Every few years the entire humpback population of the South Pacific abandons its song and adopts a brand-new one that has drifted in from thousands of miles away—the only known case of a culture-wide fashion sweeping a species that isn't human

On a humpback breeding ground, nearly every singing male performs the same long, structured song. The sounds change gradually as the season passes, but the singers continue matching one another closely enough that each population retains a recognisable shared version.

Then, in parts of the South Pacific, the old song can disappear. A substantially different arrangement arrives from a neighbouring population, spreads among the males and replaces the existing display. Researchers call the event a song revolution.

The phenomenon is not the entire humpback population abandoning a song at once. Adult males produce the display, and the new versions have moved between regional breeding populations in stages. Even with those qualifications, the scale is unusual: repeated cultural waves have been traced eastward from Australia to French Polynesia and as far as Ecuador.

A song is built from nested sequences

Humpback song is not a single call. Individual sounds, known as units, are arranged into phrases. Repeated phrases form themes, and a series of themes performed in a consistent order makes the song. A male may repeat the full sequence many times.

The display is socially learned. Within a population, most or all singing males usually perform the same song type at any one time, continually incorporating small changes while maintaining close agreement with the other singers.

This ordinary process is described as cultural evolution. A phrase may alter, a unit may be substituted, or a theme may shift over time. The song remains recognisably descended from the version heard before it.

A revolution is different.

The first documented revolution replaced an Australian song

Michael Noad and colleagues documented the first clear case in Nature in 2000. Between 1995 and 1998, humpbacks migrating along Australia’s east coast rapidly adopted a song associated with the west Australian population, whose migration lies in the Indian Ocean.

The foreign material first appeared alongside the established east-coast song. It then spread until the previous song had been replaced. The authors described the change as rapid and complete, apparently triggered by a small number of singers carrying the western song into the eastern population.

That observation separated a song revolution from normal annual change. The new display was not a modified local arrangement. It belonged to a different lineage and displaced what had been there before.

The mechanism was not directly observed. Movement between populations, shared Antarctic feeding areas and acoustic contact during migration could all allow songs to cross what appear to be separate breeding communities.

Eleven song types moved east in cultural waves

A larger pattern emerged when Ellen Garland and colleagues compared recordings gathered across the South Pacific. Their 2011 paper in Current Biology followed song change over 11 years in eastern Australia, New Caledonia, Tonga, American Samoa, the Cook Islands and French Polynesia.

The team identified 11 song types. New versions typically appeared first in eastern Australia, then reached other breeding populations farther east in later seasons. One song could move from eastern Australia to French Polynesia in roughly two years, passing through populations separated by thousands of kilometres.

The movement was not perfectly mechanical. Some songs failed to reach every location, and the recordings available for each population varied by year. Yet the repeated direction was clear enough for the researchers to describe the songs as cultural ripples moving east.

Each receiving population was not copying a broadcast from Australia across the entire ocean. Sound does not remain usable over that distance, and the whales occupy different regions at different times. The song appears to pass through contact between whales, one population and season at a time.

Hybrid songs show how a revolution may be learned

Complete replacement creates an obvious learning problem. A humpback song can contain a large, ordered repertoire, so how does a male exchange one sequence for another without losing its structure?

Garland’s team examined rare hybrid performances in a 2017 Proceedings of the National Academy of Sciences paper. Across more than 9,300 phrases from two song revolutions, they found singers combining themes from the outgoing and incoming song types.

The hybrids were not random mixtures. Whales appeared to insert or substitute segments in ways that preserved the transition rules of the display. In some cases, acoustically similar themes occupied corresponding positions in the old and new sequences.

These recordings provide brief intermediate stages between two otherwise distinct cultural products. They support the idea that males can learn a new song in sections before conforming to it fully. They do not reveal whether novelty is attractive to females, advantageous in competition between males, or simply copied because other singers are adopting it.

Migration may create the contact network

The breeding populations are geographically separated, but their annual movements create opportunities to hear unfamiliar singers. Humpbacks can meet on shared feeding grounds, along migration routes or when individuals visit a breeding ground other than the one they usually use.

A 2019 study in Royal Society Open Science compared songs recorded at the Kermadec Islands with material from wintering grounds between New Caledonia and the Cook Islands. Similar themes and hybrid songs suggested that the Kermadecs, a migratory stopover northeast of New Zealand, may help connect populations that are acoustically separated on their breeding grounds.

That conclusion is based on matching song structure and known migration patterns, not on tracking the moment one identified whale taught another. Humpbacks are wide-ranging animals, and the relevant encounters are difficult to observe directly.

Models offer an additional possibility. A 2021 global cultural-evolution model found that differences in population size could help produce directional waves, with variants more likely to move from larger populations into smaller ones. The simulations also produced periods without revolutions, so the pattern is not a clock that must reset every fixed number of years.

The song network reaches across an ocean basin

In 2022, Josephine Schulze and colleagues extended the record into the eastern South Pacific. Their Royal Society Open Science study compared songs recorded in French Polynesia from 2016 to 2018 with songs recorded off Ecuador during the same period.

Themes present in French Polynesia appeared in Ecuador in 2018, consistent with a delayed continuation of the eastward wave. The link added roughly 8,000 kilometres beyond French Polynesia and connected song populations separated by about 14,000 kilometres from eastern Australia to Ecuador.

The authors described a vocal culture rivalled in geographic extent only by our own. That is narrower than saying humpbacks are the only non-human animals with culture. Socially learned traditions occur in birds, primates, cetaceans and other animals, while sperm whales and killer whales maintain culturally transmitted vocal systems of their own.

What may be unique is the combination seen in South Pacific humpbacks: a complex behavioural display, replaced rapidly across nearly all performers in one population, then transmitted repeatedly through several populations at ocean-basin scale. The 2017 PNAS paper called it a very rare, possibly unique case of rapid population-wide horizontal cultural transmission.

The fashion comparison captures the appetite for a new shared form, but it does not explain why the change happens. For now, the recordings establish the spread more securely than they establish the motive.