ArticlePDF Available

Vocal sequences in narwhals ( Monodon monoceros )

  • University of Windsor (

Abstract and Figures

Sequences are indicative of signal complexity in vocal communication. While vocal sequences are well-described in birds and terrestrial mammals, the extent to which marine mammals use them is less well understood. This study documents the first known examples of sequence use in the narwhal (Monodon monoceros), a gregarious Arctic cetacean. Eight female narwhals were fitted with animal-borne recording devices, resulting in one of the largest datasets of narwhal acoustic behaviour to date. A combination of visual and quantitative classification procedures was used to test whether subjectively defined vocalization patterns were organized into sequences. Next, acoustic characteristics were analyzed to assess whether sequences could disclose group or individual identity. Finally, generalized linear models was used to investigate the behavioural context under which sequences were produced. Two types of sequences, consisting of “paired” patterns and “burst pulse series,” were identified. Sequences of burst pulse series were typically produced in periods of high vocal activity, whereas the opposite was true for sequences of paired patterns, suggesting different functions for each. These findings extend the set of odontocetes which are known to use vocal sequences. Inquiry into vocal sequences in other understudied marine mammals may provide further insights into the evolution of vocal communication.
Content may be subject to copyright.
Vocal sequences in narwhals (Monodon monoceros)
Sam F. Walmsley, Luke Rendell, Nigel E. Hussey, and Marianne Marcoux
Citation: The Journal of the Acoustical Society of America 147, 1078 (2020); doi: 10.1121/10.0000671
View online:
View Table of Contents:
Published by the Acoustical Society of America
Repeated downsweep vocalizations of the Araguaian river dolphin, Inia araguaiaensis
The Journal of the Acoustical Society of America 147, 748 (2020);
Lombard effect: Minke whale boing call source levels vary with natural variations in ocean noise
The Journal of the Acoustical Society of America 147, 698 (2020);
Comparison of fin whale 20Hz call detections by deep-water mobile autonomous and stationary recorders
The Journal of the Acoustical Society of America 147, 961 (2020);
Song production by the North Pacific right whale, Eubalaena japonica
The Journal of the Acoustical Society of America 145, 3467 (2019);
Effects of multiple exposures to pile driving noise on harbor porpoise hearing during simulated flights—An
evaluation tool
The Journal of the Acoustical Society of America 147, 685 (2020);
Simulation of humpback whale bubble-net feeding models
The Journal of the Acoustical Society of America 147, 1126 (2020);
Vocal sequences in narwhals (Monodon monoceros)
Sam F. Walmsley,
Luke Rendell,
Nigel E. Hussey,
and Marianne Marcoux
Sea Mammal Research Unit, School of Biology, University of St Andrews, Sir Harold Mitchell Building, St Andrews, KY16 9TH, Scotland
Integrative Biology, University of Windsor, 401 Sunset Avenue, Windsor, Ontario, N9B 3P4, Canada
Arctic Aquatic Research Division, Fisheries and Oceans Canada, Winnipeg, Manitoba, R3T 2N6, Canada
Sequences are indicative of signal complexity in vocal communication. While vocal sequences are well-described in
birds and terrestrial mammals, the extent to which marine mammals use them is less well understood. This study
documents the first known examples of sequence use in the narwhal (Monodon monoceros), a gregarious Arctic ceta-
cean. Eight female narwhals were fitted with animal-borne recording devices, resulting in one of the largest datasets
of narwhal acoustic behaviour to date. A combination of visual and quantitative classification procedures was used
to test whether subjectively defined vocalization patterns were organized into sequences. Next, acoustic characteris-
tics were analyzed to assess whether sequences could disclose group or individual identity. Finally, generalized lin-
ear models were used to investigate the behavioural context under which sequences were produced. Two types of
sequences, consisting of “paired” patterns and “burst pulse series,” were identified. Sequences of burst pulse series
were typically produced in periods of high vocal activity, whereas the opposite was true for sequences of paired pat-
terns, suggesting different functions for each. These findings extend the set of odontocetes which are known to use
vocal sequences. Inquiry into vocal sequences in other understudied marine mammals may provide further insights
into the evolution of vocal communication. V
C2020 Acoustical Society of America.
(Received 16 July 2019; revised 4 January 2020; accepted 13 January 2020; published online 13 February 2020)
[Editor: Rebecca A. Dunlop] Pages: 1078–1091
Species in many taxa combine vocalizations according
to stereotyped organizational principles, producing vocal
sequences (Kershenbaum et al., 2016). Sequences can be
composed of repetitions of the same unit or combinations
of different units (Collier et al., 2014; Kershenbaum et al.,
2016) and may be produced by a single individual or multi-
ple individuals, as in the overlapping songs of great tits
(Parus major;Krebs et al., 1981). Many species’ vocal
production capacities are largely innate, meaning they can
only use a small repertoire of sounds (Podos, 1996).
However, a set of vocal signals combined into structured
sequences can contain more information than the same
sounds produced in isolation. As such, constraints on vocal
production are hypothesized to be an evolutionary driver
sequence use, as in Campbell’s monkeys (Cercopithecus
campbelli; Ouattara et al., 2009). In species with more
flexible vocal production, sequences may still contribute to
information-rich communication, as is the case for human
language. While possible relationships between the use of
sequences and the evolution of language are not resolved
(Kershenbaum et al., 2014;Ouattaraet al., 2009; Scott-
Phillips et al., 2014), it is agreed that vocal sequences are
indicative of signal complexity. Even when individual
units do not have specific meanings, the “developmental
stress hypothesis” states that the complexity of a sequence
(e.g., birdsong) can serve as an otherwise arbitrary indica-
tor of mate quality (Nowicki and Searcy, 2004). Thus,
without attention to the sequential structure of signals, the
complexity and functions of vocal repertoires may be
underestimated (Collier et al., 2014;Kershenbaum et al.,
ıs et al., 2018).
Toothed whales are important model systems for
studying the evolution of vocal communication. For example,
bottlenose dolphins [Tursiops species (sp)] are the only non-
human species for which vocal production learning, func-
tional reference, signal innovation, and the capacity to
understand syntax have all been demonstrated (Herman,
2006;Janik, 2013). However, in contrast to the complex
sequences described in songbirds, terrestrial mammals, and
mysticetes (Payne and McVay, 1971;McDonald et al., 2006;
Berwick et al., 2011), most examples of vocal sequences in
odontocetes are repetitions of a single call type. Existing
work has identified nonrandom transitions between calls of
different types in short-finned pilot whales (Globicephala
macrorhynchus;Sayigh et al.,2013), killer whales (Orca
orcinus;Saulitis et al.,2005), and bottlenose dolphins
(Ferrer-i-Cancho and McCowan, 2012). Bottlenose dolphins
Portions of this work were presented in “Using biologging technology to
study vocal sequences in narwhals” at the Easter meeting of the Association
for the Study of Animal Behaviour, University of York, York, England,
2019, and “Rhythmically repeated patterns of pulsed vocalizations in wild
narwhals,” MSc dissertation, University of St. Andrews, 2018.
Electronic mail:, ORCID: 0000-0001-8577-6884.
ORCID: 0000-0002-1121-9142.
1078 J. Acoust. Soc. Am. 147 (2), February 2020 V
C2020 Acoustical Society of America0001-4966/2020/147(2)/1078/14/$30.00
are known to produce feeding-associated bray sequences com-
posed of repeated pulsed units, as well as “multi-looped”
sequences of signature whistles (Janik, 2000;Lu
ıs et al., 2018;
Esch et al.,2009). Finally, sperm whales (Physester macroce-
phalus) are known to produce temporally stereotyped click
codas (Watkins and Schevill, 1977). However, the extent to
which other odontocete species organize their vocalizations
into sequences is poorly understood (Janik, 2009).
Technology has traditionally been a limiting factor in
the study of marine mammal communication (Lammers and
Oswald, 2015, p. 125), given that most behaviour occurs
underwater, and vocalizations are often produced at frequen-
cies beyond the range of human hearing (Mann, 2000). This
has resulted in biases for the study of accessible (e.g.,
coastal) species or species that are readily kept in captivity.
It is possible, then, that the apparent lack of vocal sequences
in odontocetes is an artifact of methodological limitations.
Alternatively, an advanced ability to produce and learn new
signals (e.g., Abramson et al., 2018) might allow odonto-
cetes to produce large enough repertoires of singular signals
to satisfy their communication needs without recourse to
Here, we used high-bandwidth animal-borne recording
devices to investigate the vocal behaviour of individual nar-
whals (Monodon monoceros), a species for which the nature
and function of communication signals remain understudied
(Morisaka, 2012;Morisaka et al., 2013). In winter, the nar-
whal’s environment is characterized by dense ice-cover and
darkness (Laidre and Heide-Jørgensen, 2011), making
acoustic signals critical for both sensing the environment
and social interaction. Narwhals have been recorded produc-
ing a range of plausibly social sounds, typically classified as
“pulsed calls” composed of clicks with very short inter-click
intervals, as well as “tonal calls” or whistles, and “mixed
calls,” which include overlaid pulsed and tonal components
(Marcoux et al., 2012;Stafford et al., 2012;Shapiro, 2006).
Initial exploration of a subset of our recordings revealed two
types of possible sequences, one consisting of “paired” pat-
terns of pulsed calls, and the other consisting of sets of short
burst pulses, hereafter, termed “burst pulse series.”
To investigate the vocalizations in more depth, we first
asked whether they were produced according to stereotyped
organizational principles, satisfying the definition of sequen-
ces. Without being able to locate an individual using multi-
ple simultaneous recording devices, it is challenging to
attribute vocalizations to specific individuals, even when
instrumented—for example, an animal swimming alongside
a tagged individual might have its vocal production appara-
tus closer to the tag hydrophone than the tagged individual
itself (Sayigh et al., 2013). This makes it difficult to distin-
guish between sequences produced by single individuals and
those produced in call exchanges (Kershenbaum et al.,
2014). For this reason, we then investigated frequency and
amplitude characteristics of these patterns to understand
whether differences in calls recorded from different individ-
uals could be indicative of individual or group-specific sig-
nals. Finally we used generalized linear models (GLMs) to
investigate the behavioural context of sequences in order to
make inferences about their function.
A. Study system and data collection
Narwhals were captured and equipped with instrument
packages that recorded acoustic and depth data (Acousonde
Model 3B tags, Santa Barbara, California, USA) in Tremblay
Sound, Nunavut (7228049.558800 N, 8054018.259200 W)
between August 13 and September 11, 2017. Tags were
attached adjacent to the left side of the dorsal ridge by suction
cup, allowing them to be released after several days and float
to the surface for recovery. The Acousonde
tags were pro-
grammed to alternate every 30 min between low-frequency
(LF) and high-power (HP) channels with sampling rates of
25.8 kHz and 232.3 kHz, respectively (see Wiggins, 2013,for
details). The HP channel included a 22 dB low-pass anti-alias
filter at 100 kHz, allowing for high quality representations of
the primary bandwidth of narwhal clicks (20–70 kHz; Marcoux
et al., 2012;Koblitz et al., 2016). Two tags (NW08, NW09)
were programmed with an additional 20 dB gain, which was
subtracted to match the relative amplitudes of the other six tags
prior to acoustic analyses.
