Skip to content
NatureHQ

Marine lifespecies group

Salmon

Salmonidae

Salmon do smell their way home — for the last few kilometres. A river’s scent does not reach the open Pacific, and something else brings them to the right coast.

Salmon hatch in fresh water, spend years at sea, and return to the river they came from. "They smell their way home" is half the answer: smell explains the last few kilometres and cannot explain the thousands before it.

The salmon journey is two navigation problems wearing one name, and they need different mechanisms because they happen at different scales. The first is oceanic. A fish that went to sea from a particular river has to come back to the right piece of coastline from somewhere in the open Pacific or Atlantic, where no river odour reaches — dilution alone rules it out. The second is the river choice, from an estuary onwards, where the water genuinely does carry the chemistry of the catchment it drained. Olfactory imprinting answers the second and answers it decisively: young salmon exposed to a synthetic compound during the smolt stage returned eighteen months later to whichever stream had been dosed with that compound, which is not homing to a place but homing to a smell. The ocean half is less settled. The leading account is that salmon also imprint on the magnetic signature of the region they left, and the evidence for it is a fifty-six-year record of which side of Vancouver Island returning sockeye chose, tracking the drift of the field at the two entrances. That is a correlation rather than an experiment, and it is stated here as one — but it is the same imprinting logic that has experimental support in sea turtles, which is why it is taken seriously.

Developed coverage · 48% complete · reviewed 2026-09-03

What this page covers

Atlantic salmon and the Pacific salmon of the genus Oncorhynchus. The homing experiments are mostly on coho and sockeye; the details differ between species and between hatchery and wild fish.

Often confused with: Trout, which are close relatives and mostly do not make the same journey; Eels, which migrate in the opposite direction — river to ocean to breed

Quick facts

What imprinting fixes
The stream, learned as a smell during a juvenile window
What smell cannot do
Reach the open ocean — the first thousands of kilometres need something else
The ocean candidate
Magnetic signature of the natal region, on correlational evidence
The decisive experiment
A synthetic odour no river contains — and the fish came to it

What salmon can detect

Assembled from the evidence graph. Each entry is a link the corpus can justify, with the limits it was recorded under.

One journey, two navigation problems

Diagram

Which cue is doing the work, and where

Schematic. Distances are not to scale and the boundary is not a line.

Which cue is doing the work, and whereSchematic. Not to scale, and the boundary is not a line.Open oceanRiver odour: absent at anyusable concentration.Candidate: magnetic signatureevidence: correlationalCoastBoth available;first traces offresh water.RiverLearned odourdominates.evidence: testedthousands of kilometresa few kilometres“Salmon smell their way home” keeps only the right-hand box.
The same explanation in words

A journey drawn from left to right in three zones. In the open ocean, thousands of kilometres from land, the label states that river odour is absent at any usable concentration and that the leading candidate is the magnetic signature imprinted on as a juvenile — marked as correlational evidence. Approaching the coast, the two overlap: field information still available, and the first traces of freshwater chemistry appearing. In the estuary and river, the label states that odour learned during the smolt stage now dominates, with the evidence marked as experimental. Beneath, a line notes that the popular account keeps only the third zone.

A river’s smell does not reach the open Pacific. Something else brings a salmon back to the right coast, and the best current candidate is the magnetic signature of the place it left.

Emerging evidence

Real findings exist, but too few or too recent to be settled.

Olfactory imprinting operates at the scale of a river mouth and cannot supply positional information across ocean basins. Fifty-six years of sockeye fishery records show the proportion using each entrance to the Fraser River tracked which entrance better matched the geomagnetic signature present when that cohort migrated to sea, consistent with geomagnetic imprinting supplying the ocean-scale component.

Who this applies to
Sockeye returning around Vancouver Island; the two-scale model is applied more widely than it has been tested.
Studied in
Oncorhynchus nerka
Why we rate it this way, and what the caveats are
Emerging evidenceModerate confidence

The negative half — that odour cannot work at ocean scale — is physically secure. The positive half rests on a decades-long correlation in catch data rather than on an experiment.

