ArticlePDF Available

Migratory convergence facilitates cultural transmission of humpback whale song


Abstract and Figures

Cultural transmission of behaviour is important in a wide variety of vertebrate taxa from birds to humans. Vocal traditions and vocal learning provide a strong foundation for studying culture and its transmission in both humans and cetaceans. Male humpback whales (Megaptera novaeangliae) perform complex, culturally transmitted song displays that can change both evolutionarily (through accumulations of small changes) or revolutionarily (where a population rapidly adopts a novel song). The degree of coordination and conformity underlying song revolutions makes their study of particular interest. Acoustic contact on migratory routes may provide a mechanism for cultural revolutions of song, yet these areas of contact remain uncertain. Here, we compared songs recorded from the Kermadec Islands, a recently discovered migratory stopover, to multiple South Pacific wintering grounds. Similarities in song themes from the Kermadec Islands and multiple wintering locations (from New Caledonia across to the Cook Islands) suggest a location allowing cultural transmission of song eastward across the South Pacific, active song learning (hybrid songs) and the potential for cultural convergence after acoustic isolation at the wintering grounds. As with the correlations in humans between genes, communication and migration, the migration patterns of humpback whales are written into their songs.
This content is subject to copyright.
Cite this article: Owen C et al. 2019 Migratory
convergence facilitates cultural transmission of
humpback whale song. R. Soc. open sci. 6:
Received: 22 February 2019
Accepted: 30 July 2019
Subject Category:
Biology (whole organism)
Subject Areas:
animal culture, cultural evolution, song, cetacean,
humpback whale, south pacific
Author for correspondence:
Ellen C. Garland
Electronic supplementary material is available
online at
Migratory convergence
facilitates cultural
transmission of humpback
whale song
Clare Owen1, Luke Rendell1,2, Rochelle Constantine3,4,
Michael J. Noad3,5, Jenny Allen5, Olive Andrews3,6,7,
Claire Garrigue3,8,9, M. Michael Poole3,10,
David Donnelly3,11, Nan Hauser3,12 and
Ellen C. Garland1,2,3
Sea Mammal Research Unit, School of Biology, University of St Andrews, St Andrews KY16
Centre for Social Learning and Cognitive Evolution, School of Biology, University of
St Andrews, St Andrews KY16 9TH, UK
South Pacific Whale Research Consortium, PO Box 3069, Avarua, Rarotonga, Cook Islands
School of Biological Sciences, Institute of Marine Science, University of Auckland, Private Bag
92019, Auckland 1142, New Zealand
Cetacean Ecology and Acoustics Laboratory, School of Veterinary Science, The University
of Queensland, Gatton, Queensland 4343, Australia
Conservation International, Pacific Islands Programme, Science Building 302, University
of Auckland, 23 Symonds Street, Auckland 1010, New Zealand
Niue Whale Research Project, Alofi, Niue
Opération Cétacés, Noumea 98802, New Caledonia
UMR ENTROPIE (IRD, Université de La Réunion, CNRS, Laboratoire dexcellence-CORAIL),
BPA5, 98848 Noumea Cedex, New Caledonia
Marine Mammal Research Program, BP 698, Maharepa, 98728 Moorea, French Polynesia
Killer Whales Australia, 8 Campbell Parade, Box Hill South, Victoria 3128, Australia
Cook Islands Whale Research, PO Box 3069, Avarua, Rarotonga, Cook Islands
LR, 0000-0002-1121-9142; JA, 0000-0003-4658-7380;
ECG, 0000-0002-8240-1267
Cultural transmission of behaviour is important in a wide
variety of vertebrate taxa from birds to humans. Vocal
traditions and vocal learning provide a strong foundation for
studying culture and its transmission in both humans and
cetaceans. Male humpback whales (Megaptera novaeangliae)
perform complex, culturally transmitted song displays that
can change both evolutionarily (through accumulations of
small changes) or revolutionarily (where a population rapidly
adopts a novel song). The degree of coordination and
conformity underlying song revolutions makes their study of
particular interest. Acoustic contact on migratory routes may
© 2019 The Authors. Published by the Royal Society under the terms of the Creative
Commons Attribution License, which permits
unrestricted use, provided the original author and source are credited.
provide a mechanism for cultural revolutions of song, yet these areas of contact remain uncertain.
Here, we compared songs recorded from the Kermadec Islands, a recently discovered migratory
stopover, to multiple South Pacific wintering grounds. Similarities in song themes from the
Kermadec Islands and multiple wintering locations (from New Caledonia across to the Cook
Islands) suggest a location allowing cultural transmission of song eastward across the South Pacific,
active song learning (hybrid songs) and the potential for cultural convergence after acoustic
isolation at the wintering grounds. As with the correlations in humans between genes,
communication and migration, the migration patterns of humpback whales are written into
their songs.
1. Introduction
Cultural traditions play a significant role in shaping human societies [1]. Culture, broadly defined as
shared behaviour or information within a community acquired through some form of social learning
from conspecifics [2,3], can also be an important process in non-human communities, as demonstrated
in multiple studies on the cultural learning of behavioural traits in primates [4]. Recent studies have
also highlighted the importance of cultural traits and social learning in cetaceans [3,5,6]. For example,
bottlenose dolphins (Tursiops sp.) demonstrate the cultural transmission of tool use [7], while
humpback whales (Megaptera novaeangliae) have multiple, independently evolving cultural traditions
including migratory destinations, feeding techniques and songs [811]. Vocal traditions and vocal
learning provide a strong foundation for studying culture and its transmission in both humans and
cetaceans [5]. For example, vocalizations that are shared within a group and maintained through
cultural transmission over decades are used to identify social structures in killer whales (Orcinus orca)
[12] and sperm whales (Physeter macrocephalus) [13]. The stability of these vocal cultures and other
cultural traditions (e.g. prey specialization) can in turn affect genetic evolution through gene-culture
coevolutionary processes [3].
A striking example of large-scale cultural transmission in a non-human animal is the transmission of
humpback whale song between populations [9]. Male humpback whales produce long, stereotyped
vocalizations [14] that function in sexual selection for mate attraction and/or to facilitate malemale
interactions [15]. Humpback whale song is hierarchically structured: sound units are grouped into
phrases that are embedded in higher-level themes [14]. Although songs are constantly evolving, most
males within a population will converge on a single song type during any particular winter breeding
season [16,17]. Songs can also be transmitted between populations. Garland et al. [9] identified
dynamic transmission of humpback whale song that extended across the South Pacific, spanning
6000 km from eastern Australia in the west to French Polynesia in the east. It is a clear example of
large-scale horizontal cultural transmission, where a population rapidly adopts a novel song
introduced from a neighbouring population [11], and then the next adjacent population adopts the
novel song, and so on in a population-level transmission chain [9].
The western and central South Pacific region can be divided into three sections: eastern Australia,
western South Pacific (New Caledonia, Tonga and American Samoa) and the central South Pacific
(Cook Islands and French Polynesia) based on previous song studies [18]. Song types take
approximately 2 years to transit across the region and as a result, populations in each section will
typically converge on different song types at any one time; these songs may be related to each other
with some shared material or be quite distinct [9,11,19,20]. However, given humpback whalesfidelity
to natal wintering grounds [10], we still have a limited understanding of the underlying mechanism(s)
driving this cultural phenomenon.
Humpback whale song is most frequently produced and recorded on the winter breeding grounds
[15] and while the whales are migrating to and (particularly in the South Pacific) from their wintering
grounds [11]. Aggregations (e.g. on feeding or wintering grounds) and shared migratory routes may
provide an opportunity for acoustic contact, which is necessary for song transmission. Four possible
contact mechanisms have been suggested: singing on shared feeding grounds, singing on shared or
partially shared migratory routes, between-season movement of individuals between populations and
within-season movement of individuals between populations [16]. All these mechanisms could occur
in the South Pacific [19], but capturing such events and/or identifying important geographical
locations in Oceania (western and central South Pacific) remains challenging given the open ocean
migratory range of humpback whales. R. Soc. open sci. 6: 190337
A recent study at the Kermadec Islands, a remote group of islands located in the western South Pacific
(figure 1), found a large number of whales were present during the Austral spring [21]. Genetic and
photo-identification matches reveal the whales originated from multiple South Pacific wintering
grounds [21]. With very few island groups south of the tropical wintering grounds, this unusual
migratory stopover, where multiple migratory corridors may overlap when the whales are almost
2000 km into their southward migration, may provide a location for acoustic contact among multiple
populations. Whales stay on average 4.6 days at Raoul Island, Kermadecs (R. C. unpublished data,
2015), which potentially allows for the exposure to song for several days. Such contact could facilitate
the cultural convergence of song in the western South Pacific populations, and the easterly
transmission of song into the central South Pacific.
theme 7
theme 7
theme 2
theme 15
frequency (kHz)
10 20 30 40
time (s)
50 60 70 80 90
theme 16 theme 17
theme 6 theme 1 theme 3 theme 5 theme 4
theme 10
theme 9 theme 12 theme 13 theme 14 theme 11
theme 11 theme 9 theme 8
song 1a
song 1b
song 2
song 3
Figure 1. Map of the South Pacific with the wintering grounds, with the Kermadec Islands migratory stopover and the Antarctic
summer feeding areas noted. The spectrograms of song types 1a and 1b (themes 714), song type 2 (themes 16) and song type 3
(themes 1517) are colour coded to show where each song type was present. The distance between eastern Australia and French
Polynesia is approximately 6000 km. xdenotes the location of the eastern Australian recordings. Map modified from Garland et al.
[19]. A representative phrase for each theme is shown in the typical order they were sang. Note that some themes contained variant
phrases (termed phrase types); a single example from the theme is shown (see electronic supplementary material, table S3 for
detailed description of all phrase types). Spectrograms were 2048-point [1024-point song type 3] fast Fourier transform (FFT), Hann
window, 75% overlap, displaying 5 kHz and 90 s, generated in Adobe Audition. Corresponding audio files are provided for each song
type (electronic supplementary material, audio S1S4). R. Soc. open sci. 6: 190337
Here, we hypothesized that if males do migrate past the Kermadec Islands from multiple wintering
grounds during their southward migration (September and October 2015), we should see some evidence
of the cultural processes, song transmission and/or convergence. This would provide evidence for a
cultural mechanism allowing easterly transmission of song from the western to the central South
Pacific region. We compared recordings from the Kermadecs with song recordings collected during
the same winter breeding months (JulyOctober 2015) from six wintering grounds spread across the
South Pacific (eastern Australia, New Caledonia, Tonga, Niue, the Cook Islands and French Polynesia)
and assigned a possible origin population for each Kermadec song recording (or parts thereof ).
