Skip to content
NatureHQ

Senses and abilitiesmechanism

How we know where animals go

A ring gives you two dots on a map and nothing in between. Everything since has been an argument about how to fill in the middle.

Almost everything known about where animals go comes from six methods, each with a different blind spot. A ring tells you two points and nothing in between; a geolocator has to be recovered from the same bird a year later; radar counts animals it cannot name.

It is worth knowing how a migration map was made, because the method decides what the map can say. For most of the twentieth century the answer came from ringing: put a numbered band on a bird, and hope somebody finds it. That gives you two points — where it was ringed and where it was found — and pure guesswork in between, plus a heavy bias towards places with people in them to do the finding. Everything since has been an attempt to fill in the middle. Radio telemetry lets somebody with an aerial follow an animal that is nearby. Satellite tags report position from anywhere and are too heavy for most birds. Geolocators solve the weight problem by recording nothing but light levels, from which latitude and longitude can be reconstructed — but they transmit nothing, so the bird has to be caught again, which means every route measured this way belongs to an animal that survived. GPS tags fix position precisely and drain batteries doing it, so their sampling is a compromise between resolution and lifespan. Radar sees migration without identifying any individual, which is useless for a species question and exactly right for asking how many animals are aloft tonight. And acoustic telemetry, underwater, works by listening: an animal exists where a receiver hears it, which makes coverage a map of where the receivers are.

Early coverage · 37% complete · reviewed 2026-09-03

What this page covers

Methods rather than organisms. Coverage is uneven by design: large animals can carry heavier tags, so what is known about movement is systematically better for a swan than for a swift.

Often confused with: GPS in the everyday sense — most animal tags are not GPS, and the ones that are do not stream live positions

Quick facts

The binding constraint
Tag mass — which is why we know more about a swan than a swift
Geolocators
Must be recovered, so every route comes from a survivor
Radar
Counts what is aloft; cannot say which species
Underwater
An animal exists where a receiver is listening, and nowhere else

Six ways of finding out, and what each one misses

What is known about where animals go is mostly a story about batteries. Every time tags got smaller, a new set of species became answerable — which is why we know far more about a swan than a swift.

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

Tag mass, battery capacity and data-recovery requirements, rather than biological interest, have set which species and which questions were tractable in movement ecology; successive reductions in tag mass have repeatedly opened new taxa to direct tracking.

Who this applies to
A statement about method across terrestrial and aquatic tracking.
Studied in
Animalia
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

Not a contested empirical question: the constraint is physical, and the history of the field maps onto the history of the hardware.

How far it can be extended

The same constraint is described independently in the terrestrial and aquatic tracking literatures.

Caveats

  • Miniaturisation has continued since these reviews, so the specific limits date quickly.
  • Tag effects on the animal are an active concern in their own right, not only a sampling constraint.

Still unanswered

  • How much published movement data is distorted by the tags themselves, which is difficult to assess without untagged controls.

Last reviewed 2026-09-03

The evidence (3 studies)

Diagram

What each method actually gives you

Indicative. Resolutions vary with hardware generation, deployment and species.

Every route map inherits the blind spot of its methodwhat it recordsmain limitationRinging / bandingTwo points: marked, and foundNothing in between; needs a finderRadio telemetryPosition while in receiver rangeSomebody must be nearbySatellite tagPosition anywhere on EarthToo heavy for small animalsGeolocatorLight levels → rough positionMust be recovered from the animalGPS tagPrecise position, on a scheduleEvery fix costs batteryWeather radarHow much is aloft, over a regionCannot identify a speciesTag mass is the binding constraint — which is why we know more about a swan than a swift.
The same explanation in words

A comparison of six tracking methods across three properties: what it records, how precisely, and its main limitation. Ringing or banding records two points — the place of marking and the place of recovery — with the position exact but the route entirely unknown, and it depends on somebody finding the ring. Radio telemetry records position while a receiver is within range, to within metres, but somebody has to be near the animal. Satellite tags record position anywhere on Earth, to within tens or hundreds of metres, and are too heavy for small animals. Light-level geolocators record day length and the timing of local noon, giving positions accurate to roughly a hundred kilometres and unusable near the equinoxes, and the device must be physically recovered. GPS tags record position precisely, to within metres, but each fix costs battery, so the sampling interval is a compromise. Weather radar records how much biomass is aloft over a wide area with no ability to identify species at all. A closing note says that every route map inherits the blind spot of the method that made it.

