Content uploaded by Renata Santoro Sousa-Lima
Author content
All content in this area was uploaded by Renata Santoro Sousa-Lima on Apr 02, 2018
Content may be subject to copyright.
ANIMAL BEHAVIOUR, 2002, 63, 301–310
doi:10.1006/anbe.2001.1873, available online at http://www.idealibrary.com on
Signature information and individual recognition in the
isolation calls of Amazonian manatees, Trichechus inunguis
(Mammalia: Sirenia)
RENATA S. SOUSA-LIMA*, ADRIANO P. PAGLIA† & GUSTAVO A. B. DA FONSECA*
*Programa de Po´s-Graduac¸a˜o em Ecologia, Conservac¸a˜o e Manejo de Vida Silvetre, Departamento de Zoologia do
Instituto de Cieˆncias Biolo´ gicas, †Departamento de Biologia Geral do Instituto de Cieˆncias Biolo´ gicas,
Universidade Federal de Minas Gerais
(Received 6 November 2000; initial acceptance 3 January 2001;
final acceptance 25 April 2001; MS. number: A8925)
Acoustic signals are assumed to form the basis of manatee communication. Empirical evidence of
individual vocal recognition has been reported. If manatees can recognize one another by acoustical
means, it should be possible to identify individual vocal patterns. We recorded vocalizations of 14
individually housed Amazonian manatees and then digitized selected vocalizations, allowing seven
variables to be measured and subjected to multivariate statistical treatment. Discriminant function
analysis indicated that individuals can be separated on the basis of variables related to the fundamental
frequency and signal duration. We observed significant differences in the vocal patterns between sexes
and age classes. Females tended to have greater fundamental frequency and shorter note duration than
males. Calves had shorter note durations and greater values for the fundamental frequency range than
subadults and adults. An inverse relationship between total body length and fundamental frequency
range suggests that the fundamental frequency becomes more defined as the animal ages. The similar
individual patterns in the vocalizations of a mother and calf pair are discussed. Individual recognition by
Amazonian manatees according to their vocal patterns is suggested through a preliminary playback
experiment.
2002 The Association for the Study of Animal Behaviour
Many animals communicate specific messages accom-
panied by additional information about their motivation,
sex, age, or even their identity (Halliday & Slater 1983).
Recognizing and maintaining contact with specific indi-
viduals is more challenging for animals that are often
separated in such a way that visual contact is no longer
efficient. Selective forces to identify and locate distant
individuals could promote the evolution of an acoustic
basis for individual recognition (Rendall et al. 1996).
Thus, in contexts or periods where the transmission of
nonacoustic signals is constrained, individuals can still
recognize each other by their vocalizations (Falls 1982).
Individual recognition is especially important for social
species and in the association between parents and
offspring (Halliday 1983).
Theoretically, the development of vocal signatures
occurs when there is increased probability of misdirecting
parental care, most likely when individuals form breeding
colonies or in species where mother and offspring are
often separated (Jones et al. 1987). In mammals that have
low reproductive rates and extended parental care and
use sound-based communication, such as sirenians, one
might expect vocal recognition to be important for
mothers to maintain close contact with their offspring
and for identifying and locating their calves after
unintentional separations.
Nursing lasts 25 months in Amazonian manatees
(V. M. F. Da Silva, J. A. D’Affonseca Neto, G. E. Mattos &
R. S. Sousa-Lima, unpublished data). Given this long-term
investment in parental care, it is reasonable to assume
that there is considerable selective pressure on a mother
to nurse her own, rather than another individual’s
calf, and that some means of reliable identification is
Correspondence and present address: R. S. Sousa-Lima, Laborato´rio de
Mami´feros Aqua´ ticos, Instituto Nacional de Pesquisas da Amazoˆnia,
C.P. 478, Manaus, AM 69011-970, Brasil (email: pboi@inpa.gov.br).
A. P. Paglia is now at the Laborato´rio de Mastozoologia e Manejo de
Fauna, Departamento de Zoologia do Instituto de Cieˆncias Biolo´ gicas,
Universidade Federal de Minas Gerais, C.P. 486, Belo Horizonte, MG
30161-970, Brasil. G. A. B. Da Fonseca is now at the Center for
Applied Biodiversity Sciences, 1919 M Street NW, Suite 600,
Washington, D.C. 20236, U.S.A.