2. Material and methods
2.1. Song recordings
Humpback whale song recordings were collected near Raoul Island, Kermadecs (figure 1) in September
and October 2015 during a dedicated humpback whale research voyage. A hydrophone (HTI 96MIN,
frequency response 2 Hz to 30 kHz) suspended to approximately 10 m deep in water depths of
approximately 60 m from a 5 m RIB was used to collect acoustic recordings when humpback whales
were observed. The recordings were made on a Zoom or a Microtrack II digital recorder (WAV
format, 16 bit, sampling rate 44.1 kHz). A total of 12 recordings were collected over 13 days, all
within 5 km of Raoul Island. To maximize data quality and reduce uncertainty, we excluded all low-
quality sections of recordings where songs were unclear (electronic supplementary material, tables S1
and S2). This resulted in a disparity between the number of recordings and the total number of
individualsingers, as we were conservative in linking sections of song together within a recording
(see below).
Songs were recorded at each wintering ground (eastern Australia: Peregian Beach, Queensland; New
Caledonia: Southern Lagoon; Tonga: Tongatapu; Niue: Main Island; Cook Islands: Rarotonga; French
Polynesia: Moorea) across the western and central South Pacific in the 2015 austral winter or spring
(electronic supplementary material, table S1 and figure 1). Humpback whale song recordings from
eastern Australia were collected using a moored, T-shaped, five buoy hydrophone array in 1828 m
water depth (16 bit, WAV files, 22 kHz sampling rate) as part of concurrent field research (see Noad
et al. [22] and Dunlop et al. [23] for further details). Such an arrangement allowed accurate tracking of
singing whales within a 10 km radius [22]. Some recordings in New Caledonia were collected using a
single, moored passive acoustic recorder (SM2M + Wildlife Acoustics) in approximately 60 m water
depth (22 kHz sampling rate) as part of other research. One recording from Niue was taken from a
video collected using an underwater camera (Canon EOS 10D Digital SLR, audio extracted at a sampling
rate of 44.1 kHz, 16 bit, WAV file format). All other recordings were made using a hand-held
hydrophone (HighTech HTI 96MIN) suspended to approximately 10 m deep in water depths of 30 m to
greater than 1 km connected to a digital recorder (M-Audio Microtrack 24/96 or Microtrack II, or a
Zoom H2 or H4, recording at 44.1 kHz, 16 bit, WAV file format) from a designated research vessel or
a platform of opportunity. While we endeavoured to analyse all available recordings to capture as much
singer variability as possible, the available data represent a snapshot in time of a small (but broadly
representative, due to strong song matching [9]) sample of singers from each location.
2.2. Song transcription
Song is hierarchical: a sequence of sounds termed unitscomprises a phrase, phrases are repeated to
form a themeand a few themes are sung in a set order to form a song[14]. At the upper level,
different versions of the song that contain different themes are called song types[9]. All humpback
whale song recordings were analysed as spectrograms created in Adobe Audition (Fast Fourier
Transform [FFT] 2048 [eastern Australia: 1024], Hann window, 75% overlap, displaying 30 s and
05 kHz). The song was transcribed by a human classifier (C.O.) at the unit level (following Garland
et al. [9,20,24]). Individual units were named and identified by their visual and aural characteristics.
Every unit was transcribed and presented as a string of units for each phrase. Each stereotyped string
of units that were identified as a phrase was allocated a letter (e.g. A, B, C). Phrases containing
similar sounds in the same order were grouped into the same theme number (e.g. 1, 2, 3). The
qualitative assignment of phrase types (and higher-level theme groupings) was validated through
clustering of unit sequences (see below). When multiple singers were present in a recording or a R. Soc. open sci. 6: 190337
singer was silent for more than 3 min before the song continued, it was not always possible to confirm the
same singer was resuming. To avoid ambiguity, the subsequent phrases were labelled with a letter (e.g.
KI01S1a) and the strings of phrases were analysed separately. This resulted in 39 individualsingers in
the Kermadec dataset from the 12 initial recordings (electronic supplementary material, table S1).
2.3. Quantifying units
To reduce subjectivity and aid in a robust, repeatable unit and theme classification, the highest quality
recording at each recording location was selected to measure units for subsequent inclusion in a
quantitative analysis, aiming to maximize consistent naming/classification of units between locations
and song types. All units in the first phrase of each theme were measured for 11 frequency and
duration measurements following previously published studies on humpback whale unit classification
[2325] using Raven Pro 1.4 (Cornell Lab of Ornithology; smoothed spectrogram, 11 s long, 21.5 Hz
resolution, FFT 2048 [eastern Australia: 1024], Hann window, 75% overlap). If a unit type was not
present in these selected phrases (i.e. rare units), it was located and measured to ensure all unit types
identified at each location had a corresponding set of measurements. A selection box was used to
isolate the sound and Raven automatically generated the following measurements: duration,
minimum frequency, maximum frequency, bandwidth and peak frequency. In addition, the start and
end frequencies, frequency trend ratio (ratio of the start and end frequencies), frequency range ratio
(ratio of the minimum and maximum frequencies), number of inflections and the pulse repetition rate
were taken manually (following Dunlop et al. [23]). A Classification and Regression Tree (CART)
analysis was run to ensure consistent naming of unit types and resulted in a root node error of
93.42% (n= 1443) agreement in classification. These measurements, which represent the acoustic
features of each unit type, were used to create the weighting system employed in the quantitative
analysis (detailed below).
2.4. Calculating song and theme similarity using the Levenshtein distance metric
Given the stereotypy in the sequence of units of each phrase type (and therefore theme) and also the
sequence of themes that comprised a song type, differences among sequences were striking and could
be quantified using common sequence analysis metrics. The Levenshtein Distance (LD) [26] is a robust
and powerful edit distance that provides an understanding of song similarity at all levels within the
song hierarchy [24,2728]. The LD calculates the minimum number of alterations required to convert
string ainto string b[26]. An LD score is calculated by counting the number of insertions, deletions
and substitutions and can be normalized to account for the length of the string (the normalized
version is referred to as the Levenshtein Distance Similarity Index or LSI). The LSI produces a
measure of similarity (between 0 and 1) among multiple sequences of varying lengths and provides
an overall understanding of the similarity of all sequences [29]. The LD has been shown to outperform
other methods of comparing acoustic sequences by consistently reconstructing the already-known
contextual information, for example, clustering humpback whale songs by population [30].
All LSI analyses were run in R [31] using custom-written code (available at
ellengarland/leven). Detailed methods for calculating the LD, LSI and applying a weighting system
are provided in Garland et al. [29]. The LSI was initially run here with all unit sequence data (i.e. the
sequence of units making up each phrase, hereafter called a phrase string) from every singer and
clustered to validate the qualitative assignment of each phrase string to a phrase type (and higher-
level theme grouping). Once the theme categories were verified, we conducted two LSI analyses to
quantify song similarity at two levels within the song hierarchy. First, the similarity in the sequence of
units making up each phrase type (theme) was quantified to allow fine-scale matching among
individuals and populations. A representative (median) string was calculated for each individual
singer to ensure the typical string of units sang for each phrase type was included. The LSI was run
following Garland et al. [22] as a weighted analysis where unit substitution costs were calculated
based on the acoustic feature similarity between pairs of unit types (quantified as part of CART,
described above). Substitution costs were between 0 and 1 depending on the similarity in acoustic
feature space between the units, while all other operations (i.e. insertions and deletions) had a cost of
1 (see Garland et al. [29] for a detailed account of the method). We used a weighted LSI analysis to
allow fine-scale idiosyncrasies to be quantified, such as replacing specific units with a different but
similar sound unit within a phrase. These were characteristic to some populations and may be
signatures of cultural mechanisms (e.g. convergence on a single song norm, song learning). Second, R. Soc. open sci. 6: 190337
the similarity in the sequence of themes making up a song for each individual was quantified to compare
broad-scale differences among populations. A representative (median) string of the themes that
comprised a song was again calculated per individual singer, and the LSI was run un-weighted.
For both analyses, similarity (LSI) scores were average-linkage (UPGMA) hierarchically clustered and
bootstrapped (1000 times) in R using the hclust,pvclust and pvrect packages to ensure the resulting
structure was stable and likely to occur [24,29,32]. The approximately unbiased (AU) and bootstrap
probability (BP) values, as well as the standard error for each division in the dendrogram were
retained. An AU p-value of greater than 95% and a BP p-value of greater than 70% were considered
to be significant, stable and strongly supported by the data [24,32], whereas lower values suggest
variability in their division. The clusters deemed stable and supported by the bootstrap analysis (AU
p-value > 95%) [24] were additionally marked at the highest level with a box using the pvrect function
[32]. The cophenetic correlation coefficient (CCC) was also calculated as a further, independent test of
how well each dendrogram represented the data; scores over 0.8 were considered a good
representation of the associations within the data [33].
2.5. Alternative test of song similarity and assignment to wintering ground origin
As an alternative and independent test to the LD, we investigated song similarity using the presence of
phrase types in each wintering ground compared with each individual singer recorded at the Kermadecs.
The presence of all phrase types for each singer recorded at the Kermadecs (KI) and for each wintering
ground was noted. The percentage of phrase types from a Kermadec singer that matched the phrase
types sung at each wintering ground was calculated by:
% matched ¼number of shared phrase types with wintering ground
Total number of phrase types present in the KI individual0s recording 100:
For example, out of the five phrase types sung by Kermadec singer KI01S1, five phrase types matched
New Caledonia (therefore 100%), five matched with Tonga (100%), five with Niue (100%) and four with
the Cook Islands (80%) (table 1).