Resolutions are indicative and vary with hardware generation, deployment and species.
MethodWhat it measuresRoughly how preciseThe catch
Ringing / bandingTwo points: where it was marked, where it was foundExact at both pointsNothing at all in between, and it needs a person to find the ring
Radio telemetryPosition while a receiver is in rangeMetresSomebody has to be near the animal, often in an aircraft
Satellite tagPosition anywhere on Earth, reported remotelyTens to hundreds of metresToo heavy for most birds and all insects
Light-level geolocatorDay length and the timing of local noonAround 100–200 km, and useless near the equinoxesTransmits nothing — the device must be recovered from the animal
GPS tagPosition on a set scheduleMetresEvery fix costs battery, so resolution trades against lifespan
Weather radarHow much biomass is aloft across a regionNo individual resolution at allCannot identify a species, and insects must be filtered out
Words used here
Geolocator
A tag recording only ambient light against time. Day length gives latitude and the timing of local noon gives longitude, to within about a hundred kilometres.
Acoustic telemetry
Underwater tracking in which a tagged animal emits a coded ping and fixed receivers log it when it passes within range.

Every geolocator route belongs to an animal that came back

A geolocator stores light levels and transmits nothing. To learn anything you have to catch the same bird again the following year — so every route measured this way comes from a bird that survived.

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

Light-level geolocators record ambient light against time and must be physically recovered to be read, restricting the resulting datasets to individuals that survived the annual cycle and returned to a recapture site; positional accuracy is on the order of one to two hundred kilometres and longitude estimates fail around the equinoxes.

Who this applies to
Light-level geolocation as applied to birds.
Studied in
Aves
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

A property of the instrument rather than a finding about animals, and stated as such in the studies that use it.

How far it can be extended

The recovery requirement and the equinox limitation are properties of the method, not of any species.

Caveats

  • Survivor bias affects estimates of route and timing, not only of mortality: birds that took a fatal route are absent by construction.
  • Accuracy varies with latitude, weather and shading by the bird’s own feathers.

Still unanswered

  • How much published route variation is real and how much is geolocation error, which is rarely propagated into the maps that result.

Last reviewed 2026-09-03

The evidence (2 studies)

How we know

Waiting a year to find out where the bird went

How far does an Arctic tern actually travel in a year — measured, rather than estimated?

Arctic terns at colonies in Greenland and Iceland were fitted with light-level geolocators weighing about a gram. The device records nothing but ambient light against time and transmits nothing at all; the birds were left to migrate, and the following season the same individuals were caught again at the same colonies and the devices removed and read. Sunrise and sunset times give day length and local noon, from which latitude and longitude are reconstructed.

What happened

The birds followed winding Atlantic routes with a prolonged mid-ocean stopover, tracked south along either the African or the South American side, and covered annual distances far greater than the straight-line distance between the poles.

What it shows

The longest annual migration measured in any animal, and — more usefully — its actual shape, which is nothing like the straight line the older estimates assumed.

What it does not show

The method reports only from birds that survived and came back to the same colony, so every route here is a survivor’s. Light-level positions are good to roughly a hundred kilometres and unusable near the equinoxes, so route length carries real uncertainty. Two colonies do not describe a species.

The controls — what makes this evidence rather than a story
  • Recapture of the same marked individuals, so each track belongs to a known bird.
  • Two colonies, giving some check on whether a route is a local peculiarity.
  • Known accuracy limits applied when reconstructing positions, including the equinox periods where longitude cannot be derived.

From Tracking of Arctic terns Sterna paradisaea reveals longest animal migration

The bias this creates is not only about mortality rates. If a route is dangerous, the birds that took it are disproportionately absent from the data — so the routes we know about are, by construction, the ones that worked. That is worth remembering whenever a tracking study describes what a species does.

Seeing migration without seeing a single animal

Weather radar sees migration. It cannot tell you which birds, or where they are going — but it counts how many are in the air tonight across a continent, and that turns out to be predictable from the weather.

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

Weather-surveillance radar quantifies nightly migration intensity as biomass aloft without species identification; across twenty-three years of United States radar data, atmospheric conditions accounted for the great majority of variation in migration intensity, supporting skilful forecasts several days ahead.

Who this applies to
Nocturnal migration over the continental United States, where the radar network is dense.
Studied in
Aves
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

Two decades of data and a forecasting model with tested skill, which is a stronger form of validation than a single-season correlation.

How far it can be extended

European radar networks have produced comparable measurements; the method needs dense coverage.

