This summer, I worked with Dr. Mike Dorcas and his herpetology lab on using drones to study turtle populations. Using Studio M’s customized Phantom UAV, we conducted extensive assessments of the usability and effectiveness of drones in gathering data on freshwater turtle populations. Here’s a brief update on the project so far.
Unmanned aerial vehicles (UAVs), also known as drones, are already used in agricultural, police, geographic, news gathering, hobbyist, and military applications. Now there’s a growing interest in using drones to monitor wildlife populations. In the past, biologists have demonstrated the efficacy of drones in monitoring populations of several species, including elephants, rhinoceros, American alligators, whales, sea turtles and wading birds.
Dr. Dorcas specializes in reptile research, so that’s where we chose to focus. Many reptile species are in global decline, including freshwater turtles — an important element in the biodiversity of any ecosystem. Researchers report species loss and shifting population structures in freshwater turtle species worldwide.
Reduced abundance of turtles in aquatic systems could result in significant consequences for ecosystem function. Ecosystem conservation requires management, and relies on effective biodiversity monitoring.
Until recently, freshwater turtle population monitoring relied on several established methods. One of the most widely used is turtle trapping. Using baited hoop traps is reliable and cheap, but it’s labor intensive, and can only be used in areas accessible to humans. Researchers report bait hoop trapping is ineffective for herbivorous turtle species, and is biased toward active foragers.
Visual surveys from land and boats are used to study turtle populations spanning large areas, but again there’s the issue of accessibility. Snorkeling may provide accurate detection of some turtle species and size classes, but is limited by water conditions, difficulty in standardizing captures across observers, and human labor. Hand capture, too, is limited by accessibility and labor intensity.
Manned aerial survey overcomes those limitations, but is expensive, dangerous, and requires substantial planning.
Using drones to monitor turtle populations offers many advantages over these existing methods. Drones can be controlled remotely by a human pilot, or made to follow a pre-determined flight path using GPS waypoints. They are able to access areas that are unreachable by humans. This high accessibility factor overcomes the habitat limitations associated with most of the traditional methods.
UAV monitoring provides systematic and permanent data (i.e. abundances of multiple species and spatiotemporal distributions) that’s more detailed than temporary visual data. Drones can also survey large areas quickly, reducing labor requirements.
Other advantages include high resolution, the ability to provide time-sensitive data quickly, and less complex research design. Compared to manned aerial survey, it poses lower risk for the operator, and a reduced ecological footprint, while still providing data at high spatial and temporal resolution.
More Data Needed
Despite the recent popularity and potential of drone use in wildlife research, few published studies have tested the efficacy of drone flight techniques, and none have tried using drones to study freshwater turtles. This study aims to navigate the challenges and complexities of this novel technology for assessing turtle populations in three ways.
First, by determining how turtle detectability via drone varies under different environmental conditions. Second, by developing optimal piloting strategies (i.e. flight height and speed) for accurately detecting turtles. Third, by determining if turtle detection rates via drone are comparable to detection rates via traditional hoop-net surveys. These findings will allow us to make methodological recommendations for researchers interested in using drones as an alternative method for studying turtle populations.
Our UAV was a DJI Phantom P330, equipped with a GoProHero 3 Black camera, and a Zenmuse gimbal. We added an upgraded video transmitter to the drone in order to fly it in first-person view at greater distances than the standard setup allows. The drone had GPS built-in, and was controlled remotely by a pilot. Flight footage was recorded onboard at 2.7k resolution.
Drone flights were carried out from early July to mid September 2015. All flights are conducted between 11:00 and 15:00, after releasing the captured turtles. Flight paths were designed to maximize our recorded area, while minimizing overlapping coverage. All flights at each pond followed the same pre-determined path.
We flew the drone at four heights and four speeds to test the effect of height and speed on turtle detectability. Heights were two meters, 3 meter, 4 meters, and 5 meters above the water. The range was limited to 2-5 meters because flight below 2 meters is unsafe, and above 5 meters, observers are unable to distinguish turtles from tree branches. The speeds were two meters/second, 3 meters/second, 4 meters/second, and 5 meters/second, determined by the drone’s flight controller. For each height and speed combination, we conducted four individual flights. In totally, we conducted 64 flights at each of the three ponds.
Here is the view from the drone:
Can you tell what species they are?
Answer (from left to right): (ɐʇɔıd sʎɯǝsʎɹɥɔ) ǝןʇɹnʇ pǝʇuıɐd , (ɐʇdıɹɔs ɐʇdıɹɔs sʎɯǝɥɔɐɹʇ) ɹǝpıןs pǝıןןǝq-ʍoןןǝʎ , and (ɐuıʇuǝdɹǝs ɐɹpʎןǝɥɔ) ǝןʇɹnʇ buıddɐus.
Drones are increasingly popular not only as toys, but as useful tools in many applications, from filmmaking to wildlife monitoring. My research on using drones to monitor freshwater turtles will be one of the first papers in its area. I plan to publish my research at the end of the year.
If you are interested in my preliminary findings, here is a poster I presented at the Summer Research Symposium. If you want to see the drone at work, and try your own turtle detection skills, here’s your chance.
Class of 2017