LETTER TO THE EDITOR

..........................................................

Acoustic recording of false killer whale (Pseudorca crassidens)

from Mexico (L)
Raul Rioa)
Laboratory of Observational and Bioacoustics Technologies Applied to Biodiversity (TecBio), Department of Veterinary Medicine,
Federal University of Juiz de Fora (UFJF), Juiz de Fora, Minas Gerais, Brazil

ABSTRACT:
This study collected acoustic information on false killer whales (Pseudorca crassidens) in Mexican waters, close to
Roca Partida Island, Revillagigedo Archipelago. In total, 321 whistles were collected after we found a group with at
least ten individuals. The high prevalence of ascending contour types [upsweep (type I): 42.99%] contradicted the
idea that false killer whales mostly produce constant whistles. Lack of well-established reproducibility criteria for
whistle type categorization among studies may have generated results different from those expected for signal modu-
lation. Future acoustic and ecological studies should be conducted to help clarify these findings and expand the lim-
C 2023 Acoustical Society of America. https://doi.org/10.1121/10.0017726
ited knowledge about this species. V
(Received 14 October 2022; revised 2 March 2023; accepted 14 March 2023; published online 3 April 2023)
[Editor: Kathleen J. Vigness-Raposa] Pages: 2019–2022

I. INTRODUCTION The present study introduces the first baseline acoustic record-
ing of false killer whales in Mexican waters, contributing to
False killer whales (Pseudorca crassidens) are highly social
global knowledge on this delphinid’s acoustic repertoire.
delphinids found in the world’s tropical and semi-tropical
waters, mostly in offshore and deep waters, where they mainly
II. MATERIALS AND METHODS
feed on fish and squid (Baumann-Pickering et al., 2015).
The diversity and complexity of odontocete vocaliza- A. Study site and data collection
tions are a remarkable feature of this group, and these vocal-
Revillagigedo Archipelago (18 500 000 N, 112 500 000
izations are traditionally fit into categories of broadband
W) comprises four islands (Rio et al., 2022), namely, San
echolocation clicks, broadband burst pulses, and frequency-
Benedicto Island, Socorro Island, Clarion Island, and Roca
modulated narrowband whistles. Pulsed vocal signals, such
Partida Island (Fig. 1), where these animals were registered
as clicks, burst pulses, and “buzzes,” are known to have echo-
and data about them were collected. It belongs to a submar-
location and social functions (Baumann-Pickering et al.,
ine mountain ridge located in Eastern Pacific Ocean, in
2015), whereas whistles presumably play some sort of com-
Mexico’s Economic Exclusive Zone.
munication/social cohesion role (Thode et al., 2016). False
A group of at least ten false killer whale individuals was
killer whales produce whistles, echolocation clicks, and burst
first sighted on January 3, 2021, from an inflatable boat dur-
pulses (McCullough et al., 2021); oftentimes, their whistle
ing a bioacoustics expedition (Rio et al., 2022). The daytime
spectrograms resemble those produced by other odontocete
visual confirmation of P. crassidens resulted from the under-
species (Murray et al., 1998). However, they tend to present
water footage of a subgroup of four individuals and from
lower frequency, which is less frequency-modulated than
photographs of them taken during the practice of snorkeling.
most delphinid whistles (Barkley et al., 2019).
All data were collected from a liveaboard vessel (33.5 m
The vocal repertoire of both captive individuals (Murray
in length, 7.5 m in width, capacity: 30 people), with engines
et al., 1998; Yuen et al., 2005; Madsen et al., 2013) and free-
off, anchored at a permitted area around Roca Partida Island
living populations has been registered and analyzed since the
(Research Permission for Collection of Biodiversity Data: N.
first sound emitted by P. crassidens was recorded (Schevill
SGPA/DGVS/00823/20). The recording of the species was
and Watkins, 1962). The most important acoustic recordings
considered continuous while it was possible to visualize
of them were made in continental shelf waters (Weir et al.,
records of whistles and pulsed sounds in the spectrogram
2013), mainly in oceanic habitats, such as Hawaii (Thode
[duration (Dur): 98.3 min]. Part of this period was in sync
et al., 2016; Barkley et al., 2019) and international waters
with the visual confirmation of the species; therefore, all
(Oswald et al., 2003). Nevertheless, if one takes into consider-
signs of this period are attributed to the identified species.
ation all studies available in the literature about this topic,
Underwater recordings were made with hydrophone
knowledge about this species’ vocalizations remains limited.
system BuninTech H0220 (final sensitivity gain of 52 dB, by
GainBox: 152 dB re 1 V/lPa 6 3 dB; frequency band:
a) 5 Hz–80 kHz) placed 5 m underwater and connected to a
Also at: Ocean Sound, a non-governmental organization (NGO), Santos,
S~ao Paulo, Brazil: https://www.oceansound.org. Electronic mail: Tascam DR-100MKIII digital recorder by a 10-m cable.
oceansoundsecrets@gmail.com Recordings were made at a sample rate of 96 kHz and 24-bit