0003–3472/01/020301+10 $35.00/0 2002 The Association for the Study of Animal Behaviour301
necessary. Individual recognition by vocal signals also
implies the capacity to learn and to become familiar with
individual differences (Halliday 1983). Although the
majority of studies concerning individual recognition
have focused on parent–offspring relationships, it has
been shown that monogamous royal penguins, Apteno-
dytes patagonicus, can recognize their mates by individu-
ally distinctive vocal patterns (Robisson 1992).
Loesche et al. (1991) suggest that the recognition pro-
cess is comprised of four main components: the vocal
signature (production of unique signals by the sender);
perception (ability to perceive signal differences); deci-
sion (ability to choose the right individual); and reaction
(response according to the decision). Natural selection
can affect any of the four components by increasing
interindividual variation or decreasing intraindividual
variation to improve individual discrimination, by
enhancing the attention or the perception of the receiver
through improved receiver sensitivity, or by modifying
the decision rule and/or the response after recognition. In
the current study we focus on the first two components:
the presence of vocal signatures and the perception of
individual differences.
Caldwell & Caldwell (1965,1968) observed a great
interindividual variation and intraindividual consistency
in the spectrogram contours of bottlenose dolphin
whistles, Tursiops truncatus. Since these pioneering works,
vocal signatures have been investigated in many other
species. Falls (1982) even suggested that the occurrence of
vocal signatures is universal. However, this does not
mean that the level of variation is necessarily enough to
achieve effective individual recognition in all cases.
Individual recognition has been verified using play-
back experiments in birds (McArthur 1982;Jones et al.
1987;Abs & Jeismann 1988;Lessells et al. 1991;Loesche
et al. 1991; P. Jouventin & T. Aubin, unpublished data;
T. Lengagne & J. Lauga, unpublished data), primates
(Cheney & Seyfarth 1982;Rendall et al. 1996) and bats
(Balcombe 1990). Abs & Jeismann (1988) and Janik et al.
(1994) used a different approach to identify vocal signa-
tures by examining the specific parameters responsible
for the individualization of acoustic signals in birds and
cetaceans, respectively. Multivariate analyses have been
widely used for this kind of investigation in birds
(McArthur 1982;Jones et al. 1987;McGregor & Byle
1992), bats (Gelfand & McCracken 1986), primates
(Hammerschmidt & Todt 1995), wolves (Tooze et al.
1990), foxes (Frommolt et al. 1997), pinnipeds (Insley
1992;Sanvito & Galimberti 2000) and dolphins
(McCowan et al. 1998).
West Indian manatees, Trichechus manatus, are known
to vocalize while playing and eating. They also produce
alarm signals, vocalize as a means of maintaining contact
between mothers and calves, and often during intra- and
interspecific contact (Hartman 1979;Reynolds 1981;
Bengston & Fitzgerald 1985). Evans & Herald (1970) and
Sonoda & Takemura (1973) successfully recorded the
underwater vocalizations of Amazonian manatees,
T. inunguis. These authors suggest that the primary func-
tion of vocalization in this species is for communication
rather than as a navigational tool. Thus far, no study has
adequately proven the occurrence of vocal signatures in
sirenians. Empirical evidence of individual vocal recogni-
tion in the West Indian manatee (Reynolds 1981) and
the great variability in vocalizations among individual
dugongs, Dugong dugong (Anderson & Barclay 1995) could
be considered evidence for the existence of vocal identity.
Our goal in this paper is to test the following hypotheses:
(1) that there is some difference in vocal parameters
among individual Amazonian manatees; (2) that there is
a difference in vocal parameters between sexes and age
classes in this species; and (3) that there is an effect of
body size on the vocal parameters.