The percentage of matched phrase types to each wintering ground was calculated for each Kermadec
singer. The wintering ground with the highest percentage was presumed to be the likely wintering
ground of origin for the song (table 1). In cases where more than one wintering ground was equally
likely (as in the example above), the origin was not specified. The likely wintering ground of origin
was also estimated using the LSI at the phrase type/theme level. When the sequence of unitsthat
comprised a phrase type from a Kermadec singer was grouped in a stable cluster (AU greater than
95%) with a phrase string from only one wintering ground, that particular wintering ground was
suggested as the origin for that song. If a phrase string from a song recorded at the Kermadecs was
grouped with multiple wintering grounds or only one phrase type was present, no wintering ground
of origin was suggested. This provided a conservative assignment of Kermadec singers to an origin
3. Results
Three song types were identified in the song recordings from 52 singers collected at 6 wintering grounds
across the South Pacific [eastern Australia (n= 11), New Caledonia (n= 11), Tonga (n= 8), Niue (n= 7),
Cook Islands (n= 8) and French Polynesia (n= 7)]. Song type 1 was the dominant song in the central
Pacific (the Cook Islands and French Polynesia), song type 2 was the most prevalent in the west (New
Caledonia, Tonga and Niue) and song type 3 was only recorded in eastern Australia. These songs
were compared to 39 singersrecorded at the Kermadecs (electronic supplementary material, table S1).
3.1. Song from the wintering grounds and Kermadecs
The identification of three song types (labelled 13) across the South Pacific wintering grounds and the
Kermadec migratory stopover in 2015 was supported by both qualitative matching of themes and song
types (figure 1) and quantitative LSI classification by hierarchical clustering and bootstrapping of the
most representative song (median string) from each individual singer (figure 2). There were two
distinct versions of song type 1, 1aand 1b, based on phrase type (theme) presence (1a themes:
7,8,9,10,11 and 1b themes: 7,9,11,12,13,14). Hierarchical clustering and bootstrapping supported this R. Soc. open sci. 6: 190337
Table 1. Likely wintering ground origin for each singer recorded at the Kermadec Islands. The likely origin was determined using a combination of the proportion of phrases present which matched each wintering
location (NC, New Caledonia; TO, Tonga; NI, Niue; CI, Cook Islands, FP, French Polynesia; EA, east Australia) and the LSI similarity analyses. The likely origin for the song was determined when one wintering ground
had a higher percentage of phrases matched than the other wintering grounds. The LSI similarity origin was determined using both the ne-scale phrase level and broad-scale song-level cluster analyses: when the
median phrase string from a Kermadec Islands singer clustered (AU p-value >95%) with a string from a single wintering ground (no multi-origin clusters were included in the table). The likely origin was determined
when a singer from the Kermadec Islands was linked to one wintering ground on ne or broad-scale analysis. Kermadec singers with a single phrase type (n= 8) were excluded from the analysis and were not
assigned to a breeding ground.
Kermadec singer # total # phrases
% matched phrases LSI similarity assignment likely origin
song 1 song 2 song 3 theme
NC TO CI FP NC TO NI CI EA 1 2 3 5 6 14 phrase song
KI01S1 5 0 0 0 0 100 100 100 80 0 TO TO
KI01S1a 8 0 0 0 0 100 87.5 87.5 62.5 0 TO NC TO
KI01S1b 5 0 0 0 0 100 100 100 80 0 TO TO TO
KI02S1a 5 0 0 0 0 100 80 80 60 0 NC
KI02S2 2 0 0 0 0 100 100 50 50 0 TO TO
KI02S3 5 0 0 0 0 100 100 100 60 0 NC NC NC
KI03S3 5 0 0 0 0 100 100 100 60 0 TO TO TO TO
7 0 0 0 0 85.7 85.7 85.7 42.9 0 NI
KI03S5 2 0 0 0 0 100 100 50 50 0
3 0 0 0 0 66.7 66.7 66.7 66.7 0
KI04S1b 4 0 0 0 0 75 100 100 75 0
KI04S2 4 0 0 0 0 100 100 100 100 0 TO TO
KI04S3 5 0 0 0 0 80 80 100 40 0 NI
KI05S1 6 0 0 0 0 83.3 83.3 100 50 0 TO NI TO NC
11 0 0 0 0 90.9 81.8 90.9 54.6 0 NC TO
KI05S3 5 0 0 0 0 100 100 100 80 0 TO TO
(Continued.) R. Soc. open sci. 6: 190337
Table 1. (Continued.)
Kermadec singer # total # phrases
% matched phrases LSI similarity assignment likely origin
song 1 song 2 song 3 theme
NC TO CI FP NC TO NI CI EA 1 2 3 5 6 14 phrase song
KI06S1 10 0 0 0 0 100 80 90 50 0 TO NC NC NC
KI10S1 5 0 0 0 0 100 80 100 40 0
KI10S2b 3 0 0 0 0 100 100 100 100 0
KI11S1 2 0 0 0 0 100 100 100 50 0 TO TO
KI12S1 3 0 0 0 0 100 100 100 66.7 0 TO TO TO
KI12S2 5 0 0 0 0 100 80 80 80 0 NI NC
KI12S3 7 0 0 0 0 100 85.7 85.7 71.4 0 TO NC NC
KI12S4 10 0 0 0 0 100 90 90 60 0 NC NI CI NC
KI12S5 10 0 0 0 0 100 80 90 60 0 TO NC TO
KI12S6 6 0 0 0 0 100 66.7 66.7 50 0 NC NC NC NC
KI12S7 5 0 0 0 0 100 80 100 40 0
KI13S1 11 0 0 0 0 100 81.8 90.9 63.6 0 CI NC CI
KI14S1 10 0 0 0 0 100 80 90 60 0 NI NC NC
KI14S1a 7 0 0 0 0 83.3 100 100 83.3 0 TO NI
KI14S2 3 66.7 0 100 33.3 0 0 0 0 0 CI CI CI
Indicates a singer found to be performing a phrase exclusively found in New Caledonia and phrases which were not present in the New Caledonian recordings. See §3.2 Song transcription, for designation of
individualKermadec singers and electronic supplementary material, table S2 for further information on singers. R. Soc. open sci. 6: 190337
division of song type 1 into aand bversions (figure 2; AU p-value = 100%, BP p-value = 98%, SE < 0.01).
One (out of 8) singer from Tonga, one (out of 8) singer from the Cook Islands and all singers recorded in
French Polynesia (n= 7) sang song type 1a. Song type 1a was not present in recordings from the other
three wintering grounds or at the Kermadecs. Song type 1b was recorded only at the start of the
season in New Caledonia (3 out of 11 singers; electronic supplementary material, figure S2), while it
was present in later months in the Cook Islands (3 out of 8 singers). Two (out of 39) singers in the
Kermadecs were recorded singing song type 1b; however, one singer was only recorded singing
theme 14 and therefore was not included in song-level analyses. All singers from Niue (n= 7) sang
song type 2, while 8 (out of 11) singers from New Caledonia, 7 (out of 8) singers from Tonga and 4
(out of 8) singers from the Cook Islands were also recorded singing song type 2. Qualitative and
quantitative analysis showed each population sang a particular, population-specific fine-scale
combination of the phrase types, allowing each to be acoustically identified based on this
combination. Song type 2 was also the most commonly recorded song at the Kermadecs (36 out of 39
singers). One singer from the Kermadecs sang themes from song type 2 along with a phrase variant
of one theme from song type 1a (i.e. hybrid singer, defined as a singer who sings a song containing
themes from two song types [6,11]). Finally, song type 3 was only recorded in eastern Australia
(figure 2; electronic supplementary material, table S1).
3.2. Linking Kermadec singers to a possible origin
Certain phrase types were present at one wintering ground (population) and not recorded at others
(electronic supplementary material, table S1 and results). The Kermadec recordings contained 22
(of 38) phrase types; these included all of the phrase types present in song type 2 and three from song
type 1b, while a single phrase variant from song type 1a (8C, used only by the hybrid singer) was
only recorded in the Kermadecs. The presence/absence of phrase types in each wintering ground as a
2 son
3 son
1b son
Figure 2. Dendrogram of the similarity between the representative song (theme sequence) for each individual singer recorded at the
Kermadec Islands (KI) and at each of the six western and central South Pacific wintering grounds (EA, eastern Australia; NC, New
Caledonia; TO, Tonga; NI, Niue; CI, Cook Islands; FP, French Polynesia). The median string LSI scores were hierarchically clustered
using average-linkage clustering and bootstrapped (n= 1000). The AU values (significant p-values >95%, red dot [24,32]) indicated
the structure and divisions in the tree were stable and likely to occur. This was additionally confirmed using the Cophenetic
Correlation Coefficient, which indicated that the structure of the tree was a very good representation of the associations present
within the data (CCC = 0.97). Black dashed boxes delineate each song type. Red dashed boxes highlight where a singer from
the Kermadecs has been linked to a wintering ground within a stable cluster. Singer name is created from the wintering
ground, recording number and singer number. For example, singer code NC05S1 is New Caledonian recording number five,
singer number one. * indicates an incomplete song sequence from the Kermadecs, although only whales recorded singing
more than one theme were included. R. Soc. open sci. 6: 190337
whole was compared with each individual singer recorded at the Kermadecs to calculate the percentage
of matched phrase types. Three singers at the Kermadecs sang a phrase type recorded only in New
Caledonia (phrase 3B) and one or more phrase types that had not been recorded in New Caledonia.
For consistency, these three singers and any other singers from the Kermadecs who were only
recorded singing one theme type were not included in this part of the analysis. The percentage of
matched phrases suggested a wintering ground of origin for 13 out of 26 singers at the Kermadecs:
10 from New Caledonia, 2 from Niue and 1 from the Cook Islands (table 1). The remaining 13 singers
matched multiple wintering grounds and were thus left as unassigned.
TO03S1_3A NC04S1_3A
KI12S5_3B KI12S4_3B
Figure 3. Dendrograms representing the similarity of the median sequence of units in (a) theme 1 and (b) theme 3 from song
type 2 for each individual singer recorded at the Kermadec Islands (KI) and the four wintering grounds in which these themes were
present (NC, New Caledonia; TO, Tonga; NI, Niue; CI, Cook Islands). Each theme was split into stereotyped phrase types (theme 1: 1A
and 1B; theme 3: 3A, 3B, 3C and 3D). The median string LSI scores were hierarchically clustered using average-linkage clustering and
bootstrapped (n= 1000). The AU values (significant p-values >95%, red dot [24,32]) indicated the stability of each split in the tree.
