Context-dependent variability in blue whale acoustic behaviour

Abstract

Acoustic communication is an important aspect of reproductive, foraging and social behaviours for many marine species. Northeast Pacific blue whales (Balaenoptera musculus) produce three different call types—A, B and D calls. All may be produced as singular calls, but A and B calls also occur in phrases to form songs. To evaluate the behavioural context of singular call and phrase production in blue whales, the acoustic and dive profile data from tags deployed on individuals off southern California were assessed using generalized estimating equations. Only 22% of all deployments contained sounds attributed to the tagged animal. A larger proportion of tagged animals were female (47%) than male (13%), with 40% of unknown sex. Fifty per cent of tags deployed on males contained sounds attributed to the tagged whale, while only a few (5%) deployed on females did. Most calls were produced at shallow depths (less than 30 m). Repetitive phrasing (singing) and production of singular calls were most common during shallow, non-lunging dives, with the latter also common during surface behaviour. Higher sound production rates occurred during autumn than summer and they varied with time-of-day: singular call rates were higher at dawn and dusk, while phrase production rates were highest at dusk and night.

Full text

rsos.royalsocietypublishing.org
Research
Cite this article: Lewis LA et al.2018
Context-dependent variability in blue whale
acoustic behaviour. R. Soc. open sci. 5: 180241.
http://dx.doi.org/10.1098/rsos.180241
Received: 12 February 2018
Accepted: 3 July 2018
Subject Category:
Biology (whole organism)
Subject Areas:
behaviour
Keywords:
blue whale, Balaenoptera musculus,song,
acoustic communication, behavioural context
Author for correspondence:
Leah A. Lewis
e-mail: oceanmusic12@gmail.com
Electronic supplementary material is available
online at https://dx.doi.org/10.6084/m9.
gshare.c.4171997.
Context-dependent
variability in blue whale
acoustic behaviour
Leah A. Lewis1, John Calambokidis2,Alison
K. Stimpert3, James Fahlbusch2, Ari S. Friedlaender4,8,
Megan F. McKenna5, Sarah L. Mesnick6,Erin
M. Oleson7, Brandon L. Southall4,8,Angela
R. Szesciorka1,2 and Ana Širović1,9
1Scripps Institution of Oceanography, University of California San Diego,9500 Gilman
Drive, La Jolla, CA 92093, USA
2Cascadia Research Collective, 218 ½W4thAve.,Olympia,WA98501,USA
3Moss Landing Marine Laboratories, 8272 Moss Landing Road, Moss Landing,
CA 95039, USA
4Institute for Marine Sciences, University of California Santa Cruz, 115 McAllister Way,
Santa Cruz, CA 95064, USA
5Natural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive,
Fort Collins, CO 80525, USA
6Southwest Fisheries Science Center, National Marine Fisheries Service, NOAA, 8901 La
Jolla Shores Drive, La Jolla, CA 92037, USA
7Pacic Islands Fisheries Science Center, National Marine Fisheries Service, NOAA, 1845
Wasp Blvd., Building 176, Honolulu,HI 96818, USA
8Southall Environmental Associates, 9099 Soquel Drive, Suite 8, Aptos,CA 95003, USA
9Texas A&M University Galveston, 200 Seawolf Parkway, Galveston, TX 77554, USA
LAL, 0000-0002-1770-6427;BLS,0000-0002-3863-2068
Acoustic communication is an important aspect of
reproductive, foraging and social behaviours for many
marine species. Northeast Pacific blue whales (Balaenoptera
musculus) produce three different call types—A, B and D
calls. All may be produced as singular calls, but A and B
calls also occur in phrases to form songs. To evaluate the
behavioural context of singular call and phrase production
in blue whales, the acoustic and dive profile data from
tags deployed on individuals off southern California were
assessed using generalized estimating equations. Only 22%
of all deployments contained sounds attributed to the tagged
animal. A larger proportion of tagged animals were female
(47%) than male (13%), with 40% of unknown sex. Fifty per
cent of tags deployed on males contained sounds attributed to
the tagged whale, while only a few (5%) deployed on females
did. Most calls were produced at shallow depths (less than
30 m). Repetitive phrasing (singing) and production of singular
2018 The Authors. Published by the Royal Society under the terms of the Creative Commons
Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted
use, provided the original author and source are credited.

2
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
calls were most common during shallow, non-lunging dives, with the latter also common during
surface behaviour. Higher sound production rates occurred during autumn than summer and they
varied with time-of-day: singular call rates were higher at dawn and dusk, while phrase production
rates were highest at dusk and night.
1. Introduction
Sound production is an important behavioural strategy for many species. The use of sound for
reproductive purposes in terrestrial species has been well documented, for example, in songbirds [1–
3], frogs [4,5], insects [6] and ungulates [7]. Many terrestrial species also commonly produce sounds
associated with foraging [8] and other social interactions [9–12]. In the marine environment, where
sound travels with little attenuation, sound production may play a critical role in many life functions.
For instance, sound is important for invertebrates [13], fish [14–16] and marine mammal species [17–20].
Among marine mammals, it is often used for social and communicative purposes [18]. Baleen whales
are very prolific producers of sounds [19,21,22]. The behavioural context of sound production has been
examined for a subset of calls produced by well-studied baleen whale species, including the humpback
whale (Megaptera novaeangliae)[23–26], the southern right whale (Eubalena australis)[27] and the North
Atlantic right whale (E. glacialis)[28,29].
Blue whale (Balaenoptera musculus) sounds, particularly those produced by the northeast Pacific
population, have also been extensively studied [30–34]. The acoustic repertoire of this blue whale
population consists of three main sound types: A, B and D calls [30,35–38]. The pulsed A and tonal
B sound types, each lasting approximately 15–20 s, can be produced individually at irregular intervals
as singular calls [33] or together at regular intervals as A and B units within phrases. When repeated,
these phrases form bouts of song and acoustically distinguish this population from other blue whale
populations [31,39]. The A and B sound types have only been recorded from males and are thus
considered to have a reproductive function, although this was based on a small sample of known
callers [31,33]. These sounds are detected off southern California from June to January, and peak in
SeptemberorOctober[40–43]. Blue whale D calls are shorter (less than 5 s), more frequency-modulated
sounds that have been recorded from both males and females [30,31]. These variable down-swept calls
appear to be commonly produced by different blue whale populations [44,45] and are probably used
as social calls while foraging [33]. D calls are typically recorded in southern California from April to
November, peaking during the summer [33,40,41].
Recent developments in bio-logging technology [46–49] have allowed for the collection of finer-scale
data associated with sound production in a variety of marine mammals [50], including humpback
whales [51,52], North Atlantic right whales [53], Antarctic minke whales [54], fin whales [55,56]and
blue whales [33,57]. Multi-sensor tags, which are capable of recording acoustic and dive depth data
as well as body movements, may allow for evaluation of the tagged whale’s behaviour during sound
production if a reliable method is available for determining which sounds are produced from the tagged
animal. This analysis can be particularly difficult for low-frequency baleen whale sounds [50,55]. If
sounds can be attributed to the tagged whale, the detailed behaviour of tagged whales producing sounds
may also be compared to that of whales not producing sounds while tagged, to examine differences in
behaviour. Ultimately, if long-term passive acoustic data are to be used to estimate whale distributions
and densities [58–60], the behavioural context(s) of sound production must be understood.
Previous studies into the behavioural context of sound production in blue whales have been limited
either in sample size, with a small number of tag deployments resulting in relatively few hours of
collected data [57], or in the number of sounds detected, due to natural variation in sound production
by any individual tagged whale [33]. Furthermore, because the recording durations of these tags are
inherently restricted by memory, battery capacity and attachment method [50], only a small subset of
data collected from blue whales was recorded at night [33,57], providing us with limited understanding
of how blue whale behaviour varies between day and night. However, recent development of alternative
attachment methods have resulted in longer deployments and have thus enhanced our ability to obtain
behavioural and acoustic data from tagged blue whales for durations ranging from several days to
weeks [61].
In this study, we evaluated how call and phrase production rates varied with respect to behavioural
state, location, season and time-of-day in blue whales tagged off southern California. Our dataset
included the acoustic and dive depth data collected from acoustic tags deployed on individuals over

3
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
the course of 14 years, including data recorded during several long-duration deployments. This analysis
provides the most extensive analysis into the behavioural context of blue whale calling in this region
available to date.
2. Methods
2.1. Tag data collection
For these analyses we used data collected by tags deployed on blue whales off southern California from
2002 to 2016 (electronic supplementary material, table S1). The whales were tagged as part of multiple
research efforts, including collaborations between the Cascadia Research Collective (CRC) and Scripps
Institution of Oceanography (SIO), and during the Southern California Behavioural Response Study
(SOCAL-BRS) [62]. All tag deployments were conducted in accordance with the ethical standards under
CRC’s Institutional Animal Care and Use Committee protocols (AUP-6), and under National Marine
Fisheries Service (NMFS) permits as detailed in the Animal Ethics section. No additional permissions
were needed to carry out fieldwork. Tag deployments used in this study include the deployments from
southern California used in an earlier analysis of the behavioural context of blue whale calling [33].
Three types of sound- and movement-recording tags were deployed on blue whales: Bioacoustic Probes
(Bprobes), Acousondes (both developed by Greeneridge Sciences, Inc.) and Dtags [63]. Bprobes are
capable of sampling acoustic data at rates up to 20 kHz and are equipped with ancillary sensors for
recording temperature, pressure, and, in versions produced after 2003, 2-axis acceleration. Accelerometer
data enable the derivation of instantaneous body orientation of the whale during a dive cycle [64].
In addition to the auxiliary sensors found in the Bprobe, Acousondes contain an updated 3-axis
accelerometer, a compass and the ability to collect higher-frequency acoustic data. The sampling
capabilities of Dtags are similar to those of Acousondes, as they record depth, accelerometer and acoustic
data, although they each had different hydrophone sensitivities [63]. The sampling rates for acoustic,
auxiliary and accelerometer data varied with tag type and deployment (electronic supplementary
material, table S1). Across all deployments, sampling rates were in the range of 1024–240000 Hz for
acoustic data, 1–800 Hz for accelerometer data and 1–50 Hz for auxiliary data (electronic supplementary
material, table S1).
Tags were deployed on blue whales in multiple regions [33]; however, we focused our analysis only
on deployments conducted off southern California (figure 1) during ship-based efforts and shore-based
tagging operations. Blue whales were tagged opportunistically, typically based on the ability to locate
and track them visually. When an individual was chosen for tagging, the whale was approached using a
rigid-hull inflatable boat (RHIB) and a tag was deployed using a metal or fibreglass pole. For the majority
of deployments, the tag was attached to the whale with suction cups. However, starting in 2016, longer-
duration tag attachment methods (darts instead of suction cups) were used [61]. When feasible, photo
identification and biopsy samples were also collected from tagged whales.
Upon tag retrieval, digital data were downloaded from the tag to a computer. Only deployments
with at least 15 min of high-quality acoustic data were included in this analysis [33]. As part of the
SOCAL-BRS, some of the animals included in this analysis were exposed to simulated Navy sonar or
pseudo-random noise [62,65,66] (electronic supplementary material, table S1); however, in those cases,
data from during the exposure and 3h after the exposure were excluded from analysis to eliminate
potential impact of the exposure on the tagged whale’s natural behaviour. Only acoustic and pressure
data were used for this study because 3-axis accelerometer data were not collected on 44 of the 121
deployments available for analysis (they were only available for Acousonde and Dtag deployments).
2.2. Acoustic data analysis
We reviewed the acoustic data collected from all Bprobe and Acousonde deployments in spectrogram
form in Trit o n , a Matlab-based (www.mathworks.com) software package [67]. We manually detected
all blue whale A, B and D sounds based on visual and aural inspection of the spectrogram (calculated
with 1 Hz and 0.1 s resolutions, using a Hanning window), and logged the start and end times of each
sound. The acoustic data collected using Dtag deployments were decimated to a 600 Hz sample rate
before plotting the spectrogram using custom Matlab software (calculated with 1.17Hz and 0.02 s time
resolution, using a Hamming window) and manually analysed for blue whale sounds, also logging start
times of each sound.