J. Acoust. Soc. Am. 153 (4), April 2023 0001-4966/2023/153(4)/2019/4/$30.00 C 2023 Acoustical Society of America
V 2019
 https://doi.org/10.1121/10.0017726

FIG. 1. Map of the study site in the

Revillagigedo Archipelago islands,
Mexico, Eastern Pacific Ocean; bathyme-
try and landform data of the four islands
forming the archipelago: San Benedicto
Island, Socorro Island, Roca Partida
Island, and Clari on Island. Acoustic
recordings of sounds produced by a
group of false killer whales (P. crassi-
dens) were obtained in a site close to
Roca Partida Island.

resolution. All acoustic data were recorded in 5-min record- to negative aspect or vice versa) were the extracted spectral
ing files as time-stamped wav files. and temporal parameters.
All parameters were manually extracted through selec-
B. Acoustic and statistical analysis tion boxes in Raven Pro 1.6.1 software. Selection box
All whistles with good signal-to-noise (SNR) ratio and boundaries were used to extract BanF, MinF, MaxF, and
complete, clear spectral contours were first visually and Dur, whereas StaF, EndF, and InfP were additionally
aurally identified; then they were manually selected for anal- marked to properly represent their value. Qualitative visual
ysis purposes. Spectrograms were plotted in the spectrogram analysis was used to categorize whistle types.
view of Raven Pro 1.6.1 at 1024 fast Fourier transform Descriptive statistical analysis was applied to all fre-
(FFT) size, Hanning window, and 50% overlap. quency and temporal parameters. Emission rate was calcu-
The term “whistle type” was used to ascribe all whistles lated by dividing the number of whistles by the number of
of a particular frequency modulation pattern or contour to minutes of recorded whistle time. The recorded whistle time
predetermined categories based on human visual evaluations. was defined as the time interval between the first and last
The whistles were classified into six categories, according to acoustic signal (clicks or whistles), either with or without
the contour (Azevedo et al., 2007); they took into consider- visual confirmation. All statistical analyses were performed
ation the presence or lack of modulations, the modulation in GraphPad 8 at 95% significance level.
direction (ascendant or descendent), and the number of
changes in direction (inflexion points). Flat whistles were III. RESULTS
taken as relatively constant signals without inflection points; In total, 321 whistles were extracted from the 98.3-min
they changed less than 1 kHz throughout more than 90% of recorded whistle time, which led to an emission rate of 3.27
their Dur. Only the overall whistle contour was taken into whistles/min. Temporal and spectral parameters of the ana-
account in the present study, rather than the minor contour lyzed whistles are displayed in Table I. The descriptive analy-
classifications or subcategories, which take into consideration sis has shown mean whistle Dur of 0.56 6 0.25 s [mean 6
small variations at the beginning or at the end of a whistle. standard deviation (SD)]—values ranged from 0.16 to 2.17 s.
The following categories were used: upsweeps (type I); down- Frequency parameters recorded for whistles’ acoustic reper-
sweeps (type II); inverted U-shapes (or ascending-descending) toire have shown mean MinF and MaxF of 5.36 6 1.45 kHz and
(type III); U-shapes (or descending-ascending) (type IV), 8.60 6 1.55 kHz, respectively. The highest recorded frequency
wavering sinusoidal whistles (type V); and flat (type VI). was 12.95 kHz, and the lowest one was 2.69 kHz. Mean BanF
The main and classical acoustic parameters of all ana- was 3.24 6 1.41 kHz, whereas StaF (6.30 6 1.87 kHz) and
lyzed whistles were measured to feature this species’ acous- EndF (7.82 6 1.98 kHz) recorded the most similar mean values.
tic repertoire. Start (StaF), end (EndF), minimum (MinF), Based on the whistle modulation type analyses, approxi-
and maximum (MaxF) frequency, bandwidth (BanF), Dur, mately 42.99% (138/321) of emitted whistles were of the
and number of inflection points (InfP) (change from positive ascending contour type (type I). Overall, all other whistle
2020 J. Acoust. Soc. Am. 153 (4), April 2023 Raul Rio
 https://doi.org/10.1121/10.0017726