METHODS
Study Animals
We recorded eight male and six female Amazonian
manatees in the facilities of three research institutions
that maintain this species in captivity in Brazil: the
Laborato´rio de Mami´feros Aqua´ ticos of the Instituto
Nacional de Pesquisas da Amazoˆnia (LMA-INPA),
Manaus, state of Amazonas; the Centro de Preservac¸a˜o e
Pesquisa de Mami´feros Aqua´ ticos of Manaus Energia
(CPPMA), Balbina, state of Amazonas; and the Parque
Zoobotaˆnico of the Museu Paraense Emi´lio Goeldi
(MPEG), Bele´m, state of Para´(Table 1).
The animals are kept either in groups or isolated
depending on the current management objectives. There
are three round pools at LMA (10 m in diameter, 4 m in
depth, with a capacity for 200 m
3
of water each) con-
nected by two smaller ones (3.52.5 1.5 m). At CPPMA
there are three octagonal pools (10 m in diameter, 3 m in
depth, with a capacity for 200 m
3
of water each) con-
nected by two smaller ones (3.52.5 1.5 m), and at
MPEG the housing resembles a natural ‘igapo´’ (typical
Amazonian flooded forest) and has an area of 198 m
2
with depth varying from 1.2–2 m. Note that all of these
facilities were constructed according to the federal laws of
animal welfare and the advice of the Brazilian Society
of ZOOs.
Recordings
We made the recordings between 11 August and
9 September 1998 on TDK SA60 cassette tapes (type II)
using a Sony Walkman Pro (WM-D6C; flat audio fre-
quency response 40 Hz–15 kHz3 dB) and an omnidirec-
tional hydrophone (model 50Ca of the Cetacean Research
Technology) with a sensitivity of 161 dB re 1 Pa and a
frequency range response of 0.01–310 kHz.
Ethical note
It was necessary to isolate individuals during each
recording session to ensure that each vocalization could
be identified to individual and that all vocalizations were
produced in a similar context.
We isolated adult and subadult individuals either in
pools with the sides and bottom covered with rubber
tarpaulins (to avoid sound contamination from the con-
necting or nearby pools) or placed them in totally isolated
302 ANIMAL BEHAVIOUR, 61, 2
pools. We transferred the calves on a stretcher to smaller
fibreglass or rubber pools (2.81.8 0.8 m) that were
completely isolated. To ensure the animals’ well being
and to avoid disruption of normal nursing bouts (nursing
bout interval= 1 h), we limited the separation period
between the mother and calf pair (Boo and Ereˆ) to 30 min
(15 min for recording each animal). We conducted
25-min recording sessions for all individuals (except Boo
and Ereˆ) and registered their behaviour continuously
during the entire session.
Sound Analysis
We selected the recordings to be analysed following the
criteria of high signal-to-noise ratio and their occurrence
in a similar behavioural context. We chose a maximum of
10 sounds at random for each individual, digitized the
samples (sample rate 44100 Hz; sample size 16 bits), and
then analysed them using the program Canary 1.2.1
(Cornell Laboratory of Ornithology, Ithaca, New York,
U.S.A.; filter bandwidth 699.40 Hz; frame length 256
points). To minimize measurement error, we sampled
time and frequency from the most intense harmonic that
was clearly visible along the length of the signal. We then
divided these measurements by the appropriate factor
(number of the harmonic) to yield the value of the
fundamental frequency.
Statistical Analysis
We considered only two age classes (calves and others)
because the distinction between adults and subadults is
arbitrary (Hartman 1979). We performed two discrimi-
nant analyses, one for each age class, to test for differ-
ences in individual vocal patterns. In both cases we
treated each individual as a group (Manly 1994). We
excluded the variable ‘mean duration of the intervals
between notes’ from these analyses because it presented
no variation between some individuals. We performed
two separate multiway analyses of variance (MANOVAs)
using the seven measured variables (Table 3) to test for
differences in the vocal pattern between sexes and
between age classes (Everitt & Dunn 1992). We used the
mean values of the variables from each individual in a
linear regression to examine the correlation between total
body length and each of the seven vocal variables (Sokal
& Rohlf 1995).
