We have prepared an accompanying field guide, outlining how to install a lured trail camera in the field, for the purpose of detecting predators.
Background
The use of trail cameras has increased across the conservation industry in recent years, as they have become more affordable and accessible, and as it has become increasingly evident that they are a highly sensitive tool for survey, monitoring and detection relative to other available methods (Smith & Weston 2017). Cameras also provide a wealth of data, allowing researchers and practitioners alike to make inferences wider than simple presence/absence of a species (Dilks, Sjoberg, & Murphy, 2020) — such as animal behaviour (Caravaggi et al., 2017) and activity patterns (Kays et al., 2009).
ZIP has learned that a network of trail cameras paired with our ‘MotoLure’ automated lure dispensers is an effective method for detecting possums, rats and stoats, particularly for projects where it’s critical to detect relatively few animals in a large landscape.
This method is now the basis of a detection network we have established in the Perth River valley to provide the first indication that one (or more) of these predators is present within the field site.
Cameras also have the added benefit of detecting other species. The Perth River network has also detected native birds such as kea, kakaruai (South Island robin) and ruru (morepork), along with game animals such as tahr, deer and chamois.
Purpose
The primary purpose of this Finding is to describe ZIP’s current approach to possum, rat and stoat detection using lured trail cameras. This approach will continue to evolve on the basis of what we and others learn about this technique, and as new tools become available.
ZIP’s current approach to the use of trail cameras
This camera, though covered by snow and ice, was still functioning well in the depths of winter in the Perth River valley (5th August 2019)
Preferred Camera
Our preferred trail camera is the Dark Ops range by Browning. We have generally found these cameras to be more reliable, more waterproof, and easier to use than other brands we have trialled; although the harsh wet conditions of the Perth River valley can occasionally result in moisture damage.
Over the last three years we have used four slightly different models as Browning have evolved their product range. The Dark Ops models we have used in the Perth River valley are the Extreme (model BTC-6HDX), 940 HD (BTC-6HD-940), the Elite HD (BTC-HDE), and most recently, the 2020 Dark Ops Extreme Max HD (model BTC-6HD-MAX).
This model has a maximum image size of 18 MP, and a minimum delay trigger setting of 1 second. This is very useful, because valuable imagery can be missed with longer trigger delays. It is also slightly more compact than the other models (10.6 x 7.50 x 6.25 cm).
All of these cameras have similar specifications, including a maximum video size of 1280 x 720p (low compression), infrared LED night-time illumination, infrared motion detectors, and adjustable trigger speeds and detection range.
Browning Dark Ops cameras use standard SD cards, and run on six AA batteries.
Limitations of the PIR Browning camera sensor zone
Our preferred practice is to place cameras at a distance of 1.2-2.5 metres from the lure.
On the Browning camera, the PIR sensor zone is bow-tie shaped, and detects animals more readily when they approach from the left or right of the camera’s field of vision. We have found that animals approaching from the top, bottom, or corners are unlikely to trigger the PIR until they near the centre of the camera’s field of vision (ZIP, unpublished data).
The diagram to the left illustrates the estimated extent of the PIR sensor zone, at a distance of 2.1 metres. At this distance, an animal must be within the area outlined in red to trigger the PIR sensor.
SD cards
We use Sandisk 16GB Ultra 48 MB/s or 80 MB/s SD cards. We have found that these cards offer good value for money, without compromising on storage size. A 16GB card can hold up to 30,000 still images, which is sufficient to last at least two months in the field with our preferred settings. Sandisk cards are also waterproof, drop-proof and magnet-proof, which makes them very reliable in the field. It is good practice to format SD cards after removing footage, to ensure there are no residual files that could affect future recording.
Batteries
We currently use Panasonic Eneloop PRO 2450 mAh Ni-MH rechargeable AA batteries. We have found these to be a reliable option in the field. The batteries retain a high charge for significantly longer than other rechargeable batteries we have used, and work well in the cold conditions of the Perth River valley.
In general, batteries will last approximately 2-3 months in the field depending on the number of trigger events; some have been known to last over 6 months.
To ensure our cameras are continually functional and ready to detect invading possums, rats or stoats, our current practice is to check the batteries every 4-6 weeks, and change batteries when the charge drops below two bars.
We have also found that it is a good practice to fully discharge the batteries, using the discharge function on our chargers, before charging them again. We use Tenergy TN160 chargers, which have capacity to charge up to 12 batteries at a time. A full discharge and recharge cycle takes up to 12 hours, while a normal charge straight from the field takes 4-5 hours, depending on the remaining charge in the batteries.
Lure
The ZIP MotoLure
In order to maximize the chance of detecting possums, rats and stoats in a low-predator or predator-free environment, we pair trail cameras with the ZIP MotoLure, which dispenses a pre-set amount of fresh mayonnaise each night to attract possums, rats and stoats into a camera’s field of view. Dispensing mayonnaise lure each night encourages repeat visits by predators.
