How Scientists Study Animals: Methods and Techniques

Scientists study animals through a surprisingly wide toolbox of methods, from strapping tiny satellite transmitters onto migrating eagles to filtering DNA out of pond water. The techniques range from old-fashioned sit-and-watch fieldwork to machine-learning algorithms that identify species from audio recordings alone. What ties them together is a push toward studying animals with as little interference as possible, gathering data that would have been unimaginable a few decades ago while keeping the subjects alive, unstressed, and unaware they are being watched.

Satellite Telemetry and GPS Tracking

One of the most powerful ways to understand where animals go and why is to follow them, and satellite telemetry makes that possible over continental distances. Researchers trap an animal, fit it with a transmitter, and release it. The device reports its location to orbiting satellites, allowing scientists to reconstruct routes, stopover sites, and seasonal ranges. A study of Steppe Eagles fitted 16 birds with satellite transmitters, mostly in Saudi Arabia, and revealed that half of the eagles trapped during autumn migration never crossed into Africa at all, wintering instead on the Arabian Peninsula. Those that did reach Africa returned north via a completely different corridor, funneling through the Suez-Eilat area at the top of the Red Sea.¹ Without satellite tracking, that asymmetry between outbound and return routes would have stayed invisible.

The same approach works for very different species. Researchers documented the spring migration of mallards marked in Arkansas by fitting them with satellite transmitters, tracing where the ducks stopped, how long they rested, and where they ultimately nested.² In aquatic settings, passive acoustic telemetry fills a similar role. Fish and marine animals carry transmitters that emit coded signals picked up by underwater receiver networks, letting scientists build movement maps in environments where GPS signals cannot penetrate.³

Camera Traps and Motion-Activated Photography

Camera traps are among the most familiar tools in wildlife research. A motion-activated camera is strapped to a tree or post along a trail, and every warm body that walks past triggers a photograph or a short video clip. The beauty of the method is its patience: a camera trap can sit in the field for weeks or months, recording nocturnal species, rare visitors, and animals that would flee from a human observer. Camera traps are widely used to survey threatened mammal species and to estimate habitat occupancy.⁴

Camera-trap data feed into occupancy models, statistical frameworks that account for the fact that an animal can be present and still go undetected. Because a single camera pass is rarely a perfect census, researchers deploy grids of cameras and analyze the detection histories to estimate what fraction of an area a species actually occupies.⁵ A practical challenge is that detections can be clumped in time: if the same leopard walks past the camera three times in ten minutes, those are not three independent observations. Autocorrelated detections can bias results if models do not account for them.⁶

Drones and Thermal Imaging From Above

Drones have opened up a complementary aerial perspective. Equipped with thermal cameras, they can detect the heat signatures of warm-blooded animals from the sky, even at night and even when the animals are hidden in dense canopy. A study comparing drone and ground surveys for nocturnal tree-dwelling mammals in southeastern Australia found that drones consistently recorded more species and more individuals than ground-based spotlight surveys. Ground crews often missed rare specialist species like the endangered southern greater glider when populations were thin, while the drone surveys picked them up more reliably.⁷

Cost is a real consideration. Thermal drone surveys for koalas in Australia required substantial upfront investment in hardware, training, and licensing, but once that was covered, they detected more koalas per dollar spent than other methods and required the lowest average survey effort at a landscape scale.⁸ In marine settings, thermal drones have been used to survey dolphins, covering the same area in roughly one and a half to three hours that would take eight to twelve hours by boat, with less disturbance to the animals.⁹

Listening In With Passive Acoustic Monitoring

Not every animal is easy to see, but many are easy to hear. Passive acoustic monitoring uses weatherproof recorders left in the field to capture the soundscape of an area continuously, sometimes for months at a time. These recordings are then analyzed to identify which species are calling and how frequently. The approach has become a major tool for large-scale biodiversity programs and is considered increasingly important for global biodiversity targets.¹⁰

