an injured eye, scarring, a broken tail and a probable bot fly infection.
Seventeen mammal species were photographed by ground cameras (Appendix S1, Supporting information). Only four of these species were also photographed in the canopy, three of which were photographed on the ground only once (Didelphis marsupialis (canopy events = 1), Cebus apella (canopy events = 4) and Marmosa sp. (canopy events = 1), which was photographed on a liana near the ground), and one of which was photographed twice [Sciurus spadiceus (canopy events = 54)]. However, differences in camera placement on the ground vs. the canopy (the description of which is beyond the scope of the present study) preclude direct comparisons of detection indices or capture rates.
Camera Trap Avoidance
For four of the five species with more than 100 passing events (A. nigriceps, P. flavus, B. alleni and Caluromys lanatus), the regression of passing events over time indicated no significant reduction in events (Fig. 6). For Coendou cf. ichillus, the regression analysis approached significance (P = 0·06, Fig. 6), with events increasing slightly over time (slope = 0·0033).
Triggers of Non-Target Stimuli
In the canopy, NTS were responsible for 97·8% of all photograph events (total NTS events = 68 968), the majority of which occurred during the day (diurnal events = 58 118; nocturnal = 10 849). Of the 11 occasions on which the 4 GB memory cards filled, nine were likely due to leaf movement. In these nine cases, the cards filled within 8·7 ± 5·1 days, and in 93·8 ± 6·6% of the photographs, moving leaves were obvious in the photoframe. A paired t-test of the number of NTS events per day by seven cameras before and after leaf removal demonstrated a significant reduction in the rate of NTS events (before leaf removal mean = 130·9 ± 129·8 events daily; after leaf removal mean = 25·9 ± 29·0 events daily; t = 1·94; P = 0·02) after leaf removal treatment. Nocturnal NTS triggers did not appear to be caused by moving vegetation but may have been caused by insects.
On the ground, NTS were responsible for 5·6% of all photoevents (total NTS events = 1157 on 28 cameras), with more during the day than at night (diurnal events = 1081; nocturnal = 76). There was a significantly higher daily rate of NTS events in the canopy after leaves were removed vs. NTS events on the ground (N = 16 camera pairs; arboreal camera mean = 7·5 ± 11·0 events per day; ground camera mean = 0·8 ± 0·7 events per day; t = 1·75, P = 0·01). As indicated above, ground cameras generally had little vegetation in front of them because the RoW was kept open to transport personnel and machinery during construction. This, combined with the relatively stable attachment points of ground cameras (i.e. the base of the trunk), likely contributed to fewer NTS events.
Battery and Memory Card Capacity
Including vertebrates and NTS, a total of 70 493 events (215 151 photographs) were recorded over 3608 trap nights. Memory card capacity in general depends on the resolution of the photograph, as well as whether the photograph is black and white vs. colour. When filled, the 4GB memory cards used here contained a mean of 7249 photographs (N = 11, range = 5001–18 332). None of the 16 GB cards ever reached capacity, with up to 48 471 (8·19 GB) photographs included on a single card.
With the exception of cameras that were malfunctioning, flooded or damaged by animals, each time the cameras were checked (every 56 days on average), the LCD screen always displayed 99% battery life remaining. Therefore, the batteries were not changed until March 2013. During this period, the maximum number of days a camera functioned with the same batteries was 175 with a maximum number of photographs taken of 60 920 (42·1% colour, i.e., without flash). For ground cameras, we used only 6 AA lithium batteries; all were still functioning after 5 months and a maximum of 10 627 photographs (99·0% without flash).
General Cases of Camera Malfunction and Damage
Over 6 months, arboreal cameras malfunctioned or stopped taking photographs for various reasons including: full memory card due to NTS triggers (N = 8), failed day/night exchanger (the sensor inside the camera that senses light levels and determines whether the infra-red flash is needed) resulting in black night photographs (N = 8), failed mainboard which caused the camera to take photographs non-stop until the card was full (N = 1), camera failure due to insect infestation (N = 1, but additional cameras were minimally damaged by insects) and damage to/movement of cameras by Sphiggurus cf. ichillus (gnawing and opening the cameras resulting in water damage, etc., N = 4). The main camera problem experienced during the first 6 weeks was the filling of the memory cards due to NTS triggers, although there was one mainboard failure and 2 day/night exchanger failures. Failures of the day/night exchangers became increasingly common after this point, suggesting that they may be attributable to increased time in the tree, that is, increased exposure. Wildlife-induced damage also became more of a problem after the first 6 weeks.
During the 6-month period, a total of 30 different cameras were used in the canopy. Therefore, with the day/night exchanger and mainboard malfunctions, 30% of the arboreal cameras malfunctioned. Camera traps were used for a slightly shorter period on the ground (4·5 vs. 6 months), but there was a similar percentage of malfunction: 27% of 37 cameras. (Reconyx RC55 Rapidfire™ cameras were also used on the ground with a higher percentage of malfunction 54% of 24 cameras.) In Table 2, we provide suggestions for combating some common camera malfunction problems.