Because one aim of the overall tagging effort was to
investigate year-round movements, narwhals recorded in this
study were also fitted with “backpack” telemetry tags during
capture, which were secured with pins through the dorsal
ridge. Remotely deployed tags on similar species tend to last
only a few weeks to months (Andrews et al.,2008), making
net-based capture and tagging an important tool in their study
(Gonzalez, 2001;Blackwell et al.,2018). Acousonde
were sometimes secondarily secured by cable to these back-
pack tags, although the resulting telemetry data were not
used in this analysis.
B. Exploratory analyses
1. Sound auditing and initial call selection
Sound files of the HP channel were visually and aurally
audited in 15-s viewing frames in Raven Pro 1.3 [2007; fast
Fourier transform (FFT) length 4096, 50% overlap, Hann
window maximum frequency display 80 kHz]. This audit
informed the design of a simple pre-classification procedure,
where calls with sufficient signal-to-noise ratio (SNR) to be
visually and aurally discernable were annotated and
extracted. A small amount of overlap by surfacing noise,
clicks, or other vocalizations was permitted (Kaplan et al.,
2014). Initial call classes included vocalizations broadly
classified as tonal, pulsed, or mixed (containing both tonal
and pulsed components). Echolocation trains and buzzes
were readily distinguishable from other vocalizations by dif-
ferences in click rate (Lammers et al., 2004;Arranz et al.,
2016) and were typified by clicks at relatively low inter-
click intervals (approximately 200 ms) speeding up to very
high repetition rates with inter-click intervals of approxi-
mately 3 ms (Rasmussen et al.,2015;Blackwell et al.,2018).
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1079
Given our interest in identifying repeated call types,
some of which are produced when animals are isolated
(Janik and Slater, 1998), we included all recordings from
the point of release onward for these analyses. We had no
knowledge regarding social affiliations between individuals.
Furthermore, while all recordings used in this analysis came
from tags attached to female narwhals, our inability to fully
discriminate between focal and non-focal vocalizations
meant that we made no assumptions regarding the sex of
calling animals.
2. Two types of patterns: Classification of paired
patterns and burst pulse series
Given that patterns of pulsed vocalizations have not yet
been described in narwhals, we had to create bespoke defini-
tions of sequence types. We defined paired patterns as two
stereotyped, pulsed (i.e., click-based) units that co-occur in
the same order within a short time interval (<2 s), the pair
of which is repeated at least three times within 30 s (e.g.,
“A-B–A-B–A-B”; Fig. 1).
We also identified a series of short pulsed vocalizations,
possibly similar to the “chitter” described by Stafford et al.
(2012). These were termed as burst pulse series (Fig. 2). We
limited our analysis to sets of at least three burst pulse series,
each produced within 10 s of the next. Subunits greater than
0.5 s apart were assumed to belong to separate series.
Time and repetition thresholds were informed by
exploratory analyses as well as the timing of rhythmically
repeated vocalizations in other odontocetes (Janik and
Sayigh, 2013;Riesch et al., 2006;Zwamborn and
Whitehead, 2017). Quantitative support for the distinction
between paired patterns and burst pulse series was verified
retrospectively with a two-sample t-test comparing the
duration of subunits of each type (i.e. the AB units in Fig. 1
compared to the sub-units in Fig. 2).
C. Testing stereotyped organization of units
1. Visual classification of paired patterns
We used a visual classification task with independent
human observers to test our prediction that narwhals combined
recognizable types of “A-B” pairs into repetitive sequences
(see Appendix A for details of the classification task). Fleiss’
kappa statistic was used to quantify agreement between
observers (irr Package in R;Gamer et al.,2015;Landis and
Koch, 1977). To assess the distribution of call types across
time and recording devices, we labelled each pair according to
the type matched by the majority of raters if at least 4/6 agreed
on the same type. Any remaining patterns were excluded from
further analyses.
2. Discriminant function analysis of paired patterns
In principle, animals could produce vocalizations con-
taining gaps of silence that are still perceived as single calls.
Support for the interpretation that paired patterns are combi-
nations of units would come from the finding that their con-
stituent units are also produced alone. To this end, we
performed an initial classification of any “lone” calls that
appeared to match a single unit of a paired pattern. These
matches were verified with a discriminant function analysis
(DFA), classifying the “A” units of all stereotyped patterns
(as classified in the visual classification task), including the
additional “lone units.”
For the DFA, we first filtered each A unit using a
Butterworth four-pole bandpass filter bounded by user-
defined LF and high-frequency limits from the selection pro-
cess, similar to Marcoux et al. (2012). Individual clicks
FIG. 1. Sample spectrogram showing definition of repeated “paired” vocalizations (FFT length, 4096; window, Hann; overlap, 50%).
1080 J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al.
were located using the “findpeaks” function from the Signal
Processing Toolbox in MATLAB (R2017b; The
MathWorks, Natick, MA). Starting and ending pulse repeti-
tion rate (PRR, measured in Hz) were then calculated from
the initial and final quartiles of the call, respectively. These
PRR measurements and duration (s) were checked for nor-
mality and homogeneity of variance and then entered into a
DFA. (HH Package in R;Heiberger, 2018). Only paired pat-
tern types representing four or more A-B pairs were used so
that the number of variables in the DFA did not exceed the
size of the smallest group.
3. Levenshtein distance analysis of burst pulse series
Using a custom MATLAB (R2017b) script, we manu-
ally marked the start- and end-times of each subunit in each
series. These data were transformed into a binary string that
represented, with one value per recording sample, the pre-
cise timing of the burst pulse series as “on” (1) or “off” (0).
These strings were then down-sampled to every 500th value,
allowing us to reduce the computational load of subsequent
analyses while retaining high temporal resolution (465 val-
Levenshtein distance (LD) is the minimum number of
insertions, deletions, and substitutions, required to transform
one sequence into another, and has been used to compare
sequences as diverse as deoxyribonucleic acid (DNA), lan-
guage (Petroni and Serva, 2010), and humpback song
(Garland et al.,2012;Kershenbaum and Garland, 2015). To
test for repetition in sequences of burst pulse series, we calcu-
lated LD between series using the binary call representations,
first for observed transitions, defined as adjacently produced
burst pulse series within originally identified sequences and
then also for an equal number of random comparisons,pair-
wise comparisons of the same dataset of burst pulse series
reshuffled in random order. Random comparisons were calcu-
lated within samples from each tag to accommodate for any
individual-specific differences in vocal behaviour, which
could bias subsequent estimates of stereotypy. We interpreted
lower LD (i.e., higher similarity) values for observed com-
pared to random transitions as evidence of repeated use of
stereotyped sequences.
We fit a Poisson family GLM using the canonical link
function to compare LDs between observed and random
transitions according to the following equations:
LD ¼exp b0þ/Randomb1
LD PoissonðLDÞ;
where LD represents actual measures of LD and LD repre-
sents expected measures. b0is the model intercept, repre-
senting mean LD for observed transitions between burst
pulse series. b1is a contrast representing the difference
between LD measures of random comparisons relative to
observed transitions. We also fit an identical model with the
quasi-Poisson family to check for overdispersion in LD val-
ues. Only recordings with more than one sequence of burst
pulse series were included.
D. Inferring patterns of production across individuals
1. Discriminating between sequences and call
A lack of overlap between repeated signals is often
used to support inferences that a sequence is produced by a
FIG. 2. Sample spectrogram showing the definition of the burst pulse series (FFT length, 4096; window, Hann; overlap, 50%).
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1081
single individual (Lu
ıs et al., 2018;Sayigh et al., 2013;
Zwamborn and Whitehead, 2017). However, some species
produce call exchanges with precise and stereotyped timing
(Mann et al., 2006;Pika et al., 2018), meaning that nonover-
lapping sequences can also result from exchanges between
multiple individuals (Miller et al., 2004). To assess whether
sequences of vocalizations are produced by a single animal,
we examined not only overlap, but also variation in ampli-
tude and frequency of pulsed vocalizations produced in
sequences, expecting that the exceptionally directional
clicks of narwhals (Koblitz et al., 2016) should not be con-
sistently recorded with the same acoustic characteristics
from multiple individuals.
2. Exclusion of patterns produced by non-focal
Clicks recorded from suction-cup-tagged cetaceans are
hypothesized to include additional LF components that are
conducted through the body (Johnson et al., 2006;Johnson
et al., 2009;Zimmer et al., 2005;Blackwell et al., 2018).
We examined the frequency spectra of clicks from each call
sequence to determine which were unlikely to be produced
by the focal individual (see Appendix B for details of the
discrimination method). We compared the remaining
“possibly focal” sequences across tag recordings to consider
evidence of individual- or group-specific call characteristics.
E. Modeling the behavioural context of sequence
We fit GLMs to explore the context, and thereby poten-
tial function, of narwhal call sequences. Models were fit
according to the following equations:
Srepresents the expected probability of a sequence
occurring in a given 60 second period. b0is the intercept of
the linear predictor. b1represents the effect of the number of
other calls (vÞ,b2represents the effect of depth (dÞin meters,
and b3represents the difference in model intercept when
buzzes were detected. Call counts and depth values were
-transformed as the raw data were positively skewed. We
expected that a logarithmic representation would be biologi-
cally appropriate because a change in depth of 1 m vs 5 m rep-
resents a proportionally much larger difference in pressure
than 100 m vs 105 m, for example. Depth profiles were cali-
brated by zero-offset correction using a custom script in
MATLAB from the Animal Tag Tools Calibration Toolkit.
This model structure was applied separately to patterns
of pulsed calls and burst pulse series, and for each, accord-
ing to the 60 s preceding a call (“previous activity models”),
and the 60 s following a call (“subsequent activity models”).
For pseudo-absences, we randomly selected 60 additional
60-s periods from the same tags where the sequences in
question were identified but in which no sequences were
heard. These pseudo-absences were only selected from peri-
ods following the first vocalization identified on a tag
recording and prior to tag detachment as determined from
the dive profile. All four models were also fit using GLMs
with quasi-binomial families to check for overdispersion,
and diagnostic plots were used to validate all model struc-
tures prior to consideration of estimated parameters.
Statistical analyses were done in R3.4.2 (R Core Team
2017). Analysis scripts and basic datasets are available via
the Open Science Framework.