4
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
35°N
34°N
33°N
32°N
34
°N
121°W
120°30’W 119°30’W 119°W 118°W
33°30’N
34°N
120°W 119°W 118°W 117°W
inshore-S
inshore-C
inshore-N
offshore
SBC
LA
LASBC
30’
30’
30’
acousonde
Bprobe
Dtag
tagged caller
calls -not tagged
no calls
Figure 1. Locations of tagging events for all Acousondes (stars), Bprobes (crosses) and Dtags (squares) deployed on blue whales o
southern California during 2002–2016. Tags that contained sounds that were attributed to the tagged whale are marked in red, while
tags that contained sounds not assigned to the tagged whale are marked in blue. Tags that did not contain any sounds are plotted in
black. The four areas used to classify tag deployment locations for statistical analyses are marked with bold, black-hashed lines. Two
smaller areas that contained high densities of tag deployments (within the Santa Barbara Channel and o Los Angeles and Long Beach,
CA) are marked in orange and shown as inset maps on the bottom.
When studying the behavioural context of sound production, it is important to note whether the
tagged animal produced each sound recorded on the tag or not. However, there is no standard method
for accomplishing this task. Previous studies have relied upon universal signal-to-noise ratio (SNR) cut-
offs in order to assign sounds to tagged humpback whales [68], while, more recently, accelerometer data
have been used to determine caller identity in tagged fin whales [55,56]. In general, the use of source-
level estimates to confirm caller identity is problematic because tag placement on the animal can vary
both within and between deployments, and because the anatomical location of the blue whale sound
production mechanism remains unknown [69,70]. We could not rely on accelerometer data because they
were not available for many deployments and, additionally, there has been no evidence to indicate
that the same procedure used for fin whales will work for blue whales or for tag types other than
Dtags. Furthermore, while the assignment of sounds to an individual is critical for many studies, our
behavioural analysis is focused on evaluating the broader behavioural contexts associated with sound
production, and the precise assignment of sounds to the tagged individual is less crucial. Therefore, by
using a combination of high relative root-mean-square received levels (RLrms) and SNRs to attribute
sound production to the tagged whale, we feel confident that the sounds that were used for analysis
in this study were either produced by the tagged whale or a whale that was close to the tag, at
approximately the same depth and possibly engaging in the same behaviour [71], though not necessarily
the same sex. Tagged blue whales in this study were generally single or closely paired with another
individual (electronic supplementary material, table S1).

5
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
We calculated the RLrms and the SNR for each detected sound according to the following processes.
We applied an infinite impulse response (IIR) bandpass filter to the data, with the bandpass frequencies
based on the call type (A, B or D) and corresponding to the band of peak frequency for each call. The
three frequency bands were 70–100 Hz for A calls, 38–55 Hz for B calls and 25–100 Hz for D calls. We used
the third harmonic of B calls and the higher-frequency pulsed harmonics of A calls because these were
louder and simpler to discern from low-frequency flow noise. We calculated RLrms over the duration of
each sound, where the duration was calculated based on 90% of the energy of the sound [29]. We also
measured noise levels in 500 ms intervals during the 10 s prior to each call, and subtracted the lowest
of these noise-level measurements from the RLrms of the call to determine the SNR [68]. As transfer
functions were not known for all tags, we calculated only relative values, and compared them to other
relative values within a given deployment or for the same tag. We did not compensate for the built-in
high-pass filters in the Dtag hardware, nor did we correct for other system sensitivities of any tag types.
We performed all calculations in Matlab (www.mathworks.com).
To attribute sound production to the tagged (or nearby) whale as opposed to distant animals, we
calculated the mean and standard deviations of RLrms and SNRs for each call type and deployment.
If the RLrms or the SNR of an individual call was higher than the mean minus one standard deviation
calculated for that deployment, we assigned the call to the tagged whale, otherwise we removed it from
further analysis. For deployments that contained only one or two calls of one call type, we used the mean
and standard deviations of RLrms and SNRs calculated for all deployments of the same particular tag. In
all other cases, this method resulted in at least some calls being assigned to the tagged whale on every
deployment where calls were recorded. However, the method did reliably exclude very faint calls that
were probably produced by distant animals. For our purposes, we use both ‘produced’ and ‘attributed
to tagged whale’ as shorthand and in reference to sounds that were presumed to be produced by the
tagged individual or nearby whale as determined based on these methods.
As we were interested in acoustic behaviour, we evaluated whether any individual A or B sound was
produced as a singular call or as a unit within a phrase. First, we sorted all A and B detections based on
the logged start time. We calculated inter-call intervals, measured as the time from the start of one call
to the start of the next, for all A and B sounds. We classified all blue whale A and B detections that were
not produced in a pattern with regular inter-call intervals as singular calls rather than phrase units [41].
We defined phrases as sequences of A and B calls where the start of one unit was followed by another
within 49 ±10 s for A–B units and within 51 ±14s for B–B units [40]. We defined repetitive phrases as
sequences of AB phrases where the interval between the ending B unit of one phrase and the leading A
unit of the next was 70 ±29 s [40]. We grouped all single AB phrases, i.e. those that were neither preceded
nor followed by another phrase, and repetitive phrases together into the same ‘phrase’ category. We also
grouped them regardless of phrase composition (i.e. the ratio of A to B units within the phrase). We
classified all D call detections as singular D calls.
2.3. Pressure data analysis and behavioural classication
We used pressure (i.e. depth) data as a proxy for the behavioural state of the tagged blue whale.
The pressure data from each deployment was loaded into the AcqKnowledge software (v. 3.9.1, Biopac
Systems, Inc.), and individual dives were identified based on changes in pressure over time. A dive
was defined as a submergence that exceeded 10 m in depth. For each recorded dive, we marked the
following dive characteristics using the program’s manual selection tools: dive start time (in local time;
defined as the start time of submergence from the surface); dive duration (defined as the time between
submergence from and re-emergence to the surface); maximum dive depth (defined as the maximum
depth reached during the dive); time spent at bottom of dive (defined as the time between the whale’s
descent from and ascent to the surface, and based on the depth change over time); and the number
of vertical lunges present within a dive. Although vertical as well as horizontal lunges can be easily
identified using accelerometer data [64,72], our dive analyses were limited to the use of pressure data
because accelerometer data were not available for all deployments. Thus, for our study, only vertical
lunges were identified, and these were defined as vertical excursions in excess of 5m from the bottom of
a dive [57].
If singular A, singular B or D calls, or phrases attributed to the tagged blue whale were detected within
a dive, we recorded the number and type of sounds present. We determined the depth of production for
each sound based on pressure data and the time at the start of the sound. Many of the Bprobe and
Acousonde deployments exhibited systematic offsets in depth measured at the surface. To correct this,
we calculated the average surface depth for each deployment and when it differed from 0m, we applied

6
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
this value as a correction factor to all sound production depths. We excluded dives during which the tag
fell off (i.e. the final dive of each deployment) from all analyses.
We used a combination of the maximum dive depth and the presence of vertical lunges detected
within a dive to broadly classify each dive into one of five behavioural states: shallow non-lunging,
shallow lunging, deep non-lunging, deep lunging and surface behaviour. We classified all dives without
vertical lunges that did not exceed the 50 m depth as shallow, non-lunging dives [33]. Dives that exceeded
the 50 m depth without vertical lunges were classified as deep, non-lunging dives. We classified all
dives that did not exceed the 50 m depth but contained vertical lunges as shallow lunging dives,
and all dives that contained vertical lunges at a depth exceeding 50 m as deep lunging dives. If the
tagged blue whale spent extended time near the surface, without any identifiable diving behaviour, we
classified this behaviour as surface behaviour. Bouts of surface behaviour were distinguishable from
normal surface breath intervals based on the duration of time spent within 5m of the surface (generally
in excess of 10 min). For these behavioural categories, we use lunging and non-lunging as shorthand
to describe the presence or absence of vertical lunges only; no other changes in the tagged whale’s
orientation were recorded because 3-axis accelerometer data were not available on all deployments.
Furthermore, although lunges are generally indicative of foraging [57,64,72], we cannot exclude that
feeding occurred also during non-lunging states because we did not analyse available accelerometer
data (for reasons noted above). As we were primarily focused on identifying broad behavioural
contexts, our behavioural state classifications should not be considered as a precise evaluation of a
tagged whale’s feeding or non-feeding behaviour. Similarly, because our dive analyses were limited to
changes in pressure data over time, we could not evaluate whether bouts of surface behaviour included
feeding events.
2.4. Sound production and behavioural context analysis
To statistically evaluate the effect of location, season, behavioural state and time-of-day on sound
production, we modelled the occurrence of singular A, singular B, and D calls, and AB phrases using
generalized estimating equations (GEEs) [73]. We used this approach because GEEs allow for estimates
of population average parameters from correlated or clustered data [73]. Thus, by clustering the data
into units based on individual deployments, we were able to account for differences between individual
tagged whales as well as autocorrelation within an individual deployment.
We classified deployments based upon the location (latitude and longitude) of the initial tagging
event into one of four groups: inshore-south (south of 33° N, In-S), inshore-central (between 33° and
34° N, In-C), inshore-north (north of 34° N, In-N) and offshore (offshore of Santa Catalina Island and
south of the Channel Islands) (figure 1). Seasonal trends in call and phrase production rates were
evaluated by classifying data based on month of initial deployment: spring, for deployments occurring
between March and May; summer, for deployments occurring from June to August; and autumn, for tags
deployed between September and November. However, because just one deployment occurred during
the spring (electronic supplementary material, table S1), seasonal tests were conducted between summer
and autumn only. Diel patterns in call and phrase production rates were similarly analysed by classifying
data into four time-of-day periods: dawn, day, dusk and night. We used the definitions of these periods
as described by Wiggins et al.[32], based on times of nautical twilight, sunrise and sunset at the location
each tag was deployed. We binned all data based on the average dive duration into 12 min intervals. If a
12 min bin crossed between two time-of-day periods, we classified the bin into one of the four categories
based on the majority of minutes spent in a particular period. We defined the behavioural state for each
bin based on the five dive categories described previously. If multiple diving behaviours were recorded
within a 12 min interval, we classified the behavioural state as the behaviour that took up the majority of
time. Finally, the number of singular A, singular B and D calls, and phrases attributed to the tagged whale
were counted over each 12 min interval. Hourly call and phrase production rates were then calculated
and used as the response variable in the GEE, and location, season, behavioural state and time-of-day
were used as covariates. Mean hourly call and phrase production rates per behavioural state were also
calculated for each tagged caller, and mean production rates across all tagged callers were then plotted
across the five behavioural states.
We used individual whales as the clustering unit for the GEE, and modelled the call rate using
Poisson distribution and log-link function with an autoregressive correlation structure to account for
temporal correlation between bins within a single deployment. We used the standard robust sandwich
variance estimate for all reported results [73]. We performed all analyses using the geeglm() function of
the ‘geepack’ package [71,74] in the R Studio (v. 1.0.153) statistical software platform [75].