A study in the Western Ghats of southern India deployed acoustic recorders along a gradient of forest recovery, from actively restored sites to mature benchmark forests, and assessed all vocalizing fauna simultaneously rather than surveying one species at a time. The work showed that different groups of animals responded differently to habitat recovery over two decades, highlighting why a multi-species approach matters when evaluating whether restoration programs are working.¹¹ In the ocean, passive acoustic monitoring tracks marine mammals over vast spatial and temporal scales, picking up whale songs and dolphin clicks that would be impossible to observe visually in deep or remote waters.¹²

The bottleneck used to be analysis: months of recordings generate enormous audio files that no human team can listen to in full. Machine learning has changed that. Deep-learning algorithms can now classify recordings by species, by individual animal, by call type, and even by behavioral state.¹³ Open-source annotation tools allow researchers to label training datasets so that these algorithms keep improving for conservation-grade applications.¹⁴

Environmental DNA and Genetic Sampling

One of the most striking advances in recent years is the ability to detect animals without ever seeing, hearing, or photographing them. Every organism sheds DNA into its surroundings through skin cells, mucus, feces, and urine. Filter a water sample from a pond and you can extract short fragments of DNA that reveal which species live there. This environmental DNA, or eDNA, approach was demonstrated early on with a frog species, showing that specific DNA fragments could be amplified from natural wetlands to confirm the frog’s presence even when the animals themselves were hard to spot.¹⁵

The method has expanded rapidly. A review of eDNA as a survey tool concluded that it often offers greater sensitivity than traditional methods for detecting rare or invasive aquatic species, and when combined with next-generation sequencing, entire faunas can be identified from a single water sample.¹⁶ Researchers studying sea turtles developed species-specific assays that could detect and quantify turtle DNA in ocean water and nesting-beach sand. The sand samples even yielded enough genetic material for population-level genomic studies and pathogen monitoring, tasks that previously required blood draws or tissue biopsies.¹⁷

Fecal samples are another genetic goldmine. Swabbing scat lets researchers extract DNA to identify individual animals and track populations without capturing anyone. A study of East African mammals tested a simplified field method using silica-dried scat swabs and successfully sequenced mitochondrial DNA from giraffe, impala, oryx, and lion, though success rates varied by species. Taking multiple swabs from a single sample helped overcome the lower success rates for some species.¹⁸ The approach works for reptiles too: genotypes have been obtained from scat swabs of herbivorous reptiles collected under natural field conditions, though repeated amplifications are needed to confirm individual identity.¹⁹

Measuring Stress and Physiology Without a Lab

Understanding what an animal is experiencing internally, not just where it goes, matters for conservation and welfare alike. One common approach is measuring glucocorticoids, the stress hormones that spike when an animal faces a threat. These hormones and their breakdown products can be measured from feces, which means researchers can assess physiological stress without ever handling the animal. Fecal cortisol metabolite analysis has been validated as a reliable indicator of adrenal activity in species from farmed blue foxes to wild primates.²⁰

There is a catch, though. Fecal samples left in the field are exposed to sun, rain, and microbial activity, all of which can alter hormone concentrations and lead to inaccurate readings. Researchers need to account for how long a sample sat before collection and under what conditions.²¹

For more continuous physiological data, biologgers are the tool of choice. These small devices are implanted subcutaneously or attached externally and can record body temperature, heart rate, and heart-rate variability around the clock. A study of Lidia cattle used subcutaneous biologgers to document circadian rhythms and short-term responses to summer heat.²² More broadly, the array of sensors used in animal biotelemetry now includes accelerometers that detect feeding and spawning behavior, depth sensors for diving animals, and even devices that register proximity to other individuals, predators, or prey.²³

Behavioral Observation and Ethograms

Technology cannot fully replace a trained pair of eyes. Behavioral observation, the oldest method in the toolbox, remains central to animal science. The basic approach involves choosing a focal animal and recording what it does at set intervals or continuously over a period. A study of zebu cattle in a slaughterhouse lairage used the focal-animal sampling method with recordings taken every five minutes per animal to assess how transport distance and resting time affected behavior.²⁴

For complex behavioral repertoires, researchers develop ethograms: standardized catalogs of every distinct behavior a species performs, each carefully defined so that different observers code the same action the same way. In a study of Indian rhinoceros breeding behavior, scientists used focal and ad libitum sampling to document scent-marking, tactile stimulation, and vocalizations, correlating those observed behaviors with hormonal data from fecal and blood analysis. The resulting ethogram served as a reference tool for categorizing and comparing behaviors across individuals and time.²⁵