A. General features of recordings
Eight female narwhals were equipped with hydrophone
tag packages. We identified 3261 vocalizations from the
resulting high-frequency recordings, including tonal calls,
pulsed calls, mixed calls, and patterns of pulsed calls (Table
I). Most tags (7/8) detached from the narwhal prior to maxi-
mum recording duration, resulting in varying recording
lengths per tag (Table I). Periods of multiple overlapping
vocalizations at rates of up to 3 calls/s were identified on 4/
8 tags, and presumed to be the result of narwhals aggregat-
ing into larger herds. Consistent with other studies, pulsed
vocalizations outnumbered whistles (Stafford et al., 2012;
Miller et al., 1995), although we did note several periods
when many overlapping whistles were produced, as
described by Ford and Fisher (1978).
Contrary to suggestions that narwhals tend not to feed
in the summertime (Stafford et al., 2012), buzzes indicative
of prey capture attempts were identified on all but one tag
(7/8; Table I). Buzzes tended to occur in series with a short
gap after the terminal buzz, and occurred at very high rates
on some recordings (up to 11 buzzes/min). Narwhals did not
appear to withhold from vocalizing immediately after tag-
ging, as has been found in other studies. Even when sam-
pling 50% of possible recordings (only the HP channel), we
found that narwhals used buzzes comparatively soon after
tagging [8.5 62.8 [standard error (SE)] hours; N¼7,
excluding one animal that did not buzz at all] compared to
22.7 65.1 h of post-release silence as found by Blackwell
et al. (2018;N¼6).
B. Summary of identified patterns
We identified eight plausible call types consistent with
our definition of paired patterns. We also included one pat-
tern of two stereotyped pulsed calls with an additional small
set of clicks preceding a clear pair of calls (type I; Fig. 3)
and one pattern for which both units were linked by a tonal
“bridge” (type VI; Fig. 3). Each of these ten patterns was
included as a template in the subsequent matching task. This
can be seen in Mm. 1.
Mm. 1. Combined sound file of ten templates. This is a file
of type “wav” (13.3 Mb).
1082 J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al.
Auralassessmentat0.1speed confirmed that the burst
pulse series consisted of sets of clicks interspersed with clearly
defined periods of silence. This can be seen in Mm. 2. Burst
pulse series were detected on nearly all tag recordings (7/8)
but were distributed unevenly. Sequences of at least three burst
pulse series with intervals of less than 10 s between each (e.g.,
Fig. 2) were identified on fewer tag recordings (3/8) but still
comprised approximately 40% of all burst pulse series
detected. Within these sequences, we also noticed that single
subunits matching the amplitude, frequency, and duration of
subunits in the more typical burst pulse series were sometimes
produced. These singletons, while rare (7% of all burst pulse
series), were also included in the analysis.
Mm. 2. Example of burst pulse series. This is a file of type
“wav” (3.6 Mb).
In total, we isolated 36 sequences comprising varying
numbers of burst pulse series (N
¼212). Aural assess-
ment suggested that PRR was relatively consistent within
and across the burst pulse series. Some sequences were com-
posed of series with different numbers and temporal organi-
zation of subunits. However, most sequences appeared to be
repetitive with small modifications such as the addition or
subtraction of a single set of clicks (Fig. 2). The length of
individual subunits in the burst pulse series often appeared
to increase over the course of the call, and some (approxi-
mately 14%) of the burst pulse series were initiated by an
especially short subunit (<50 ms).
Subunits of paired patterns [mean ¼515.8 640.8 (SE)
ms; N¼21] were significantly longer than those of the
burst pulse series [mean 104.9 61.5 (SE) ms; N¼900;
two-sample t-test; t(20.06)¼10.05, p<0.001], supporting
TABLE I. Details of female M. monoceros fit with Acousonde
acoustic tags in Tremblay Sound, Nunavut. Here, “pattern” refers to any plausible stereo-
typed patterns of pulsed calls as well as series of burst pulses. Note that “time to vocalize” was calculated using the LF channel as well, so may exceed the
duration of the HP channel recording. No buzzes were detected on the recording from NW14.
Narwhal ID Deployment date
Recording length
of HP channel
Time to vocalize (min) Calls detected
Call Buzz Tonal Pulsed Mixed Pattern
NW08 Aug 13, 2017 41 h 53 30 15 110 0 199
NW09 Aug 16, 2017 16 h 38 246 2 17 0 0
Aug 30, 2017 4 h <5 158 4 36 1 23
Sept 2, 2017 4 h <5 754 408 696 240 583
NW14 Sept 3, 2017 12h 10 N/A 45 91 8 46
Sept 11, 2017 11 h 34 870 12 99 5 10
Sept 11, 2017 8h 38 271 23 41 0 5
Sept 11, 2017 16 h 31 1231 13 119 27 32
Associated with a calf.
Captured together.
FIG. 3. Templates of paired pattern types used in the visual classification task. All templates are shown with identically calculated spectrograms, 80 kHz
maximum frequency, and 3 s duration (FFT length, 4096; window, Hann; overlap, 50%).
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1083
our decision to split them into distinct classes of vocal
C. Test of stereotyped organization of units
Six participants naive to the original order and context
of vocalizations consistently matched paired patterns to ten
templates, although interobserver agreement varied between
types (Table II). Quantitative classification was also suc-
cessful: using jackknifed cross-validation, the DFA classi-
fied paired patterns with 80.6% accuracy [Wilk’s K¼0.04,
X2 ¼205.5, DF (degrees of freedom) ¼21.0, p<0.001].
Here, classification accuracy was dependent on pattern type:
four classes had very high accuracies (I, 100%; II, 100%;
IV, 100%; VIII, 94%; IX, 88.9%), and three classes were
not accurately discriminated (III, 0%; V, 50%; VI, 0%).
We identified 29 sequences of paired patterns, consist-
ing of 72 pairs in total. Classifiers confirmed that each
sequence contained just a single type of A-B pair (Fig. 3).
We identified lone units matching the A units of two types
of paired patterns (II and IX). These were classified with
their expected types with 100% accuracy in the DFA,
confirming that individual units of the paired patterns are
sometimes produced alone. We detected no lone “B” units
for any patterns. Lone A units were sometimes rhythmically
repeated (Fig. 4) and often produced in close proximity to
full “pairs” (Fig. 5). When both units of a pattern were pro-
duced together, they were always produced in the same
order. We also noticed that the mean PRR of the B unit was
higher than that of the A unit for 9/10 types.
Measures of LD between burst pulse series were overdis-
persed (dispersion parameter 104.9), so we interpreted effect
coefficients from the quasi-Poisson GLM. Pairs of burst pulse
series from observed transitions were more similar than pairs
from randomly generated transitions (mean
¼150.7, mean LD
¼200.7, p<0.001).
Burst pulse series did not appear to change gradually across
the study period,
indicating that narwhals combine burst
pulse series into repetitive “sequences of series.”
D. Patterns of production across individuals
Frequency and amplitude were highly consistent
between units of paired patterns and across subsequent A-B
TABLE II. Summary statistics of Fleiss’ kappa test applied to multiple observer classification of patterns of M. monoceros paired patterns. Tag specificity
indicates the recording on which each pattern was identified with the number of A-B pairs in parentheses. Bolded tag identifications (IDs) designate calls
that were likely produced by the tagged animal. Reliability descriptors are taken from Landis and Koch (1977).
kz pReliability descriptor Tag specificity (N)
I 0.813 31.472 <0.001 Near-perfect NW08 (8)
II 0.921 35.683 <0.001 Near-perfect NW12 (7)
III 0.399 15.469 <0.001 Fair NW12 (5)
IV 0.757 29.316 <0.001 Substantial NW12 (5)
V 0.660 25.551 <0.001 Substantial NW12 (8)
VI 0.492 19.060 <0.001 Moderate NW12 (1), NW14 (5)
VII 0.533 20.633 <0.001 Moderate NW08 (4)
VIII 0.772 29.893 <0.001 Substantial NW08 (16), NW08 (1), NW11 (1)
IX 0.607 23.525 <0.001 Substantial NW11 (6), NW12 (2)
X 0.373 14.439 <0.001 Fair NW12 (3)
No match 0.321 12.447 <0.001 Fair (All)
All classes 0.584 63.4 <0.001 Substantial
FIG. 4. Spectrogram showing lone A units of a stereotyped pattern of calls repeated in rapid succession (type IX; FFT length, 4096; window, Hann; overlap,
50%). Classified visually (kappa statistic 0.607, p<0.001) and by DFA (88.9% classification for type IX).
1084 J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al.
pairs in the same sequence. There was one exception: a
sequence of sub-type VIII included one pair produced with
clicks of higher frequency than others in the sequence.
did not identify any instances of overlapping A and B units
of paired patterns, nor was there any overlap of repeated
pairs. Sequences of burst pulse series were generally pro-
duced with consistent frequencies and amplitudes, although
we detected two sequences with overlapping series.
When excluding calls attributable to non-focal individu-
als, 98% (45/46) of remaining paired patterns were identified
on a single recording only (bold sequences in Table II).
These tag-specific patterns appeared to be stable over our
recording durations: types I, II, VIII, and IX were identified
7.2, 10.2, 38.6, and 40.7 h apart within their respective
recordings. Although sequences of burst pulse series (as
defined above) were only detected on three tags, and only
two tags had more than one sequence, we detected tag-
specific differences in burst pulse series characteristics: series
recorded from NW08 had fewer subunits than series recorded
from NW12 (two-sample t-test; mean subunits
mean subunits
¼4.79; t(207) ¼6.1, p<0.001).
E. Behavioural context of sequence production
Relative to randomly selected samples, paired patterns
were 50% more likely, and burst pulse series 77% less
likely, respectively, to be followed by buzzes, although
these estimates were associated with very large standard
error and were not statistically significant (Table III). Depth
did not appear to influence the likelihood of detecting vocal
sequences, matching our observation that sequences were
produced at various positions in the water column, except
the deepest dives (i.e., below 300 m; Table III, Fig. 6).
A one-unit increase in log
-transformed counts of
vocalizations in a 60-s recording sample corresponded to a
330% increase in the probability of the sample being pre-
ceded (and 460% increase of being followed) by a burst
pulse series, suggesting they are associated with vocal
exchanges generally (Table III). In contrast, paired patterns
were significantly less likely to be produced in the context
of other calls (Fig. 7). Three sequences of burst pulse series
from one tag (NW12) were excluded from behavioural con-
text models as pressure sensors on the tag failed midway
through the deployment.
We have shown that narwhals produce at least two
kinds of vocal sequences. Paired patterns are composed of
distinct units combined into strictly ordered sets, providing
parallels to vocal sequences in avian and other mammalian
species. Burst pulse series, similar to vocalizations in other
odontocetes, are shown to be combined into repetitive
sequences of series.
