Attaching a device to an animal and receiving its positions produces enormous quantities of data, and turning that into understanding is a field in itself.
What is recorded
Position, at intervals from seconds to days depending on the device and battery.
Increasingly, accelerometry recording body movement in three axes, which allows behaviour to be inferred.
Temperature, heart rate in some devices, and depth for diving animals.
Which together allow reconstruction of what an animal was doing as well as where it was.
Home range
The area an animal uses, which is less straightforward to define than it sounds.
Early methods drew a polygon around all locations, which includes areas never visited.
Kernel density methods estimate use intensity across space.
Movement-based methods account for the fact that consecutive positions are not independent, which the earlier methods ignored.
Which means home range figures from different eras are not comparable.
Behavioural inference
Movement patterns differ by activity.
Slow tortuous movement suggests foraging. Fast directed movement suggests travel. Stationary periods suggest rest.
Statistical models classify segments of a track into behavioural states, which allows time budgets to be estimated without observation.
Accelerometry improves this substantially, distinguishing behaviours that produce similar movement patterns.
Habitat selection
Comparing where animals go with what was available.
Which requires a definition of availability, and that choice affects the results substantially.
Step selection analysis compares each actual movement with alternatives the animal could have made, which handles this more rigorously.
Mortality
Devices with mortality sensors transmit when an animal stops moving, allowing carcasses to be located.
Which permits cause of death to be established, and it has revealed mortality sources that were not suspected.
Poaching, vehicle collision, disease and predation rates have all been quantified this way for species where direct observation was impossible.
The sample problem
Tracking studies use few animals, since devices are expensive and capture is difficult.
Which means individual variation can dominate, and generalising from a handful of animals to a population requires care.
Studies increasingly report individual variation explicitly rather than averaging it away.
Capture effects
Fitting a device requires capturing the animal, which is stressful and carries risk.
Which means data from the period immediately after capture may not represent normal behaviour, and studies generally exclude it.
Mortality during capture is reported in the literature and is a genuine ethical cost weighed against the knowledge gained.
Data sharing and sensitivity
Aggregated tracking data allows analyses across species and regions that individual studies cannot support.
Which has produced repositories and standards.
Location data for threatened species is sensitive, since it could assist poaching, and access controls are applied.
Cases where published or leaked location data was used to target animals have occurred, which is why the caution exists.
What it has changed
Migration routes, habitat requirements, mortality causes and human interaction have all been quantified for species where they were previously guessed.
Which has directly informed protected area design, corridor identification and mitigation of specific threats.
Device attachment
Collars for many mammals, harnesses for birds, glued or implanted devices for others.
Which each carry welfare considerations, and attachment methods have been refined following studies of their effects.
Drop-off mechanisms allowing collars to release after a set period avoid the need to recapture, and they are now standard for many applications.
Battery and data limits
Determine sampling frequency and duration, which is the central design trade-off.
Frequent positions give detailed movement and drain the battery faster.
Which means study design must decide between duration and resolution, and both are needed for different questions.
Solar charging and improved efficiency have relaxed this constraint considerably.
Combining with other data
Tracking data becomes more powerful combined with environmental layers — habitat, weather, human activity.
Which allows movement to be related to conditions, and it is where most of the ecological insight comes from.
Public engagement
Tracked animals with public-facing maps generate substantial interest.
Which builds support and raises the sensitivity question, since precise real-time locations of vulnerable animals are a risk.
Delayed or generalised public data addresses this while retaining the engagement.
What questions it answers well
Movement, habitat use, migration, home range and mortality.
What it answers poorly
Population size, since tracking a few individuals says little about how many exist.
Interactions between individuals, unless many are tracked simultaneously.
And diet, unless combined with other methods.
Which is why tracking is generally one component of a study rather than the whole of it.
Ethics review
Studies involving capture and attachment require approval from animal welfare committees in most institutions.
Which assesses the justification, the methods and the expected impact against the knowledge gained.
Reporting of adverse events has improved and remains inconsistent across the literature.
Cost
Devices, capture operations and data subscriptions make tracking expensive per animal.
Which limits sample sizes and means the method suits questions where a few individuals genuinely answer them.
Falling device costs have widened its use considerably over the past decade.