Table 2. Arboreal camera trapping trouble-shooting guide
Lithium batteries may not work as well under some climatic conditions (e.g. cold); consider other battery types.
False trigger problems
Wind, sun and leaves
False day photographs
Clear leaves away from in front of the camera as much as possible (at least 1·5 m)
Place camera in large tree, close to the tree trunk and on a large branch
Place the camera in as protected a location as possible (i.e. reduce wind and sun exposure)
Check memory cards and batteries after windy days
Do not aim camera towards palm fronds
Sensitivity setting may be reduced to medium, but events may go undetected
Insects inside camera
False day and night photographs which can fill the SD card and damage to inside of camera to the point of full malfunction
Put fine stainless steel wool and/or Vaseline at the base of the camera where it contacts the tree
Put fine stainless steel wool and/or Vaseline all around the outside housing of the camera where there are openings (only on the housing, not on the inside of gaskets)
Insects on branches in front of the camera
Photographs of insects
Do not use deterrent substances on branch because they may also deter target species
Check memory cards and batteries frequently
Use large memory cards
Condensation inside the camera
Corrosion on battery terminals
Perform camera maintenance on dry days
Place small packets of silica gel inside camera
Animals not properly triggering motion detector
Subjects difficult to identify (e.g. just the tail in photograph)
Place camera facing down the branch, allowing as much time as possible within the detection bands (Fig. 3)
Be sure to refer to your camera's instruction manual to locate the detection bands (they may be in just the lower half of the screen)
Rodents damaging/opening camera
Damage to camera housing, flooding of inside of camera
Secure camera with padlock or wrap wire around camera housing
Camera mount restrictive
Mount does not allow view of focal point
Use a mount with one or two ball joints, allows camera to be angled perpendicular to branch of attachment (Fig. 5)
Small memory card
Card fills before batteries die
Use memory cards that are as large as possible (e.g. 16 GB may take as many as 60 920 photographs)
Short battery life
Camera dies before card fills
Use maximum number of batteries (e.g. 12 AA lithium batteries – may last as long as 175+ days)
For cameras that take C batteries, use C to AA adapters and AA lithiuma
Non-rechargeable batteries recommended
Damage to straps supporting camera
Camera falling from canopy
Use a backup copper wire to hold the camera
Delay between sets of photographs
Trajectory/behaviour of animal is not clearly interpretable
Unique to camera model, but consider increasing the ‘number of pictures per trigger’
This study demonstrates that while there are unique obstacles to arboreal camera trapping, many of these issues can be addressed and challenges can be minimized to produce unique ecological data. This is part of a larger investigation in which mammal use of natural canopy bridges was monitored for over a year with relatively little investment in camera placement and maintenance. In this context, where animals were not sufficiently habituated to allow direct observations of bridge use, the method allows all crossing locations to be continuously monitored. Furthermore, our results suggest that animals do not avoid the cameras, making this technique one of the least invasive data collection tools available for arboreal animals. We also discovered that various behaviours can be documented in detail and in close proximity with camera trap photographs. Use of video functions common in most camera traps would further enhance behaviour-based research possibilities in the canopy (although memory card space would become a more serious concern).
Although initial placement of cameras in the canopy can be labour intensive, maintenance requires significantly less time than placement. While hoisting cameras up into the canopy with ropes might be less time-consuming than climbing, we do not believe that it would be nearly as effective in trapping animals because a great deal of accuracy and precision are necessary to test various angles and ensure that animals will pass through the camera's detection zone. This requires manual manipulation in the canopy. However, new products are becoming available that may significantly reduce the effort involved in checking cameras once placed in the canopy. Perhaps the most promising is the introduction of cellular-phone-enabled camera trap models from at least a few popular camera trap brands (e.g. Bushnell®, SpyPoint®, Reconyx®, Uway®). These cameras allow download of low-resolution photographs through the GSM cellular phone network. This technology is sure to become more common and more refined in the near future, yet it is currently limited for large remote field applications by the need for cellular phone coverage, as well as the extra expense of purchasing cellular phone plans for each camera. Additionally, there are now SD cards with built-in wireless connectivity (e.g. Toshiba FlashAir™), allowing remote download of photographs over relatively short distances. Depending on camera height, this may be a feasible option. In the case of the cellular-enabled cameras, trips to the field to check cameras could be scheduled based on updated knowledge about the status of each camera. With wireless download SD cards, climbing into the canopy would only be required if photographs downloaded from the SD card indicated potential problems.