TABLE I. Descriptive analyses [mean 6 standard deviation (SD); minimum (Min) and maximum (Max) values; coefficient of variation (CV)] of frequency
(kHz) and temporal (s) parameters recorded for whistles (n ¼ 321) emitted by false killer whales (P. crassidens) close to Roca Partida Island, Revillagigedo
Archipelago, Pacific Ocean, Mexico.

StaF EndF MinF MaxF BanF Dur

Mean 6 SD 6.30 6 1.87 7.82 6 1.98 5.36 6 1.45 8.60 6 1.55 3.24 6 1.41 0.56 6 0.25
Min 2.69 2.85 2.69 4.41 0.68 0.16
Max 12.32 12.95 11.03 12.95 7.29 2.17
CV (%) 29.64 25.30 27.12 18.02 43.39 44.42

types presented proportions similar to each other; the simple similar mean StatF-EndF and MinF-MaxF values. These val-
flat structure (type VI) was the least prevalent contour [9.35% ues, in turn, show markedly narrow frequency range—some-
(30/321)]—it was followed by type II [10.59% (34/321)], times, they only reach 1.1 kHz (Weir et al., 2013).
type IV [12.46% (40/321)], type III [12.77% (41/321)], and When it comes to spectral modulation, this species is
type V [11.84% (38/321)], respectively. Figure 2 shows known to mostly vocalize flat whistles (Weir et al., 2013;
examples of whistle spectrograms and their respective Thode et al., 2016). These particular bioacoustic features, on
adopted modulation categories. the one hand, favor acoustic identification between delphinid
species; on the other hand, they impair the identification of
intraspecific variations. False killer whale whistles can be
IV. DISCUSSION
properly species-classified based on the highest correct classi-
Featuring a given species’ vocal repertoire is critical for the fication in comparison to other dolphin species (Oswald et al.,
subsequent analysis of signal functionality, geographical varia- 2007; McCullough et al., 2021); however, their time-
tion, social relevance, and transmission. Bioacoustics efforts to frequency whistle measurements have failed to accurately
study P. crassidens are, somehow, complicated due to inherent classify encounters of three genetically differentiated false
challenges associated with the assessed animals, since they killer whale populations in Hawaii (Barkley et al., 2019).
mostly live in deep ocean waters; consequently, scientific acous- Despite BanF data, mean values recorded for frequency
tic knowledge about them remains scarce. Accordingly, the pre- parameters and for the temporal production of whistles recorded
sent study is the first to collect acoustic information about false for Mexican oceanic false killer whales were consistent with those
killer whales (P. crassidens) distributed in Mexican oceanic found in few studies that have also shown full descriptive results
waters, close to Roca Partida Island, Revillagigedo Archipelago. (Oswald et al., 2003; Oswald et al., 2007; Weir et al., 2013).
Overall, based on the available literature, false killer Oceanic studies conducted by Oswald et al. (2003)
whales often produce relatively short-Dur, little-modulation and Oswald et al. (2007) presented similar results, regard-
and low-frequency whistles (4–10 kHz) (Murray et al., 1998; less of whether they were conducted with samples compris-
Oswald et al., 2003; Weir et al., 2013; Thode et al., 2016) at ing 69 (Oswald et al., 2003) or 340 whistles (Oswald et al.,

FIG. 2. Examples of whistle spectrograms and their respective adopted modulation categories emitted by false killer whales (P. crassidens) close to Roca
Partida Island, Revillagigedo Archipelago, Pacific Ocean, Mexico. Frequency (kHz) is on the y axis, and it ranges from 0 to 20 kHz. Time (s) is on the x
axis, and it represents 5 s. Scaling was the same for all items. Spectrogram settings: FFT size ¼ 1024; Hanning window; overlap ¼ 50%. Numerical informa-
tion type at the top of each whistle represents its classification based on the adopted categories: upsweeps (type I); downsweeps (type II); inverted U-shapes
(or ascending-descending) (type III); U-shapes (or descending-ascending) (type IV); wavering sinusoidal whistles (type V); and flat (type VI).