How far it can be extended

Geomagnetic imprinting has independent experimental support in sea turtles, and the olfactory half is established across salmonids.

Caveats

  • Correlational, from catch records rather than tracked fish.
  • Other things that co-vary with the field over decades — ocean temperature, currents — are hard to exclude completely.
  • No magnetoreceptor has been identified in salmon.

Still unanswered

  • Whether hatchery rearing, which happens inside buildings and pipes, disrupts magnetic imprinting as it can disrupt olfactory imprinting.

Last reviewed 2026-09-03

The evidence (3 studies)

How we know

Fifty-six years of salmon catches, and which way round an island they came

If smell only works near the river, what brings a salmon back to the right coast after years in the open ocean?

Sockeye returning to the Fraser River must pass Vancouver Island on one side or the other. Fishery records document which route each year’s run took, going back more than half a century. Those proportions were compared with the drifting geomagnetic signatures of the two entrances, testing whether each cohort favoured the entrance whose field best matched the signature present when that cohort left the river as juveniles.

What happened

The proportion of fish using each entrance tracked which one matched the magnetic signature of their outward year.

What it shows

That something magnetic, learned on the way out, is guiding the ocean-scale part of the return — the part smell cannot explain. It is the second half of a two-scale answer whose first half is olfactory imprinting.

What it does not show

It is correlational, and built from catch statistics rather than tracked fish: no salmon was followed. Other things drift with the field over decades — currents, temperature — and cannot be entirely excluded. No magnetoreceptor has been identified in a salmon.

The controls — what makes this evidence rather than a story
  • A record long enough — fifty-six years — that field drift produces genuine variation to test against.
  • Field values taken from independent geomagnetic models rather than from the fishery data.
  • Sea-surface temperature examined as a competing explanation.

From Evidence for Geomagnetic Imprinting as a Homing Mechanism in Pacific Salmon

How it knows where to go

A salmon returning to spawn

  • Magnetic signature of the natal region

    Proposed

    The ocean-scale part: which stretch of coast to come back to

    Fifty-six years of sockeye fishery records track the drift of the field at the two entrances to the Fraser River.

  • Odour learned as a smolt

    Demonstrated

    The river and the stream, once the fish is close enough to detect it

    Juveniles exposed to a synthetic compound returned eighteen months later to whichever stream had been dosed with it.

  • River flow and current

    Supported

    The direction to swim once inside fresh water

    Long observed in spawning runs; the least contested and least interesting part.

Words used here
Smolt
The stage at which a young salmon changes physiologically for salt water and moves downstream. It is also the window in which home-water odour is learned.
Imprinting
Learning that happens only during a particular window of development and is then difficult to overwrite.

Homing to a smell that belongs to no river

How we know

Teaching salmon a smell that exists in no river

Do salmon return to a place, or to a smell they learned there? The two are impossible to separate — unless you supply a smell no place has.

Young hatchery coho salmon were exposed to morpholine, a synthetic chemical found in no natural stream, during the smolt stage when they would normally be learning their home water. A second group was exposed to nothing. Both were released. Eighteen months later, as the adults returned to spawn, streams were dosed with morpholine and traps set at scented and unscented streams alike.

What happened

The exposed salmon turned up overwhelmingly at whichever stream had been scented with morpholine. The unexposed controls distributed themselves elsewhere.

What it shows

That salmon learn a chemical signature during a juvenile sensitive period and use that memory to choose a stream as adults. Substituting an artificial odour is what makes it decisive: these fish did not go home, they went to a smell.

What it does not show

It says nothing about the ocean. A river’s odour does not reach the open Pacific, so this explains the final stream choice and not the thousands of kilometres before it. Hatchery fish released as a cohort may also imprint differently from wild fish, and one synthetic compound is a poor model of natural stream water, which is a mixture nobody has fully described.

The controls — what makes this evidence rather than a story
  • Unexposed fish released at the same time and place, which should show no preference for the scented stream.
  • A synthetic compound absent from natural waters, so a fish arriving at it cannot be following anything else.
  • Traps at multiple streams, including unscented ones, so a preference can be measured rather than assumed.