7
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
3. Results
A total of 874.1 h of acoustic and dive profile data collected from 121 tags (mean attachment
duration =7.2 h; s.d. =18.0 h; median =2.2 h) deployed on blue whales off southern California were
analysed (electronic supplementary material, table S1). A total of 13 individuals were tagged more than
once but none of those tags contained calls attributable to the tagged animal. Of all tag deployments,
22.3% (27 tags) contained sounds that were attributed to the tagged whale or the tagged whale’s group,
an additional 12.4% (15 tags) contained blue whale sounds that were not attributed to the tagged whale
and 65.3% (79 tags) contained no blue whale sounds (figure 1; electronic supplementary material, table
S1). Among all tagged animals, 13% (16 whales) were male, 47% (57 whales) were female, for 5% (six
whales) sex could not be determined from the available sample and 35% (42 blue whales) were without
skin samples for sex determination (electronic supplementary material, table S1). Within the 16 tags
deployed on male blue whales (electronic supplementary material, table S1), 50% (eight tags) contained
sounds attributed to the tagged individual, 25% (four tags) contained sounds that were not attributed to
the tagged whale and 25% (four tags) contained no blue whale sounds. For the 56 tags that were deployed
on females (electronic supplementary material, table S1), 5% (three tags) contained sounds attributed to
the tagged individual, 13% (seven tags) contained sounds that were not attributed to the tagged whale
and 82% (46 tags) contained no blue whale sounds.
Overall, out of the total of 4514 blue whale sounds detected from 42 acoustic records, 73% (3308) were
attributed to 27 tagged individuals. The majority of all sounds attributed to tagged individuals were
phrases (880; comprising 880 A units and 1361 B units) and D calls (550). Similar numbers of singular A
and B calls were produced (229 and 288, respectively). The majority of singular A calls (117) and singular
B calls (175) were produced around the same time as phrases, albeit at longer and more irregular intervals
than A and B phrase units. Sounds attributed to tagged male blue whales consisted predominately of
phrases (81; comprising 81 A units and 85 B units), singular A calls (74) and singular B calls (25), although
D calls (131) were also produced in high numbers, especially by one male tagged in 2010 (with 117
attributed D calls) (electronic supplementary material, table S1). Sounds attributed to female blue whales,
on the other hand, were fewer and consisted of eight singular A calls, one singular B call and 24 D calls
(electronic supplementary material, table S1). However, all of the singular A and B calls, as well as one D
call, were recorded by tags deployed on three females that were all tagged while in close association with
another blue whale. Furthermore, the RLrms and SNR values calculated for these calls were relatively low
in comparison to calls of the same type recorded during other tag deployments. Therefore, it is possible
that these calls may have been produced by the nearby, associated whales rather than the tagged females.
The remaining 23 D calls were recorded on a tag deployed on a female in a tightly associated pair whose
sex was determined based on being sighted with a calf in a previous year (electronic supplementary
material, table S1). Finally, we should note that the majority (354) of D calls were produced by a single
individual of unknown sex.
Most tag deployments, and the largest number of acoustic records containing calls, occurred within
the inshore-central and inshore-north locations (figure 1), particularly off Los Angeles/Long Beach and
within the Santa Barbara Channel. However, the effect of location on call and phrase production rates
varied with sound type (figure 2,table 1). Singular B call and phrase production rates recorded from
blue whales tagged in the inshore-north region were significantly lower than the rates from individuals
tagged in other regions (B calls: p=0.001, phrases: p=0.043) (figure 2,table 1). Singular A call rates were
significantly lower in the offshore region than in any other tagging region (p=0.010) (figure 2,table 1).
D call rates were higher from blue whales tagged in the inshore central region than in any other region
(figure 2,table 1).
Deep lunging dives comprised the majority of daytime tag data, while shallow non-lunging dives
and surface behaviour dominated at night (figure 3). However, the proportion of time spent within these
behavioural states differed by month (figure 3). Specifically, during the day, the per cent of time spent in
deep lunging dives decreased between the summer and autumn, while shallow non-lunging and surface
behaviours increased (figure 3). With the exception of the single March deployment, tagged blue whales
spent the least amount of time in the shallow lunging dive state each month (figure 3).
Blue whale call and phrase production rates recorded during the autumn were higher than those
recorded during the summer, though only differences in singular A and D call production rates were
significant (A calls: p=0.010, D calls: p=0.026) (table 1). This seasonal variation may correspond to the
increased per cent of time that blue whales spent within shallow non-lunging and surface behavioural
states during the autumn (figure 3), because we observed increased call and phrase production by blue
whales within these behavioural states. About 75% of all sounds attributed to tagged individuals were

8
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
location
singular Asingular BD callsphrases
partial fit partial fit partial fit partial fit
behaviour time-of-day
2.0
1.0
0
–1.0
–2.0
–3.0
2.0
1.0
0
–1.0
–2.0
1.5
1.0
0.5
0
–0.5
1.0
0
–1.0
–2.0
–3.0
–4.0
2.0
3.0
1.0
0
–1.0
1.0
0.5
0
–0.5
–1.0
0
–1.0
–2.0
–3.0
–4.0
–5.0
–6.0
3.0
2.0
1.0
0
–1.0
–2.0
–3.0
2.0
1.5
1.0
0.5
0
0.5
0
–0.5
–1.0
3.0
2.0
1.0
0
0
–1.0
–2.0
–3.0
–4.0
inshore-
north
inshore-
south
offshore deep-
no lunges
shallow-
lunging
shallow-
no lunges
surface
behaviour
day nigh
t
dusk
Figure 2. Partial t plots output from the generalized estimating equations (GEEs)used to model blue whale call and phrase production
rates as a function of location, behavioural state and time-of-day. The horizontal lines at zero in each plot represent the reference level
of each factor: inshore-central for location; deep lunging dives for behavioural state; and dawn for time-of-day.
Tabl e 1. Resultsfrom the generalizedestimating equations (GEEs)usedto assess spatial,temporal and behaviouralvariability in singular
call and phrase production rates. Foreach sound t ype,coecient parameter estimates (Cp. est), robust standarderrors (s.e.) and p-values
for each of the levels within the four factor variables (location, season, behavioural stateand time -of-day (TOD)) are presented. Variables
and associated levels with signicant p-values are marked with an asterisk and italicized. Statistics are based upon the dierence from
the reference level of each factor: inshore-central for location; autumn for season; deep lunging divesfor behavioural state; and dawn for
time-of-day.
sound type factor level Cp. est s.e. p-value
singular A location In-N −0.786 0.702 0.2630
...................................................................................................................................
In-S −0.020 0.922 0.983
...................................................................................................................................
O −1.738 0.686 0.010*
............................................................................................................................................................................
(Continued.)

9
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
Table 1. (Continued.)
sound type factor level Cp. est s.e. p-value
season summer −1.399 0.543 0.010*
..................................................................................................................................................................................
behaviour Dp-no lunges 0.309 0.392 0.431
.............................................................................................................................................
Sh-lunges 0.818 0.657 0.213
.............................................................................................................................................
Sh-no lunges 0.596 0.505 0.238
.............................................................................................................................................
surface −0.935 0.579 0.106
..................................................................................................................................................................................
TOD day −0.152 0.320 0.636
.............................................................................................................................................
dusk 0.621 0.431 0.150
.............................................................................................................................................
night 0.410 0.420 0.329
.........................................................................................................................................................................................................................
singular B location In-N −2.552 0.785 0.001*
.............................................................................................................................................
In-S −0.575 0.869 0.508
.............................................................................................................................................
O −0.337 0.504 0.504
..................................................................................................................................................................................
season summer −1.271 0.678 0.061
..................................................................................................................................................................................
behaviour Dp-no lunges 1.660 0.501 0.001*
.............................................................................................................................................
Sh-lunges 0.171 0.805 0.832
.............................................................................................................................................
Sh-no lunges 2.131 0.415 2.90 ×10−7*
.............................................................................................................................................
surface 0.343 0.363 0.344
..................................................................................................................................................................................
TOD day −0.404 0.320 0.207
.............................................................................................................................................
dusk −0.189 0.591 0.748
.............................................................................................................................................
night −0.358 0.344 0.297
.........................................................................................................................................................................................................................
D calls location In-N −1.965 0.731 0.007*
.............................................................................................................................................
In-S −3.155 0.882 3.50 ×10−4*
.............................................................................................................................................
O −4.793 0.248 <2.00 ×10−16 *
..................................................................................................................................................................................
season summer −1.731 0.776 0.026*
..................................................................................................................................................................................
behaviour Dp-no lunges 0.869 0.358 0.015*
.............................................................................................................................................
Sh-lunges −0.161 0.35 0.645
.............................................................................................................................................
Sh-no lunges 1.122 0.428 0.009*
.............................................................................................................................................
surface 1.524 0.464 0.001*
..................................................................................................................................................................................
TOD day 0.239 0.191 0.21
.............................................................................................................................................
dusk 1.202 0.279 1.60 ×10−5*
.............................................................................................................................................
night 0.471 0.328 0.15
.........................................................................................................................................................................................................................
phrases location In-N −2.460 1.217 0.043*
.............................................................................................................................................
In-S −1.884 1.231 0.126
.............................................................................................................................................
O −1.566 0.880 0.075
..................................................................................................................................................................................
season summer −0.180 0.870 0.836
..................................................................................................................................................................................
behaviour Dp-no lunges 1.656 0.467 3.90 ×10−4*
.............................................................................................................................................
Sh-lunges 1.037 0.849 0.222
.............................................................................................................................................
Sh-no lunges 2.652 0.490 6.20 ×10−8*
.............................................................................................................................................
surface 0.826 0.471 0.079
..................................................................................................................................................................................
TOD day −0.719 0.234 0.002*
.............................................................................................................................................
dusk 0.583 0.100 6.80 ×10−9*
.............................................................................................................................................
night −0.086 0.138 0.532
.........................................................................................................................................................................................................................