RESULTS
Overall Characteristics of the Vocalizations
Amazonian manatee vocalizations consist of one to
four notes (Fig. 1). The vocalizations are mostly har-
monic, consisting of 1–12 frequency bands. Some vocal-
izations present non-harmonic frequency components
resulting in a harsh sound quality. We also recorded
broadband click-like sounds. However, due to the relative
ease of analysis of the harmonic signals, those were the
ones investigated for this study. The mean signal dura-
tion (duration of the notes plus duration of the intervals
between them) was 242107 ms (range 50–500 ms). The
mean fundamental frequency was 41.2 kHz. In many
cases, the harmonics were more intense than the
fundamental frequency band (Fig. 1).
Each individual has a single type of harmonic isolation
call. The intraindividual variation is exemplified in Fig. 1.
The number of notes may vary in the same individual,
but the overall characteristics of the notes remain the
same. There is intra- and interindividual variation on the
duration of the signals, while the fundamental frequency
values are more conservative within individuals.
Table 1. Identification of the animals studied
Individual Institution* Sex
Age
class†
Origin
(location/state)‡
Body
length
(cm)
Ac¸ai´ CPPMA M C Codaja´s/AM 103.0
Anama˜ INPA M C Anama˜ Lake/AM 112.5
Aria´ INPA F C Ariau´ ’s Lake/AM 113.0
Boo INPA F A Unknown 211.0
Carimbo´ CPPMA M S Vigia/AM 199.0
Ereˆ INPA M C INPA (Boo’s calf) 112.0
Guarani INPA M S Cuieiras River/AM 174.0
Macuxi INPA F C Tacutu´ River/RR 112.0
Manaca´ INPA F S Manacapuru´ /AM 176.0
Marcelo CPPMA M C Barcelos/AM 135.0
Santinha CPPMA F C Santare´ m/PA 100.0
Preto CPPMA M C Rio Preto da Erva/AM 130.0
Puru´ INPA M S Manacapuru´ /AM 160.0
Uiara MPEG F C Breves/PA 100.0
*CPPMA: Centro de Preservac¸a˜o e Pesquisa de Mami´feros Aquaticos; INPA: Instituto Nacional de Pesquisas da
Amazoˆnia; MPEG: Museu Paraense Emi´lio Goeldi.
†M: male; F: female; C: calf; S: subadult; A: adult.
‡AM: Amazonas; RR: Roraima; PA: Para´.
303SOUSA-LIMA ET AL.: IDENTITY IN MANATEE CALLS
Data Discrimination
Discriminant function analysis using only calves
grouped the vocal variables by individual and sex (Fig. 2).
The first axis of the discriminant function showed a
strong negative correlation with the mean, minimum and
maximum fundamental frequencies (Table 2). Therefore,
female calves had higher fundamental frequency values
than males. The second axis was inversely correlated with
the ‘mean note duration’ and with ‘signal duration’
(Table 2), separating the male calf Ereˆ (with longer
vocalizations) from the other calves.
Purú (male subadult)
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
Anamã (male calf)
5
500
20
10
15
100 200 300 400
5
500
20
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
Erê (male calf)
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
Boo (female adult)
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
Manacá (female subadult)
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
Ariá (female calf)
5
500
20
0
10
15
100 200 300 400
5
500
20
10
15
100 200 300 400
5
500
20
10
15
100 200 300 400
Uiara (female calf)
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
5
500
20
0
10
15
100 200 300 400
Time (ms)
Frequency (kHz)
Figure 1. Three examples of spectrograms of vocalizations from each animal recorded. Note the similarity between the calls of Boo and Ereˆ
(mother and calf pair).
304 ANIMAL BEHAVIOUR, 61, 2
Discriminant function analysis also indicated individ-
ual vocal patterns within the considered age class that
grouped subadults and adults (Fig. 3), while no sex
discrimination was evident. The first axis of the discrimi-
nant function showed a strong positive correlation with
the mean, minimum and maximum fundamental fre-
quencies. This axis can be considered a function of
frequency, isolating the subadult male Puru´, with lower-
pitched vocalizations, from the rest of the animals. The
second axis was inversely correlated with ‘mean note
duration’ and ‘signal duration’ (Table 2). Note that Boo
and her calf (Ereˆ) had the longest signal duration of their
respective groups (Figs 2,3).