From Imprinting to Chemical Cues: The Basis for Home Stream Selection in Salmon

Salmon exposed to a synthetic chemical as juveniles returned years later to whichever stream had been scented with it. They were homing to a smell they had learned, not to a place.

Established

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

Coho salmon exposed to morpholine during the smolt stage and released were recaptured predominantly at streams artificially scented with morpholine eighteen months later, while unexposed controls were not, demonstrating olfactory imprinting during a juvenile sensitive period as the basis of home-stream selection.

Who this applies to
Hatchery coho salmon in a Great Lakes system, using a synthetic odorant.
Studied in
Oncorhynchus kisutch
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

Substituting an odour that occurs in no stream removes the alternative explanations at a stroke: a fish arriving at morpholine cannot be following anything else.

How far it can be extended

Olfactory imprinting is supported across salmonids, though the decisive substitution experiment was done in this species.

Caveats

  • One synthetic compound stands in for natural stream odour, which is a complex mixture nobody has fully characterised.
  • Hatchery fish released as a cohort; the timing and content of wild imprinting may differ.
  • Explains the final choice of stream, and nothing about the ocean crossing that precedes it.

Still unanswered

  • What natural stream odour actually consists of, and how stable it is across years of changing flow and land use.

Last reviewed 2026-09-03

The evidence (2 studies)

Substituting the odour is what makes this decisive rather than suggestive. A fish that returns to its home stream might be doing so for any number of reasons that correlate with the stream. A fish that returns to morpholine has told you what it was following, because morpholine is in no river anywhere and the only place it had ever encountered the compound was a hatchery tank eighteen months earlier.

The sense doing the work

What is still unsettled

  • What is a stream’s odour actually made of?

    Why it matters: The imprinting result is fifty years old and the chemistry it depends on has never been characterised, which makes it impossible to say how stable a home smell is across years of changing flow and land use.

    What would settle it: Chemical profiling of natal waters across seasons, combined with behavioural tests of the components.

  • Does hatchery rearing disrupt magnetic imprinting?

    Why it matters: Hatcheries are full of pipes, pumps and reinforcing steel, all of which distort the local field. If the ocean-scale mechanism is magnetic, hatchery fish may be imprinting on a signature that exists nowhere in the sea.

    What would settle it: Comparing return rates and routes of fish reared with and without magnetic distortion, which is a tractable experiment and has barely been done.

  • How much straying is normal, and is it useful?

    Why it matters: Not all salmon return to their natal stream, and that error rate is what lets salmon colonise new rivers after glaciers or dams. It is usually treated as failure rather than as a strategy.

    What would settle it: Long-term marking studies quantifying straying rates against environmental change.

Claims about this, checked

Things people have heard, and what the evidence actually supports.

The research behind this page

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

2015Current Biology

Evidence for Geomagnetic Imprinting and Magnetic Navigation in the Natal Homing of Sea Turtles

Where magnetic signatures of adjacent coastal areas converged over time, nesting density increased; where they diverged, it decreased.

2013Current Biology

Evidence for Geomagnetic Imprinting as a Homing Mechanism in Pacific Salmon

The proportion of fish using each route tracked which entrance more closely matched the magnetic signature present when that cohort left the river as juveniles.

1976Science

Imprinting to Chemical Cues: The Basis for Home Stream Selection in Salmon

Salmon exposed as smolts returned overwhelmingly to the stream scented with that chemical, while unexposed controls did not.

1951The American Naturalist

Discrimination of Stream Odors by Fishes and Its Relation to Parent Stream Behavior

Fish could be trained to tell the two waters apart, and lost the ability when their sense of smell was blocked.

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 48% completeness against what we would call a finished subject, and was last reviewed on 2026-09-03. It carries 2 claims and answers 1 mapped search questions.

  • no research from the last few years is attached — check for newer work
  • Only the homing story is told. Salmon life history, the physiological change at the smolt stage, and the post-spawning death of Pacific species are absent.
  • Atlantic salmon, which do not necessarily die after spawning, are named rather than treated separately.
  • Aquaculture, hatcheries and their genetic consequences are outside the page except where they bear on imprinting.