This was additionally confirmed using the Cophenetic Correlation Coefficient, which indicated that the structure of both trees was a
very good representation of the associations present within the data [(a) theme 1 CCC = 0.96, (b) theme 3 CCC = 0.94]. Red dashed
boxes highlight where a singer from the Kermadecs has been linked to a wintering ground within a stable cluster. Singer name is
created from the wintering ground, recording number, singer number and theme name. For example, singer code KI12S2_1A is
Kermadecs recording number 12, singer number two, theme 1A. R. Soc. open sci. 6: 190337
Using the most representative song (i.e. composed of a sequence of themes) from each individual
singer, the LSI song scores were clustered. If a stable cluster (AU p-value > 95%; represented by a red
dot, figure 2) contained a Kermadecs singer with just one wintering ground, a likely origin population
was determined. Within song type 2, stable clusters linked two Kermadec singers with New
Caledonia and one Kermadec singer with Niue (red boxes, figure 2). The remaining Kermadec singers
within the song type 2 cluster were part of multi-origin clusters, although an eastwest divide was
evident (New Caledonia and Tonga versus Niue and Cook Islands) in all remaining stable Kermadec
and wintering ground groups (figure 2). Of the two Kermadec song type 1b singers, one singer was
consistently grouped with song sequences from the Cook Islands, although this was not quite a stable
cluster (AU = 89, figure 2). The second singer was represented by a single theme [14] and was
therefore excluded from figure 2; however, fine-scale analysis linked this Kermadec recording with the
Cook Islands (see below; electronic supplementary material, figure S1).
To assess the fine-scale (phrase type) similarity between songs recorded at the Kermadecs and each
wintering ground, a representative (median) unit sequence for each phrase type per singer was compared
for each theme. Hierarchical clustering and bootstrapping of the similarity scores for all individuals
recorded singing themes 1, 2, 3, 5, 6 and 14 linked singers from the Kermadecs to New Caledonia,
Tonga, Niue and the Cook Islands (table 1, figure 3a,b; electronic supplementary material, figure S1).
Themes 4, 7, 8 and 12 produced a varied picture of phrase type origin as Kermadec singers were
placed with singers from multiple wintering grounds. The remaining themes present at the wintering
grounds (9, 10, 11, 13, 15, 16, 17) were not recorded in any of the Kermadec songs.
In summary, the percentage matched phrase analyses suggested that 10 Kermadec singers were likely
to have originated from New Caledonia, 2 from Niue and 1 from the Cook Islands, whereas the fine-scale
(phrase) LSI analyses suggested that 11 of the Kermadec singers were from Tonga, 2 from New
Caledonia and 2 from the Cook Islands (table 1). The broad-scale (song) LSI analysis suggested two
Kermadec singers originated from New Caledonia and one from Niue (table 1). Combining all lines of
evidence, there were three occasions where all analyses agreed on the likely wintering ground of
origin for a Kermadec singer: KI12S6 and KI14S1 were assigned to New Caledonia, and KI14S2 was
assigned to the Cook Islands (table 1). All other singersorigins were left unassigned due to a lack of
agreement among analyses.
4. Discussion
Whales from many different populations were passing the Kermadecs at the same time (table 1),
providing the opportunity for song learning and easterly transmission of song across the South
Pacific. This is clearly one of potentially multiple important locations for the cultural transmission of
humpback song. Owing to fine-scale differences identified in the song recordings from the wintering
grounds analysed here, it was possible to match the songs recorded at the Kermadecs to New
Caledonia, Tonga, Niue and the Cook Islands, a pattern mirrored by data on genetically and
photographically identified individual whales in the same area [21]. These song analyses also
suggested that singers from both French Polynesia and eastern Australia were unlikely to visit this
migratory location. In general, little is known about the migratory routes of South Pacific humpback
whales. The temporary aggregation of whales at a migratory stopover may be related to humpback
whalesstrong urge to socially aggregate [34] and/or due to following oceanographic landmarks on
migration [35]. Regardless of the underlying driver, the stopover provides an opportunity for acoustic
connectivity among multiple migratory streams. The hybrid song recorded at the Kermadecs is
consistent with the hypothesis of song learning on migration [6,11], and although such an aggregation
of whales at a migratory stopover is temporary, it may be a major driver facilitating the easterly
transmission of song across the South Pacific. Furthermore, as humpback whales sing less frequently
on the feeding grounds [19], and as far as we know, do not aggregate at the Kermadecs during the
northward migration, the song learned during the southward migration needs to be remembered until
the next breeding season (akin to song memory in oscine songbirds [36]).
4.1. Possible origins of Kermadec singers
Analyses suggested the songs recorded at the Kermadecs matched all sampled South Pacific wintering
grounds except eastern Australia and French Polynesia, despite photo-identification and genetic
studies demonstrating the occasional between-season movement of individuals and thus connectivity R. Soc. open sci. 6: 190337
across the wider South Pacific region [3739]. A concurrent genetic and photo-identification study from
the Kermadecs [21] strongly reinforced our findings: individuals from multiple populations (excluding
eastern Australia and French Polynesia) were passing the Kermadecs on their southward migration.
Thus far, genetic and satellite tagging studies have not found a migratory link or route between
eastern Australia and the Kermadecs [4042], and although the exact location of the summer feeding
ground(s) or the migratory routes for the French Polynesian population remains elusive, photo-
identification and genetic matches have been found with the Antarctic Peninsula (South America) to
the east [43]. Migrating via the Kermadecs would represent a large and potentially unnecessary
deviation for both eastern Australian and French Polynesian whales.
4.2. Acoustic contact allows easterly transmission
Owing to the consistent easterly transmission of song types across the South Pacific [9,1819], multiple
song types were identified. The low frequency of song type 1 (the oldestsong type) in the recordings
from the western South Pacific region follows previous patterns [9], as the newersong type 2 from
the west was rapidly replacing it. Song type 1b was only recorded early in the season in New
Caledonia prior to singers switching to the newsong type 2 (electronic supplementary material,
figure S2), while it was recorded throughout the season at the Cook Islands. As expected, there was
also a low frequency of song type 1 at the Kermadecs. The strong pairing of themes (figure 3) from
the two whales singing song type 1b, indicated that both these Kermadec songs (and thus singers)
likely originated from the Cook Islands. This complements evidence from photo-identification and
genetic matches [21], that some whales from the Cook Islands, a central South Pacific population,
pass the Kermadecs on their southwardly migration. It is still unclear, however, whether the Cook
Islands represents a migratory corridor or a distinct winter breeding ground [39,44]. Regardless,
whales sampled in the Cook Islands are genetically most similar to those wintering in Tonga, but in
previous studies, acoustically more similar to those from French Polynesia [9,18,20,39,44]. Central
South Pacific populations (the Cook Islands and French Polynesia) need to come into contact at some
point in the year with whales from the western group (e.g. New Caledonia or Tonga) to allow
cultural transmission of the song. The Kermadecs provide a location where eastern whales may be
exposed to new song types from western populations, emphasizing the importance of shared
migratory routes or stopovers for the cultural transmission of song.
4.3. Song learning and potential for convergence
One song recorded at the Kermadecs mostly contained song type 2 themes but also one theme from song
type 1 (i.e. a hybrid song). This follows the pattern observed in the wintering grounds, where whales in
the western South Pacific region have switched, or are in the process of switching, to song type 2. Hybrid
songs are rare and likely short lived [6,11], so this hybrid song, with which we have likely captured some
part of the process by which singers change their song display from an older to a new song version,
suggests that the Kermadecs are a location where song learning occurs. Whales stay on average 4.6
days at Raoul Island (R.C. unpublished data, 2015), which potentially allows for exposure to song for
several days. Such song exposure may allow both the learning and transmission of population-specific
phrase types among populations and acquisition of new songs facilitating the easterly transmission of
these songs across the South Pacific. While this addresses how songs may spread east, the question of
why songs only spread in an easterly direction remains unresolved. One possible explanation is the
difference in population sizes; the large eastern Australian population may influence the smaller
South Pacific populations more then they influence it [9]. Future research examining population size,
density and transmission, potentially using agent-based models [45], may help to answer this question.
On the wintering grounds, South Pacific humpback whales are acoustically isolated (song only travels
effectively for tens of kilometres at most [46]) from other wintering grounds due to their natal site
fidelity. Previous South Pacific song studies have noted population-level differences in phrase type
presence [20]. Song from three Kermadec singers contained a phrase type which was sampled only
from New Caledonia and one or more phrases which were not identified in New Caledonian
recordings during the preceding season. During temporary acoustic isolation at the separate wintering
grounds, progressive song evolution creates fine-scale population differences [16,20]. The fine-scale LSI
analysis is consistent with the hypothesis that the reestablishment of acoustic contact at the
Kermadecs may have initiated the merging of these population-level differences in song (type 2) to a
single norm, resulting in a degree of cultural convergence. R. Soc. open sci. 6: 190337
Conforming to a cultural norm is important in many taxa [47]. In humans, conformity stems from a
motivation to copy others to fit into a social norm [48]; children will abandon behavioural preferences in
favour of imitating peers [49]. Cultural boundaries, often created by language differences, give humans a
means of establishing who to socialize and cooperate with [50]. Vocal conformity and dialect boundaries
are both prominent features of humpback whale song, and these features are also coupled with a
constant evolution of the trait. How cultural evolution and sexual selection each contribute to this
cultural phenomenon remains an open question.
5. Conclusion
Here, our results are consistent with the conclusion that song learning, transmission and potentially
convergence occurs at the Kermadecs, where migratory routes from multiple populations overlap. This
contact is likely to be a driver of horizontal transmission of song across the South Pacific. While
convergence and transmission have been shown within a population during migration and on their
wintering grounds [9,11], song exchange and convergence on a shared migratory route, and the
location of such an event, remained elusive. Song themes from multiple wintering grounds matched
songs recorded at the Kermadecs, including a hybrid of two songs, suggesting that multiple humpback
whale populations from across the South Pacific are travelling past these islands and song learning
may be occurring. Although it is not possible to discount other potential mechanisms of song
transmission [16], our results are consistent with the hypothesis of song learning on a shared migratory
route, a mechanism that could drive the eastern transmission of song across the South Pacific [9].