FIG. 5. Spectrogram showing a stereotyped pattern of calls followed by two additional lone A units (type II; FFT length, 4096; window, Hann; overlap,
50%). Classified visually (kappa statistic 0.921, p<0.001) and by DFA (100% classification).
TABLE III. Parameter estimates for GLMs relating the occurrence of
sequences of paired patterns (n¼29) or burst pulse series (n¼21) produced
by M. monoceros to the behavioural context preceding and following the
call. Call occurrences were supplemented with 60 randomly selected 60-s
samples to provide pseudo-absences.
Estimate SE zp
Paired patterns
Previous activity model
Intercept 0.304 0.488 0.624 0.532
Calls 0.865 0.393 2.202 0.028
Depth 0.005 0.351 0.014 0.989
Feeding buzzes 0.489 0.688 0.711 0.477
Subsequent activity model
Intercept 0.550 0.493 1.116 0.265
Calls 0.476 0.388 1.228 0.219
Depth 0.205 0.327 0.626 0.531
Feeding buzzes 0.407 0.574 0.709 0.478
Burst pulse series
Previous activity model
Intercept 1.790 0.679 2.637 0.008
Calls 1.231 0.499 2.469 0.014
Depth 0.484 0.496 0.975 0.329
Feeding buzzes 0.488 0.912 0.535 0.593
Subsequent activity model
Intercept 2.281 0.819 2.786 0.005
Calls 1.526 0.533 2.863 0.004
Depth 0.835 0.495 1.688 0.091
Feeding buzzes 1.472 0.868 1.697 0.090
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1085
Consensus between the visual classification task and the
DFA for most types of paired patterns demonstrates that
they are stereotyped and rhythmically repeated. The finding
that the first units of two types (II, IX) were also produced
alone provides further support for the interpretation of these
as combinations of calls, rather than single calls with gaps
of silence (Kershenbaum et al., 2016). We are unable to
conclude with certainty that other types are also composed
of divisible units as we did not identify any candidate cases.
However, this may have largely been an artefact of the rarity
FIG. 6. Distributions of depths registered on the tags (left) compared to depths at which sequences of paired patterns (middle) and burst pulse series (right)
were produced.
FIG. 7. Marginal effects of other vocalizations preceding the occurrence of sequences of pulsed vocalizations. Effects were estimated using binomial family
1086 J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al.
of certain types as the two patterns for which we identified
lone units were relatively common (Table I). To our knowl-
edge, there are only two examples of ordered sequences in
odontocetes: the “N7-N8” calls of Northern resident killer
whales and the bray sequences of bottlenose dolphins (Ford,
ıs et al., 2018). Structurally similar combinations
of vocalizations are found more commonly in other taxa,
where they have been linked to communicative complexity.
For example, meerkats (Suricata suricatta) produce similar
“di-drrr” calls composed of two units, the pair of which are
rhythmically repeated, and interpreted as evidence for hier-
archical communication (Collier et al., 2014). Putty-nosed
monkeys (Cercopithecus nictitans) produce “pyow” and
“hack” calls, which are regularly combined into ordered
sequences but are also sometimes produced alone. Playback
experiments have demonstrated that the ordering of these
calls conveys distinct meaning and is considered a rare
example of syntax in a wild organism (Arnold and
uhler, 2008;Kershenbaum et al., 2016). As such, the
finding that B units of paired patterns were only ever found
following A units is noteworthy as it is indicative of a
“finite-state” grammar similar to those documented in pri-
mate and avian taxa (Berwick et al., 2011;Fitch and
Hauser, 2009;Shettleworth, 2010). While inflexible A-B
grammars are fundamentally different from the recursive
grammatical structures that support human language
(Berwick et al., 2011;Fitch and Hauser, 2009), they can be
important vehicles for the transfer of information (Bradbury
and Vehrencamp, 1998, p. 395). Of course, it remains to be
tested whether the rule-like structure of narwhal paired pat-
terns supports compositional communication, i.e., whether
A and B units convey distinct information when combined
uhler, 2018).
Whereas burst pulse series are varied (e.g., in number
of subunits), narwhals combine similar series into repetitive
sequences. Vocalizations similar to burst pulse series have
been detected in recordings of northern right whale dolphins
(Lissodelphis borealis;Rankin et al., 2007), dusky dolphins
(Lagenorhynchus obscurus;Vaughn-Hirshorn et al., 2012),
pacific white-sided dolphins (Lagenorhynchus obliquidens;
Henderson et al., 2011), and Heaviside’s dolphins
(Cephalorhynchus heavisidii;Martin et al., 2019), although
there are nontrivial differences between species. For exam-
ple, the “patterned burst pulses” in northern right whale dol-
phins are the only other series that appear to be rhythmically
repeated, and the “burst pulse sequences” produced by
dusky dolphins rarely had more than two subunits. Calves of
Araguaian river dolphins (Inia araguaiaensis) produce calls
similar to the narwhals’ burst pulse series, although it
remains to be tested whether they are combined into sequen-
ces of series (Melo-Santos et al., 2019).
We are unable to rule out the possibility that sequences
were sometimes produced by multiple individuals in coordi-
nated call exchanges, which has important implications for
the interpretation of communicative complexity (as dis-
cussed above). However, several findings were inconsistent
with this explanation. First, the strong directionality of
narwhal clicks makes it unlikely that multiple individuals
would be able to consistently produce clicks resulting in
near-identical frequency ranges on an animal-attached tag
(Koblitz et al., 2016). Second, overlap between burst pulse
series was very rare and was not identified in paired pat-
terns, supporting the interpretation that they were produced
by a single individual (Sayigh et al., 2013). If these sequen-
ces were the result of call exchanges, they would imply a
high degree of temporal synchrony, akin to duetting or even
“turn taking,” given the highly stereotyped time intervals
between units (Pika et al., 2018). Therefore the most parsi-
monious interpretation is that sequences of paired patterns
were produced by single individuals. The same can be con-
cluded for sequences of burst pulse series, whereas the
detection of some cases of overlap suggests that they may
also be produced in call exchanges.
If sequences are typically produced by a single individ-
ual, the variation in sequence characteristics between tags
opens the possibility that they support individual or group
recognition, as has been suggested for other narwhal vocal-
izations (Shapiro, 2006). For example, call types associated
with specific tag recordings have been taken as preliminary
evidence for individual-specific signals in short-finned pilot
whales (G. macrorhynchus;Quick et al., 2018). We identi-
fied possible individual specificity in the paired patterns,
which were re-produced on the same tags up to 40 h apart.
However, two considerations suggest that these patterns
should not be interpreted as “signature sequences.” First, we
would expect a signature signal to be the predominant vocal-
ization in an individual’s repertoire (Cook et al., 2004),
whereas paired patterns were relatively rare. Second, for
species with large vocal repertoires, a subsampling should
result in recording-specific differences, even if repertoire
composition is identical across individuals, making us cau-
tious to label these as individual specific without further
inquiry. We were surprised to find that burst pulse series
recorded from two different individuals had different and
continuous (i.e., not obviously incomplete or under-sam-
pled) distributions of numbers of subunits. Nevertheless,
these patterns seem unlikely to disclose individual identity,
given the wide range of numbers of subunits used.
In addition to the consideration of signal properties,
linking acoustic cues to behaviour can help to provide
insights into their function (Papale et al., 2017). Burst pulse
series were more likely to be produced in contexts of high
vocal activity. This matches well with theory that graded
signals are primarily used in social contexts where animals
are in close proximity (Ford, 1988). We detected the oppo-
site effect for paired patterns. These calls were often pro-
duced in very quiet periods. Accordingly, it may be that the
highly stereotyped paired patterns may serve a role in
longer-distance communication. Neither paired patterns nor
burst pulse series were significantly related to prey-capture
attempts, and were not recorded when individual narwhals
were below 300 m (Fig. 6), suggesting that they are not
related to feeding, as are the “bray sequences” of bottlenose
dolphins (Janik, 2000;King and Janik, 2015). Combining
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1087
our findings with visual observations (e.g., from shore or
uncrewed aerial vehicle) might reveal further associations
between context and sequence use. For example, call
sequences may facilitate contact between mothers and
calves (Smolker et al., 1993), like the “type A” pulsed
vocalizations of belugas (Vergara et al., 2010). Female nar-
whals give birth in midsummer (Furgal and Laing, 2012;
Heide-Jørgensen, 2009), and paired patterns were recorded
on all tags attached to narwhals observed with calves during
capture (Table I).
Both paired patterns and burst pulse series are charac-
terized by rhythmic repetition. While the energetic cost of
vocalizing underwater may not be a significant limitation
(Jensen et al., 2012), the risk of detection by eavesdropping
acoustic predators (e.g., killer whales) is likely important for
narwhals (Deecke et al., 2002;Breed et al., 2017;Furgal
and Laing, 2012;Laidre et al., 2006), making the use of
repeated signals surprising. One benefit of repetitive signal-
ing is to increase the likelihood of successful transmission,
perhaps especially important in the icy, reverberation-prone
environments inhabited by narwhals (Brumm and Slater,
2006;Ey and Fisher, 2009;Vergara et al., 2010). For exam-
ple, links between noise and signal redundancy have been
identified in blue whales (Balaenoptera musculus;Miller
et al., 2000) and killer whales (Foote et al., 2004).
Alternatively, the repetitive nature of these sequences may
suggest that call production is sustained until a response is
achieved in a target animal or the number of repetitions
itself encodes specific information (Payne and Pagel, 1997;
Janik and Sayigh, 2013;Sloan and Hare, 2004). For exam-
ple, the number of “dee” units in call sequences of black-
capped chickadees (Poecile atricapilla) is inversely propor-
tional to the size of an encroaching predator (Templeton
et al., 2005). Encoding information in the temporal structure
of a signal is also consistent with the optimization of trans-
mission success in noisy habitats; these features should
degrade less severely than other acoustic properties over
long distances (Bradbury and Vehrencamp, 1998, p. 129).
This reasoning is proposed for the use of temporally stereo-
typed vocalizations in white-eyelid mangabeys (Cercocebus
sp.; Waser, 1982, p. 134), sperm whales (Gero et al., 2016),
and forest-dwelling chingolo sparrows (Zonotrichia capen-
sis;Bradbury and Vehrencamp, 1998, p. 136).