10
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
100
75
50
25
0
Mar Apr May June
9.6 h
1 tag
92.8 h
3 tags
67.6 h
10 tags
128.4 h
5 tags
298.4 h
19 tags
July Aug Sep Oct all data
behaviour
deep-lunging
shallow-lunging
surface behaviour
deep-no lunges
shallow-no lunges
Mar Apr Ma
y
June Jul
y
Au
g
Sep Oct all data
100
75
50
night
per cent of time spent in each behaviour per month (%) per cent of time spent in each behaviour per month (%)
day
25
0
3.2 h
1 tag
17.2 h
4 tags
151.5 h
15 tags
224.5 h
58 tags
165.1 h
38 tags
14.2 h
5 tags
575.7 h
121 tags
(b)
(a)
Figure 3. Thepercentof time tagged bluewhales spent in each behaviouralstateduringthe day (a) and night(b)per month. The number
of tags containing day- or night-time data, and the total hours of data available during each diel period are also listed per month. No tags
were deployed on blue whales o southern California during the months of April or May.
produced during shallow, non-lunging dives (figure 4), and another 10% were produced at shallow
depths during deep non-lunging dives. The fewest number of sounds were produced during shallow
lunging dives (around 1% of all sounds; figure 4). In comparison to the proportion of time that blue
whales spent in deep lunging and non-lunging dives overall (figure 3), the number of sounds produced
within these behavioural states were far fewer than the number of sounds produced during shallow
lunging and surface behavioural states (figure 4).

11
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
50
per cent of sounds produced during each behavioural state (%)
25
0all sounds singular A singular B D calls phrases
behaviour
deep-lunging
shallow-lunging
deep-no lunges
shallow-no lunges
surface behaviour
75
100 n= 1,947 n= 229 n= 288 n= 550 n= 880
Figure 4. The per cent of singular A, singular B and D calls, and phrases that wereproduced within dierent behavioural states. The total
number of detections for each sound type are listed, with each count of phrases possibly combining multiple A and B units. Only sounds
that were attributed to tagged blue whales are included.
Phrase production rates in particular were significantly higher during shallow non-lunging and deep
non-lunging dives than during other dive behaviours (figures 2and 5,table 1). During the production
of A and B units within phrases, dives were consistently shallower (less than 35 m in maximum
depth) than during the production of singular calls (figure 6). Apart from surface breath intervals, blue
whales producing repetitive phrases, or singing (n=11 individuals), often regularly dived to a relatively
consistent depth for each dive over a period of hours, displaying a similar behaviour throughout the
duration of the song bout (figure 6). During these bouts, singing individuals also consistently ended each
dive with a B unit before surfacing (figure 6). Blue whale behaviours exhibited during the production of
single phrases, those neither preceded nor followed by another phrase, were less consistent (figures 6
and 7). In some cases, single phrases were produced towards the end or beginning of deeper dives by
an individual that would later begin singing (figure 6), while during other deployments, the tagged blue
whale produced both single phrases and singular calls within a short time frame (figure 7).
During the production of singular A, B and D calls, blue whale behaviour was more variable (figure 7),
with dives frequently extending beyond 35 m, although call production generally still occurred within
the upper 30 m (figures 7and 8). Both singular B and D call rates were higher during non-lunging dives
than during lunging dives (figures 2,4and 5,table 1). D call rates were also higher during extended
surface behaviour than during lunging dives (figures 2,4and 5,table 1), and D call production rates
were the highest of any sound type during these surface bouts (figures 2,4and 5,table 1). Differences
in singular A call production rates between different behavioural states were insignificant (figures 2and
5,table 1). However, A calls were produced in higher numbers than other sound types during shallow
lunging dives (figure 4).
Regardless of dive behaviour, the majority of all blue whale sounds attributed to tagged animals were
produced in the top 30m of the water column (figure 8). Both singular A and B calls most frequently
occurred between 15 and 20 m. When produced as parts of a phrase, A and B units were most commonly
produced between 25 and 30 m, but they were also frequently produced between 15 and 20 m. In general,
D calls were regularly produced at shallow depths ranging from 5 to 20m.

12
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
8
6
4
2
0
deep-lunging deep-no lunges shallow-no lunges surface behaviour
behavioural state
mean production rate (sounds h−1)
shallow-lunging
singular A
sound type
singular B
D calls
phrases
Figure 5. Meanhourly production ratesof singular A(orange),singular B (blue)and D calls (green), and phrases(purple) within dierent
behavioural states. Bars represent the standard error of the mean, calculated for each sound type. Only data from tag deployments
containing calls attributed to the tagged whale are included. Production rates were calculated using only calls that were assigned to
the tagged individuals.
Time-of-day had a significant impact on blue whale call and phrase production rates. In general, the
production rates of all sound types showed crepuscular increases compared to other time-of-day periods.
Singular A call rates were lowest during the day and highest at dusk, but differences were not significant
(figure 2,table 1). Singular B call rates, on the other hand, were lower during the day and at night than
during dawn or dusk, but differences were not significant (figure 2,table 1). D call production rates
were significantly higher at dusk and similar across the other three periods (figure 2,table 1). Phrase
production rates were also highest at dusk, but significantly lower during the day than during the night
or at dawn (figure 2,table 1).
4. Discussion
We observed significant variability in blue whale call and phrase production rates with respect to
different behavioural states, in addition to spatial and temporal patterns, and across sexes. The majority
of all sounds attributed to tagged blue whales were produced at shallow depths (less than 30m) during
shallow non-lunging dives. However, there were distinct differences between the behaviours associated
with the production of singular calls versus phrases. Singular A, B and D calls were more frequently
produced during non-lunging dives or during bouts of surface behaviour, and phrases were typically
produced during shallow non-lunging dives. Furthermore, while the large majority of the tagged blue
whales in this study were females, very few of those tags had sounds attributed to them (5%), while the
attribution of sounds to tagged male blue whales (50%) was much more common.
The consistent behaviour exhibited during the production of repetitive phrases suggests an exclusive
behavioural state for blue whales. Similar to the observations made by Stimpert et al. [56]onfin
whales, tagged blue whales that were producing repetitive phrases, or singing, repeatedly made
long, stereotypically ‘U-shaped dives’ to a consistent shallow depth, often for extended periods of
time. In general, the behaviours associated with song production in male blue whales have been

13
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
0
50
100
150
depth (m)
200
250
16 August 2011
10.30
14.30 18.30 22.30 02.30 17 August 2011
08.15
18.3016.30
40
30
20
10
0
singular A
singular B phrase: B unit
phrase: A unit
local time (hh.mm)
Figure 6. Blue whale dive prole recorded from an Acousonde deployed on 16 August 2011, showing extended singing behaviour. Two
hoursof the repetitive phrase, or song,boutareshown in the inset.Nightis shaded ingrey.Only soundsattributedto the tagged individual
are included.
described as solitary and travelling, not feeding [33]. When foraging, blue whales exhibit lunge-feeding
behaviours [64,72,76,77], often exploiting prey patches at depth [33,78]. The absence of dives containing
vertical lunges during song bouts, coupled with the shallow production depths of repetitive song
phrases, supports the hypothesis that feeding and singing behaviours in blue whales may be mutually
exclusive states [32,33,41,79]. As song has only been recorded from male blue whales and is thought to
be associated with reproduction [31,33], the repetitive production of phrases to form long song bouts in
the absence of foraging could also be used as an indicator of the singing male’s condition to potential
mates [80–82].
Contrary to the consistent diving behaviour exhibited by singing blue whales, the behaviours that we
observed from tagged individuals producing singular calls were much more variable, which suggests
that singular A and/or B calls may have a distinct behavioural purpose from A and B sounds produced
as units within phrases. Blue whales in both the Pacific and Atlantic have been shown to exhibit different
behaviours when producing singular calls versus singing, with the former more frequently engaging in
feeding, milling and resting, and the latter in travelling behaviours [33,83]. Additionally, the infrequent
production of A and/or B calls may be used by males to maintain pair bonds during feeding [33]. The
use of foraging-associated calls has been well documented in many bird and mammalian species, and in
several cases, these calls may be used for reproductive benefits [8]. For example, Krunkelsven et al. [84]
observed that captive male bonobos used food-associated calls to attract females to a food source, which
ultimately led to copulations between the females and calling males.
Although singular A and B calls were previously only recorded from male blue whales [33], we also
recorded singular A and B calls on tags deployed on three female individuals (electronic supplementary
material, table S1). The depths at which these singular calls were recorded were similar to those observed
in other deployments, and the tagged females exhibited no unusual diving behaviour. While numerous
examples exist in insect and frog species of females producing similar signals to those produced by
males [4,85,86], focal follow data collected during and after each of these deployments indicate that
all three females were tagged while interacting closely with another whale as part of a pair (electronic
supplementary material, table S1). Although these calls were attributed to the tagged females based on