J. Acoust. Soc. Am. 153 (4), April 2023 Raul Rio 2021
 https://doi.org/10.1121/10.0017726

2007). These values were always below the mean values such as habitat, genetic or cultural drift, group size, group
recorded herein for all assessed acoustic parameters. composition, and behavior (Murray et al., 1998; May-
Sometimes, there were minimum differences in StaF [Oswald Collado and Wartzok, 2008). However, these factors were
et al. (2003), StaF 5.20 kHz; Oswald et al. (2007), StaF not controlled herein.
5.77 kHz; our results, StaF 6.30 kHz] and MinF [Oswald Bioacoustics emissions can be used as a tool in both evo-
et al. (2003), MinF 4.70 kHz; Oswald et al. (2007), MinF lutionary studies and applied ecology or species conservation
5.28 kHz; our results, MinF 5.36 kHz], and sometimes, there studies (Laiolo, 2010). In the future, attempts to categorize
was greater difference in them but always similarity to each the whistle types of false killer whales and other studies
other, for example, mean MaxF [Oswald et al. (2007), mean focused on acoustic monitoring based on environmental and
MaxF 6.95 kHz; our results, MaxF 8.60 kHz] and EndF behavioral context will help broaden bioacoustics knowledge
[Oswald et al. (2003), mean EndF 5.80 kHz; Oswald et al. about this species and contribute to their conservation.
(2007), mean EndF 6.27 kHz; our results, EndF 7.82 kHz].
On the other hand, parameters presenting the most discrepant ACKNOWLEDGMENTS
mean results (MaxF and EndF), in comparison to results by
Oswald et al. (2003) and Oswald et al. (2007), were the most This study was funded by the NGO Ocean Sound
similar to results observed by Weir et al. (2013), who (https://www.oceansound.org) through the scientific project
recorded false killer whales (n ¼ 20 whistles) in continental entitled “Ocean Sound Secrets.”
shelf waters in Gabon and C^ote d’Ivoire.
Azevedo, A. F., Oliveira, A. M., Dalla Rosa, L., and Lailson-Brito, J. (2007).
Mean duration result (0.56 s) recorded in the current “Characteristics of whistles from resident bottlenose dolphins (Tursiops
study was higher than that of previously documented oce- truncatus) in southern Brazil,” J. Acoust. Soc. Am. 121(5), 2978–2983.
anic [0.4 s (Oswald et al., 2003) and 0.44 s (Oswald et al., Barkley, Y., Oleson, E. M., Oswald, J. N., and Franklin, E. C. (2019).
2007)] and neritic [0.32 s (Weir et al., 2013)] encounters. “Whistle classification of sympatric false killer whale populations in
Hawaiian waters yields low accuracy rates,” Front. Mar. Sci. 6, 645.
Thus, the lowest and the highest temporal differences were Baumann-Pickering, S., Simonis, A. E., Oleson, E. M., Baird, R. W., Roch,
observed in a study that used a sample size similar to the M. A., and Wiggins, S. W. (2015). “False killer whale and short-finned
present one (Oswald et al., 2007) (n ¼ 340 whistles) and in pilot whale acoustic identification,” Endang. Species Res. 28, 97–108.
the study that recorded the smallest whistle sample (Weir Laiolo, P. (2010). “The emerging significance of bioacoustics in animal
species conservation,” Biol. Conserv. 143(7), 1635–1645.
et al., 2013) (n ¼ 20 whistles), respectively. Small sample Madsen, P. T., Lammers, M., Wisniewska, D., and Beedholm, K. (2013).
sizes reduce the possibility of observing less frequent events “Nasal sound production in echolocating delphinids (Tursiops truncatus
that, after all, reduce the amplitude of the observed results. and Pseudorca crassidens) is dynamic, but unilateral: Clicking on the
The current mean BanF (mean 3.24 6 1.41 kHz) repre- right side and whistling on the left side,” J. Exp. Biol. 216, 4091–4102.
May-Collado, L., and Wartzok, D. A. (2008). “Comparison of bottlenose