Ethics. The University of St. Andrews School of Biology Ethics Committee approved this study.
Data accessibility. The datasets supporting this article (comprising raw song transcripts and unit measurements) have been
uploaded as part of the electronic supplementary material.
Authorscontributions. C.O. transcribed acoustic samples, conducted data analysis and drafted the manuscript (with
supervision from E.C.G. and L.R.); E.C.G. conceived and designed the study, contributed acoustic data, supervised
data analysis, assisted with interpretation of results and critically revised the manuscript; L.R. assisted with
interpretation of results and critically revised the manuscript; R.C. conceived and designed the study, contributed
acoustic data and revised the manuscript; M.J.N. conceived and designed the study, contributed acoustic data and
revised the manuscript; J.A., C.G., D.D., O.A., N.H. and M.M.P. contributed acoustic data and revised the
manuscript. All authors contributed to the editing of the manuscript and were responsible for the approval of the
final manuscript.
Competing interests. We declare we have no competing interests.
Funding. C.O. was partially supported by the Sidney Perry Foundation, and the NERC Sea Mammal Research Unit
made a contribution towards the write up of this study. E.C.G. was supported by a Royal Society Newton
International Fellowship and a Royal Society University Research Fellowship. L.R. was supported by the
MASTS pooling initiative (The Marine Alliance for Science and Technology for Scotland) and their support is
gratefully acknowledged. MASTS is funded by the Scottish Funding Council ( grant reference HR09011) and
contributing institutions. J.A. was supported by an Australian Government Research Training Program
Scholarship, Australian American Association University of Queensland Fellowship and the Sea World Research
and Rescue Foundation Inc. Research in the Kermadec Islands was primarily supported by the New Zealand
Ministry for Primary Industries BRAG Fund, a University of Auckland FRDF Grant, the Australian Antarctic
Division, and the Pew Charitable Trusts. Data collection in eastern Australia was funded by the E&P Sound and
Marine Life Joint Industry Program (JIP) and the US Bureau of Ocean Energy Management as part of the
BRAHSS project. M.M.P. was partially supported by the National Oceanic Society and Dolphin & Whale
Watching Expeditions.
Acknowledgements. We thank Blue Planet Marine, Conservation International, the crew of the RV Braveheart, the research
teamRémi Dodémont, Becky Lindsay, James Tremlett, the Raoulies and our colleagues at the South Pacific Whale
Research Consortium, for supporting research in the Kermadec Islands. Many thanks to Ngāti Kuri and Te
Aupōuri for allowing us to work with their taonga. Research was conducted under a Department of Conservation
Permit #44388-MAR to R.C. The BRAHSS project thanks Rebecca Dunlop, Doug Cato and the numerous staff and
volunteers who have assisted with acoustic data collection. Thanks to Patrice Plichon and the Province Sud, as well
as Véronique Perard, Rémi Dodémont and Solene Derville from Operation Cétacés for their help with data
collection in New Caledonia. The Niue Whale Research Project thanks our project partner Fiafia Rex of Oma Tafuà
for collecting song data and without whom this work could not happen in Niue. Cook Islands research would like
to thank Joan Hauser Daeschler, Rose Cottage Outreach, Alyssa Stoller and Team. D.D. would like to thank Blue
Water Explorer Ltd. Fieldwork in French Polynesia was conducted under a permit issued to M.M.P. from French
Polynesias Ministry of the Environment. The authors thank Patrick Miller and two anonymous reviewers for
comments that improved previous versions of this manuscript. R. Soc. open sci. 6: 190337
1. Ramsey G. 2013 Culture in humans and other
animals. Biol. Philos. 28, 457479. (doi:10.
2. Rendell L, Whitehead H. 2001 Culture in whales
and dolphins. Behav. Brain Sci. 24, 309324.
3. Whitehead H. 2017 Geneculture coevolution in
whales and dolphins. Proc. Natl Acad. Sci. USA
114, 78147821. (doi:10.1073/pnas.
4. Whiten A, Ayala FJ, Feldman MW, Laland KN.
2017 The extension of biology through culture.
Proc. Natl Acad. Sci. USA 114, 77757781.
5. Janik VM. 2014 Cetacean vocal learning and
communication. Curr. Opin. Neurobiol. 28,
6065. (doi:10.1016/j.conb.2014.06.010)
6. Garland EC, Rendell L, Lamoni L, Poole MM,
Noad MJ. 2017 Song hybridization events
during revolutionary song change provide
insights into cultural transmission in humpback
whales. Proc. Natl Acad. Sci. USA 114,
78227829. (doi:10.1073/pnas.1621072114)
7. Krützen M, Mann J, Heithaus MR, Connor RC,
Bejder L, Sherwin WB. 2005 Cultural
transmission of tool use in bottlenose dolphins.
Proc. Natl Acad. Sci. USA 102, 89398943.
8. Allen J, Weinrich M, Hoppitt W, Rendell L. 2013
Network-based diffusion analysis reveals cultural
transmission of lobtail feeding in humpback
whales. Science 340, 485488. (doi:10.1126/
9. Garland EC, Goldizen AW, Rekdahl ML,
Constantine R, Garrigue C, Hauser ND, Poole
MM, Robbins J, Noad MJ. 2011 Dynamic
horizontal cultural transmission of humpback
whale song at the ocean basin scale. Curr. Biol.
21, 687691. (doi:10.1016/j.cub.2011.03.019)
10. Baker CS, Palumbi SR, Lambertsen RH, Weinrich
MT, Calambokidis J, OBrien SJ. 1990 Influence
of seasonal migration on geographic-
distribution of mitochondrial-DNA haplotypes in
humpback whales. Nature 344, 238240.
11. Noad MJ, Cato DH, Bryden MM, Jenner MN,
Jenner KCS. 2000 Cultural revolution in whale
songs. Nature 408, 537. (doi:10.1038/
12. Deecke VB, Barrett-Lennard LG, Spong P, Ford
JK. 2010 The structure of stereotyped calls
reflects kinship and social affiliation in resident
killer whales (Orcinus orca). Naturwissenschaften
97, 513518 (doi:10.1007/s00114-010-0657-z)
13. Rendell LE, Whitehead H. 2003 Vocal clans in
sperm whales (Physeter macrocephalus).
Proc. R. Soc. Lond. B 270, 225231. (doi:10.
14. Payne RS, McVay S. 1971 Songs of humpback
whales. Science 173 585597. (doi:10.1126/
15. Herman LM. 2017 The multiple functions of
male song within the humpback whale
(Megaptera novaeangliae) mating system:
review, evaluation, and synthesis. Biol. Rev. 92,
17951818. (doi:10.1111/brv.12309)
16. Payne R, Guinee LN. 1983 Humpback whale
(Megaptera novaeangliae) songs as an indicator
of stocks. In Communication and behavior of
whales (ed. R Payne), pp. 333358. Boulder,
CO: Westview Press.
17. Payne K, Payne R. 1985 large-scale changes
over 19 years in songs of humpback whales in
Bermuda. J. Comp. Ethol. 68,89114.
18. Garland EC et al. 2015 Population structure of
humpback whales in the western and central
South Pacific Ocean as determined by vocal
exchange among populations. Conserv. Biol. 29,
11981207. (doi:10.1111/cobi.12492)
19. Garland EC, Gedamke J, Rekdahl ML, Noad MJ,
Garrigue C, Gales N.2013 Humpback whale
song on the Southern Ocean feeding grounds:
implications for cultural transmission. PLoS ONE
8, e79422. (doi:10.1371/journal.pone.0079422)
20. Garland EC et al. 2013 Quantifying humpback
whale song sequences to understand the
dynamics of song exchange at the ocean basin
scale. J. Acoust. Soc. Am. 133, 560569. (doi:10.
21. Riekkola L et al. 2018 Application of a multi-
disciplinary approach to reveal population
structure and Southern Ocean feeding grounds
of humpback whales. Ecol. Indic. 89, 455465.
22. Noad MJ, Cato DH, Stokes MD. 2004 Acoustic
tracking of humpback whales: measuring
interactions with the acoustic environment. In
Paper presented at the Annual Conference of the
Australian Acoustical Society,Surfers Paradise,
QLD, Australia,35 November 2004.
23. Dunlop RA, Noad MJ, Cato DH, Stokes D. 2007
The social vocalization repertoire of east
Australian migrating humpback whales
(Megaptera novaeangliae). J. Acoust. Soc. Am.
122, 28932905. (doi:10.1121/1.2783115)
24. Garland EC, Lilley MS, Goldizen AW, Rekdahl ML,
Garrigue C, Noad MJ.2012 Improved versions of
the Levenshtein distance method for comparing
sequence information in animalsvocalisations:
tests using humpback whale song. Behaviour
149, 14131441. (doi:10.1163/1568539X-
25. Rekdahl ML, Dunlop RA, Noad MJ, Goldizen AW.
2013 Temporal stability and change in the social
call repertoire of migrating humpback whales.
J. Acoust. Soc. Am. 133, 17851795. (doi:10.
26. Levenshtein VI. 1966 Binary codes capable of
correcting deletions, insertions, and reversals.
Soviet Physics Doklady 10, 707710.
27. Helweg DA, Cato DH, Jenkins PF, Garrigue C,
McCauley RD. 1988 Geographic variation in
South Pacific humpback whale songs. Behaviour
135,127. (doi:10.1163/156853998793066438)
28. Tougaard J, Eriksen N. 2006 Analysing
differences among animal songs quantitatively
by means of the Levenshtein distance measure.
Behaviour 143, 239252. (doi:10.1163/
29. Garland EC, Rendell L, Lilley MS, Poole MM,
Allen JA, Noad MJ. 2017 The devil is in the
detail: quantifying vocal variation in a complex,
multi-levelled, and rapidly evolving display.
J. Acoust. Soc. Am. 142, 460472. (doi:10.1121/
30. Kershenbaum A, Garland EC. 2015 Quantifying
similarity in animal vocal sequences: which
metric performs best? Methods Ecol. Evol. 6,
14521461. (doi:10.1111/2041-210X.12433)
31. R Development Core Team. 2015 R Foundation
for statistical Computing, Vienna, Austria.