14
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
18 September to 19 September 2016
0
50
100
150
depth (m)depth (m)
200
250
18.57
02.45
02.00
05.45
19 September to 20 September 2016
11.00
0
50
100
150
200
250
19.00 02.00 11.00 19.00
singular A
singular B
D call
phrase: A unit
phrase: B unit
local time (hh.mm)
50
40
30
20
10
0
(b)
(a)
19.00
Figure 7. Blue whale dive prole recorded from an Acousonde deployed on 18 September 2016, showing variability between the
behaviours associated with singular call and phrase production. Only 48 h from the full 5 d deployment are shown: (a)therst24h
from 18 September to 19 September 2016; and (b) 19 September to 20 September 2016. Three hours from 19/09/2016 are highlighted in
the inset to illustrate behaviour during singular call and single phrase production. Night is shaded in grey. Only sounds attributed to the
tagged individual are included.
our methods, the calculated RLrms and SNR values were lower than the majority of other singular calls.
Therefore, it is possible that the recorded sounds could have been produced by the nearby, untagged
individuals of unknown sex rather than the tagged females.
We found production rates of D calls, which have previously been associated with foraging [33,41], to
be significantly higher during shallow non-lunging dives and periods of surface behaviour than during
other dive types. This increase in D calling at shallower depths during non-lunging dives, coupled with
the fact that D calls were apparently recorded from both males and females [33], suggests that D calls
may be used in multiple contexts, although we reiterate that our methods did not allow for evaluation
of whether bouts of surface behaviour included feeding events, so foraging could have been occurring
during surface behaviour. Many species may use food-associated calls as broadcast advertisements [8],
but most mammals and birds produce such calls at reduced amplitudes, durations and rates compared
to social calls [87]. Blue whale D calls, which are both shorter in duration and produced less frequently
than A or B calls, also have a reduced propagation range compared to the propagation capability of B
calls [31,88,89]. Thus, D calls are probably used as shorter-range social calls between nearby individuals.
However, it is important to point out that a single blue whale tag deployment from September 2016
contained approximately 66% (364 of 550) of all D calls (figure 8; electronic supplementary material,
table S1). This particular whale spent extended amounts of time at or near the surface (22.9 out of 102.5 h
of data collected), so it is possible that the observed differences in D call production rates may be largely
driven by this individual’s behaviour. Additionally, although D calls have been attributed to female blue
whales in previous studies [33], we only recorded D calls on two tags deployed on females (electronic

15
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
80
singular A calls singular B calls D calls
phrase: A units
no. sounds detected within depth range
phrase: B units
100 140
120
100
80
60
40
20
0
400
250
300
200
150
100
50
0
300
200
100
0
0
10 20 30 40 50 60
0
10 20 30 40 50 60
depth at sound start (m)
depth at sound start (m)
depth at sound start (m)depth at sound start (m)depth at sound start (m)
80
60
40
20
0
010 20 30 40 50 60 01020
30 40 50 60
60
40
20
0
0 10203040 6050
Figure 8. Histogramsof depth atthe start of soundproduction for allsingular A, singular B, D calls, and phraseunits attributed to tagged
individuals. Note dierent y-axis scales.
supplementary material, table S1). This difference could be due to differences in study locations, because
our analysis included only data collected from individuals tagged off southern California and many of
them near shore, or to the assigning of calls to the tagged animal. Alternatively, this could also imply
that females do not produce sounds very often during the summer, when most tag data were collected.
Based on these data, we suggest that there are potential behavioural advantages for sound production
at shallower depths. Despite differences observed between dive type and behavioural state during the
production of singular calls and phrases, tagged blue whales produced the majority of sounds at shallow
depths, generally within 30 m of the surface. This depth range is consistent with the average depths
of blue whale calling (20–30 m) that Oleson et al. [33] reported using a subset of these data, and is
also similar to, albeit a bit deeper than, the average depths of fin whale calling (10–15 m) recorded by
Stimpert et al. [56] from tag deployments off southern California. Additionally, our results are consistent
with the average depth of B call production proposed in a theoretical model of blue whale sound
production by Aroyan et al. [70]. Oleson et al. [33] speculated that signal output may be maximized
at these shallow calling depths. In their model of fin whale calling, Weirathmueller et al. [90] show that
fin whale 20 Hz pulses increased in amplitude and range when produced between 25–30 m, as well as
between 65–70 m. For male blue whales that are seeking mates or producing male–male displays [91]and
thus singing for extended periods of time, the ability to communicate over longer ranges with minimal
energy expenditure would be advantageous. The gradual changes in dive depths during the production
of repetitive song phrases observed in the data collected from one individual (figure 6), which were
not associated with any systematic data drifts, may illustrate an inadvertent upward drift of the animal
while singing and could be related to body condition [92]. Similar dive drifts have also been linked
to foraging and the ratio of fat to lean body mass in elephant seals [93]. This drift was not noted in
other whales diving to similar depths. Although details regarding the sound production mechanism in
baleen whales remain largely unknown, a recent model presented by Dziak et al.[69] suggests a pulsed-
air mechanism for B call production in blue whales that is anatomically similar to the model proposed
for humpback whales [94]. Simulations run on humpback whale songs recorded off Hawaii suggest that

16
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
male humpbacks may select the optimum singing depth and frequency in order to maximize propagation
potential [95]. The consistent behaviour that we recorded from tagged individuals during bouts of song
production certainly indicates that blue whales may be achieving the same benefit by producing calls at
these particular depths.
The seasonal and diel differences that we observed in blue whale sound production rates may be
correlated to temporal changes in behaviour. Production rates of all sound types were higher during the
autumn than during the summer, although only differences between singular A and D call rates were
significant. The production of sounds by tagged individuals during each month of tag deployments, with
the exception of March, is similar to the occurrence of B calls reported from long-term studies [41,96]. In
contrast to the summer peak in D calling recorded in these studies [33,41], D call production rates in
our data were greater during the autumn. Our analysis also showed significant trends in blue whale
call production rates with respect to time-of-day: singular B and D call rates were highest at dusk,
while singular A call rates increased both at dawn and dusk. These patterns are similar to the B calling
peaks reported during twilight hours [32]. Blue whale phrase production rates in our dataset were also
highest at dusk and significantly lower during the day than at night or dawn. As blue whales typically
prey upon diel-migrating euphausiids, which are found in high concentrations at depth during the
day [76,77,97,98], individuals may time periods of singing behaviour to coincide with periods of reduced
prey availability [32,33]. Similar behavioural trade-offs have been observed during the day in European
robins [99,100], and in nightingales [101], who adjust their nightly singing rates at dusk according to the
energy reserves they have accumulated during the day.
The proportion of time blue whales spent in different non-acoustic behavioural states also shifts
seasonally. The amount of time that tagged blue whales spent in deep diving states, specifically deep
lunging dives indicative of foraging [33,76–78], was generally greatest during the summer months,
between June and August. The single blue whale tagged in March of 2015 also primarily exhibited
deep and shallow lunging dive behaviours. However, shallow non-lunging behaviours, which might
be more commonly associated with song production [33], comprised the majority of all hours of data
collected during the autumn, between September and October. In many mammalian species, including
elephants [102], pinnipeds [103] and ungulates [104–107], the limitation or cessation of food consumption
by males during the breeding season is common—the motivation for which appears to be the reduction
of foraging time in favour of breeding efforts. While it has previously been accepted that both sexes
of several baleen whale species also limit food intake during their annual migrations to breeding
grounds [89], blue whales may exhibit both foraging and reproductive behaviours on their low-latitude
breeding grounds [108,109]. Therefore, it is also possible that multiple behavioural states may also be
exhibited at feeding sites. Although southern California is primarily considered to be a seasonal feeding
ground for blue whales, our analysis indicates a temporal separation between two behavioural states for
blue whales in this region: individuals may begin feeding as early as spring, continuing through summer
until the autumn, at which time reproductive behaviours begin to dominate. Unfortunately, little-to-no
data were collected from late autumn to spring, so it is difficult to assess whether these behavioural
trends extend to adjacent months and seasons.
We recorded significant behavioural differences between calling and singing blue whales, in addition
to both spatial and temporal patterns in call and phrase production rates, but also discovered several
substantial biases in tagging effort. First and foremost, although the overall sex ratio of the Northeast
Pacific population of blue whales is estimated to be about even, our data show a clear bias towards
tagging of female blue whales, because 72% (57 out of 79) of the individuals for which sex determination
occurred were females. In addition, the majority (95%) of all tag deployments occurred inshore of the
Channel Islands. In mammals, males generally disperse more than females [110–113], while females of
many species are typically more social [114]. Our sex bias in tag deployments off southern California
suggests that female blue whales may spend more time inshore, and are thus more likely to be
encountered, compared to males. Furthermore, the ability to tag blue whales varies with respect to the
individual whale’s behavioural state: it is much easier to deploy a tag on an animal that is resting or
milling than on an animal that is travelling quickly. Because deployment requires a close approach to an
individual, how different sexes respond to close approach could bias towards a particular sex. Previous
studies on blue whales off southern California have observed both slow, shallow diving behaviour in
females [31] as well as increased singing from solitary males travelling offshore [33]. Therefore, it is
possible that the tagging bias arose not only from differential use of habitat by male and female blue
whales in this area but also from sexual differences in behaviour or approachability.
In addition to the sex bias that we discovered in tag deployments, the data analysed in our study
were collected from opportunistic tagging efforts that occurred primarily between July and September.

17
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
Therefore, the autumn peak in D call production in our dataset may be due to the paucity of data
collected during late-spring and early-summer deployments, which is when D calls have been shown
to increase in other studies [33,40,41]. However, approximately 87% (481 out of 550) of all D calls
were produced by two blue whales (eight total individuals produced D calls) tagged in September of
2010 and September of 2016, so the increase in D call production rates during the autumn is biased
by the behaviour of these two individuals. Furthermore, over the years of tag deployments, different
survey efforts were targeting animals in specific behavioural states, or under different environmental
conditions and geographical locations for different studies. Therefore, our collection of tag data cannot
be considered a truly random sample of the population. Overall, understanding these biases is critical to
allowing us to better understand the behavioural context of sound production rates, which is needed for
better interpretation of the ecology and habitat use of blue whales from passive acoustic data.
The data collected through the use of multi-sensor tags can allow for the assessment of the behavioural
context of sound production. However, there are several limitations associated with studying blue whale
acoustic behaviour through tag deployments. Most importantly, the assignment of any recorded sound
to the tagged individual rather than a nearby whale is not a straightforward task. Recent studies on fin
whales have indicated that caller identity can be confirmed based on detection of sounds in accelerometer
data [55,56]; however, this method is only applicable to tags capable of recording high-sample-rate
accelerometer data and, further, has proved less successful for the longer-duration and higher-frequency
sounds produced by blue whales [115]. Owing to these issues, and to standardize analysis among our
dataset that included a large proportion of tags without high-frequency three-dimensional accelerometer
sampling (36%), we assigned recorded sounds to tagged animals based on calculated relative RLrms and
SNR for all sounds detected within a given tag deployment. Based on these methods, it is possible that
some of the sounds that we attributed to tagged individuals may have been produced by another whale
swimming nearby, particularly for whales associated in pairs as is typical for blue whales; however, if
this were the case, it is possible that the adjacent whale was engaging in the same general behaviour
as the tagged individual [116]. In addition, based on our methods, deployments containing many calls
that were all relatively faint would still have had calls attributed to the tagged individual. An alternative
approach would have been to use a universal RLrms or SNR cut-off to attribute calls to tagged blue
whales; however, our data were complicated by the use of multiple tag types, each with its own unique
system specifications, so RLrms levels were often not comparable. We chose not to use a universal
SNR cut-off because the SNR also varies by tag type and deployment given differences in hardware
configuration. Thus, because we cannot say with absolute certainty that tagged individuals produced all
of the calls attributed to them, our reported call and phrase production rates should not be considered
absolute production rates from any particular individual. However, we are confident that the broad
behavioural contexts associated with calling in this analysis give insight into general blue whale acoustic
behaviour across sexes and behavioural states.
There are a number of threats facing the northeast Pacific population of blue whales, particularly
ship strikes and anthropogenic noise [65,117–119]. Understanding baseline behaviours of this population
is critical to being able to evaluate possible impacts of anthropogenic activity to this and other
animal populations. Although the sound types produced by blue whales in this area have been well
described [31–33,39,79], still relatively little is known about the behaviours associated with sound
production. Overall, our analysis provides valuable insight into how blue whale acoustic behaviour
varies on different temporal, spatial and, most importantly, behavioural scales. Understanding this
variability and establishing a baseline for the behaviours, in addition to recognizing the biases and
limitations associated with the use of tag data, is critical to allowing us to better understand the
behavioural context of sound production. Such a baseline will, in turn, allow for better interpretation
of the ecology and habitat use of blue whales from passive acoustic data and will provide a broader
context for interpretation of data from studies on the impact of human activities on this population.
Ethics. All tag deployments were conducted in accordance with the ethical standards under CRC’s Institutional Animal
Care and Use Committee protocols (AUP-6), and under NMFS permits 14346 and 19116 (B.L.S., principal investigator)
and 540-1502-00, 540-1811 and 16111 (J.C., principal investigator). No additional permits were required to carry out
fieldwork.
Data accessibility. The datasets supporting this article have been uploaded as part of the electronic supplementary
material.
Authors’ contributions. L.A.L. analysed all the data, conducted statistical analyses and wrote the manuscript. J.C.
participated in data collection and design of the study. A.K.S. participated in some data collection and facilitated
analysis of acoustic data. J.F. participated in some data collection. A.S.F. participated in some data collection. M.F.M.
participated in some data collection and facilitated analysis of acoustic data. S.L.M. facilitated access to sex data.