Sex and Age Differences
The MANOVAs showed significant differences in the
vocal patterns between sexes and age classes. Significant
differences between males and females for almost all the
variables, except ‘signal duration’, were found. Females
had greater values of mean, maximum and minimum
fundamental frequencies, ‘duration of the intervals
between notes’, and ‘fundamental range’, and smaller
values only in ‘mean note duration’.
Significant differences between age classes were found
in ‘signal duration’ and ‘fundamental range’ (Table 3).
Calves had smaller values of ‘mean note duration’ and
greater values of ‘fundamental range’. The variable ‘maxi-
mum fundamental frequency’ was greater in calves,
although it was only marginally significant (level of
significance of 5%).
Body Size Effect
There was no sex difference in body size. The only
inverse relationship found among all of the variables
measured was between the range of the fundamental
frequency and total body length (r=0.54; F
1,12
=4.9,
P=0.046; Fig. 4). However, the regression explained only
30% of the variation, giving it relatively weak predictive
power.
DISCUSSION
The data gathered for this study can be easily compared to
data on the characteristics of Amazonian manatee vocal-
izations available in the literature. The duration limits
for these kinds of vocalizations have been previously
–12 16
6
Discriminant axis 1
Discriminant axis 2
–16 12840–8
4
2
0
–2
–4
–6
–8
–4–12
–10
Santinha
Ariá
Marcelo
Açai
Uiara
Macuxi
Erê
Anamã
Preto
Figure 2. Plot of the two first discriminant functions using calves
only. Open symbols indicate females.
Table 2. Factor loading of each variable on the two canonical functions of the discriminant function analysis
Variables
Calves Subadults/adults
Function 1 Function 2 Function 1 Function 2
Mean note duration 0.09 −0.88 −0.06 −0.75
Signal duration −0.05 −0.63 −0.001 −0.45
Mean fundamental frequency −0.79 −0.14 0.86 −0.11
Maximum fundamental frequency −0.41 −0.23 0.58 −0.03
Minimum fundamental frequency −0.62 −0.03 0.76 −0.19
Fundamental range −0.08 −0.18 0.07 0.10
Eigenvalue 66.04 10.38 87.07 27.90
Cumulative proportion of variance 81% 94% 71% 94%
Wilks’ λ0.0001 0.0076 0.00002 0.002
χ
2
740.6 397.8 401.9 234.1
df 48 35 24 15
Pvalue <0.001 <0.001 <0.001 <0.001
The variables with the highest loading are highlighted in bold.
–10 10
12
Discriminant axis 1
Discriminant axis 2
–20 50–5–10–15
10
8
6
4
2
0
–2
–4
–6
–8
Carimbó
Boo
Manacá
Guarani
Purú
Figure 3. Plot of the two first discriminant functions using subadults
and adults only. Open symbols indicate females.
305SOUSA-LIMA ET AL.: IDENTITY IN MANATEE CALLS
reported as 0.15–0.22 s (Evans & Herald 1970) and 0.30–
0.40 s (Sonoda & Takemura 1973). In the present study
we found limits of 0.05–0.50 s. Similarly, the fundamen-
tal frequency range in the present study was increased
from 6–8 kHz (Evans & Herald 1970) and 2–4 kHz
(Sonoda & Takemura 1973) to 1.07–8 kHz. The larger
sample size of 14 individuals used in this investigation,
compared with only one or two in the previous studies, is
probably responsible for the significant improvement in
the accuracy of describing the basic characteristics of
Amazonian manatee vocalizations.
Intraindividual Variation
The vocalization of one individual varied more in dur-
ation than in fundamental frequency (Fig. 1). Although
discriminant axis 2 (related to duration, Figs 2,3) also
contributed to the isolation of individual vocal patterns,
the fundamental frequency appears to be a more
conservative parameter than signal duration.
Insley (1992) verified that the temporal characteristics
of two species of pinnipeds are also more variable within
an individual than those variables related to the fre-
quency domain. Abs & Jeismann (1988) showed that the
state of arousal influences the temporal pattern in court-
ship songs of birds. Note duration can codify the emo-
tional state in fish (Davis 1988 cited in Lengagne et al.
1997). Bottlenose dolphins produce longer signature
whistles when stressed (Caldwell et al. 1990). Contextual
information can be related to the alteration of some vocal
parameter that would not affect the individual informa-
tion present in the contour of the whistle (Janik et al.