Caveats

  • Radar cannot identify species, and insects must be filtered out of the signal.
  • It measures how much is aloft, not where any of it is going or whether it arrives.
  • Coverage is a function of the radar network, so the picture is continental where the network is and absent where it is not.

Still unanswered

  • How to attribute radar-measured biomass to species groups, which would connect these measurements to population trends.

Last reviewed 2026-09-03

The evidence (1 study)

Diagram

What a weather radar sees after dark

Schematic. The bloom pattern is illustrative of the phenomenon, not a real radar image.

An instrument built to watch rain, watching migrationSchematic. Not a real radar image.dusknothing aloftan hour after sunsetdeparture bloomsmiddle of the nighta broad driftIt counts biomass aloft. It cannot say which birds, or where they are going.
The same explanation in words

A sequence of three radar sweeps through one night. At dusk the display is nearly empty. An hour after sunset, expanding rings bloom outward from roosting areas as birds take off, filling much of the display. By the middle of the night the pattern has smoothed into a broad drift moving in one direction across the whole area. Labels note that the radar is measuring biomass aloft rather than identifying anything, that insects have to be filtered out of the signal, and that the same instrument built to watch rain turns out to watch migration.

This is the method that makes migration a forecastable phenomenon rather than an anecdote about a garden. Two decades of radar over the United States show that how much is aloft on a given night is largely predictable from the weather — which means a city can be told, several days ahead, that tonight is a night to turn the lights off.

Where the methods get used

What is still unsettled

  • How much do tags change the behaviour they measure?

    Why it matters: Every tracked route is a route flown by an animal carrying something it did not evolve to carry. Effects on energetics and survival are documented and rarely propagated into conclusions.

    What would settle it: Paired studies comparing tagged and untagged individuals on the same measures, which are difficult because the untagged animals are by definition unmeasurable.

  • Can radar be taught to identify what it is seeing?

    Why it matters: Connecting continental-scale migration measurements to species-level population trends is currently impossible, and it is the biggest gap in migration monitoring.

    What would settle it: Combining radar with acoustic monitoring or with dual-polarisation signatures, which is an active area.

  • How biased are the resulting maps towards well-studied places?

    Why it matters: Receivers, radars and ringing effort are concentrated in wealthy countries, so gaps in coverage look like gaps in animal movement.

    What would settle it: Coordinated deployment in under-covered regions, which is mostly a funding question rather than a technical one.

The research behind this page

5 studies, newest first. Each one has a page explaining what it found and what it could not show.

2018Science

A continental system for forecasting bird migration

Nightly migration intensity is predictable from weather, with atmospheric conditions accounting for the great majority of variation, and forecasts skilful up to a week ahead.

2015Science

Terrestrial animal tracking as an eye on life and planet

Tag mass, battery life and data recovery, rather than biology, have set the boundaries of what was knowable about animal movement, and each reduction in tag size opened a new class of species and question.

2015Science

Aquatic animal telemetry: A panoramic window into the underwater world

Water blocks the signals terrestrial tracking depends on, so aquatic tracking works either by recording and later transmitting, or by listening for tagged animals with fixed receiver arrays — which makes coverage a function of where receivers happen to be.

2010Proceedings of the National Academy of Sciences

Tracking of Arctic terns Sterna paradisaea reveals longest animal migration

Birds followed a winding route with an extended stopover in the North Atlantic and a southbound track down either the African or the South American side, covering annual distances far greater than the straight-line pole-to-pole distance.

2009Science

Tracking Long-Distance Songbird Migration by Using Geolocators

Birds travelled far faster on spring migration than existing estimates assumed, with individual daily rates several times higher than had been inferred from ringing recoveries.

This page is a stop on a longer route

A guided journey reads several subjects in a deliberate order, with an argument for why one follows another. You can join in the middle.

Where to go from here

Each of these follows from something on this page — a relationship in the evidence, a claim people ask about, or the next mechanism along.

How complete this page is, and what it is still missing

NatureHQ publishes its own gaps. This page is at 37% completeness against what we would call a finished subject, and was last reviewed on 2026-09-03. It carries 4 claims and answers 2 mapped search questions.

  • no research from the last few years is attached — check for newer work
  • more experiments could be explained in plain English
  • no popular claim about this subject has been checked yet
  • Stable isotope analysis, which infers where an animal grew its feathers rather than following it, is not covered.
  • Acoustic telemetry is described in one paragraph despite being the main method for fish and sharks.
  • The ethics and permitting of tagging are mentioned only as a limitation.