Environmental scientists rely on a growing toolkit of technologies to track vulnerable species, from GPS collars that follow individual animals across continents to DNA traces filtered from a scoop of river water. The choice depends on the species, the habitat, and the question researchers need to answer. Here’s how each major technology works and where it fits in conservation.
GPS and Satellite Tracking
GPS collars and satellite tags remain the most direct way to follow individual animals in real time. A collar transmits location data to orbiting satellites, which relay it to researchers. Modern devices weigh as little as 5.3 ounces, light enough for a striped skunk to wear safely, though they’re still too heavy for very small animals like bats. Some newer designs harvest energy from the animal’s own movement, theoretically lasting an entire lifetime without a battery swap. One such tracker on an Exmoor pony was still transmitting after 147 days with no signs of slowing.
Two satellite networks dominate wildlife tracking: Argos and Iridium. Argos has been in use since 1978 and runs on six to seven polar-orbiting satellites. Its main weakness is coverage gaps near the equator, where many vulnerable tropical species live. Satellites may pass overhead only a few times a day, leaving several hours with no connection at all. The Iridium network, built from 66 satellites, provides continuous global coverage and can transmit roughly ten times more data per signal (340 bytes versus 32). When programmed to collect more than 24 GPS fixes per day, Iridium tags significantly outperform Argos-based tags in data quality. This matters for species like sea turtles that surface briefly and unpredictably.
Environmental DNA Sampling
Not every vulnerable species can be collared. Many are too small, too elusive, or too rare to find reliably. Environmental DNA, or eDNA, sidesteps that problem entirely. Every organism sheds DNA into its surroundings through skin cells, mucus, waste, and other biological material. Scientists collect a water sample, typically one to two liters from a stream or lake surface, filter it, and extract whatever genetic material is present. The filtered DNA is then amplified and matched against reference databases to identify which species are in the area.
The technique is sensitive enough to detect critically endangered fish in large river systems. Researchers have used it to confirm the presence of the Mekong giant catfish and the rare green sturgeon without ever seeing either animal. For species at very low densities, advanced amplification methods can pick up traces that standard techniques miss, making eDNA especially valuable for confirming whether a vulnerable species still occupies a particular waterway. Filtering larger volumes of water improves the odds of detection for the rarest species.
Camera Traps and AI Image Recognition
Camera traps are motion-triggered cameras placed along wildlife trails, near water sources, or at other strategic points. They generate millions of images, and manually sorting through them to identify species used to be a major bottleneck. Deep learning algorithms now handle much of that work. At locations where models have been trained on local wildlife, the best systems achieve around 95.6% accuracy in identifying species. Some setups reach 98% under ideal conditions.
Performance drops sharply in unfamiliar settings. The same top-performing model fell to 68.7% accuracy when classifying species at locations it hadn’t been trained on. This gap is one of the biggest practical challenges in the field. Combining multiple models into an ensemble improves results slightly, pushing accuracy to about 71% in new locations and 96% in familiar ones. For conservation programs monitoring a specific reserve or habitat over time, camera traps paired with trained AI are highly effective. For broad surveys across new terrain, human review is still essential.
Passive Acoustic Monitoring
Autonomous recording units can be deployed in forests, oceans, or wetlands to capture the sounds of a habitat continuously for weeks or months. This approach is particularly useful for species that are heard far more often than they’re seen: birds calling from dense canopy, whales communicating underwater, or frogs vocalizing at night. The recorders run unattended, making them a low-cost option for long-term studies.
The main challenge is data volume. A single recorder generates enormous audio files that need to be processed and annotated, identifying which calls belong to which species against a background of wind, rain, insects, and human noise. Machine learning models are increasingly used to automate this annotation, though building accurate classifiers still requires substantial training data for each target species.
Thermal-Imaging Drones
Drones equipped with thermal cameras detect animals by their body heat, making them effective for surveying wildlife at night or in dense vegetation where visual counts fail. Drone surveys typically detect more animals than ground-based methods, partly because they cause less disturbance. Ground surveys often startle animals at or just before the moment of detection, which skews the data. A drone launched from a parked vehicle at a strategic distance can observe animals behaving naturally.
In one comparison studying deer, traditional spotlight surveys from vehicles appeared to show strong road avoidance. Drone data told a different story: deer weren’t necessarily avoiding roads. They were likely reacting to the approaching survey vehicle and moving away before researchers could count them. After researchers identify a flight altitude that doesn’t disturb the target species, drones can observe with less behavioral interference than a person on foot or in a truck.
Lidar for Habitat Mapping
Tracking animals is only half the picture. Scientists also need to understand and map the habitat those animals depend on, especially for canopy-dwelling species whose homes are inaccessible from the ground. Airborne lidar fires laser pulses from an aircraft and measures how they bounce back, building a precise three-dimensional model of forest structure: canopy height, layering, gaps, and structural complexity.
This technology proved critical for the marbled murrelet, a seabird that nests in old-growth forest canopy. USGS researchers found that incorporating lidar data into habitat models dramatically improved their ability to predict where murrelets were nesting. Both the specific measurements and the overall pattern pointed to the same conclusion: murrelets reliably select nesting sites based on canopy structural complexity. Lidar is currently the only method capable of quantifying three-dimensional vegetation structure at the fine scale needed to map habitat for canopy-associated species across large, structurally complex forests.
PIT Tags for Fish and Small Animals
Passive integrated transponders, or PIT tags, are tiny glass capsules (about 12.5 millimeters long) containing a microchip wrapped in a copper coil. They have no battery. When a tagged fish swims through an antenna that emits an electromagnetic field, the coil powers the chip just long enough to transmit a unique ID code. This makes PIT tags inexpensive, long-lasting, and suitable for both small and large fish. They’re widely used to monitor migration patterns of vulnerable freshwater species at dam passages, stream junctions, and other chokepoints.
The tradeoff is range. A PIT tag only transmits when the animal passes within reach of a fixed antenna, so it doesn’t provide continuous location data. For species that follow predictable migration routes through narrow corridors, this limitation matters less. Researchers install antenna arrays at key points and build a picture of movement timing, survival rates, and population size over successive seasons.
Keeping Tracking Data Safe From Poachers
Real-time location data for endangered animals creates an obvious security risk. If poachers intercept GPS coordinates for a rhino or a vulture roost, the same technology meant to protect a species could accelerate its decline. Conservation teams now treat tracking data with protocols more common in cybersecurity: encrypting transmissions, restricting access through formal data-sharing policies, and working with telemetry manufacturers to prevent sabotage or unauthorized exploitation of signals. GPS-tagged vultures and wolves have even been used as “sentinels,” their movements flagging suspicious activity like poisoning events or illegal kills that affect other species in the same area.