We conclude that narwhals produce at least two kinds
of vocal sequences. The paired patterns appear to be one of
few ordered, multiunit sequences in odontocetes, and to our
knowledge, the first in the Monodontidae. Whereas we pro-
pose that they may play a role in long-distance communica-
tion, their specific function remains unknown. The use of
repetition in successive burst pulse series suggests that their
signaling benefit outweighs the costs associated with redun-
dant signaling. Further inquiry into the function and
population-distribution of sequence use (e.g., whether they
are produced by narwhals in East Greenland and Western
Hudson Bay) could lead to insights regarding the ecological,
genetic, and cultural development of narwhal communica-
tion (Garland et al., 2011;Janik, 2000;May-Collado and
Wartzok, 2008;Rendell and Whitehead, 2001). Together,
our findings provide a small contribution to the understand-
ing of the phylogenetic distribution of vocal sequence pro-
duction and suggest that methodological limitations may
explain the apparent paucity of vocal sequences in other
odontocetes. We hope that future inquiry will continue to
elucidate the function(s) of these sequences in narwhals and
widen the set of species in which sequence use is tested, ulti-
mately supporting comparative studies to better understand
evolution of vocal communication.
We thank Patrick Miller and Katie Harris for providing
helpful comments on early iterations of this work. We also
thank Mikhail Barabanov, Vanessa Simons, Rowan Gardner,
Nick Jones, and Diana Mirella Pab
on Figueroa for
participating in the visual classification task. Funding for
S.F.W. was provided by an R and A Ransome Scholarship.
Data used in this analysis were collected by the Ecosystem
Approach to Tremblay Sound project led by the Department of
Fisheries and Oceans, Canada. We thank Robert Hodgson and
all participants involved in this project and the associated
funding bodies, including the Nunavut Wildlife Management
Board, the World Wildlife Fund, Golder Associates, and
Ocean Wise. We are grateful to the Mittimatalik Hunters &
Trappers Organization and the Mittimatalik Inuit Community
for their support and involvement in this project. We also
thank the Polar Continental Shelf Program, the Government of
Nunavut, Parks Canada, the Ocean Tracking Network,
a Rimouski, the University of
Windsor, the University of Manitoba, the University of
Calgary, l’Universit
e de Montr
eal, the University of British
Columbia, and the University of New Hampshire for their
partnership. Finally, we thank two anonymous reviewers
whose comments improved the manuscript.
Visual classification has been shown to be an effective
method for the classification for vocal sequences (Janik and
Sayigh, 2013;Kershenbaum et al., 2016). While spectro-
graphic visualizations of pulsed signals are highly contin-
gent on FFT and time dimensions used, differences in click
rate (the expected form of stereotypy for paired patterns)
should be consistently distinguishable as long as spectro-
gram parameters are kept constant.
One randomly selected pair from each of the paired pat-
tern types we defined served as a template (Fig. 3). We then
printed 100 patterns of pulsed calls to be matched to these
templates. This number of patterns was chosen as a balance
between including adequate variation to assess patterns
across tags, while not being overly cumbersome for human
classifiers, which could result in reduced performance
(Rendell and Whitehead, 2003). These included the
1088 J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al.
remaining paired patterns from the original sequences, as
well as possible pair-like patterns identified elsewhere in the
classification procedure, but which were not rhythmically
repeated. Given the propensity of narwhals to aggregate in
groups where many (and often overlapping) vocalizations
are heard, this test sample was likely to include pulsed calls
that were produced close together in time by chance, i.e.,
“null” patterns.
Six participants matched call patterns to one of ten pos-
sible templates or indicated that no match was found.
Participants had varying degrees of expertise in bioacoustics
(range 1–20 yr), although none had previous experience
classifying narwhal vocalizations. Participants were
instructed to ignore any background noise or differences in
frequency range for the classification task. Index numbers
for calls were randomized prior to the task.
To improve our ability to make inferences regarding the
individual specificity of call types, we devised a threshold
for “unusual” low-sound frequency based on previous
reports of narwhal click frequency ranges. Estimates of the
peak frequency of narwhal clicks using standard hydro-
phones include 19 kHz (Miller et al., 1995) and 12–20 kHz
(Marcoux et al., 2012). Others have reported that narwhals
produce clicks at variable frequency ranges, the lowest hav-
ing a peak frequency of 3.5–5 kHz and no energy below
3 kHz [Fig. 2(a) in Stafford et al., 2012]. In contrast, record-
ings of tagged narwhals contain clicks with substantial
energy below these frequencies (Shapiro, 2006;Blackwell
et al., 2018). As such, pulsed patterns with clearly visible
energy below 3 kHz were identified as possibly focal.
On average, clicks produced by non-focal individuals
should have lower amplitudes than those plausibly produced
by the tagged animal, acknowledging that non-focal vocaliza-
tions may sometimes produce higher received levels
(Johnson et al.,2009). We used a two-sample t-testtotestthe
prediction that pulsed vocalizations with visible LF energy
(<3 kHz) should have greater root mean square (RMS)
amplitudes, using the first unit of each paired pattern. A sig-
nificant difference would provide further support for the
hypothesis that click-based sounds lacking energy below
3 kHz are produced by non-focal whales. Units were filtered
with a 3 kHz high-pass filter to remove the confounding
effects of the LF energy, itself, influencing amplitude. As pre-
dicted, we found that pulsed vocalizations with energy below
3 kHz had higher amplitudes than calls lacking energy below
3 kHz (two-sample t-test; mean amplitude
¼156.8 dB
RMS, mean amplitude
¼142.4 dB RMS, t(108)
¼9.57, p<0.001), even when energies below 3 kHz were fil-
tered out, supporting the hypothesis that calls lacking LF
energy were produced by non-focal whales.
That being said, it is thought that high-amplitude pulsed
vocalizations produced by nearby, non-focal individuals can
also result in additional LF energy on a tag recording
(Blackwell et al., 2018). As in-depth consideration of the
conduction and propagation of LF energy was beyond the
scope of this analysis, we assumed that our discrimination
process should allow for an increased (but not total) ability
to relate sequences of calls to specific individuals.
See¼tagwiki:tools:calibration (Last
viewed 1 June 2018).
See (Last viewed 31 January 2020).
See supplementary material at for dis-
tributions of subunit durations; boxplots of all paired pattern units show-
ing distributions of duration, number of clicks, and mean PRR; the plot of
dissimilarity between burst pulse series as a function of temporal spacing;
and spectral density curves.
Abramson, J. Z., Hern
andez-Lloreda, M. V., Garc
ıa, L., Colmenares, F.,
Aboitiz, F., and Call, J. (2018). “Imitation of novel conspecific and human
speech sounds in the killer whale (Orcinus orca),” Proc. R. Soc. B Biol.
Sci. 285, 20172171.
Andrews, R. D., Pitman, R. L., and Ballance, L. T. (2008). “Satellite track-
ing reveals distinct movement patterns for Type B and Type C killer
whales in the southern Ross Sea, Antarctica,” Polar Biol. 31, 1461–1468.
Arnold, K., and Zuberb
uhler, K. (2008). “Meaningful call combinations in
a non-human primate,” Curr. Biol. 18, 202–203.
Arranz, P., DeRuiter, S. L., Stimpert, A. K., Neves, S., Friedlaender, A. S.,
Goldbogen, J. A., Visser, F., Calambokidis, J., Southall, B. L., and Tyack,
P. L. (2016). “Discrimination of fast click-series produced by tagged
Risso’s dolphins (Grampus griseus) for echolocation or communication,”
J. Exp. Biol. 219, 2898–2907.
Berwick, R. C., Okanoya, K., Beckers, G. J. L., and Bolhuis, J. J. (2011).
“Songs to syntax: The linguistics of birdsong,” Trends Cogn. Sci. 15,
Blackwell, S. B., Tervo, O. M., Conrad, A. S., Sinding, M. H. S., Hansen,
R. G., Ditlevsen, S., and Peter, M. (2018). “Spatial and temporal patterns
of sound production in East Greenland narwhals,” PLoS One 13, 1–23.
Bradbury, J. W., and Vehrencamp, S. L. (1998). “Sound propagation,” in
Principles of Animal Communication (Sinauer Associates, Sunderland,
MA), pp. 113–139.
Breed, G. A., Matthews, C. J. D., Marcoux, M., Higdon, J. W., LeBlanc, B.,
Petersen, S. D., Orr, J., Reinhart, N. R., and Ferguson, S. H. (2017).
“Sustained disruption of narwhal habitat use and behavior in the presence
of Arctic killer whales,” Proceedings of the National Academy of
Sciences 114(10), 2628–2633..
Brumm, H., and Slater, P. J. B. (2006). “Ambient noise, motor fatigue, and
serial redundancy in chaffinch song,” Behav. Ecol. Sociobiol. 60,
Collier, K., Bickel, B., van Schaik, C. P., Manser, M. B., and Townsend, S.
W. (2014). “Language evolution: Syntax before phonology?,” Proc. R.
Soc. B 281, 20140263.
Cook, M. L. H., Sayigh, L. S., Blum, J. E., and Wells, R. S. (2004).
“Signature-whistle production in undisturbed free-ranging bottlenose dol-
phins (Tursiops truncatus),” Proceedings of the Royal Society of London.
Series B: Biological Sciences 271, 1043–1049.
Deecke, V. B., Slater, P. J. B., and Ford, J. K. B. (2002). “Selective habitua-
tion shapes acoustic predator recognition in harbour seals,” Nature 729,
Esch, H. C., Sayigh, L. S., and Wells, R. S. (2009). “Quantifying parameters
of bottlenose dolphin signature whistles,” Mar. Mammal Sci. 25,
Ey, E., and Fischer, J. (2009). “The ‘acoustic adaptation hypothesis’—a
review of the evidence from birds, anurans and mammals,” Bioacoustics
19, 21–48.
Ferrer-i-Cancho, R., and McCowan, B. (2012). “The span of correlations in
dolphin whistle sequences,” J. Stat. Mech. Theory Exp. 2012, P06002.
Fitch, W. T., and Hauser, M. D. (2009). “Computational constraints on syn-
tactic processing in a nonhuman primate,” Science 377, 377–380.
Foote, A. D., Osborne, R. W., and Hoelzel, A. R. (2004). “Whale-call
response to masking boat noise,” Nature 428, 910.
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1089
Ford, J. K. (1988). “Acoustic behaviour of resident killer whales (Orcinus
orca) off Vancouver Island, British Columbia,” Can. J. Zool. 67,
Ford, J. K. B., and Fisher, H. D. (1978). “Underwater acoustic signals of the
narwhal (Monodon monoceros),” Can. J. Zool. 56(4), 552–560.
Furgal, C., and Laing, R. (2012). “A synthesis and critical review of the tra-
ditional ecological knowledge literature on narwhal (Monodon mono-
ceros) in the eastern Canadian Arctic fish,” Ocean. Canada, Can. Sci.
Advis. Secr., Fisheries and Oceans Canada, Canadian science advisory
secretariat, Vol. 3848, 47 pp.