sented more than twice the means reported in the literature dolphin whistles in the Atlantic Ocean: Factors promoting whistle var-
[Oswald et al. (2003), mean BanF 1.40 6 1.3 kHz; Weir et al. iation,” J. Mamm. 89(5), 1229–1240.
(2013), mean BanF 1.10 6 0.72 kHz]. According to Weir McCullough, J. L. K., Simonis, A. E., Sakai, T., and Oleson, E. M. (2021).
et al. (2013), their results were comparable to those observed “Acoustic classification of false killer whales in the Hawaiian Islands
based on comprehensive vocal repertoire,” JASA Express Lett. 1, 071201.
by Oswald et al. (2003); small sample sizes preclude any Murray, S. O., Mercado, E., and Roitblat, H. L. (1998). “Characterizing the
meaningful discussion about vocalizations produced by false graded structure of false killer whale (Pseudorca crassidens) vocal-
killer whales. Furthermore, the only study capable of provid- izations,” J. Acoust. Soc. Am. 104, 1679–1688.
ing acoustic parameter values of BanF comparable to the pre- Oswald, J. N., Barlow, J., and Norris, T. F. (2003). “Acoustic identification
of nine delphinid species in the eastern tropical Pacific Ocean,” Mar.
sent ones did not reveal its mean data (Oswald et al., 2007). Mamm. Sci. 19, 20–37.
Whistle modulation analysis [type I: 42.99% (138/321)] Oswald, J. N., Rankin, S., Barlow, J., and Lammers, M. O. (2007). “A tool
results contradicted the idea that false killer whales mostly for real-time acoustic species identification of delphinid whistles,”
produce constant whistles (Murray et al., 1998; Oswald et al., J. Acoust. Soc. Am. 122, 587–595.
Rio, R., Rosales-Nanduca, H., Piuma, L., Piuma, J., Piuma, M., Redecker,
2003; Weir et al., 2013; Thode et al., 2016). Lack of well- G., and Hoffmann, L. S. (2022). “First report of signature whistles in an
established reproducibility criteria for whistle type categori- oceanic common bottlenose dolphin (Tursiops truncatus) population from
zation among studies may have generated results different Revillagigedo Archipelago, Mexico,” Mar. Mamm. Sci. 38(4), 1308–1324.
from those expected for signal modulation. These findings Schevill, W. E., and Watkins, W. A. (1962). Whale and Porpoises Voices:
A Phonograph Record (Woods Hole Oceanographic Institution, Woods
would explain why our results presented mean values for Hole, MA).
acoustic parameters similar to those found in the literature, Thode, A., Wild, L., Straley, J., Barnes, D., Bayless, A. R., O’Connell, V.,
but with clear and unprecedented prevalence of ascending Oleson, E., Sarkar, J., Falvey, D., Behnken, L., and Martin, S. W. (2016).
whistles. Although false killer whale whistles may seem visu- “Using line acceleration to measure false killer whale (Pseudorca crassi-
dens) click and whistle source levels during pelagic long-line depre-
ally flat in the spectrogram, they oftentimes show subtle slope; dation,” J. Acoust. Soc. Am. 140, 3941–3951.
however, when they are carefully analyzed, they can be cate- Weir, C., Collins, T., Cross, T., Gill, A., Elwen, S., Unwin, M., and Parnell,
gorized as type I. This finding clearly showed increased prev- R. (2013). “False killer whale (Pseudorca crassidens) sightings in conti-
alence of upsweep contours (type I) that, consequently, nental shelf habitat off Gabon and C^ ote d’Ivoire (Africa),” Mar.
Biodivers. Rec. 6, E65.
reduced values recorded for flat whistles (type VI). Yuen, M. M. L., Nachtigall, P. E., Breese, M., and Supin, A. Y. (2005).
Finally, it is important to take into consideration that “Behavioral and auditory evoked potential audiograms of a false killer whale
our acoustic results could be influenced by many factors, (Pseudorca crassidens),” J. Acoust. Soc. Am. 118(4), 2688–2695.

2022 J. Acoust. Soc. Am. 153 (4), April 2023 Raul Rio