18
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
E.M.O. participated in some data collection and facilitated analysis of acoustic data. B.L.S. facilitated access to
SOCAL-BRS data. A.R.S. participated in some data collection and helped with analysis of dive data. A.Š. designed
and coordinated the study, participated in some data collection, guided statistical analyses and helped draft the
manuscript. All the authors reviewed and contributed to the final document edits. All the authors gave their final
approval for publication.
Competing interests. The authors declare no competing interests.
Funding. Tag deployment efforts were supported over the years by funding from the Office of Naval Research (ONR)
Marine Mammal Program to different research programmes for J.C., E.M.O. and A.Š., as well as the SOCAL BRS field
effort, which was primarily supported by the U.S. Navy’s Living Marine Resources (LMR) Program, with additional
support from the Office of Naval Research. Analysis was supported by ONR (Dr Michael Weise) grant no. N00014-
14-1-0414. L.A.L. was supported with funding from the National Science Foundation Graduate Research Fellowship
Program under grant no. DGE 1144086.
Acknowledgements. The authors would like to thank everyone involved at the CRC and the entire SOCAL BRS team,
especially Ann Allen and Jeremy Goldbogen, for their help with data collection, organization and analysis throughout
the course of this project. We also thank John Hildebrand for his support of data collection and analysis over the
years. We acknowledge the funding support of Dr Michael Weise and Dana Belden (Office of Naval Research Marine
Mammal Program).
References
1. Catchpole CK, Slater PJ. 2003 Bird song: biological
themes and variations. Cambridge, UK: Cambridge
University Press.
2. Kroodsma DE, Miller EH. 1996 Ecology and
evolution of acoustic communicationin birds.
Ithaca, NY: Cornell University Press.
3. Marler PR, Slabbekoorn H. 2004 Nature’s music: the
science of birdsong. London, UK: AcademicPress.
4. Emerson SB, Boyd SK. 1999 Mating vocalizationsof
female frogs: control and evolutionary
mechanisms. Brain Behav.Evol. 53, 187–197.
(doi:10.1159/000006594)
5. Gerhardt HC, TannerSD, Corrigan CM, Walton HC.
2000 Female preference functions based on call
duration in the gray tree frog(Hyla versicolor).
Behav.Ecol. 11, 663–669. (doi:10.1093/beheco/
11.6.663)
6. ForrestT. 1983 Calling songs and mate choice in
mole crickets. In Orthopteran mating systems:
sexual competition in a diverse groupof insects (eds
DT Gwynne, GK Morris), pp. 185–204. Boulder,CO:
Westview Press.
7. Charlton BD, Reby D, McComb K. 2007 Female red
deer prefer the roars of larger males. Biol.Lett. 3,
382–385. (doi:10.1098/rsbl.2007.0244)
8. Clay Z, Smith CL, BlumsteinDT. 2012
Food-associated vocalizationsin mammals and
birds: what do these calls really mean? Anim.
Behav.83,323–330.(doi:10.1016/j.anbehav.2011.
12.008)
9. Mitani JC, Nishida T. 1993 Contextsand social
correlates of long-distance calling by male
chimpanzees. Anim. Behav.45, 735–746.
(doi:10.1006/anbe.1993.1088)
10. Zuberbühler K, Noë R, Seyfarth RM. 1997 Diana
monkey long-distance calls: messages for
conspecics and predators. Anim.B ehav.
53, 589–604. (doi:10.1006/anbe.1996.0334)
11. Evans CS, Evans L, Marler P. 1993 On the meaning
of alarm calls: functional reference in an avian
vocal syste m. Anim.Behav. 46, 23–38.
(doi:10.1006/anbe.1993.1158)
12. Bee MA, Bowling AC. 2002 Socially mediated pitch
alteration by territorial male bullfrogs,Rana
catesbeiana.J. Herpetol.36, 140–143. (doi:10.1670/
0022-1511(2002)036[0140:smpabt]2.0.co;2)
13. Montgomery JC, Jes A, Simpson SD, Meekan M,
Tindle C. 2006 Sound as an orientation cue for the
pelagic larvae of reef shes and decapod
crustaceans. In Advances in marine biology,
pp. 143–196. Elsevier.(doi:10.1016/s0065-2881(06)
51003-x)
14. Amorim MCP. 2006 Diversity of sound production
in sh. In Communicationin shes (eds F Ladich, SP
Collin, P Moller, BG Kapoor), pp. 71–104. Eneld,
NH: Science Publishers.
15. Hawkins A. 1986 Underwater sound and sh
behaviour.In The behaviour of teleost shes (ed. TJ
Pitcher), pp. 114–151. New York, NY: Springer.
16. PopperAN, Fay RR. 1993 Sound detection and
processing by sh: critical review and major
research questions (Part 1 of 2). BrainB ehav.Evol.
41,14–25.(doi:10.1159/000113821)
17. RichardsonWJ,GreeneCRJ,MalmeCI,Thomson
DH. 1995 Marine mammal sounds.SanDiego,
CA: Academic Press.
18. TyackPL, Clark CW. 2000 Communication and
acoustic behavior of dolphins and whales. In
Hearing by whales and dolphins (eds WW Au,
AN Popper,RN Fay), pp. 156–224. New York,NY:
Springer.
19. Thompson TJ,Winn HE, Perkins PJ. 1979 Mysticete
sounds. In Behavior of marine animals,
pp. 403–431. Berlin, Germany: Springer.
20. Au WW. 2012 The sonar of dolphins.NewYork,NY:
Springer.
21. Clark CW. 1990 Acoustic behavior of mysticete
whales. In Sensory abilities of cetaceans (eds J
Thomas, R Kastelein), pp. 571–583. New York,NY:
Springer.
22. Edds-Walton PL. 1997 Acoustic communication
signals of mysticete whales. Bioacoustics 8, 47–60.
(10.1080/09524622.1997.9753353)
23. PayneRS, McVay S. 1971 Songs of humpback
whales. Science 173, 585–597.
(doi:10.1126/science.173.3997.585)
24. Winn HE, Winn LK. 1978 The song of the humpback
whale Megaptera novaeangliae in the WestIndies.
Mar.Biol. 47,97–114.(doi:10.1007/bf003
95631)
25. D’VincentCG,NilsonRM,HannaRE.1985
Vocalizationandcoordinatedfeedingbehaviorof
the humpback whale in southeastern Alaska. Sci.
Rep. Whales Res.Inst. 36,41–47.
26. SilberGK. 1986 The relationship of social
vocalizations to surface behavior and aggression in
the Hawaiian humpback whale (Megaptera
novaeangliae). Can. J. Zool.64, 2075–2080.
(doi:10.1139/z86-316)
27. Clark CW. 1983 Acoustic communication and
behavior of the southern right whale (Eubalaena
australis). In Communicationand behavior of
whales (ed. R Payne),pp. 163–198. Boulder, CO:
Westview Press.
28. Parks SE, Tyack PL. 2005 Sound production by
North Atlantic right whales (Eubalaena glacialis)in
surfaceactivegroups.J.Acoust. Soc. Am. 117,
3297–3306. (doi:10.1121/1.1882946)
29. Parks SE, Hamilton PK, Kraus SD, Tyack PL. 2005
The gunshot sound produced by male North
Atlantic right whales (Eubalaena glacialis)andits
potential function in reproductive advertisement.
Mar.Mamm. Sci. 21, 458–475.
(doi:10.1111/j.1748-7692.2005.tb01244.x)
30. Thode AM, D’Spain GL, Kuperman WA. 2000
Matched-eld processing,geoacoustic inversion,
and source si gnature recovery of b lue whale
vocalizat ions. J. Acoust.S oc.Am. 107, 1286–1300.
(doi:10.1121/1.428417)
31. McDonald MA, Calambokidis J, Teranishi AM,
Hildebrand JA. 2001 The acoustic calls of blue
whales o California with gender data. J. Acoust.
Soc. Am. 109,1728–1735.(doi:10.1121/1.1353593)
32. WigginsSM, Oleson EM, McDonald MA,
Hildebrand JA. 2005 Blue whale (Balaenoptera
musculus) diel call patterns oshore of Southern
California. Aquat. Mamm. 31, 161–168.
(doi:10.1578/am.31.2.2005.161)
33. Oleson E, Calambokidis J, BurgessW, McDonald M,
LeDuc C, Hildebrand J. 2007 Behavioral context of
call production by eastern North Pacic blue
whales. Ma r. Ecol. P rog. S er. 330,269–284.
(doi:10.3354/meps330269)
34. Staord KM, Nieukirk SL, Fox CG. 2001 Geographic
and seasonal variation of blue whale calls in the
North Pacic. J. Cetacean Res. Manag. 3, 65–76.
35. ThompsonPO, Findley LT, Vidal O, Cummings WC.
1996 Underwater sounds of blue whales,