Gamer, M., Lemon, J., Fellows, I., and Singh, P. (2015). “Package irr:
Various coefficients of interrater reliability and agreement,” a package for
R software (R Foundation for Statistical Computing, Vienna, Austria)
Garland, E. C., Goldizen, A. W., Rekdahl, M. L., Constantine, R., Garrigue,
C., Hauser, N. D., Poole, M. M., Robbins, J., and Noad, M. J. (2011).
“Dynamic horizontal cultural transmission of humpback whale song at
the ocean basin scale,” Curr. Biol. 21, 687–691.
Garland, E. C., Lilley, M. S., Goldizen, A. W., Rekdahl, M. L., Garrigue,
C., and Noad, M. J. (2012). “Improved versions of the Levenshtein dis-
tance method for comparing sequence information in animals’ vocalisa-
tions: Tests using humpback whale song,” Behaviour 149, 1413–1441.
Gero, S., Whitehead, H., and Rendell, L. (2016). “Individual, unit and vocal
clan level identity cues in sperm whale codas,” R. Soc. Open Sci. 3,
Gonzalez, N. (2001). Inuit Traditional Ecological Knowledge of the
Hudson Bay Narwhal (Tuugaalik) Population, Fisheries and Oceans
Canada. Iqaluit, Nunavut, Canada, pp. 1–26.
Heiberger, R. M. (2018). HH: Statistical Analysis and Data Display:
Heiberger and Holland, a package for R software,
Heide-Jørgensen, M. P. (2009). “Monodon monoceros,” in Encyclopedia of
marine mammals, 2nd ed., edited by W. F. Perrin, B. W
ursig, and H.
Thewissen (Academic Press, San Diego, California), pp. 754–758.
Henderson, E. E., Hildebrand, J. A., and Smith, M. H. (2011).
“Classification of behavior using vocalizations of Pacific white-sided dol-
phins (Lagenorhynchus obliquidens),” J. Acoust. Soc. Am. 130, 557–567.
Herman, L. M. (2006). “Intelligence and rational behaviour in the bottle-
nosed dolphin,” in Rational animals?, edited by S. Hurley and M. Nudds
(Oxford University Press, Oxford, UK), pp. 439–467.
Janik, V. M. (2000). “Food-related bray calls in wild bottlenose dolphins
(Tursiops truncatus),” Proc. Biol. Sci. 267, 923–927.
Janik, V. M. (2009). “Acoustic communication in delphinids,” Adv. Study
Behav. 40, 123–157.
Janik, V. M. (2013). “Cognitive skills in bottlenose dolphin
communication,” Trends Cogn. Sci. 17, 157–159.
Janik, V. M., and Sayigh, L. S. (2013). “Communication in bottlenose dol-
phins: 50 years of signature whistle research,” J. Comp. Physiol. A
Neuroethol. Sens. Neural Behav. Physiol. 199, 479–489.
Janik, V. M., and Slater, P. J. B. (1998). “Context-specific use suggests that
bottlenose dolphin signature whistles are cohesion calls,” Anim. Behav.
56, 829–838.
Jensen, F. H., Beedholm, K., Wahlberg, M., Bejder, L., and Madsen, P. T.
(2012). “Estimated communication range and energetic cost of bottlenose
dolphin whistles in a tropical habitat,” J. Acoust. Soc. Am. 131, 582–592.
Johnson, M., De Soto, N. A., and Madsen, P. T. (2009). “Studying the
behaviour and sensory ecology of marine mammals using acoustic record-
ing tags: A review,” Mar. Ecol. Prog. Ser. 395, 55–73.
Johnson, M., Madsen, P. T., Zimmer, W. M. X., de Soto, N. A., and Tyack,
P. L. (2006). “Foraging Blainville’s beaked whales (Mesoplodon densir-
ostris) produce distinct click types matched to different phases of
echolocation,” J. Exp. Biol. 209, 5038–5050.
Kaplan, M. B., Aran Mooney, T., Sayigh, L. S., and Baird, R. W. (2014).
“Repeated call types in Hawaiian melon-headed whales (Peponocephala
electra),” J. Acoust. Soc. Am. 136, 1394–1401.
Kershenbaum, A., Blumstein, D. T., Roch, M. A., Akc¸ay, C¸., Backus, G.,
Bee, M. A., Bohn, K., Cao, Y., Carter, G., C
asar, C., Coen, M., Deruiter,
S. L., Doyle, L., Edelman, S., Ferrer-i-Cancho, R., Freeberg, T. M.,
Garland, E. C., Gustison, M., Harley, H. E., Huetz, C., Hughes, M.,
Hyland Bruno, J., Ilany, A., Jin, D. Z., Johnson, M., Ju, C., Karnowski, J.,
Lohr, B., Manser, M. B., McCowan, B., Mercado, E., Narins, P. M., Piel,
A., Rice, M., Salmi, R., Sasahara, K., Sayigh, L., Shiu, Y., Taylor, C.,
Vallejo, E. E., Waller, S., and Zamora-Gutierrez, V. (2016). “Acoustic
sequences in non-human animals: A tutorial review and prospectus,”
Biol. Rev. 91, 13–52.
Kershenbaum, A., Bowles, A. E., Freeberg, T. M., Jin, D. Z., Lameira, A.
R., and Bohn, K. (2014). “Animal vocal sequences: Not the Markov
chains we thought they were,” Proc. R. Soc. B Biol. Sci. 281(1792),
Kershenbaum, A., and Garland, E. C. (2015). “Quantifying similarity in ani-
mal vocal sequences: Which metric performs best?,” Methods Ecol. Evol.
6, 1452–1461.
Kershenbaum, A., Ilany, A., Blaustein, L., and Geffen, E. (2012).
“Syntactic structure and geographical dialects in the songs of male rock
hyraxes,” Proc. R. Soc. B Biol. Sci. 279, 2974–2981.
King, S. L., and Janik, V. M. (2015). “Come dine with me: Food-associated
social signalling in wild bottlenose dolphins (Tursiops truncatus),” Anim.
Cogn. 18, 969–974.
Koblitz, J. C., Stilz, P., Rasmussen, M. H., and Laidre, K. L. (2016).
“Highly directional sonar beam of narwhals (Monodon monoceros) mea-
sured with a vertical 16 hydrophone array,” PLoS One 11(11), e0162069.
Krebs, J. R., Aschcroft, R., and Orsdol, K. V. (1981). “Song matching in
the great tit Parus major,” Anim. Behav. 29, 918–923.
Laidre, K. L., and Heide-Jørgensen, M. P. (2011). “Life in the lead:
Extreme densities of narwhals Monodon monoceros in the offshore pack
ice,” Mar. Ecol. Prog. Ser. 423, 269–278.
Laidre, K. L., Heide-Jørgensen, M. P., and Orr, J. R. (2006). “Reactions of
narwhals, Monodon monoceros, to killer whale, Orcinus orca, attacks in
the eastern Canadian Arctic,” Can. Field-Naturalist 120, 457–465.
Lammers, M. I., and Oswald, J. I. (2015). “Analyzing the acoustic commu-
nication of dolphins,” in Dolphin communication and cognition: Past,
present, and future, edited by D. L. Herzing and C. M. Johnson (MIT
Press, Cambridge, MA).
Lammers, M. O., Au, W. W. L., Aubauer, R., and Nachtigall, P. E. (2004).
“A comparative analysis of the pulsed emissions of free-ranging
Hawaiian spinner dolphins,” in Echolocation Bats Dolphins, edited by J.
Thomas, C. Moss, and M. Vater (University of Chicago Press, Chicago,
IL), pp. 414–419.
Landis, J. R., and Koch, G. G. (1977). “The measurement of observer agree-
ment for categorical data,” Biometrics 33, 159–174.
ıs, A. R., Alves, I. S., Sobreira, F. V., Couchinho, M. N., and dos Santos,
M. E. (2018). “Brays and bits: Information theory applied to acoustic com-
munication sequences of bottlenose dolphins,” Bioacoustics 4622, 1–11.
Mann, J. (2000). “Unraveling the dynamics of social life,” in Cetacean
Societies: Field Studies of Dolphins and Whales, edited by J. Mann, R. C.
Connor, P. L. Tyack, and H. Whitehead (University of Chicago Press,
London), pp. 45–64.
Mann, N. I., Dingess, K. A., and Slater, P. J. B. (2006). “Antiphonal four-
part synchronized chorusing in a neotropical wren,” Biol. Lett. 2, 1–4.
Marcoux, M., Auger-M
e, M., and Humphries, M. M. (2012).
“Variability and context specificity of narwhal (Monodon monoceros)
whistles and pulsed calls,” Mar. Mammal Sci. 28, 649–665.
Martin, M. J., Elwen, S. H., Kassanjee, R., and Gridley, T. (2019). “To
buzz or burst-pulse? The functional role of Heaviside’s dolphin,
Cephalorhynchus heavisidii, rapidly pulsed signals,” Anim. Behav. 150,
May-Collado, L. J., and Wartzok, D. (2008). “A comparison of bottlenose
dolphin whistles in the Atlantic Ocean: Factors promoting whistle var-
iation,” J. Mammal. 89, 1229–1240.
McDonald, M. A., Mesnick, S. L., and Hildebrand, J. A. (2006).
“Biogeographic characterisation of blue whale song worldwide: Using
song to identify populations,” J. Cetacean Res. Manag. 8, 55–65.
Melo-Santos, G., Figueiredo Rodrigues, A. L., Tardin, R. H., de S
a Maciel,
I., Marmontel, M., Da Silva, M. L., and May-Collado, L. J. (2019). “The
newly described Araguaian river dolphins, Inia araguaiaensis
(Cetartiodactyla, Iniidae), produce a diverse repertoire of acoustic sig-
nals,” PeerJ 7, e6670.
Miller, L. A., Pristed, J., Møshl, B., and Surlykke, A. (1995). “The click-
sounds of narwhals (Monodon monoceros) in Inglefield Bay, Northwest
Greenland,” Marine Mammal Science 11(4), 491–502.
Miller, P. J. O., Biassoni, N., Samuels, A., and Tyack, P. L. (2000). “Whale
songs lengthen in response to sonar,” Nature 405, 903.
Miller, P. J. O., Shapiro, A. D., Tyack, P. L., and Solow, A. R. (2004).
“Call-type matching in vocal exchanges of free-ranging resident killer
whales, Orcinus orca,” Anim. Behav. 67, 1099–1107.
1090 J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al.
Morisaka, T. (2012). “Evolution of communication sounds in odontocetes:
A review,” Int. J. Comp. Psychol. 25, 1–20.
Morisaka, T., Yoshida, Y., Akune, Y., Mishima, H., and Nishimoto, S.
(2013). “Exchange of ‘signature’ calls in captive belugas (Delphinapterus
leucas),” J. Ethol. 31, 141–149.
Nowicki, S., and Searcy, W. A. (2004). “Song function and the evolution of
female preferences: Why birds sing, why brains matter,” Ann. N. Y.
