We hypothesize that song type 1 (blue) spread from French Polynesia to Ecuador no earlier than 2016. The French Polynesian version of this song added three new themes between 2015 and 2016, and one of those themes (7a) was present in all song type 1 (blue) songs recorded in Ecuador in 2018 [58] (table 1; electronic supplementary material, table S1). Song may be transmitted between populations through three mechanisms [55]: song exchange on shared or partially shared feeding grounds or migratory routes, individuals using different breeding grounds in different years, or individuals moving between breeding grounds within a breeding season [55,57,58,60,64,72,83]. We hypothesize that the song was most likely transmitted between French Polynesia and Ecuador through song exchange on feeding grounds. Humpback whales from Ecuador (and its neighbouring population in breeding stock G, Colombia) feed near the Western Antarctic Peninsula (Antarctic Feeding Area I) based on genetic and photo-ID studies [84–86] (electronic supplementary material, figure S8). It is still uncertain, however, where the summer feeding grounds for French Polynesian whales are located, but there have been observations of some French Polynesian individuals and other South Pacific humpbacks near the Western Antarctic Peninsula [73,84,87]. Humpback whales sing mainly on breeding grounds and during migration, but multiple studies have recorded songs on foraging grounds around the globe [64,75,76,83]. This suggests that song learning is indeed possible during summer months when whales from different breeding populations may feed in similar areas (e.g. [64]). South Pacific breeding populations to the west of French Polynesia have been observed in Antarctic Feeding Areas V, VI and I [77,88], while overlap between foraging grounds has been found between Oceanian and Colombian breeding populations [84,89] (electronic supplementary material, figure S8). Genetic and photo-ID studies have observed individuals from French Polynesia and the Samoan islands near the West Antarctic Peninsula Area I [63,84,90], the typical feeding area for both the Colombian and Ecuadorian populations [84–86] (electronic supplementary material, figure S8). Clearly, there is potential for vocal exchange between French Polynesia and Ecuador due to overlap in the feeding grounds of at least a portion of the populations, and the occasional presence of song [83].
While song transmission on shared feeding grounds is the most likely scenario for song exchange in the current study, movement of whales from one breeding population to another between seasons has been observed between French Polynesia and other neighbouring populations in the central South Pacific [63] and, although rare, between French Polynesia and Colombia [91]. It cannot be ruled out that the song revolution from French Polynesia to Ecuador was the result of immigration of one or more whales into the Ecuadorian population. Finally, song may have spread east from French Polynesia through locations where whales have been sighted in small numbers (i.e. the Pitcairn Islands, Easter Island, Galapagos Islands [62]) in a stepping stone model. Future studies should incorporate opportunistic song recordings taken in remote central and eastern South Pacific locations to investigate this hypothesis. While song has continued to spread east, as predicted from previous empirical and modelling studies [41,70,92], we acknowledge that our sample size is small and spans only three years. However, these data produced a robust result in that song themes first present in French Polynesia appeared in Ecuador in a subsequent year replacing the existing song. Future studies should investigate whether this easterly transmission pattern holds through time given the apparent difference in population sizes between the smaller French Polynesian population and that of the larger stock G (Ecuador, Colombia, Costa Rica and Panama) [69], which may impact song dynamics [70]. Agent-based models indicate that songs are likely to spread from larger to smaller populations and as a result of rare interactions outside of the breeding grounds [70].
Finally, hybrid songs provide insights into the song learning process by which singers transition from singing an old song type to a new, novel song [60]. In the current study, two hybrid singers (singers 7 and 8) sung primarily song type 1 (blue) while mixing in themes from song types 2 (green) and 3 (orange) (table 1). Hybrid songs had only previously been documented a handful of times on the eastern Australian and French Polynesian breeding grounds and during the southbound migration past the Kermadec Islands in the western South Pacific [58,60]. However, while recording a hybrid song has historically been rare, season-long autonomous recorders in locations where song sharing, and learning, is occurring are more regularly capturing hybrid songs. Humpback whales are thought to learn songs through segmentation, and through the transition or substitution of themes influenced by sound sequence similarity [60]. Previously described hybrid songs contained themes of one song type that was commonly placed into the middle of the other song [60]. Here, singer 7 sang themes from song 1 (blue) and 3 (orange) early in the season, while singer 8 combined themes from song 1 (blue) and 2 (green) later in the season (table 1). This learning mechanism of picking up a novel song through segmentation has also been shown in songbirds near song dialect boundaries [93]. Hybrid songs occurred in village indigo bird (Vidua chalybeate) and orange-tufted sunbird (Nectarinia osea) males along dialect boundaries who displayed songs from either dialect [23,94]. While corn buntings also have distinct dialects and some males along dialect boundaries learn both, no song hybridization has been reported [12,24,25]. The dialects of corn buntings can be compared to humpback whale songs in a different aspect. Their dialects change from year to year with all males conforming to new versions much like how humpback whale songs progressively evolve [24,26]. In both species, changes to a current song through evolution are thought to occur through learning or production errors or innovation by individuals that are then adopted by the population [12,24,92]. By contrast, song revolutions in humpback whales are unlikely to be triggered by production errors [95] but rather represent the appearance of a new song type in a population [12]. A relevant comparison to song revolutions in songbirds may be that of white-throated sparrows (Zonotrichia albicolis) [96], where a novel song was documented spreading eastwards replacing the existing song across Canada [96]. This showed a replacement of song as documented for humpback whales in the South Pacific, but the time span over which song was replaced (two decades as opposed to merely two years in humpbacks) was different by an order of magnitude [41,96].
Here, we documented both song evolution and revolution, and the presence of hybrid singers which support previous studies on South Pacific humpback whale song dynamics [41,56,58,60,61,72]. The clear transmission of song themes from French Polynesia to Ecuador is the first evidence of song sharing between the two populations, although it is not known how this song will evolve in the next year, whether song revolutions in Ecuador occur repeatedly, and at what time intervals. The study of humpback whale song culture not only draws parallels to songbird song characteristics, but sheds light on the underlying mechanisms of social learning and cultural evolution in animals ranging from fish [97] to other cetacean species [10,98,99] through to humans [100]. Culture is an important aspect of animal communities that can increase the survival and reproductive success of individuals through socially learnt behaviours and traditions [6], and may increase their plasticity of responding to emerging threats.
5. Conclusion
Here, we documented the first evidence of eastward cultural transmission of songs from the central South Pacific breeding population of French Polynesia across to the eastern South Pacific breeding population of Ecuador. Three song themes were shared between the populations, with the song type being traced back as far as Tonga in the western South Pacific. This demonstrates that humpback whales are vocally connected across the ocean basin. Song transmission between French Polynesia and Ecuador is likely facilitated on overlapping feeding grounds around the West Antarctic Peninsula. This study extends our understanding of the extent of cultural transmission among humpback whale populations in the South Pacific and adds to unravelling the underlying mechanisms of song learning. Further studies are required to confirm whether song revolutions regularly spread from French Polynesia to Ecuador, and if so, at what interval. Further, to evaluate the extent of this cultural phenomenon, future studies should investigate whether songs continue to transmit eastwards from Ecuador (eastern South Pacific) to Brazil (western South Atlantic), and onwards around the Southern Hemisphere. Understanding this cultural phenomenon will provide valuable comparative perspectives to the evolution of complex communication including the evolution of human language and culture. As with humans, the patterns of migration are written into the songs of humpback whales.