32. Suzuki R, Shimodaira H. 2006 Pvclust: an R
package for assessing the uncertainty in
hierarchical clustering. Bioinformatics 22,
15401542. (doi:10.1093/bioinformatics/btl117)
33. Sokal RR, Rohlf FJ. 1962 The comparison of
dendrograms by objective methods. Taxon 11,
3340. (doi:10.2307/1217208)
34. Clapham PJ, Zerbini AN. 2015 Are social
aggregation and temporary immigration driving
high rates of increase in some Southern
Hemisphere humpback whale populations? Mar.
Biol. 162, 625634. (doi:10.1007/s00227-015-
35. Garrigue C, Clapham PJ, Geyer Y, Kennedy AS,
Zerbini AN. 2015 Satellite tracking reveals novel
migratory patterns and the importance of
seamounts for endangered South Pacific
humpback whales. R. Soc. open sci 2, 150489.
36. Catchpole C, Slater PJB. 2008 Bird song:
biological themes and variations, 2nd edn.
Cambridge, UK: Cambridge University Press.
37. Garrigue C et al. 2011 Movement of individual
humpback whales between wintering grounds
of Oceania (South Pacific), 1999 to 2004.
J. Cetac. Res. Manage. 3, 275281.
38. Garrigue C et al. 2011 First assessment of
interchange of humpback whales between
Oceania and the East coast of Australia. J. Cetac.
Res. Manage. 3, 269274.
39. Steel D et al. 2017 Migratory interchange of
humpback whales (Megaptera novaeangliae)
among breeding grounds of Oceania and
connections to Antarctic feeding areas based on
genotype matching. Polar Biol. 41, 653662.
40. Gales N, Double MC, Robinson SA, Jenner C,
Jenner M, King ER, Gedamke JA, Paton DA,
Raymond B. 2009 Satellite tracking of
southbound East Australian humpback whales
(Megaptera novaeangliae): challenging the feast
or famine model for migrating whales. Int Whal
Comm: SC61/SH17.
41. Constantine R et al. 2014 Remote Antarctic
feeding ground important for east Australian
humpback whales. Mar. Biol. 161, 10871093.
42. Schmitt NT et al. 2014 Mixed-stock analysis of
humpback whales (Megaptera novaeangliae)on
Antarctic feeding grounds. J. Cetac. Res.
Manage. 14, 141147.
43. Albertson GR et al. 2018 Temporal stability and
mixed-stock analyses of humpback whales
(Megaptera novaeangliae) in the nearshore
waters of the Western Antarctic Peninsula. Polar
Biol. 41, 323340. (doi:10.1007/s00300-017-
2193-1) R. Soc. open sci. 6: 190337
44. Hauser N, Zerbini AN, Geyer Y, Heide
Jørgensen MP, Clapham P. 2010 Movements of
satellite monitored humpback whales,
Megaptera novaeangliae, from the Cook
Islands. Mar. Mammal Sci. 26, 679685.
45. Mcloughlin M, Lamoni L, Garland EC, Ingram S,
Kirke A, Noad MJ, Rendell L, Miranda E. 2018
Using agent-based models to understand the
role of individuals in the song evolution of
humpback whales (Megaptera novaeangliae).
Music Sci. 1,117. (doi:10.1177/205920431
46. Cato DH. 1991 Songs of humpback whales: the
Australian perspective. Mem. Qld Mus. 30,
47. Aplin LM, Farine DR, Morand-Ferron J, Cockburn
A, Thornton A, Sheldon BC. 2015 Experimentally
induced innovations lead to persistent
culture via conformity in wild birds.
Nature 518, 538541. (doi:10.1038/
48. Whiten A, Horner V, De Waal FB. 2005
Conformity to cultural norms of tool use in
chimpanzees. Nature 437, 737740. (doi:10.
49. Haun DB, Rekers Y, Tomasello M. 2014 Children
conform to the behavior of peers; other great
apes stick with what they know. Psychol. Sci. 25,
21602167. (doi:10.1177/0956797614553235)
50. Nettle D, Dunbar RI. 1997 Social markers and
the evolution of reciprocal exchange. Curr.
Anthropol. 38,9399. (doi:10.1086/204588) R. Soc. open sci. 6: 190337
... Secondly, whales from the same ocean interact with each other by sharing the same themes, considered as regional dialects. In a few cases, songs changed faster, with new units introduced by humpback whales from another area, although in fact the mechanism is still being disentangled and may include song learning on migratory routes leading to their characterization as cultural revolution (Noad et al., 2000;Allen et al., 2019;Owen et al., 2019). However, different populations often emit different themes, as do whales within the same population across spans of several years. ...
... With these comparisons, we would like to show if the whales emit the same sequences or not. We applied the Levenshtein distance similarity index (LSI) (Levenshtein, 1966), previously used in different studies, especially to assess the changes in humpback whale songs across multiple sites and over years (Eriksen et al., 2005;Rekdahl et al., 2018;Owen et al., 2019). This index allows to evaluate the number of changes needed to transform one song to another. ...
... A change could be an insertion, a deletion, or a substitution of one unit in the songs. Furthermore, the representative song (median string) was computed for each stimulus (Owen et al., 2019). In our study, we chose to provide the normalized version of the Levenshtein distance on the unit sequences emitted after each concrete sound element. ...
Full-text available
We describe an art–science project called “Feral Interactions—The Answer of the Humpback Whale” inspired by humpback whale songs and interactions between individuals based on mutual influences, learning process, or ranking in the dominance hierarchy. The aim was to build new sounds that can be used to initiate acoustic interactions with these whales, not in a one-way direction, as playbacks do, but in real interspecies exchanges. Thus, we investigated how the humpback whales generate sounds in order to better understand their abilities and limits. By carefully listening to their emitted vocalizations, we also describe their acoustic features and temporal structure, in a scientific way and also with a musical approach as it is done with musique concrète , in order to specify the types and the morphologies of whale sounds. The idea is to highlight the most precise information to generate our own sounds that will be suggested to the whales. Based on the approach developed in musique concrète , similarities with the sounds produced by bassoon were identified and then were processed to become “concrete sound elements.” This analysis also brought us to design a new music interface that allows us to create adapted musical phrases in real-time. With this approach, interactions will be possible in both directions, from and to whales.
... They show strong maternally directed site fidelity to breeding and feeding grounds, with occasional movement among locations [39][40][41][42]. Song sharing among populations is suggested to occur as a result of three mechanisms [31], which have been demonstrated to varying degrees around the world [39,[42][43][44][45][46][47][48][49]. Song sharing between populations can occur through males visiting more than one wintering ground in consecutive years, by males visiting more than one wintering ground within a breeding season, and finally through song sharing on shared feeding grounds and/or on shared or partially shared migratory routes [31]. ...
... Humpback whales predominantly sing during the breeding season, although some song has also been heard during migration and on the feeding grounds [48,65]. Here, during migration and on the feeding grounds, rare interactions between individuals of different populations may provide opportunities for exposure to song from neighbouring populations [44,45]. To mimic these patterns of singing behaviour and interactions in our model, each year individuals went through 10 song learning events (learning epochs). ...
Full-text available
Humpback whale song is an extraordinary example of vocal cultural behaviour. In northern populations, the complex songs show long-lasting traditions that slowly evolve, while in the South Pacific, periodic revolutions occur when songs are adopted from neighbouring populations and rapidly spread. In this species, vocal learning cannot be studied in the laboratory, learning is instead inferred from the songs' complexity and patterns of transmission. Here, we used individual-based cultural evolutionary simulations of the entire Southern and Northern Hemisphere humpback whale populations to formalize this process of inference. We modelled processes of song mutation and patterns of contact among populations and compared our model with patterns of song theme sharing measured in South Pacific populations. Low levels of mutation in combination with rare population interactions were sufficient to closely fit the pattern of diversity in the South Pacific, including the distinctive pattern of west-to-east revolutions. Interestingly, the same learning parameters that gave rise to revolutions in the Southern Hemisphere simulations gave rise to evolutionary patterns of cultural evolution in the Northern Hemisphere populations. Our study demonstrates how cultural evolutionary approaches can be used to make inferences about the learning processes underlying cultural transmission and how they might generate emergent population-level processes. This article is part of the theme issue ‘Vocal learning in animals and humans’.
... Singing humpback whales have captured the interest of researchers because of the apparent complexity (Allen et al., 2018;Suzuki et al., 2006), and variability (Garland et al., 2013;Owen et al., 2019), of the songs they produce. Several past scientific studies have emphasized how populations of humpback whales change the structural properties of songs across years (Cato, 1991;Garland et al., 2011;Maeda et al., 2000;Mercado et al., 2005;Payne & Payne, 1985;Winn & Winn, 1978). ...
... Historically, explanations for why humpback whales continuously change characteristics of their songs share the assumption that singers modify songs to enhance their communicative value: to avoid boring listeners (Winn & Winn, 1978), to establish a singer's creativity (Garland & McGregor, 2020;Payne, 2000), or to provide an acoustic logo of temporary group membership (Darling et al., 2006;Owen et al., 2019;Rekdahl et al., 2018). This assumption comes from cross-taxa comparisons with songbirds, and from a general belief that whales vary songs to increase the complexity and novelty of those songs, thereby increasing a singer's chances of producing offspring. ...
Observations of animals’ vocal actions can provide important clues about how they communicate and about how they perceive and react to changing situations. Here, analyses of consecutive songs produced by singing humpback whales recorded off the coast of Hawaii revealed that singers constantly vary the acoustic qualities of their songs within prolonged song sessions. Unlike the progressive changes in song structure that singing humpback whales make across months and years, intra-individual acoustic variations within song sessions appear to be largely stochastic. Additionally, four sequentially produced song components (or “themes”) were each found to vary in unique ways. The most extensively used theme was highly variable in overall duration within and across song sessions, but varied relatively little in frequency content. In contrast, the remaining themes varied greatly in frequency content, but showed less variation in duration. Analyses of variations in the amount of time singers spent producing the four themes suggest that the mechanisms that determine when singers transition between themes may be comparable to those that control when terrestrial animals move their eyes to fixate on different positions as they examine visual scenes. The dynamic changes that individual whales make to songs within song sessions are counterproductive if songs serve mainly to provide conspecifics with indications of a singer’s fitness. Instead, within-session changes to the acoustic features of songs may serve to enhance a singer’s capacity to echoically detect, localize, and track conspecifics from long distances.