This study confirms that arboreal cameras can produce significantly higher frequencies of non-target stimulus events than terrestrial cameras. However, results presented here provide some guidelines for camera placement and programming to increase the efficiency and effectiveness of arboreal trapping (Table 2). Perhaps the most critical recommendation is that cameras should be checked within a few days of placement to see whether NTS events are a significant problem or to see whether cameras are malfunctioning. Most problems are detectable very soon after placement. In the case of the Reconyx PC800s with 12 AA lithium batteries, battery life was not a limiting factor in the number of trap nights. Therefore, it is best to make sure that the memory card is large (i.e. 16 + GB). Under our circumstances, after the camera was set and checked, it could be left for multiple weeks or even months without maintenance.
Here, we found that leaves in front of a camera can lead to high numbers of NTS photographs during the day. Therefore, leaves should be removed to the fullest extent possible when installing and maintaining cameras. We had few further problems with memory cards filling with NTS events after leaves were removed. Minimizing branch movement by selecting large trees and large attachment branches close to the trunk also likely minimizes camera movement due to wind. The inability to use the detection test function (‘walk test’) often available on cameras prevents exact testing, so particular attention should be paid to the camera model's detection zone specifications. It is also good practice to bring a digital camera (that uses the same type of memory card as the camera trap) when placing the camera to review test photographs and confirm an appropriate field of view before descending and leaving the camera.
While arboreal cameras were highly exposed to wind and sun and experienced more damage from insects and porcupines, it is encouraging that failure rates in the canopy were no higher than they were on the ground. Camera failure rates overall were lower in this study than rates reported elsewhere for the tropical forest (e.g. Kays et al. 2011: 70%).
Arboreal camera trapping has the most data gathering potential when the goal is to detect and monitor activity at a very specific location where animal use or movement is likely. When placing a camera, it is important to consider that rather than capturing subjects on a plane, as in terrestrial camera trapping, subjects are using a three-dimensional network of lines, which affects capture challenges. Selecting an appropriate focal branch may be the biggest challenge in successful arboreal camera trapping. The success of this study in gathering over 1,500 photoevents of 47 vertebrate species may be attributable to the fact that we were monitoring potentially highly ‘attractive’ locations: the only pipeline clearing crossing points within hundreds of metres. In situations where specific structures/crossings are not available, photoevents will likely be more challenging to acquire. To maximize success, we recommend selecting a branch that is attractive to animals for purposes such as navigation, resource acquisition, social interaction or resting, and this may require some observation prior to camera installation.
The success we have experienced in documenting wildlife in the canopy with camera traps provides support for using this method in place of traditional trapping or shooting and collecting of wildlife where possible. Camera trapping seems to cause minimal to no stress or injury to animals. Furthermore, it does not require a collection or animal handling permit, nor does it require training in trapping and animal handling. Finally, while this study was not designed to assess mammal diversity in the canopy, our results suggest that there is much potential for effectively gathering such data in a less invasive and potentially more rapid manner than physical trapping.
Camera trapping shows enormous potential as a non-invasive tool for studying elusive arboreal wildlife as well as for monitoring use of focal areas of interest. Major advances in battery life, memory card capacity and camera technology now allow cameras to be left in the canopy without the need for frequent, demanding canopy ascents, and as the technology continues to improve, challenges related to non-target stimuli may become less problematic. However, as demonstrated here, simple steps can be taken to minimize NTS events. We look forward to this baseline work spurring a new wave of creative applications of this exciting tool to provide a clearer window into life in the forest canopy.
We thank five anonymous reviewers for their valuable comments on previous drafts of this manuscript. We thank Stanford W. Gregory, Jr. for camera tree mount design support, Sophia Celino, Guillermo Joo Novoa, Matthijs Schuring and Drew Hart for valuable support in photoprocessing, and Sulema Castro, Tatiana Pacheco, and Marcel Costa for administrative and logistics support. For identification of mammals and birds, we thank Louise Emmons, Robert Voss, Daniel Lane and Grace Servat. We also thank our field nurses, support staff and local guides from the communities of Camisea and Shivankoreni. This is contribution 16 of the Peru Biodiversity Program. We thank Repsol Exploración Perú for financial and logistical support. This research was conducted under the Peruvian government's Dirección General Forestal y de Fauna Silvestre Resolución Directoral No. 0221-2011-AG-DGFFS-DGEFFS, No. 0197-2012-AG-DGFFS-DGEFFS and No. 0265-2012-AG-DGFFS-DGEFFS.
Data for the non-target stimulus analyses can be found at Dryad entry doi:10.5061/dryad.2236n (Gregory et al. 2014).
Corrections added on 08 May 2014, after first online publication: Sapajus apella changed to Cebus apella and Sphiggurus cf. ichillus changed to Coendou cf. ichillus throughout text; Wilson & Reeder (2005) deleted from the text and References.
Corrections added on 08 May 2014, after first online publication: Coendou ichillus changed to ‘N’, and Sciurus ignitus changed to ‘D’.