Acad. Sci. 1016, 704–723.
Ouattara, K., Lemasson, A., and Zuberbuhler, K. (2009). “Campbell’s mon-
keys concatenate vocalizations into context-specific call sequences,”
Proceedings of the National Academy of Sciences 106(51), 22026–22031.
Papale, E., Perez-Gil, M., Castrillon, J., Perez-Gil, E., Ruiz, L., Servidio,
A., Tejedor, M., Giacoma, C., and Mart
ın, V. (2017). “Context specificity
of Atlantic spotted dolphin acoustic signals in the Canary Islands,” Ethol.
Ecol. Evol. 29, 311–329.
Payne, R. J. H., and Pagel, M. (1997). “Why do animals repeat displays?,”
Anim. Behav. 54, 109–119.
Payne, R. S., and McVay, S. (1971). “Songs of humpback whales,” Science
173, 587–597.
Petroni, F., and Serva, M. (2010). “Measures of lexical distance between
languages,” Physica A: Statistical Mechanics and its Applications 389,
Pika, S., Wilkinson, R., Kendrick, K. H., and Vernes, S. C. (2018). “Taking
turns: Bridging the gap between human and animal communication,”
Proc. R. Soc. B Biol. Sci. 285, 20180598.
Podos, J. (1996). “Motor constraints on vocal development in a songbird,”
Anim. Behav. 51, 1061–1070.
Quick, N., Callahan, H., and Read, A. J. (2018). “Two-component calls in
short-finned pilot whales (Globicephala macrorhynchus),” Mar. Mammal
Sci. 34, 155–168.
R Core Team (2014). “R: A language and environment for statistical
computing,” R Foundation for Statistical Computing, Vienna, Austria.
Rankin, S., Oswald, J., Barlow, J., and Lammers, M. (2007). “Patterned
burst-pulse vocalizations of the northern right whale dolphin,
Lissodelphis borealis,” J. Acoust. Soc. Am. 121, 1213–1218.
Rasmussen, M. H., Koblitz, J. C., and Laidre, K. L. (2015). “Buzzes and
high-frequency clicks recorded from narwhals (Monodon monoceros)at
their wintering ground,” Aquat. Mamm. 41, 256–264.
Raven Pro. (2007). “Raven Pro: Interactive Sound Analysis Software,”
(Cornell Lab of Ornithology, Ithaca, NY)
Rendell, L., and Whitehead, H. (2001). “Culture in whales and dolphins,”
Behav. Brain Sci. 24, 309–382.
Rendell, L. E., and Whitehead, H. (2003). “Comparing repertoires of sperm
whale codas: A multiple methods approach,” Bioacoustics 14, 61–81.
Riesch, R., Ford, J. K. B., and Thomsen, F. (2006). “Stability and group
specificity of stereotyped whistles in resident killer whales, Orcinus orca,
off British Columbia,” Anim. Behav. 71, 79–91.
Saulitis, E. L., Matkin, C. O., and Fay, F. H. (2005). “Vocal repertoire and
acoustic behavior of the isolated AT1 killer whale subpopulation in south-
ern Alaska,” Can. J. Zool. 83, 1015–1029.
Sayigh, L., Quick, N., Hastie, G., and Tyack, P. (2013). “Repeated call
types in short-finned pilot whales, Globicephala macrorhynchus,” Mar.
Mammal Sci. 29, 312–324.
Scott-Phillips, T. C., Gurney, J., Ivens, A., Diggle, S. P., and Popat, R.
(2014). “Combinatorial communication in bacteria: Implications for the
origins of linguistic generativity,” PLoS One 9, e95929.
Shapiro, A. D. (2006). “Preliminary evidence for signature vocalizations
among free-ranging narwhals (Monodon monoceros),” J. Acoust. Soc.
Am. 1695, 1695–1705.
Shettleworth, S. J. (2010). Cognition, Evolution, and Behaviour, 2nd ed.
(Oxford University Press, New York).
Sloan,J.L.,andHare,J.F.(2004). “Monotony and the information con-
tent of Richardson’s ground squirrel (Spermophilus richardsonii)
repeated calls: Tonic communication or signal certainty?,Ethology
110, 147–156.
Smolker, R. A., Mann, J., and Smuts, B. B. (1993). “Use of signature whis-
tles during separations and reunions by wild bottlenose dolphin mothers
and infants,” Behav. Ecol. Sociobiol. 33, 393–402.
Stafford, K. M., Laidre, K. L., and Heide-Jørgensen, M. P. (2012). “First
acoustic recordings of narwhals (Monodon monoceros) in winter,” Mar.
Mammal Sci. 28, E197–E207.
Templeton, C. N., Greene, E., and Davis, K. (2005). “Allometry of alarm
calls: Black-capped chickadees encode information about predator size,”
Science 308, 1934–1938.
Vaughn-Hirshorn, R. L., Hodge, K. B., W
ursig, B., Sappenfield, R. H.,
Lammers, M. O., and Dudzinski, K. M. (2012). “Characterizing dusky
dolphin sounds from Argentina and New Zealand,” J. Acoust. Soc. Am.
132, 498–506.
Vergara, V., Michaud, R., and Barrett-Lennard, L. G. (2010). “What can
captive whales tell us about their wild counterparts? Identification, usage,
and ontogeny of contact calls in belugas (Delphinapterus leucas),” Int. J.
Comp. Psychol. 23, 278–309.
Waser, P. M. (1982). “The evolution of male loud calls among mangabeys
and baboons,” in Primate Communication, edited by C. T. Snowdon, C.
H. Brown, and M. R. Petersen (Cambridge University Press, New York),
pp. 117–143.
Watkins, W. A., and Schevill, W. E. (1977). “Sperm whale codas,”
J. Acoust. Soc. Am. 62, 1485–1490.
Wiggins, S. M. (2013). Acousonde Acoustic Tag Calibration at the
Transducer Evaluation Center (TRANSDEC), a technical report (La Jolla,
California), pp. 1–28.
Zimmer, W. M. X., Johnson, M. P., Madsen, P. T., and Tyack, P. L. (2005).
“Echolocation clicks of free-ranging Cuvier’s beaked whales (Ziphius
cavirostris),” J. Acoust. Soc. Am. 117, 3919–3927.
uhler, K. (2018). “Combinatorial capacities in primates,” Curr.
Opin. Behav. Sci. 21, 161–169.
Zwamborn, E. M. J., and Whitehead, H. (2017). “Repeated call sequences
and behavioural context in long-finned pilot whales off Cape Breton,
Nova Scotia, Canada,” Bioacoustics 26, 169–183.
J. Acoust. Soc. Am. 147 (2), February 2020 Walmsley et al. 1091
... Introduction Narwhals (Monodon monoceros) are deep-diving, medium-sized toothed whales endemic to the Arctic waters of Greenland and Northeastern Canada [e.g., 1,2]. Like other toothed whales, narwhals are a vociferous species, using echolocation to navigate and forage within their environment [3][4][5][6][7][8] and produce a diverse repertoire of sound types for communication with conspecifics [3,6,[9][10][11][12]. Sound types are classified into general groups of signals which share similar characteristics [13]. ...
... Like other toothed whales, narwhals are a vociferous species, using echolocation to navigate and forage within their environment [3][4][5][6][7][8] and produce a diverse repertoire of sound types for communication with conspecifics [3,6,[9][10][11][12]. Sound types are classified into general groups of signals which share similar characteristics [13]. Those occurring in the narwhal vocal repertoire include tonal and pulsed sounds [3,6,[9][10][11][12] and sounds which combine both tonal and pulsed components [10,12]. ...
... Like other toothed whales, narwhals are a vociferous species, using echolocation to navigate and forage within their environment [3][4][5][6][7][8] and produce a diverse repertoire of sound types for communication with conspecifics [3,6,[9][10][11][12]. Sound types are classified into general groups of signals which share similar characteristics [13]. Those occurring in the narwhal vocal repertoire include tonal and pulsed sounds [3,6,[9][10][11][12] and sounds which combine both tonal and pulsed components [10,12]. ...
Full-text available
Narwhals ( Monodon monoceros ) are gregarious toothed whales that strictly reside in the high Arctic. They produce a broad range of signal types; however, studies of narwhal vocalizations have been mostly descriptive of the sounds available in the species’ overall repertoire. Little is known regarding the functions of highly stereotyped mixed calls (i.e., biphonations with both sound elements produced simultaneously), although preliminary evidence has suggested that such vocalizations are individually distinctive and function as contact calls. Here we provide evidence that supports this notion in narwhal mother-calf communication. A female narwhal was tagged as part of larger studies on the life history and acoustic behavior of narwhals. At the time of tagging, it became apparent that the female had a calf, which remained close by during the tagging event. We found that the narwhal mother produced a distinct, highly stereotyped mixed call when separated from her calf and immediately after release from capture, which we interpret as preliminary evidence for contact call use between the mother and her calf. The mother’s mixed call production occurred continually over the 4.2 day recording period in addition to a second prominent but different stereotyped mixed call which we believe belonged to the narwhal calf. Thus, narwhal mothers produce highly stereotyped contact calls when separated from their calves, and it appears that narwhal calves similarly produce distinct, stereotyped mixed calls which we hypothesize also contribute to maintaining mother-calf contact. We compared this behavior to the acoustic behavior of two other adult females without calves, but also each with a unique, stereotyped call type. While we provide additional support for individual distinctiveness across narwhal contact calls, more research is necessary to determine whether these calls are vocal signatures which broadcast identity.
... The present study aimed to characterize the movements of Arctic char within Tremblay Sound, Nunavut, Canada, a model ecosystem for investigating the impacts of sea-ice phenology on consumer movements. Spring ice breakup here draws not only char but high biomass of other top-level consumers such as narwhal Monodon monoceros and Greenland sharks Somniosus microcephalus into the system (Heide-Jørgensen et al. 2002, Barkley et al. 2020, Walmsley et al. 2020. Moreover, char are abundant in Tremblay Sound in the summer, and harvesting by local communities provides an important subsistence resource. ...
... It is possible the marine migration distance of Arctic char is at least partially determined by proximity to preferential summer feeding habitat and abundant resources. For example, the seasonal migrations of Arctic char, narwhal (Heide-Jørgensen et al. 2002, Walmsley et al. 2020), and Greenland shark (Barkley et al. 2020) could sug-gest high food availability within Tremblay Sound. The fact that external migrants enter and remain in Tremblay Sound following an extended migration (relative to resident fish) provides further evidence of the potential regional importance of this system in terms of either heightened resource availability or possibly other factors such as providing refuge from predation etc. Shallow coastal regions such as Tremblay Sound also provide ideal feeding habitat for char (Spares et al. 2015, and the quantity of food observed in char stomachs suggests fish are readily exploiting available resources (Hammer 2021). ...
