Post provided by Meredith Palmer, Chris “Akiba” Wang, Jacinta Plucinski & Robert M. Pringle

Camera traps are a valuable tool in ecological research, especially for capturing large quantities of information on the behaviour of an array of wildlife within an ecological community. Camera traps are seldom used to experimentally testing key animal behaviour hypotheses, despite the potential offered by the non-invasive technology. In this blog post, Dr. Meredith Palmer and co-authors discuss the application of the ‘BoomBox’ camera trap module that allows researchers to conduct a unique suite of manipulative experiments on free-living species in complex environments, as published in their Methods in Ecology and Evolution article ‘BoomBox: An Automated Behavioral Response (ABR) Camera Trap Module for Wildlife Playback Experiments’.

Wildlife Interactions: The Network Structuring Ecological Systems

In a landscape populated with potential predators, competitors, mates, and mutualists, the interactions between and among species shape all facets of their ecology. Examining wildlife interactions provides valuable insight into spatial and temporal patterns of animal activity, their distribution and abundance, and the ability for species to coexist. Furthermore, the nature of these interactions is rapidly changing in response to anthropogenic pressures, necessitating urgent study of how species respond to each other both in intact and modified systems.

Studying Animal Behavior in the Wild

Ideally, a researcher would be able to document sufficient naturally occurring encounters between focal species to draw strong inference on the underlying drivers of behavior decision-making. For many species, this would take exorbitant amount of effort – decades of dedicated observation. The next best approach would be to conduct controlled experiments on free-ranging animals to rigorously uncover mechanisms shaping their responses to one another.

This is not as easy as it may first appear: experimenting on free-ranging wildlife populations can be prohibitively difficult. Playback studies and other wildlife-simulation experiments have been the go-to approach for decades, in which a researcher deploys an audio, visual, or olfactory cue and monitors the target’s subsequent response. However, researchers often struggle to locate sufficient focal animals to draw inference, particularly if the study subjects are cryptic, rare, nocturnal, or otherwise difficult to encounter in the wild. Once a subject has been found, there is always the risk that the presence of the human experimenter – rather than the cues presented – is driving responses to experimental stimuli. As a result, key questions in animal ecology and conservation biology have hitherto gone unanswered.

There are non-intrusive means study animals in the wild. Camera traps (also known as trail cameras) are remote cameras that automatically take images or video of animals that pass in front of them, triggering their heat and motion sensors. By placing these cameras in the environment, researchers can collect continuous data on entire wildlife communities without the fear that human presence is disrupting their behaviors. Camera traps run continuously and indiscriminately, generating large volumes of data on entire wildlife communities. However, camera traps are used to passively monitor wildlife, collecting observational data that – while able to provide valuable correlational inference – cannot rigorously establish inference on the mechanisms underlying species interactions and decisions-making.

Automated Behavioral Response Systems – the best of both worlds?

Recently, innovative researchers have combined these approaches to automatically manipulate and remotely record the behavioral responses of free-ranging animals. The “Automated Behavior Response” (ABR) methodology uses camera traps to film the reactions of wildlife to audio stimuli that are triggered by passing wildlife. This approach blends the benefits of camera trapping (e.g., lack of human observers, continuous monitoring of entire wildlife communities) with the power of playback experiments (e.g., manipulating the environment to establish strong mechanistic inference). This new design can capture responses of numerous species to a user-defined set of cues, generating sample sizes sufficient to evaluate ecological hypotheses that would be untestable using standard methods.

BoomBox – An Accessible and Affordable Open-Source ABR Solution

Our goal in designing the ‘BoomBox’ was to create an ABR system that was open-source, low-cost, and highly accessible to people with a broad range of technical experience. The BoomBox is a modular add-on for commercial camera traps that connects directly to the camera’s own triggering mechanisms. When passing wildlife activates the camera trap, the BoomBox turns on within fractions of a millisecond to deploy an audio cue. As the camera trap records, the playback experiment commences: animal responses to the programmed sounds are captured on video and can be later analyzed to understand animal responses to predators, competitors, humans, or novel situations.

The BoomBox is created using low-cost commercial materials that can be purchased at a local electronics store or easily obtained online. We plucked stereo components from Bluetooth speakers and MP3 decoders from kid’s toys; in fact, the original electronic underpinnings of the BoomBox was plucked from an interactive children’s book written and designed by one of the authors. At the time of writing, the total cost of a BoomBox (excluding camera trap) is ~USD75. To further support novel electrician ecologists, we provide illustrated instructional guides and video tutorials that walk the user through every step of the assembly process. 