1994), such as duration.
The evidence that other kinds of information can be
related to signal duration lead us to conclude that,
because signal duration is a more variable vocal par-
ameter, it can be related either to individuality, or to the
motivation of the sender and the context in which the
signal is produced.
Taking into account that the main function of manatee
vocalization is the maintenance of proximity between
mother and calf (Hartman 1979), if signal duration is in
fact related to stress and the intensity of the encoded
message, then the physical isolation of the mother and
calf pair (Boo and Ereˆ) during the recording sessions (the
first time they had been separated) might have caused
enough distress to result in the production of longer
signals. However, the similarity between mother and calf
call duration (see Fig. 1) could also be evidence of genetic
relatedness or the result of a learning process.
Individual Recognition
The successful grouping of the data by individual in the
discriminant function analysis indicates the presence of
individually stereotyped vocalizations, a prerequisite for
individual recognition by sound (Falls 1982;Jones et al.
1987;Insley 1992). Vocalizations were discriminated
based on five characteristics of the signal: maximum,
mean and minimum fundamental frequencies, mean
note duration and signal duration (Table 2). McArthur
(1982) was unable to identify the vocal parameter that
would allow individual recognition in pin˜on jays, Gym-
norhinus cyanocephalus, and suggested that recognition
Table 3. Mean±SD of each variable used in the MANOVA and univariate Ftest between age classes and sexes
Variable
Age class
(λ
7,126
=0.732, P<0.001)
Sex
(λ
7,126
=0.437, P<0.001)
Calves
(N=90)
Others
(N=44) F
1,132
Plevel
Males
(N=74)
Females
(N=60) F
1,132
Plevel
Mean note duration 172.6±119.7 224.4±140.8 4.91 0.03 215.5±118.7 157.6 ±134.4 7.0 0.009
Mean interval duration* 11.9±18.7 11.5 ±21.8 0.02 0.899 7.3±17.6 17.3 ±20.8 9.2 0.002
Signal duration† 234.6±103.6 255.7 ±113.6 1.16 0.281 241.6 ±100.0 241.5±115.9 0.0 0.99
Mean fundamental frequency‡ 3.9±1.18 3.9 ±1.19 0.01 0.901 3.2 ±1.11 4.8±0.43 109.2 <0.001
Maximum fundamental frequency 4.9±1.71 4.3 ±1.32 3.99 0.05 3.8 ±1.48 5.9±0.81 95.2 <0.001
Minimum fundamental frequency 3.3±1.25 3.4 ±1.15 0.30 0.583 2.6 ±1.10 4.2±0.60 103.1 <0.001
Fundamental range§ 1.7±0.90 0.9 ±0.40 24.5 <0.001 1.2 ±0.67 1.7±0.95 11.9 <0.001
*Mean duration of the intervals between notes.
†Duration of the notes plus duration of the intervals.
‡Calculated as the mean value of three intervals of 0.008–0.01 s over the duration.
§Difference between maximum and minimum frequencies.
0.4 220
3.2
Total body length (cm)
Range of fundamental
frequency (kHz)
80 200180160140100
2.8
2.4
2.0
1.6
1.2
120
0.8
Male
Female
r = –0.54
P = 0.046
Figure 4. Regression between total body length and range of
fundamental frequency.
306 ANIMAL BEHAVIOUR, 61, 2
might be based on ‘gestalt’ perception of calls, incorpo-
rating simultaneous variation in many parameters. As we
have found for the manatees, characteristics of the funda-
mental frequency are important in many species of birds
(Robisson 1992) and mammals (Gelfand & McCracken
1986;Tooze et al. 1990;Insley 1992;Frommolt et al.
1997) in which vocal signatures have been verified.