... S7 and 8). Theme sequences were used to account for the skewing that tends to occur in LSI analyses based on length 55 , following similar studies 14,37,56 . To identify population-level differences within each song type, separate similarity analyses were also conducted between every pair of song cycles per song type (Purple: N = 46, Light Purple: N = 54, Brown: N = 61, Light Brown: N = 48, Teal: N = 72, Orange: N = 72, Figs. ...
Full-text available
Among animal species, the songs of male humpback whales (Megaptera novaeangliae) are a rare example of social learning between entire populations. Understanding fine-scale similarity in song patterns and structural features will better clarify how accurately songs are learned during inter-population transmission. Here, six distinct song types (2009–2015) transmitted from the east Australian to New Caledonian populations were quantitatively analysed using fine-scale song features. Results found that New Caledonian whales learned each song type with high accuracy regardless of the pattern’s complexity. However, there were rare instances of themes (stereotyped patterns of sound units) only sung by a single population. These occurred more often in progressively changing ‘evolutionary’ songs compared to rapidly changing ‘revolutionary’ songs. Our results suggest that populations do not need to reduce complexity to accurately learn song patterns. Populations may also incorporate changes and embellishments into songs in the form of themes which are suggested to be learnt as distinct segments. Maintaining complex song patterns with such accuracy suggests significant acoustic contact, supporting the hypothesis that song learning may occur on shared feeding grounds or migration routes. This study improves the understanding of inter-population mechanisms for large-scale cultural transmission in animals.
... The most recent analyses of Australian songs show revolutions occurring every other year on average (Allen et al., 2018). This surfeit of "revolutions" (~50 over 15 years) has expanded speculations about when cross-population song sharing happens, with it now assumed to occur in feeding grounds or during migrations when singers from different populations are more likely to come into acoustic contact (Owen et al., 2019;Warren et al., 2020;Zandberg et al., 2021). Garland and colleagues (2020Garland and colleagues ( , 2017Garland and colleagues ( , 2018 proposed that cultural revolutions in songs occur when a single singer innovates or hears singers from another population, whereas cultural evolution (more gradual change) results from copying errors. ...
Full-text available
Singing humpback whales constantly modify their songs over hours, days, months, and years, throughout their adult lives. Intriguingly, humpbacks appear to vary songs in concert, with most singers in a population producing similar songs at any given time. The convergent vocal dynamics of singing humpbacks have convinced many that songs are vocal customs, passed from singer to singer through vocal imitation. This interpretation has recently been challenged, however, by the discovery that singers not in acoustic contact may sing highly similar songs, and also appear to change their songs along similar trajectories. How could singers that cannot hear each other culturally conform? Here, it is argued that the changes humpback whales make to songs are inconsistent with either communal copying or competitive improvisation. Instead, singers apar to be continuously morphing the acoustic properties of songs in predictable ways both within and across songs, even in the absence of cross-population interactions. There is no direct evidence that singing humpback whales learn songs by copying other singers. The fact that groups of singers change songs in similar ways is not evidence of vocal imitation, cultural transmission, or cultural evolution. So called “cultural revolutions” in humpback whale songs, which have been touted as the clearest and most impressive evidence of culture in any nonhuman animal, actually provide evidence against vocal culture in humpback whales. Vocal complexity and convergence can arise through mechanisms other than cultural transmission via vocal imitation, and in the case of humpback whales, genetic predispositions and ecological conditions may be more relevant to determining how singers collectively change songs over time.
... The behavioural variant-the song type-is transmitted among individuals and subsequently populations. Clear evidence of complete song types appearing and rapidly replacing the existing song in its entirety within a population has been repeatedly shown across the South Pacific region [20,49,54,55,58,64,66,67,78,79]. As controlled social learning experiments are currently unfeasible in this species, agentbased models that explicitly test social learning and cultural transmission of humpback song at the individual level have provided compelling evidence for these underlying cultural processes [57,58]. ...
Culture presents a second inheritance system by which innovations can be transmitted between generations and among individuals. Some vocal behaviours present compelling examples of cultural evolution. Where modifications accumulate over time, such a process can become cumulative cultural evolution. The existence of cumulative cultural evolution in non-human animals is controversial. When physical products of such a process do not exist, modifications may not be clearly visible over time. Here, we investigate whether the constantly evolving songs of humpback whales ( Megaptera novaeangliae ) are indicative of cumulative cultural evolution. Using nine years of song data recorded from the New Caledonian humpback whale population, we quantified song evolution and complexity, and formally evaluated this process in light of criteria for cumulative cultural evolution. Song accumulates changes shown by an increase in complexity, but this process is punctuated by rapid loss of song material. While such changes tentatively satisfy the core criteria for cumulative cultural evolution, this claim hinges on the assumption that novel songs are preferred by females. While parsimonious, until such time as studies can link fitness benefits (reproductive success) to individual singers, any claims that humpback whale song evolution represents a form of cumulative cultural evolution may remain open to interpretation. This article is part of a discussion meeting issue ‘The emergence of collective knowledge and cumulative culture in animals, humans and machines’.
... Effects of earlylife experience can be identified by analyzing the site fidelity of animals to their breeding ground (Weinrich, 1998) and by comparing the migration patterns of offspring to those of their mother's (Colbeck et al., 2013). Finally, comparing the movement of cultural groups, especially if sympatric, can help to assess the effect of culturally transmitted information on animals' space use (Kendal et al., 2018;Owen et al., 2019). ...
Full-text available
Integrating diverse concepts from animal behavior, movement ecology, and machine learning, we develop an overview of the ecology of learning and animal movement. Learning-based movement is clearly relevant to ecological problems, but the subject is rooted firmly in psychology, including a distinct terminology. We contrast this psychological origin of learning with the task-oriented perspective on learning that has emerged from the field of machine learning. We review conceptual frameworks that characterize the role of learning in movement, discuss emerging trends, and summarize recent developments in the analysis of movement data. We also discuss the relative advantages of different modeling approaches for exploring the learning-movement interface. We explore in depth how individual and social modalities of learning can matter to the ecology of animal movement, and highlight how diverse kinds of field studies, ranging from translocation efforts to manipulative experiments, can provide critical insight into the learning process in animal movement.
Marine mammalsMarine mammals exhibit a wide range of variation in both individualIndividual and group behavior. Here, we explore how individuals respond to risk and uncertainty and how their social interactionsInteraction and the social systems that they construct can support resilienceResilience, but may also generate vulnerabilityVulnerability. We describe how populationPopulation level processes that are important for conservationConservation are functions of the aggregated behavior of individuals. Understanding the diversity of individuals and their interactions as components in larger complex systems can inform both what is conserved and how conservationConservation is undertaken. Inter-individualIndividualvariationInter-individual variation and the roles of some key individuals within marine mammal societies help us understand why the removal of certain individuals can sometimes be more consequential for conservationConservation than a single unit decline in census populationPopulation size. Further, the social learningSocial learning within marine mammal populations and the emergence of unique culturesCulture has implications for defining populationPopulationsegmentsPopulation segmentfor conservationConservation. This is particularly the case where these cultures interface with resource use, for example, in culturally transmitted foraging strategies. Key to understanding the conservationConservation significance of variation in individualIndividual and group behavior is unraveling how the underlying processes influence vital ratesVital rates(survivalSurvival, reproductionReproductionand dispersalDispersal). Social learningSocial learning and culture can influence range recoveryRecovery, responses to anthropogenic threatsAnthropogenic threats such as climate changeClimate change, and can exacerbate what we describe here as “anthropo-dependenceAnthropo-dependence.” But, social learning can also provide opportunities for increased ecological resilienceResilience, by providing a behavioral buffer to ecologicalEcological buffer change. By exploring different behavioral processes, from the individualIndividual, up to the group and the populationPopulation level, we provide insights for conservationConservation policy and practice, including distinguishing distinct populationPopulationsegmentsPopulation segment. We highlight some key areas for future researchResearch and note the immense value of longitudinal datasets for trackingTracking some of these processes.
Access to resources shapes species’ physiology and behaviour. Water is not typically considered a limiting resource for rainforest-living chimpanzees; however, several savannah and savannah-woodland communities show behavioural adaptations to limited water. Here, we provide a first report of habitual well-digging in a rainforest-living group of East African chimpanzees ( Pan troglodytes schweinfurthii ) and suggest that it may have been imported into the community’s behavioural repertoire by an immigrant female. We describe the presence and frequency of well-digging and related behaviour, and suggest that its subsequent spread in the group may have involved some degree of social learning. We highlight that subsurface water is a concealed resource, and that the limited spread of well-digging in the group may highlight the cognitive, rather than physical, challenges it presents in a rainforest environment.
As discussed in the introduction, the search for a Northwest Passage (NWP) from the Atlantic to the Pacific was a 500-year endeavor, beset by failure and disaster. Since the voyages by Columbus to the Caribbean islands, which he mistakenly took to be part of Asia, most Europeans had the false impression that the distance between Europe and Asia, traveling west, was relatively small. As the crow flies, traveling eastward from Lisbon to Hong Kong is a journey of about 11,000 km. By comparison, traveling westward from Lisbon to Hong Kong is a journey of more than 18,500 km. The length of the NWP itself is roughly equal to the distance between Nice and Amsterdam. The NWP spans roughly 1450 km from the North Atlantic entrance north of Baffin Island in the east to the Beaufort Sea north of Alaska in the west. The route is located entirely within the Arctic Circle, less than 1900 km from the North Pole (Fig. 8.1).
Full-text available
Male humpback whales produce hierarchically structured songs, primarily during the breeding season. These songs gradually change over the course of the breeding season, and are generally population specific. However, instances have been recorded of more rapid song changes where the song of a population can be replaced by the song of an adjacent population. The mechanisms that drive these changes are not currently understood, and difficulties in tracking individual whales over long migratory routes mean field studies to understand these mechanisms are not feasible. In order to help understand the mechanisms that drive these song changes, we present here a spatially explicit agent-based model inspired by methods used in computer music research. We model the migratory patterns of humpback whales, a simple song learning and production method coupled with sound transmission loss, and how often singing occurs during these migratory cycles. This model is then extended to include learning biases that may be responsible for driving changes in the song, such as a bias towards novel song, production errors, and the coupling of novel song bias and production errors. While none of the methods showed population song replacement, our model shows that shared feeding grounds where conspecifics are able to mix provide key opportunities for cultural transmission, and that production errors facilitated gradually changing songs. Our results point towards other learning biases being necessary in order for population song replacement to occur.