We wanted the ABR device to be easily programmable and highly flexible. The software is based on the Arduino programming language – an open-source electronics platform designed for beginner programmers that provides a simple and accessible user experience. Users can update the audio playlist, the delay between when the camera begins recording and when the cue is played, and the frequency at which the experiment is performed (to prevent, for example, an animal from habituating to playbacks by repeatedly triggering the BoomBox ABR). Our open-source libraries and sketches (code) allow users to customize these materials to suit the needs of a wide range of research programs.

Furthermore, the small size, modularity, durability, and power efficiency of the BoomBox makes them easy to transport and set up in the field and can operate continuously for months at a time. The BoomBox can connect to many of the most popular brands of off-the-shelf camera trap, including Bushnell, Reconyx, Browning, and Spypoint – cameras that are not only highly accessible, but may already be owned by an active wildlife researcher. 

BoomBox in the Wild

To evaluate their effectiveness, we tested the BoomBox in two locations – one pristine open grassland ecosystem (Serengeti-Mara Ecosystem in Tanzania) and one recovering forested savanna habitat (Gorongosa National Park in Mozambique). Each location contained similar large herbivore species (the targets of our experiment), but vastly different habitat types and predator communities. Deploying the BoomBox in the field is a straightforward process, similar to setting up a standard camera trap for capturing data on the select focal species. However, we strongly recommend (1) practicing setting up the device in the home or office (pets make great test subjects!) and (2) treating the first 1-2 weeks of deployment as a ‘pilot period’, with frequent checks to adjust and update settings to optimize field performance.

The pilot period provided a few lessons learned; however, overall – the deployments were highly successful. Over a period of < 8 weeks per site, we collected thousands of videos containing discernable animal responses to predator cues (compared, for example, to the ~350 trials collected over a period of three years using more traditional approaches). From these videos, we were able to evaluate the occurrence, latency, frequency, and duration of different categories of anti-predator response to each predator species. As the BoomBox samples the entire wildlife community, we were able to rigorously evaluate the responses of over three dozen species to six predator and control categories, allowing for unprecedented evaluation of wild animal anti-predator behavior.

Next Steps – Evolving the BoomBox Platform

In our test cases, we programmed the BoomBox with different predator audio cues to evaluate predator-prey interactions; other researchers are currently using this device to study predator-predator, predator-mesopredator, and human-wildlife conflict dynamics. The ABR methodology can be paired with other data-gathering techniques, such as deploying the devices within a standard camera trap monitoring survey or targeting animals equipped with GPS collars, accelerometers, or biologgers to obtain additional information individual state and landscape-level decision making.

The BoomBox itself can also be modified to accommodate peripheral devices that expand the number of questions that can be asked during these behavioral experiments. For instance, weight plates or infrared temperature sensors for recording animal mass or body temperature could be attached, allowing researchers to understand how animal condition alters their response to external stimuli. New advances in TinyML may soon provide the capacity for onboard AI to only trigger the peripheral devices when the camera has been triggered by a species of interest.

New Avenues for Animal Behavior Studies

These new tools allow researchers to conduct a manipulate playback experiments on a wide array of free-living wildlife, enhancing our ability to identify the drivers of species behavior and interactions in natural systems. With our open-source design, we aspire to encourage innovation and adaptation of this device to suite a diverse range of research questions. More broadly, our goal with this project was to transform the camera trap from a passive data collecting device to a sensor platform that actively perturbs and reacts to the environment to provide new insight into the inner workings of the natural world.

To read the full Methods in Ecology and Evolution article, ‘BoomBox: An Automated Behavioural Response (ABR) camera trap module for wildlife playback experiments’, click here.

To watch instructional and sample videos, visit the FreakLabs YouTube channel here.

About the author

Meredith is a behavioral ecologist and conservation biologist at Princeton University, conducting a National Science Foundation-funded postdoctoral fellowship in the lab of Dr. Rob Pringle. Her research uses emerging technologies to study predator-prey interactions, particularly how terrestrial herbivores survive in landscapes with complex large carnivore guilds. These studies are used to inform on-going conservation efforts in Africa and North America.