19
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
Balaenoptera musculus, in the Gulf of California,
Mexico. Mar. Mamm. Sci. 12, 288–293.
(doi:10.1111/j.1748-7692.1996.tb00578.x)
36. ClarkCW, Fristrup KM. 1997 Whales ‘95: A
combinedvisualandacousticsurveyofblueand
n whales o southern California. Rep.Int.
Whaling Comm. 47, 583–600.
37. Rivers JA. 1997 Blue whale, Balaenopteramusculus,
vocalizations from the waterso central
California. Mar.Mamm. Sci. 13, 186–195.
(10.1111/j.1748-7692.1997.tb00626.x)
38. Staord KM, Nieukirk SL, Fox CG. 1999 An acoustic
link between blue whales in the eastern tropical
Pacic and the northeast Pacic. Mar.Mamm. Sci.
15, 1258–1268.
(doi:10.1111/j.1748-7692.1999.tb00889.x)
39. McDonaldMA, Mesnick SL, Hildebrand JA. 2006
Biogeographic characterization of blue whale song
worldwide: using song to identify populations.
J. Cetacean Res. Manag. 8, 55–65.
40. Lewis LA, Širović A. 2017 Variabilityin blue whale
acoustic behavior o southern California. Mar.
Mamm. Sci. 34, 311–329 (doi:10.1111/mms.12458)
41. Oleson EM, Wiggins SM, Hildebrand JA. 2007
Temporalseparation of blue whale call types on a
southern California feeding ground. Anim. Behav.
74, 881–894. (doi:10.1016/j.anbehav.2007.01.022)
42. Burtenshaw JC, Oleson EM, Hildebrand JA,
McDonald MA, Andrew RK, Howe BM, MercerJ.
2004 Acoustic and satellite remotesensing of blue
whale seasonality and habitat in the northeast
Pacic. Deep Sea Res. PartII Top. Stud. Oceanogr.51,
967–986. (doi:10.1016/s0967-0645(04)00095-5)
43. Širovic A, Rice A, Chou E, Hildebrand J, Wiggins S,
Roch M. 2015 Seven years of blue and n whale
call abundance in the Southern California Bight.
Endanger.Species Res. 28,61–76.
(doi:10.3354/esr00676)
44. Mellinger DK, Clark CW. 2003 Blue whale
(Balaenoptera musculus) sounds fromthe Nor th
Atlantic. J. Acoust. Soc. Am. 114, 1108–1119.
(doi:10.1121/1.1593066)
45. Rankin S, Ljungblad D, Clark C, Kato H. 2005
Vocalisations of Antarctic blue whales,
Balaenoptera musculus intermedia,recorded
during the 2001/2002 and 2002/2003 IWC/SOWER
circumpo lar cruises, Area V, Antarc tica. J. Cetacean
Res. Manag. 7,13–20.
46. Cooke S. 2008 Biotelemetry and biologging in
endangered species research and animal
conservation: relevance to regional,national, and
IUCN Red List threat assessments. Endanger.
Species Res. 4,165–185.(doi:10.3354/esr00063)
47. Rutz C, Hays GC. 2009 New frontiers in biologging
science. Biol. Lett.5, 289–292.
(doi:10.1098/rsbl.2009.0089)
48. Bograd S, Block B, Costa D, Godley B. 2010
Biologging technologies: new tools for
conservation. Introduction. Endanger.Species Res.
10,1–7.(doi:10.3354/esr00269)
49. Block BA. 2005 Physiological ecology in the 21st
century: advancements in biologging science.
Integr.Comp. Biol. 45, 305–320.
(doi:10.1093/icb/45.2.305)
50. Johnson M, Aguilar de Soto N, Madsen P. 2009
Studying the behaviour and sensory ecology of
marine mammals using acoustic recording tags: a
review. Ma r. Ecol . Prog . Ser. 395, 55–73.
(doi:10.3354/meps08255)
51. Stimpert AK, Wiley DN, Au WWL, Johnson MP,
Arsenault R. 2007 ‘Megapclicks’: acoustic click
trainsandbuzzesproducedduringnight-time
foraging of humpback whales (Megaptera
novaeangliae). Biol.Lett. 3,467–470.
(doi:10.1098/rsbl.2007.0281)
52. Parks SE, Cusano DA, Stimpert AK, Weinrich MT,
Friedlaender AS, Wiley DN. 2014 Evidence for
acoustic communication among bottom foraging
humpback whales. Sci. Rep. 4, 75–108.
(doi:10.1038/srep07508)
53. Parks S, Searby A, Célérier A, Johnson M, Nowacek
D, TyackP. 2011 S ound production behaviorof
individual North Atlantic right whales:
implications for passive acoustic monitoring.
Endanger.Species Res. 15,63–76.
(doi:10.3354/esr00368)
54. Risch D et al. 2014 Mysterious bio-duck sound
attributed to the Antarctic minke whale
(Balaenoptera bonaerensis).Biol. Lett. 10,20140175.
(doi:10.1098/rsbl.2014.0175)
55. Goldbogen JA, Stimpert AK, DeRuiter SL,
Calambokidis J, Friedlaender AS, Schorr GS,
Moretti DJ, TyackPL, Southall BL. 2014 Using
accelerometers to determi ne the calling
behavior of tagged baleen whales. J. Exp.
Biol. 217, 2449–2455.(doi:10.1242/jeb.103259)
56. Stimpert AK et al. 2015Soundproductionand
associated behavior of tagged n whales
(Balaenoptera physalus) in the Southern California
Bight. Anim. Biotelem. 3,23.
(doi:10.1186/s40317-015-0058-3)
57. Calambokidis J, Schorr GS, Steiger GH, Francis J,
Bakhtiari M, Marshall G, Oleson EM, Gendron D,
Robertson K. 2007 Insights into the underwater
diving, feeding, and calling behavior of blue
whales from a suction-cup-attached
video-imaging tag (Crittercam). Mar.Technol. Soc.
J. 41,19–29.(doi:10.4031/002533207787441980)
58. Küsel ET, Mellinger DK, Thomas L, Marques TA,
Moretti D, WardJ. 2011 Cetacean population
density estimation from single xed sensors using
passive acoustics. J. Acoust.Soc. Am. 129,
3610–3622. (doi:10.1121/1.3583504)
59. MarquesTA,ThomasL,WardJ,DiMarzioN,Tyack
PL. 2009 Estimating cetacean population density
using xed passive acoustic sensors: an example
with Blainville’s beaked whales. J. Acoust. Soc. Am.
125, 1982–1994. (doi:10.1121/1.3089590)
60. Marques T, Munger L, Thomas L, WigginsS,
Hildebrand J. 2011 Estimating North Pacic right
whale Eubalaena japonica density using passive
acoustic cue counting. Endanger. Species Res. 13,
163–172. (doi:10.3354/esr00325)
61. Szesciorka AR, Calambokidis J, Harvey JT. 2016
Testingtag attachments to increase the
attachment duration of archival tags on baleen
whales. Anim. Biotelem. 4,18.
(doi:10.1186/s40317-016-0110-y)
62. Southall BL, Moretti D, Abraham B, Calambokidis
J, DeRuiter SL, TyackPL. 2012 Marine mammal
behavioral response studies in southern California:
advances in technology and experimental
methods. Mar.Technol. Soc. J. 46,48–59.
(doi:10.4031/mtsj.46.4.1)
63. Johnson MP, Tyack PL. 2003 A digital acoustic
recording tag for measuring the response of wild
marine mammals to sound. IEEE J. Ocean. Eng. 28,
3–12. (doi:10.1109/joe.2002.808212)
64. Goldbogen JA. 2006 Kinematics of foraging dives
andlunge-feedinginnwhales.J. Exp. Biol. 209,
1231–1244. (doi:10.1242/jeb.02135)
65. Goldbogen JA et al. 2013 Blue whales respond to
simulated mid-frequency military sonar. Proc. R.
Soc. B 280, 20130657. (doi:10.1098/rspb.2013.0657)
66. Southall B, Nowacek D, Miller P, TyackP. 2016
Experimental eld studies to measure behavioral
responses of cetaceans to sonar. Endanger. Species
Res. 31,293–315.(doi:10.3354/esr00764)
67. Wiggins SM, Hildebrand JA. 2007 High-frequency
Acoustic Recording Package(HARP) for
broad-band, long-term marine mammal
monitoring. In 2007 Symp.on Underwater
Technology andWorkshop on Scientic Use of
Submarine Cables and Related Technologies,Tokyo,
Japan, 17–20 April. IEEE.
(doi:10.1109/ut.2007.370760)
68. Parks SE, Johnson M, Nowacek D, Tyack PL. 2010
Individual right whales call louder in increased
environmental noise. Biol.Lett. 7, 33–35.
(doi:10.1098/rsbl.2010.0451)
69. Dziak RP, Haxel JH, Lau T-K,Heimlich S,
Caplan-Auerbach J, Mellinger DK, Matsumoto H,
Mate B. 2017 A pulsed-air model of blue whale B
call vocalizations. Sci. Rep.7,9122.
(doi:10.1038/s41598-017-09423-7)
70. Aroyan JL, McDonald MA, Webb SC, Hildebrand JA,
Clark D, Laitman JT, Reidenberg JS. 2000 Acoustic
models of sound production and propagation. In
Hearing by whales and dolphins (eds WW Au,
AN Popper,RN Fay), pp. 409–469 New York,NY:
Springer.
71. Yan J. 2002 Geepack: yet another package for
generalized estimating equations. RNews2,
12–14.
72. Goldbogen JA, Calambokidis J, Friedlaender AS,
FrancisJ,DeRuiterSL,StimpertAK,FalconeE,
Southall BL. 2012 Underwater acrobatics
by the world’slargest predator: 360 rolling
manoeuvres by lunge-feeding blue whales.
Biol. Lett.9, 20120986. (doi:10.1098/rsbl.2012.0986)
73. Hardin JW, Hilbe JM. 2012 Generalizedestimating
equations, 2nd edn. Boca Raton, FL: CRC Press.
74. Halekoh U, Højsgaard S, Yan J. 2006 The R package
geepack for generalized estimating equations.
J. Stat. Softw. 15, 1–11. (doi:10.18637/jss.v015.i02)
75. R. Core Team. 2015 R: a language and environment
for statistical computing.Vienna,Austria:R
Foundation for Statistical Computing.
76. Croll DA et al. 1998 An integrated approach to the
foraging ecology of marine birds and mammals.
Deep Sea Res. Part II Top. Stud. Oceanogr. 45,
1353–1371. (doi:10.1016/s0967-0645(98)00031-9)
77. Acevedo-Gutierrez A, Croll D, Tershy B. 2002 High
feeding costs limit dive time in the largest whales.
J. Exp. Biol.205,1747–1753.
78. Croll DA, Acevedo-Gutiérrez A, TershyBR,
Urbán-Ramírez J. 2001 The diving behavior of blue
and n whales: is dive duration shorter than
expected based on oxygen stores? Comp.Biochem.
Phys iol., Par t A Mol. Inte gr. Physio l. 129, 797–809.
(doi:10.1016/s1095-6433(01)00348-8)
79. Staord KM, Moore SE, Fox CG. 2005 Diel variation
in blue whale calls recorded in the eastern tropical
Pacic. Anim. Behav.69, 951–958.
(doi:10.1016/j.anbehav.2004.06.025)
80. Sullivan BK, KwiatkowskiMA. 2007 Courtship
displays in anurans and lizards: theoretical and