We believe that deploying trail cameras paired with lure dispensers before a predator removal operation may increase predator visitation rates; which in turn may increase confidence in the network as a method for detecting any remaining individuals after the operation is carried out.
Corflute ‘chew cards’
For possums and rats, peanut butter-lured chew cards are an affordable alternative to the MotoLure, although these will need to be replaced more frequently as the peanut butter is eaten or degrades. This should be roughly every 3-4 weeks in summer and can be extended during the winter depending on environmental conditions. To lure trail cameras using this method, we recommend purchasing unfilled 10 cm x 10 cm chew cards, and filling them to a depth of 4-6 cm on one side with a high-oleic smooth salted peanut butter (e.g. Pic’s), mixed with enough icing sugar to thicken it to a paste consistency that clings to the back of a spoon without running off – a ratio of approximately nine parts peanut butter to one part icing sugar.
Network design
In order for the Remove and Protect approach to be economically viable across large landscapes, it is important to minimise costs associated with both the network of trail cameras and labour required to service them.
In the Perth River valley, we are currently trialling a network of lured trail cameras at a density of one camera every 35 hectares, across all predator habitat within the field site – including tussock and alpine scrub, exposed ridges, open forest and wet valleys.
We are very confident this density is sufficient to detect individual possums and stoats, so that our team can initiate a response to remove these individuals (using traps, toxins and hunting with a predator dog). We are still determining whether or not this density is sufficient to quickly alert us to the presence of an emergent population of rats, while this population is still spatially restricted and can be removed using a targeted approach.
Footage review
We have developed a software tool that:
makes it easier, more efficient and more accurate to manually review trail camera footage to record detections of target species;
can be used offline, as our team often need to review footage in remote backcountry, with limited internet service, in order to enable a timely response to a detection; and
is able to process both image and video files.
The software we have developed meets all of our requirements. When in use, the software: (i) enables a reviewer to record detections using ‘quick keys’, each of which is associated with one of up 26 target species or other items of interest; (ii) automatically extracts metadata associated with each detection (e.g. time and date); and (iii) is capable of exporting results into a spreadsheet (in CSV format) and folders for specific targets (for easy reference at a later date). Somewhat unimaginatively, we currently refer to this tool as “The Classifier”, and we are continuing to refine it.
Similar software tools already exist (e.g. Snoopy, ViXen and FastPhotoTagger). However, none of these tools met enough of our specific requirements for us to easily apply them in our context.
A project to improve the efficiency of the use of cameras to detect predators
ZIP’s prototype detection device, currently undergoing field testing in the Perth River valley
While trail cameras are increasingly recognised as the most sensitive tool available for detecting the presence of predators, it can be very expensive to maintain a network of trail cameras – particularly in remote locations. This is largely due to the high labour cost associated with changing batteries and SD cards every 3 to 6 weeks, and with processing footage from the cameras. Manually servicing trail cameras in remote areas does not necessarily provide timely notification of the detection of individual predators, given their mobility and/or reproductive capability.
Consequently, for the past several months, the ZIP team, with the help of some talented contractors, has been developing a new predator detection device. The work has been funded by DOC, NEXT Foundation, Predator Free 2050 Limited, and the Biological Heritage National Science Challenge. The team also acknowledges the pioneering work of Grant Ryan, from The Cacophony Project, to develop a similar device particularly suited to more accessible areas.
The ZIP device will comprise a highly sensitive thermal camera, on-board artificial intelligence (A.I.) video analysis software, and remote reporting capability. The device will be need to be rugged, and use little power in order to run without servicing for as long as possible; ZIP’s target is for the battery to last more than 6 months, but that will depend upon the number of triggers. It will need to be able to reliably identify possums, rats and stoats to a very high level of accuracy, and also be able to remotely report target detections to conservation managers.
We’ll provide an update about progress with developing this device in the coming weeks. For now though, some more information is available in this short video.
Reference
Caravaggi A, Banks P, Cole Burton A, Finlay CMV, Haswell PM, Hayward MW, Rowcliffe MJ, Wood MD 2017. A review of camera trapping for conservation behaviour research. Remote Sensing in Ecology and Conservation 3(3): 109-122.
Dilks P, Sjoberg T, & Murphy E 2020. Effectiveness of aerial 1080 for control of mammal pests in the Blue Mountains, New Zealand. New Zealand Journal of Ecology 44(2): 1-7.
Kays R, Tilak S, Kranstauber B, Jansen P, Carbone C, Rowcliffe M, Fountain T, Eggert J, He Z 2011. Camera Traps as Sensor Networks for Monitoring Animal Communities. International Journal of Research and Reviews in Wireless Sensor Networks 1(2): 19-29
Smith DH, Weston KA 2017. Capturing the cryptic: a comparison of detection methods for stoats (Mustela erminea) in alpine habitats. Wildlife Research 44(5): 418-426.