Counting Populations With Mark-Recapture

Knowing how many animals are in a population is fundamental, but counting every individual is rarely feasible. Mark-recapture methods offer a workaround: capture a sample, mark them, release them, then capture a second sample later. The proportion of marked individuals that show up in the second sample gives you a basis for estimating total population size. A study using an intensively surveyed gecko population as a controlled benchmark tested twelve different mark-recapture estimators and found that models accounting for the fact that some individuals are harder to catch than others performed best across all accuracy criteria.²⁶

Camera traps have added a photographic twist to this method. For carnivores with unique coat patterns, like leopards or jaguars, individual identification from photographs replaces physical marking entirely. For species without unique markings, a hybrid approach works: physically collar a subset of the population with GPS tags and then use camera traps to record both collared and uncollared individuals. Researchers studying bobcats in urban southern California deployed a grid of 30 cameras alongside GPS-collared animals and compared traditional mark-recapture estimates with a new “hybrid” mark-resight model they developed, expanding the statistical toolkit for population surveys.²⁷

Citizen Science and Crowdsourced Data

Professional researchers cannot be everywhere at once, and citizen science has become an increasingly recognized way to fill the gaps. Volunteers contribute wildlife sightings, bird counts, and photo records through organized platforms and apps. An analysis of environmental impact statements in the United States found that citizen science data was incorporated into 40 percent of statements by 2022, up from just 3 percent in 2012, reflecting growing confidence in the quality and utility of crowdsourced biodiversity data.²⁸ The sheer geographic and temporal coverage that thousands of volunteers provide is something no professional survey budget can replicate, though data quality controls and species identification accuracy remain active areas of discussion.

Ethics, Welfare, and the Cost of Studying Animals

Every method described here involves a tradeoff between the knowledge gained and the potential impact on the animal. Capturing birds to attach transmitters, trapping mammals for marking, even the presence of a drone overhead can cause stress. Research on captive wild animals has documented numerous stressors including cage restraint, human presence, unfamiliar environments, and artificial lighting.²⁹ Repeated trapping can have lasting effects that go beyond temporary discomfort. A long-term study of pied flycatchers found that birds with no previous trapping experience entered nest boxes more quickly than birds that had been caught before, and that this reluctance increased as the number of prior captures accumulated over the bird’s lifetime. The practical consequence was a sampling bias toward younger birds, because older, trap-wary individuals progressively dropped out of the data.³⁰

This is one reason the push toward noninvasive methods, like eDNA, acoustic monitoring, camera traps, and fecal sampling, has been so strong. The ethical framework governing animal research in many countries revolves around the 3Rs: replacement, reduction, and refinement. Replacement means using alternatives to live animals when possible. Reduction means using the fewest animals necessary. Refinement means minimizing suffering in whatever procedure is used. A scoping review of how animal ethics committees actually make decisions found deficits in the assessment of the 3Rs and in the weighing of harms against benefits in submitted research applications.³¹ A separate analysis of Swedish ethical review applications found that information about harm, benefit, and the 3Rs was often insufficient or occasionally missing altogether.³² The gap between the principle and the practice is real, and it shapes ongoing debates about how field research should be conducted.

When Tracking Data Catches Poachers

An unexpected use of animal tracking technology has emerged in wildlife crime detection. GPS-tagged vultures and wolves, monitored for ecological research, sometimes lead investigators to the remains of illegally killed animals. When researchers visited potential feeding spots identified by vulture tracking data, they discovered shot red deer with their heads removed in Spain and Portugal, a fallow deer suspected of being poached in Bulgaria, and additional carcasses at wolf feeding sites in the Czech Republic. The animals carrying the GPS tags were effectively acting as sentinels, their scavenging behavior flagging crime scenes that would otherwise have gone undetected in remote landscapes. The approach turns routine ecological monitoring into a tool for law enforcement, though it also raises questions about whether publicly available tracking data could be exploited by poachers to locate endangered species, a concern sometimes called cyber-poaching that the research community is still working out how to manage.