20
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 180241
................................................
empirical contributions to our understanding of
costs and selection on males due to female choice.
Funct. Eco l. 21, 666–675. (doi:10.1111/j.1365-2435.
2007.01244.x)
81. VehrencampSL, Bradbury JW, Gibson RM. 1989
The energetic cost of display in male sage grouse.
Anim. Behav. 38, 885–896.
(doi:10.1016/s0003-3472(89)80120-4)
82. Amorim MCP, Simoes JM, Mendonca N, Bandarra
NM,AlmadaVC,FonsecaPJ.2010Lusitanian
toadsh song reects male quality. J. Exp.Biol. 213,
2997–3004. (doi:10.1242/jeb.044586)
83. Di Iorio D. 2009 Vocal communication in blue
whales (Balaenoptera musculus) and noise
interference. PhD dissertation, University of
Zurich, Switzerland. (doi:10.5167/uzh-30004)
84. Krunkelsven E, Dupain J, Elsacker LV,Verheyen RF.
1996 Food calling by captive bonobos (Pan
paniscus): an experiment. Int. J. Primatol.17,
207–217. (doi:10.1007/bf02735448)
85. Bailey WJ. 2003 Insect duets: underlying
mechanisms and their evolution. Phys iol. Entom ol.
28, 157–174.
(doi:10.1046/j.1365-3032.2003.00337.x)
86. Dobler S, Stumpner A, Heller K-G. 1994 Sex-specic
spectral tuning for the partner’s song in the
duetting bushcricket Ancistrura nigrovittata
(Orthoptera: Phaneropteridae). J. Comp.Physiol. Pt
A175,303–310.(doi:10.1007/bf00192989)
87. Bradbury JW, Vehrencamp SL. 2011 Principles of
animal communication. 2nd edn. Sunderland, MA:
Sinauer.
88. Staord KM, FoxCG, Clark DS. 1998 Long-range
acoustic detection and localization of blue whale
calls in the northeast Pacic Ocean. J. Acoust. Soc.
Am. 104, 3616–3625. (doi:10.1121/1.423944)
89. Balcazar NE, Klinck H, Nieukirk SL, Mellinger DK,
Klinck K, Dziak RP, Rogers TL. 2016 Using calls as an
indicator for Antarctic blue whale occurrence and
distribution across the southwest Pacicand
southeast Indian Oceans. Mar.Mamm. Sci. 33,
172–186. (doi:10.1111/mms.12373)
90. WeirathmuellerMJ, Wilcock WSD, Soule DC. 2013
Source levels of n whale 20 Hz pulses measured
in the Northeast Pacic Ocean. J. Acoust. Soc. Am.
133,741–749.(doi:10.1121/1.4773277)
91. Cholewiak DM, CerchioS, Jacobsen JK, Urbán-R. J,
Clark CW. 2018 Songbird dynamics under the sea:
acoustic interactions between humpback whales
suggest song mediates male interactions. R. Soc.
open sci. 5, 171298. (doi:10.1098/rsos.171298)
92. Miller BS, Leaper R, Calderan S, Gedamke J. 2014
Red shift, blue shift: investigating Doppler shifts,
blubber thickness, and migration as explanations
of seasonal variation in the tonality of Antarctic
blue whale song. PLoS ONE 9,e107740.
(doi:10.1371/journal.pone.0107740)
93. Crocker DE, Boeuf BJL, Costa DP. 1997 Drift diving
in female northern elephant seals: implications for
food processing. Can.J. Zool. 75,27–39.
(doi:10.1139/z97-004)
94. Cazau D, Adam O, Laitman JT, Reidenberg JS. 2013
Understanding the intentional acoustic behavior
of humpback whales: a production-based
approach. J. Acoust. Soc. Am. 134, 2268–2273.
(doi:10.1121/1.4816403)
95. Mercado E III, Frazer LN. 1999 Environmental
constraints on sound transmission by humpback
whales. J. Acoust. Soc. Am. 106, 3004–3016.
(doi:10.1121/1.428120)
96. Oleson EM, Calambokidis J, Barlow J, Hildebrand
JA. 2007 Blue whale visual and acoustic encounter
rates in the Southern California Bight. Mar.Mamm.
Sci. 23,574–597.(doi:10.1111/j.1748-7692.2007.
02303.x)
97. Croll D, Marinovic B, Benson S, Chavez F, Black N,
TernulloR, Tershy B. 2005 From wind to whales:
trophic links in a coastal upwelling system. Mar.
Ecol. Prog. Ser. 289, 117–130. (doi:10.3354/
meps289117)
98. Fiedler PC et al. 1998 Blue whale habitat and prey
in the California Channel Islands. Deep Sea Res.
Part II Top. St ud. Oceanog r. 45, 1781–1801.
(doi:10.1016/s0967-0645(98)80017-9)
99. Thomas RJ, Cuthill IC. 2002 Body mass regulation
and the daily singing routines of European robins.
Anim. Behav. 63,285–295.(doi:10.1006/anbe.2001.
1926)
100. Thomas R, Cuthill I, Goldsmith A, Burdett Proctor S,
Cosgrove D, LidgateH. 2003 The trade- o between
singing and mass gain in a daytime-singing bird,
the European robin. Behaviour140, 387–404.
(doi:10.1163/156853903321826693)
101. Thomas RJ. 2002 The costs of singing in
nightingales. Anim. Behav.63, 959–966.
(doi:10.1006/anbe.2001.1969)
102. Poole JH. 1987 Rutting behavior in African
elephants: the phenomenon of musth. Behaviour
102,283–316.(doi:10.1163/156853986X00171)
103. Coltman DW, Bowen WD, Boness DJ, Iverson SJ.
1997 Balancing foraging and reproduction in the
male harbour seal, an aquatically mating
pinniped. Anim. Behav. 54, 663–678.
(doi:10.1006/anbe.1997.0470)
104. Mysterud A, Langvatn R, Stenseth NC. 2004
Patterns of reproductiveeor t in male ungulates.
J. Zool. 264, 209–215. (doi:10.1017/s095283690
4005618)
105. Mysterud A, Bonenfant C, Loe LE, Langvatn R,
YoccozNG, Stenseth NC. 2008 Age-specic feeding
cessation in male red deer during rut. J. Zool.275,
407–412. (doi:10.1111/j.1469-7998.2008.
00453.x)
106. Pelletier F, Mainguy J, Côté SD. 2009 Rut-induced
hypophagia in male bighorn sheep and mountain
goats: foraging under time budget constraints.
Ethology 115, 141–151. (doi:10.1111/j.1439-0310.
2008.01589.x)
107. Miquelle DG. 1990 Why don’t bull moose eat
during the rut? Behav. Ecol. Sociobiol. 27, 145–151.
(doi:10.1007/bf00168458)
108. Branch TA et al.2007 Past and present distribution,
densities and movements of blue whales
Balaenoptera musculus in the Southern
Hemisphere and northern Indian Ocean. Mamm.
Rev. 37, 116–175. (doi:10.1111/j.1365-2907.
2007.00106.x)
109. Mate BR, Lagerquist BA, Calambokidis J. 1999
Movements of North Pacic blue whales during
the feeding season o southern California and
their southern fall migration. Mar.Mamm. Sci. 15,
1246–1257.
(doi:10.1111/j.1748-7692.1999.tb00888.x)
110. Greenwood PJ.1980 Mating systems, philopatry
anddispersalinbirdsandmammals.Anim. Behav.
28, 1140–1162. (doi:10.1016/s0003-3472(80)
80103-5)
111. Pusey AE, Packer C. 1987 Dispersal and philopatry.
In Primate societies (eds BB Smuts, DL Cheney,
RM Seyfarth, RW Wrangham, TT Stuhsaker),
pp. 250–266. Chicago,IL: University of Chicago
Press.
112. Moore J, Ali R. 1984 Are dispersal and inbreeding
avoidance related?Anim. B ehav.32, 94–112.
(doi:10.1016/s0003-3472(84)80328-0)
113. Jack KM, Fedigan L. 2004 Male dispersal patterns
in white-faced capuchins, Cebus capucinus.Anim.
Behav.67,771–782.(doi:10.1016/j.anbehav.2003.
06.015)
114. Clutton-Brock TH. 1989 Female transferand
inbreeding avoidance in social mammals. Nature
337,70–72.(doi:10.1038/337070a0)
115. Saddler MR, Bocconcelli A, Hickmott LS, Chiang G,
Landea-Briones R, Bahamonde PA, HowesG, Segre
PS, Sayigh LS. 2017 Characterizing Chilean blue
whale vocalizations with DTAGs: a test of using tag
accelerometers for caller identication. J.Exp. Biol.
220, 4119–4129. (doi:10.1242/jeb.151498)
116. Stimpert AK, Peavey LE, Friedlaender AS, Nowacek
DP. 2012 Humpback whale song and foraging
behavior on an Antarctic feeding ground.
PLoS ONE 7,e51214.(doi:10.1371/journal.
pone.0051214)
117. Redfern J et al. 2013 Assessing the risk of ships
striking large whales in marine spatial planning.
Conser v.B iol. 27,292–302.(doi:10.1111/cobi.12029)
118. Melcón ML, Cummins AJ, Kerosky SM, Roche LK,
Wiggins SM, Hildebrand JA. 2012 Blue whales
respond to anthropogenic noise. PLoSONE
7,e32681.(doi:10.1371/journal.pone.0032681)
119. Rockwood RC, Calambokidis J, JahnckeJ. 2017 High
mortalityofblue,humpbackandnwhalesfrom
modeling of vessel collisions on the U.S. West
Coast suggests population impacts and
insucient protection. PLoS ONE 12, e0183052.
(doi:10.1371/journal.pone.0183052)