Skip to content
NatureHQ

Marine lifespecies

Humpback whale

Megaptera novaeangliae

In two seasons, every male humpback off eastern Australia stopped singing their song and started singing someone else’s.

The humpback is the animal that proved culture can cross an ocean. Males sing a long, hierarchically structured song, every male in a population sings roughly the same one, and it changes — sometimes gradually, sometimes replaced wholesale within two seasons by a song carried in by a handful of arriving whales.

What makes humpback song extraordinary is not that it is beautiful, which is a matter of taste, but that it is shared and that it moves. A song is built in layers: units combine into phrases, phrases repeat to make themes, and themes occur in a fixed order to make a song a male may repeat for hours without a break. Every singer in a population sings essentially the same song at the same time, and the song drifts as the season goes on — with every singer drifting together, which already requires that they are listening to each other. Then there is the part that made this a landmark in the study of animal culture. In the mid-1990s a few whales singing the west Australian song appeared off eastern Australia. Within two seasons the entire east Australian population had abandoned its own song and taken up the western one — not blended, replaced. Later work showed this was not a freak event but a system: song types appear off eastern Australia and sweep eastward across the South Pacific, arriving at each population a year or two later than the last, like a wave crossing an ocean. What none of this explains is what song is for. The obvious answers — attracting females, competing with males — have surprisingly thin support, and a display that every male copies from every other male is a strange thing for sexual selection to have produced. The best-documented cultural transmission in any animal sits on top of a behaviour whose function is still genuinely unresolved.

Developed coverage · 43% complete · reviewed 2026-09-03

What this page covers

One species, distributed worldwide in populations that migrate between polar feeding grounds and tropical breeding grounds. Song is produced by males and is a property of populations rather than of individuals.

Quick facts

Song structure
Units into phrases, phrases into themes, themes in fixed order
How fast a song can be replaced
Two seasons, across an entire population
Direction of travel
Eastward across the South Pacific, one to two years per population
What song is for
Genuinely unresolved

A song with a grammar of themes

Hierarchically structured, hours long, and sung by every male in a population at once.

The structure was invisible until somebody drew it. Converted to a spectrogram and laid out across minutes, humpback sound turns out to be organised in nested repeats: short units group into phrases, phrases repeat to build a theme, and themes follow one another in a consistent order to form a song. A male may cycle through it for hours. Every singer in a population sings the same version, and that version changes through the season with all of them changing together.

Structure is not meaning. Humpback song is hierarchically organised, and that fact gets recruited constantly into claims about whale language. It should not be. A hierarchy is a statement about the form of a signal — that it has parts, and parts of parts. Nothing about it implies the parts refer to anything, and no unit of humpback song has ever been shown to carry a meaning.

Based on Several animals have one or two of the things that make human language what it is. Nothing has been shown to have the whole set.

A whole population of humpbacks can abandon its song and take up another within two years, and songs travel across an ocean in a consistent direction.

Established

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

Humpback whale song is a hierarchically structured display that changes progressively within a population and is periodically replaced entirely. Replacement has been documented following the arrival of a small number of animals singing a different song, and song types propagate directionally across the South Pacific over one to two years per population.

Who this applies to
Best documented in the South Pacific, where populations have been recorded across a decade. Sung by males on and around breeding grounds.
Studied in
Megaptera novaeangliae
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

A wholesale replacement observed directly, followed by a decade of ocean-scale sampling showing the same pattern repeating with consistent direction. Neither genetic change nor independent invention accounts for either.

How far it can be extended

Song of this kind is a humpback characteristic; nothing comparable is documented in other baleen whales.

Caveats

  • Why a novel song displaces the resident one is not established.
  • The function of song itself remains uncertain, which limits what changes in it can be taken to mean.
  • Sampling effort differs a great deal between populations.

Still unanswered

  • What song is actually for — male competition, female choice, or something else.
  • Where between populations the transmission physically happens.

Last reviewed 2026-08-31

The evidence (3 studies)

The song that crossed an ocean

A handful of immigrants, and a whole population changed what it sang.

Off eastern Australia in the mid-1990s, researchers recording the resident population noticed two whales singing something else entirely — the song of the west Australian population, on the other side of the continent. The following season a handful more sang it. The season after that, the eastern song had gone. Not merged, not blended: the whole population had switched.