Barbary macaque, Macaca sylvanus, mothers are able to
recognize their offspring by more than one signal cue,
which may improve the robustness of the recognition
system against possible distortion caused by the environ-
ment (Hammerschmidt & Todt 1995). The recognition
system proposed by these authors includes several advan-
tages; for example: differences between individuals do
not need to be extreme if they are distributed amongst a
number of parameters; one particular parameter does not
need to be very distinct in order to guarantee fair dis-
crimination; and, learning of individual characteristics is
facilitated by the fact that individuality in vocalizations is
not restricted to specific parameters, but can involve any
of them. An example of this principle is illustrated by the
mother and calf pair (Boo and Ereˆ), which differed from
the others predominantly in the parameters ‘mean note
duration’ and ‘signal duration’. A potential receiver may
easily learn this individual specificity because these par-
ameter values contrast strongly with those of the other
manatees.
In timber wolves, Canis lupus, the fundamental fre-
quency and the richness of harmonics in the vocaliza-
tions is determined primarily by individual characteristics
of the vocal apparatus (Tooze et al. 1990). Morphological
differences in the structures responsible for phonation or
in the neural control of sound production affect the
production of calls (Janik 1999). It is likely that in
manatees these traits cause sufficient alterations in vocal
characteristics such that individual discrimination can
occur even by measuring only a few parameters.
Individual information encoded in the variables
measured from the isolation calls and in other distinctive
features, such as frequency and amplitude modulations
observed on the spectrograms (Fig. 1), should be enough
evidence to suggest the existence of vocal signatures in
Amazonian manatees. Recent recordings reveal stable
individual vocal patterns over 4 years (R. S. Sousa Lima,
unpublished data). None the less, the analysis of only one
call type and the context in which the recordings were
made may be considered a limitation on the strength of
our evidence. Further investigations of the entire vocal
repertoire during intraspecific interactions are required
before these calls can be defined as vocal signatures
(sensu Caldwell et al. 1990).
The occurrence of vocal identity is suggestive, but does
not demonstrate individual recognition. The variation in
individual vocal patterns may not be sufficient to always
allow accurate discrimination (Falls 1982). The specific
parameters measured in this study could be different from
those actually used by the animal for individual vocal
recognition.
We performed one preliminary playback trial by pre-
senting two acoustic stimuli to the adult female (Boo) to
test her ability to perceive and recognize her own calf’s
vocalizations. Prior to presenting the playback, we con-
ducted a 5-min control period to record Boo’s activity,
defining a behavioural standard from which any response
could be identified. The silent control period was followed
by a 5-min playback of vocalizations either from Boo’s
own calf (Ereˆ), or from a male Amazonian manatee calf
(Anama˜) that was approximately the same size and age of
Boo’s calf. The stimuli were played from an Oceaneer
underwater speaker (DRS-8, frequency range 100 Hz–
16 kHz with peak resonance frequency at 4.5 kHz, 158 dB
re 1 Pa) connected to the same recorder used in the
acquisition of the original signals. Another 5-min period
of silence was allowed following the playback to further
monitor the animals’ behaviour and ensure that any
response detected during the playback was due to the
stimulus presented. A 5-h interval separated the two play-
back sessions. The calf was separated from the mother
during the playbacks in order to record its vocalizations.
Boo showed only a mild response to the playback of the
unrelated male calf’s vocalization. The response consisted
of Boo turning her head towards the speaker the first time
the stimulus was presented and then going back to her
previous activity of swimming around the pool. Boo’s
response to the stimulus of her calf’s vocalization play-
back was striking and immediate. Boo moved rapidly
towards the speaker and touched it with her lips before
going to the other side of the pool. Every time the
stimulus was presented, she approached the speaker and
then went back to the opposite side of the pool, until the
5 min of the playback were over. During the control
period, Boo swam around the pool and showed no special
interest in the speaker.
Although further trials are needed for absolute verifica-
tion, our preliminary playback experiments suggest the
existence of acoustically based individual recognition in
Amazonian manatees.
Mother and Calf Vocal Similarities: Learning or
Inheritance?
The similarity between the vocalization patterns of Boo
and Ereˆ(Fig. 1) is intriguing. Such an observation could
be the result of a learning process and/or of genetic
inheritance.
Janik & Slater (2000) defined different forms of social
learning in animal communication. They suggest two
main forms of social learning, contextual (comprised of
usage and comprehension learning) and production
learning. Production learning refers to cases in which an
individual modifies the form of its signal as a result of
experience with those of another individual. It can result
either in signals becoming matched, or in distinct
differences arising between individuals (Janik 1999).