Full-text available
Humpback whales (Megaptera novaeangliae) congregate to breed during the austral winter near tropical islands of the South Pacific (Oceania). It has long been assumed that humpback whales from Oceania migrate primarily to Antarctic feeding grounds directly south (International Whaling Commission Management Areas V and VI); however, there are few records of individual movement connecting these seasonal habitats. Based on genetic samples of living whales collected over nearly two decades, we demonstrate interchange between the breeding grounds of Oceania and Antarctic feeding Areas V, VI, and I (i.e., from 130°E to 60°W), as well as with the eastern Pacific (Colombia), and the migratory corridors of eastern Australia and New Zealand. We first compared genotype profiles (up to 16 microsatellite loci) of samples collected from Oceania breeding grounds to each other and to those from the eastern Pacific. The matching profiles documented 47 individuals that were present on more than one breeding ground, including the first record of movement between Oceania and Colombia. We then compared the 1179 genotypes from the breeding grounds to 777 from the migratory corridors of east Australia and New Zealand, confirming the connection of these corridors with New Caledonia. Finally, we compared genotypes from breeding grounds to 166 individuals from Antarctic feeding Areas I–VI. This comparison of genotypes revealed five matches: one between New Caledonia and Area V, one between Tonga and Area VI, two between Tonga and Area I (western edge), and one between Colombia and Area I (Antarctic Peninsula). Despite the relatively small number of samples from the Antarctic, our comparison has doubled the number of recorded connections with Oceania available from previous studies during the era of commercial whaling.
Full-text available
Southern Hemisphere humpback whales breed in tropical waters and migrate to Antarctica to forage. While the breeding grounds are well defined, the population structure on Antarctic feeding grounds is poorly described. The Western Antarctic Peninsula (WAP) is of particular interest, where rapidly changing environmental conditions could alter prey distribution or migration pathways. To examine changes in the population of whales around the WAP, we used mitochondrial DNA (mtDNA) and 15 microsatellite loci. We compared our WAP dataset to a dataset collected 18 years earlier, and identified new haplotypes for the region, but found no significant difference between the datasets. We compared whales from the WAP to breeding populations in Oceania, Colombia, and Brazil. We used an Analysis of Molecular Variance to confirm significant genetic differentiation between the WAP and each breeding ground (overall FST = 0.035/0.007 mtDNA/microsatellite, p < 0.001) except Colombia. Bayesian mixed-stock analyses showed a large apportionment to Colombia (mtDNA 93.0%; CL 91–99%; microsatellites 86%; CL 72–93%) and a small apportionment to French Polynesia/Samoan Islands (mtDNA 2.9%; CL 0.0–11.5%; microsatellites 8.9%; CL 0–22%), supporting the strong connection between Colombia and the WAP. Assignment tests allocated 81 individuals to Colombia and two to French Polynesia/Samoan Islands. No other breeding grounds had significant apportionments. Direct connectivity of French Polynesia to the WAP was confirmed with the first genotype match of French Polynesia to a feeding area. Continued genetic monitoring will highlight the complex patterns of humpbacks in this rapidly changing climate. Our results serve as a baseline for humpback whale population structure, illustrate mixed-stock analysis as a useful tool for migrating wildlife, and aid in future management considerations for humpbacks.
Full-text available
Cultural processes occur in a wide variety of animal taxa, from insects to cetaceans. The songs of humpback whales are one of the most striking examples of the transmission of a cultural trait and social learning in any nonhuman animal. To understand how songs are learned, we investigate rare cases of song hybridization, where parts of an existing song are spliced with a new one, likely before an individual totally adopts the new song. Song unit sequences were extracted from over 9,300 phrases recorded during two song revolutions across the South Pacific Ocean, allowing fine-scale analysis of composition and sequencing. In hybrid songs the current and new songs were spliced together in two specific ways: (i) singers placed a single hybrid phrase, in which content from both songs were combined, between the two song types when transitioning from one to the other, and/or (ii) singers spliced complete themes from the revolutionary song into the current song. Sequence analysis indicated that both processes were governed by structural similarity rules. Hybrid phrases or theme substitutions occurred at points in the songs where both songs contained "similar sounds arranged in a similar pattern." Songs appear to be learned as segments (themes/phrase types), akin to birdsong and human language acquisition, and these can be combined in predictable ways if the underlying structural pattern is similar. These snapshots of song change provide insights into the mechanisms underlying song learning in humpback whales, and comparative perspectives on the evolution of human language and culture.
Obtaining direct measurements to characterise ecosystem function can be hindered by remote or inaccessible regions. Next-generation satellite tags that inform increasingly sophisticated movement models, and the min-iaturisation of animal-borne loggers, have enabled the use of animals as tools to collect habitat data in remote environments, such as the Southern Ocean. Research on the distribution, habitat use and recovery of Oceania's humpback whales (Megaptera novaeangliae) has been constrained by the inaccessibility to their Antarctic feeding grounds and the limitations of technology. In this multidisciplinary study, we combine innovative analytical tools to comprehensively assess the distribution and population structure of this marine predator throughout their entire migratory range. We used genotype and photo-identification matches and conducted a genetic mixed-stock analysis to identify the breeding ground origins of humpback whales migrating past the Kermadec Islands, New Zealand. Satellite tracking data and a state-space model were then used to identify the migratory paths and behaviour of 18 whales, and to reveal their Antarctic feeding ground destinations. Additionally, we conducted progesterone assays and epigenetic aging to determine the pregnancy rate and age-profile of the population. Humpback whales passing the Kermadec Islands did not assign to a single breeding ground origin, but instead came from a range of breeding grounds spanning ∼3500 km of ocean. Sampled whales ranged from calves to adults of up to 67 years of age, and a pregnancy rate of 57% was estimated from 30 adult females. The whales migrated to the Southern Ocean (straight-line distances of up to 7000 km) and spanned ∼4500 km across their Antarctic feeding grounds. All fully tracked females with a dependent calf (n = 4) migrated to the Ross Sea region, while 70% of adults without calves (n = 7) travelled further east to the Amundsen and Bellingshausen Seas region. By combining multiple research and analytical tools we obtained a comprehensive understanding of this wide-ranging, remote population of whales. Our results indicate a population recovering from exploitation, https://doi. T and their feeding ground distribution serves as an indicator of the resources available in these environments. The unexpected Kermadec Islands migratory bottleneck of whales from several breeding grounds, variable distribution patterns by life history stage and high pregnancy rates will be important in informing conservation and management planning, and for understanding how this, as well as other whale populations, might respond to emerging threats such as climate change.
The interchange and isolation of individual humpback whales between wintering grounds of Oceania (South Pacific) and the east coast of Australia were documented by individual identification photographs collected from 1999 to 2004. Interchange was assessed using regional catalogues of fluke photographs, totalling 692 individuals from Oceania (represented by New Zealand, New Caledonia, Vanuatu, Fiji, Samoa, Tonga, Niue, Cook Island, French Polynesia and American Samoa) and 1242 individuals from Hervey and Byron Bay representing the southbound and the northbound migration along the east coast of Australia (EA). Overall, there were seven documented movements between EA and Oceania. Four instances of movement of four individuals were documented between EA and Oceania, all between EA and the closest breeding grounds of New Caledonia. A further three movements were recorded between EA and a small catalogue (n = 13) from the New Zealand migratory corridor. During this same period, 20 cases of interchange were documented among nine breeding grounds: French Polynesia, Cook Islands, Niue, American Samoa, Samoa, Tonga, Fiji, Vanuatu and New Caledonia. The low level of interchange between Oceania and the east coast of Australia and the movement across Oceania (including interchange across the boundaries of Areas V and VI) have important implication in understanding the stock structure and abundance of humpback whales in the South Pacific.
Identifying and quantifying variation in vocalizations is fundamental to advancing our understanding of processes such as speciation, sexual selection, and cultural evolution. The song of the humpback whale (Megaptera novaeangliae) presents an extreme example of complexity and cultural evolution. It is a long, hierarchically structured vocal display that undergoes constant evolutionary change. Obtaining robust metrics to quantify song variation at multiple scales (from a sound through to population variation across the seascape) is a substantial challenge. Here, the authors present a method to quantify song similarity at multiple levels within the hierarchy. To incorporate the complexity of these multiple levels, the calculation of similarity is weighted by measurements of sound units (lower levels within the display) to bridge the gap in information between upper and lower levels. Results demonstrate that the inclusion of weighting provides a more realistic and robust representation of song similarity at multiple levels within the display. This method permits robust quantification of cultural patterns and processes that will also contribute to the conservation management of endangered humpback whale populations, and is applicable to any hierarchically structured signal sequence.
Whales and dolphins (Cetacea) have excellent social learning skills as well as a long and strong mother-calf bond. These features produce stable cultures, and, in some species, sympatric groups with different cultures. There is evidence and speculation that this cultural transmission of behavior has affected gene distributions. Culture seems to have driven killer whales into distinct ecotypes, which may be incipient species or subspecies. There are ecotype-specific signals of selection in functional genes that correspond to cultural foraging behavior and habitat use by the different ecotypes. The five species of whale with matrilineal social systems have remarkably low diversity of mtDNA. Cultural hitchhiking, the transmission of functionally neutral genes in parallel with selective cultural traits, is a plausible hypothesis for this low diversity, especially in sperm whales. In killer whales the ecotype divisions, together with founding bottlenecks, selection, and cultural hitchhiking, likely explain the low mtDNA diversity. Several cetacean species show habitat-specific distributions of mtDNA haplotypes, probably the result of mother-offspring cultural transmission of migration routes or destinations. In bottlenose dolphins, remarkable small-scale differences in haplotype distribution result from maternal cultural transmission of foraging methods, and large-scale redistributions of sperm whale cultural clans in the Pacific have likely changed mitochondrial genetic geography. With the acceleration of genomics new results should come fast, but understanding gene-culture coevolution will be hampered by the measured pace of research on the socio-cultural side of cetacean biology.