That could have been an oddity. A decade of recordings across six South Pacific populations showed it was not. Song types keep appearing off eastern Australia and moving east — reaching New Caledonia, then Tonga, then the Cook Islands and French Polynesia, each a year or two after the last. It is a wave of transmission with a consistent direction, and neither genetics nor independent invention produces that pattern.

How we know

Following one whale song across an ocean

Is a population’s song replaced by chance, or does the same song travel between populations in a consistent direction?

Songs were recorded from six humpback populations spread across thousands of kilometres of the western and central South Pacific, over about a decade. Each season’s songs were classified into types by their theme content and order, and the year in which each type first appeared in each population was compared against geography.

What happened

Song types appeared first off eastern Australia and then swept eastward, reaching each population in turn a year or two after the last, all the way across the South Pacific.

What it shows

Cultural transmission at ocean-basin scale, and repeatedly. A consistent direction of travel with a consistent lag is what a wave of transmission looks like, and it is not what independent invention or genetic change looks like.

What it does not show

It does not explain why a new song displaces the old one so completely, and it does not establish what song is for — which means the direction of spread cannot be interpreted in functional terms. Sampling effort differs substantially between populations, and where the whales actually hear each other is inferred rather than observed.

The controls — what makes this evidence rather than a story
  • Multiple populations sampled in the same seasons, so a song type can be dated independently in each.
  • Classification by hierarchical theme structure rather than by ear, so "the same song" is a measurable claim.
  • Repeated over several song types across a decade, so the pattern is not one event.

From Dynamic Horizontal Cultural Transmission of Humpback Whale Song at the Ocean Basin Scale

The journey

Humpback whales of the western South Pacific

From Antarctic and sub-Antarctic feeding grounds to Tropical and subtropical breeding waters

When
Towards the tropics in the southern autumn, back towards the ice in spring; individuals return to the same feeding areas across years
How far
Among the longest migrations of any mammal, stated as a range rather than a figure because populations differ and tracked samples are small

What it steers by, in the order they take over

  1. Route fidelity learned from experience
  2. Maternal transmission of feeding-ground destinations
  3. Environmental cues that remain poorly identified

How anybody knows

Photo-identification of individuals by tail markings across decades, satellite tagging of a small number of animals, and genetic work showing that feeding-ground use is inherited down the maternal line.

The biggest thing still unknown

What they actually navigate by. Route fidelity is well documented and the mechanism behind it is not; there is no good evidence identifying the cues a humpback uses to hold a course across open ocean.

The journey, and what it has to do with the song

A new way to catch fish, spreading through a social network

Song is not the only thing humpbacks pass to each other.

In the Gulf of Maine in the 1980s a humpback was seen doing something new: slapping the surface hard with its tail before diving to blow a bubble net around a shoal. Over the following decades the technique appeared in more and more animals — and because every whale on that feeding ground is individually identifiable, it was possible to ask a sharper question than whether it spread. It was possible to ask whether it spread by learning.

A new way of feeding appeared in one humpback and spread through the population along the lines of who spent time with whom.

Well supported

Good evidence backs this, though some details remain open.

Lobtail feeding — striking the water surface with the tail before a bubble-net dive — appeared in a Gulf of Maine humpback population and spread through it over decades in a pattern predicted by association networks rather than by independent discovery.

Who this applies to
One population on one feeding ground, followed for twenty-seven years.Do not extend this beyond the taxa listed — the popular version over-reaches.
Studied in
Megaptera novaeangliae
Why we rate it this way, and what the caveats are
Well supportedModerate confidence

The network prediction is a real test rather than a description, and it held across 653 identified individuals and 27 years. Confidence stops at moderate because association at the surface is an imperfect proxy for the chance to observe, and the spread coincided with a change in prey.

How far it can be extended

A single population and a single behaviour; nothing establishes that other humpback populations transmit techniques the same way.

Caveats

  • Being seen together at the surface is a proxy for the opportunity to watch, not a record of watching.
  • The spread coincided with a shift in prey species, which the analysis accounts for but cannot remove.

Still unanswered

  • What exactly is learned — the whole technique, or one component of it.
  • Whether the technique persists now that prey availability has changed again.

Last reviewed 2026-08-31

The evidence (1 study)

How we know

Twenty-seven years of photographs, and a new way to catch fish

When a new behaviour appears in a wild population, is it spreading by learning, or is everybody working it out separately?

Twenty-seven years of sightings of individually identified humpback whales on one feeding ground were assembled into a record of who was seen with whom, and of when each whale was first observed using a novel technique — striking the water surface with the tail before diving to feed. The order in which whales acquired it was tested against the social network: if learning is the route, a whale should acquire it sooner the more time it has spent with whales that already have it.