... Whistles are narrow-band and frequency-modulated tonal vocalizations 21,28 . Finally, combined signals, or "mixed calls, " include overlaid or paired pulsed and tonal sounds 23,29,30 . Acoustic classifiers may use a single call type or multiple call types to differentiate species 31 . ...
Full-text available
Belugas ( Delphinapterus leucas ) and narwhals ( Monodon monoceros ) are highly social Arctic toothed whales with large vocal repertoires and similar acoustic profiles. Passive Acoustic Monitoring (PAM) that uses multiple hydrophones over large spatiotemporal scales has been a primary method to study their populations, particularly in response to rapid climate change and increasing underwater noise. This study marks the first acoustic comparison between wild belugas and narwhals from the same location and reveals that they can be acoustically differentiated and classified solely by echolocation clicks. Acoustic recordings were made in the pack ice of Baffin Bay, West Greenland, during 2013. Multivariate analyses and Random Forests classification models were applied to eighty-one single-species acoustic events comprised of numerous echolocation clicks. Results demonstrate a significant difference between species’ acoustic parameters where beluga echolocation was distinguished by higher frequency content, evidenced by higher peak frequencies, center frequencies, and frequency minimums and maximums. Spectral peaks, troughs, and center frequencies for beluga clicks were generally > 60 kHz and narwhal clicks < 60 kHz with overlap between 40–60 kHz. Classification model predictive performance was strong with an overall correct classification rate of 97.5% for the best model. The most important predictors for species assignment were defined by peaks and notches in frequency spectra. Our results provide strong support for the use of echolocation in PAM efforts to differentiate belugas and narwhals acoustically.
Full-text available
The recent discovery of the Araguaian river dolphin (Inia araguaiaensis) highlights how little we know about the diversity and biology of river dolphins. In this study, we described the acoustic repertoire of this newly discovered species in concert with their behaviour. We analysed frequency contours of 727 signals (sampled at 10 ms temporal resolution). These contours were analyzed using an adaptive resonance theory neural network combined with dynamic time-warping (ARTwarp). Using a critical similarity value of 96%, frequency contours were categorized into 237 sound-types. The most common types were emitted when calves were present suggesting a key role in mother-calf communication. Our findings show that the acoustic repertoire of river dolphins is far from simple. Furthermore, the calls described here are similar in acoustic structure to those produced by social delphinids, such as orcas and pilot whales. Uncovering the context in which these signals are produced may help understand the social structure of this species and contribute to our understanding of the evolution of acoustic communication in whales.
Full-text available
Four groups of toothed whales have independently evolved to produce narrowband high-frequency (NBHF) echolocation signals (i.e. clicks) with a strikingly similar waveform and centroid frequency around 125 kHz. These signals are thought to help NBHF species avoid predation by echolocating and communicating at frequencies inaudible to predators, a form of acoustic crypsis. Heaviside's dolphins produce NBHF echolocation clicks in trains and often in rapid succession in the form of buzzes. In addition, a second click type with a lower frequency and broader bandwidth was recently described, typically emitted in rapid succession in the form of burst-pulses. We investigated the relationship between buzz and burst-pulse signals and both surface behaviour (foraging, ‘interacting with the kayak’ and socializing) and group size, using a multivariable regression on the signal occurrence and signal count data. Signal occurrence and counts were not related to group size in the regression analysis. Burst-pulses were strongly linked to socializing behaviour, occurring more often and more frequently during socializing and much less during foraging. Buzz vocalizations were not strongly linked to a specific behaviour although there was some evidence of an increase in production during foraging and socializing. In addition, individual level production rates of buzzes during foraging and socializing, and burst-pulses during socializing decreased with increasing group size. Temporally patterned burst-pulse signals were also identified, often occurring within a series of burst-pulses and were directly linked to specific events such as aerial leaping, backflipping, tail slapping and potential mating. Our findings suggest Heaviside's dolphins have a more complex communication system based on pulsed vocalizations than previously understood, perhaps driven by the need to facilitate the social interactions of this species.
Full-text available
Changes in climate are rapidly modifying the Arctic environment. As a result, human activities—and the sounds they produce—are predicted to increase in remote areas of Greenland, such as those inhabited by the narwhals (Monodon monoceros) of East Greenland. Meanwhile, nothing is known about these whales’ acoustic behavior or their reactions to anthropogenic sounds. This lack of knowledge was addressed by instrumenting six narwhals in Scoresby Sound (Aug 2013–2016) with Acousonde™ acoustic tags and satellite tags. Continuous recordings over up to seven days were used to describe the acoustic behavior of the whales, in particular their use of three types of sounds serving two different purposes: echolocation clicks and buzzes, which serve feeding, and calls, presumably used for social communication. Logistic regression models were used to assess the effects of location in time and space on buzzing and calling rates. Buzzes were mostly produced at depths of 350–650 m and buzzing rates were higher in one particular fjord, likely a preferred feeding area. Calls generally occurred at shallower depths (<100 m), with more than half of these calls occurring near the surface (<7 m), where the whales also spent more than half of their time. A period of silence following release, present in all subjects, was attributed to the capture and tagging operations, emphasizing the importance of longer (multi-day) records. This study provides basic life-history information on a poorly known species—and therefore control data in ongoing or future sound-effect studies.
Full-text available
Language, humans’ most distinctive trait, still remains a ‘mystery’ for evolutionary theory. It is underpinned by a universal infrastructure — cooperative turn-taking — which has been suggested as an ancient mechanism bridging the existing gap between the articulate human species and their inarticulate primate cousins. However, we know remarkably little about turn-taking systems of nonhuman animals, and methodological confounds have often prevented meaningful cross-species comparisons. Thus, the extent to which cooperative turn-taking is uniquely human or represents a homologous and/or analogous trait is currently unknown. The present paper draws attention to this promising research avenue by providing an overview of the state of the art of turn-taking in four animal taxa — birds, mammals, insects and anurans. It concludes with a new comparative framework to spur more research into this research domain and test which elements of the human turn-taking system are shared across species and taxa.
Full-text available
Vocal imitation is a hallmark of human spoken language, which, along with other advanced cognitive skills, has fuelled the evolution of human culture. Comparative evidence has revealed that although the ability to copy sounds from conspecifics is mostly uniquely human among primates, a few distantly related taxa of birds and mammals have also independently evolved this capacity. Remarkably, field observations of killer whales have documented the existence of group-differentiated vocal dialects that are often referred to as traditions or cultures and are hypothesized to be acquired non-genetically. Here we use a do-as-I-do paradigm to study the abilities of a killer whale to imitate novel sounds uttered by conspecific (vocal imitative learning) and human models (vocal mimicry). We found that the subject made recognizable copies of all familiar and novel conspecific and human sounds tested and did so relatively quickly (most during the first 10 trials and three in the first attempt). Our results lend support to the hypothesis that the vocal variants observed in natural populations of this species can be socially learned by imitation. The capacity for vocal imitation shown in this study may scaffold the natural vocal traditions of killer whales in the wild.
Full-text available
Although predators influence behavior of prey, analyses of electronic tracking data in marine environments rarely consider how predators affect the behavior of tracked animals. We collected an unprecedented dataset by synchronously tracking predator (killer whales, [Formula: see text] = 1; representing a family group) and prey (narwhal, [Formula: see text] = 7) via satellite telemetry in Admiralty Inlet, a large fjord in the Eastern Canadian Arctic. Analyzing the movement data with a switching-state space model and a series of mixed effects models, we show that the presence of killer whales strongly alters the behavior and distribution of narwhal. When killer whales were present (within about 100 km), narwhal moved closer to shore, where they were presumably less vulnerable. Under predation threat, narwhal movement patterns were more likely to be transiting, whereas in the absence of threat, more likely resident. Effects extended beyond discrete predatory events and persisted steadily for 10 d, the duration that killer whales remained in Admiralty Inlet. Our findings have two key consequences. First, given current reductions in sea ice and increases in Arctic killer whale sightings, killer whales have the potential to reshape Arctic marine mammal distributions and behavior. Second and of more general importance, predators have the potential to strongly affect movement behavior of tracked marine animals. Understanding predator effects may be as or more important than relating movement behavior to resource distribution or bottom-up drivers traditionally included in analyses of marine animal tracking data.
Do primates have syntax-like abilities? One line of enquiry is to test how subjects respond to different types of artificial grammars. Results have revealed neural structures responsible for processing combinatorial content, shared between non-human primates and humans. Another approach has been to study natural communication, which has revealed a wealth of organisational principles, including merged compounds and sequences with stochastic, permutated, hierarchical and cross-modal combinatorial structure. There is solid experimental evidence that recipients can attend to such combinatorial features to extract meaning. The current debate is whether animal communication can also be compositional, that is, whether signallers assemble meaningful units to create utterances with novel meanings.
Acoustic call sequences are important components of vocal repertoires for many animal species. Bottlenose dolphins (Tursiops truncatus) produce a wide variety of vocalizations, in different behavioural contexts, including some conspicuous vocal sequences – the ‘bray series’. The occurrence of brays is still insufficiently documented, contextually and geographically, and the specific functions of these multi-unit emissions are yet to be understood. Here, acoustic emissions produced by bottlenose dolphins in the Sado estuary, Portugal, were used to provide a structural characterization of the discrete elements that compose the bray series. Information theory techniques were applied to analyse bray sequences and explore the complexity of these calls. Log-frequency analysis, based on bout criterion interval, confirmed the bout structure of the bray series. A first-order Markov model revealed a distinct pattern of emission for the bray series’ elements, with uneven transitions between elements. The order in these sequential emissions was not random and consecutive decreases in higher order entropy values support the notion of a well-defined structure in the bray series. The key features of animal signal sequences here portrayed suggest the presence of relevant information content and highlight the complexity of the bottlenose dolphin’s acoustic repertoire.
Short-finned pilot whales (Globicephala macrorhynchus) have complex vocal repertoires that include calls with two time-frequency contours known as two-component calls. We attached digital acoustic recording tags (DTAGs) to 23 short-finned pilot whales off Cape Hatteras, North Carolina, and assessed the similarity of two-component calls within and among tags. Two-component calls made up <3% of the total number of calls on 19 of the 23 tag records. For the remaining four tags, two-component calls comprised 9%, 23%, 24%, and 57% of the total calls recorded. Measurements of six acoustic parameters for both the low and high frequency components of all two-component calls from the five tags were compared using a generalized linear model. There were significant differences in the acoustic parameters of two-component calls between tags, verifying that acoustic parameters were more similar for two-component calls recorded on the same tag than for calls between tags. Spectrograms of all two-component calls from the five tags were visually graded and independently categorized by five observers. A test of inter-rater reliability showed substantial agreement, suggesting that each tag contained a predominant two-component call type that was not shared across tags.