What happened

The technique spread in a pattern that followed the association network closely, appearing first in the whales that had spent time with whales already using it. Asocial learning alone did not account for the timing.

What it shows

Culture demonstrated statistically rather than argued for by elimination. Social transmission makes a specific prediction about the order in which individuals acquire a behaviour, and twenty-seven years of records were enough to test that prediction and have it hold.

What it does not show

Being seen at the surface together is a proxy for the chance to observe, and an imperfect one. The spread coincided with a shift in prey, which the analysis accounts for but cannot remove. What exactly is learned — the whole technique or one component of it — is unknown.

The controls — what makes this evidence rather than a story
  • Association data collected independently of the feeding behaviour, so the network is not defined by the thing being explained.
  • Models of asocial learning fitted alongside the social model, so the two explanations compete rather than one being assumed.
  • Changes in prey availability over the period included in the analysis rather than ignored.

From Network-Based Diffusion Analysis Reveals Cultural Transmission of Lobtail Feeding in Humpback Whales

The prediction social learning makes is specific: a whale should take up the technique sooner the more time it has spent with whales that already have it. Twenty-seven years of records of who was photographed alongside whom were enough to test that, and it held. This is culture demonstrated by a statistic rather than argued for by eliminating alternatives — a stronger kind of evidence, and available only where somebody kept the records.

What is still unsettled

  • What is humpback song actually for?

    Why it matters: Female choice and male–male competition are both proposed and neither is well supported. A display every male copies from every other male is an awkward fit for either, and until the function is known no change in the song can be interpreted.

  • Why does a novel song displace the resident one so completely?

    Why it matters: Whether the attraction is novelty itself or something about the whales who bring it decides whether this is a fashion, a signal of quality, or something with no function at all.

  • Where does the transmission physically happen?

    Why it matters: Populations meet on migration, on shared feeding grounds, and at range boundaries. Which of those is the transmission route is unknown, and it determines how fast a song could possibly spread.

  • Does changing song do anything for the whale that changes?

    Why it matters: If singers gain nothing measurable, this is cultural transmission with no fitness signal attached — which would make it a much stranger phenomenon than it is usually presented as.

Whales that pass things on

Follow it further

  • Animal culture

    Where this sits among the other evidence for transmission by learning

  • Sperm whale

    A different whale, a different signal, and clans instead of songs

  • Orca

    Dialects that stay stable rather than sweeping across an ocean

  • How scientists listen to animals

    Why the structure of this song was seen before it was heard

The research behind this page

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

2014Nature

Experimentally induced innovations lead to persistent culture via conformity in wild birds

The trained solution spread through each sub-population and remained the dominant method, persisting across years and through substantial population turnover.

2013Science

Network-Based Diffusion Analysis Reveals Cultural Transmission of Lobtail Feeding in Humpback Whales

A new technique — striking the surface with the tail before diving to feed — spread through the population in a pattern that followed the social network closely, appearing first in animals that associated with whales already using it.

2011Current Biology

Dynamic Horizontal Cultural Transmission of Humpback Whale Song at the Ocean Basin Scale

Song types repeatedly appeared first off eastern Australia and then swept eastward across the Pacific, population by population, taking a year or two to cross the ocean.

2000Nature

Cultural revolution in whale songs

Within two seasons the entire east Australian population had abandoned its own song and adopted the western one, essentially wholesale rather than by gradual blending.

1999Nature

Cultures in chimpanzees

Thirty-nine behaviours, many of them involving tools, varied between sites in ways that ecology did not explain.

1971Science

Songs of Humpback Whales

The sounds are organised hierarchically and repeat: units combine into phrases, phrases repeat to form themes, and themes occur in a consistent order to form a song that a whale may repeat for hours.

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 43% completeness against what we would call a finished subject, and was last reviewed on 2026-09-03. It carries 3 claims and answers 0 mapped search questions.

  • 1 high-priority search intent(s) not yet covered
  • no research from the last few years is attached — check for newer work
  • no popular claim about this subject has been checked yet
  • Feeding behaviour, including bubble-net feeding, is mentioned only through the cultural-transmission study and deserves its own treatment.
  • Migration — among the longest of any mammal — is not covered here.
  • The recovery of humpback populations after whaling is a major conservation story that this page does not tell.
  • Non-song social sounds, which are produced by both sexes, are almost absent.