Vocal learning is considered as production learning in
the vocal domain. The similar vocal pattern between Boo
and Ereˆ could be a result of production learning by Ereˆ,
controlling the respiratory, phonatory and/or filter sys-
tems to match the mother’s template signal. Janik &
Slater (2000) state that the duration and amplitude of a
signal can be changed by the respiratory system alone,
although the control over the fundamental frequency
307SOUSA-LIMA ET AL.: IDENTITY IN MANATEE CALLS
and its modulations are done by the phonatory system
(sound-producing apparatus). Changes in the energy dis-
tribution in the vocalization are caused by the filter
system (filtering or resonance structures). Bat calls used in
mother–infant recognition are frequency-matched (Jones
& Ransome 1993). Janik & Slater (2000) suggest that
these bat calls are learned. Figure 1 shows the same
phenomenon in the vocalizations of Boo and Ereˆ.
Caldwell & Caldwell (1979) suggest that the acoustical
environment can alter vocal patterns if these signals are
learned. A young captive bottlenose dolphin studied by
these authors also developed a vocal signature very simi-
lar to that of its mother. Right after Ereˆ’s birth, he and the
mother (Boo) were isolated from the others and have not
been separated from one another since then. Besides the
obvious genetic proximity, Boo was the only vocal refer-
ence available to Ereˆ. The infant bottlenose dolphin
studied by Caldwell & Caldwell (1979) was also exposed
almost exclusively to its mother’s signature whistle.
The structures responsible for phonation may be more
morphologically similar in closely related individuals
and thus produce sounds more alike than sounds pro-
duced by unrelated animals. Therefore, the influence of
genetic relatedness and learning on the observed
vocal similarity between parent and offspring cannot be
separated yet.
Vocal Differences between Males and Females
Steel (1982) found that male Florida manatees,
T. manatus latirostris, vocalize at a higher pitch than
females. Our results show that females have higher
fundamental frequency values than male Amazonian
manatees. The Amazonian manatee breeds seasonally
(Best 1982) and gender-specific vocal cues could be useful
in mate recognition and attraction.
Relationship Between Vocal Parameters and
Body Size
The inverse relationship between total body length and
the range of the fundamental frequency in Amazonian
manatees has low predictive power, but could indicate
that maturity has some influence on the decrease in the
fundamental frequency range. Gelfand & McCracken
(1986) observed significant relationships between all of
the vocal parameters and age in bats. Robisson (1992) also
verified that some species of penguins show vocal altera-
tions that are related to maturity. He observed that
individual vocal identity is related to parameters of the
fundamental frequency, which become better defined
(more individually distinct) as the young prepare to leave
the nest.
Frequency modulation in bottlenose dolphin signature
whistles increases with age (Caldwell et al. 1990). Individ-
ual identity in bottlenose dolphins has been related to the
whistle contour (frequency modulation), rather than to
fundamental frequency values. The higher frequency
modulation in adult bottlenose dolphins suggests an
increment in individual specificity. In Amazonian mana-
tees, this specificity is achieved through well-defined
fundamental frequency values, along with other signal
characteristics. The greater values of the ‘fundamental
range’ variable in Amazonian manatee calves and the
inverse correlation between body size and fundamental
range suggest that the fundamental frequency becomes
more defined as the animal ages.
The ‘mean note duration’ in the vocalizations of sub-
adult and adult Amazonian manatees is greater than in
calves. Caldwell et al. (1990) suggest that some factor that
scales with size may limit the duration (primarily a
function of the number of loops per whistle) of sounds
that a bottlenose dolphin can produce. Longer signals
may require more energy, which is more limited in calves.
To call conspecific attention or elicit parental care,
Amazonian manatee calves may use a strategy other than
the production of long signals, such as increasing the
vocalization rate. In fact, the average production rate in
Amazonian manatee calves is six vocalizations/min,
while subadults and adults vocalize once/min (R. S.
Sousa-Lima, unpublished data).
Potential Functions of Vocal Signatures in
Sirenians
The production of acoustic signals involves a signifi-
cant energetic cost (Gerhardt 1983 and must, therefore,
have some adaptive value to the sender. What might the
functions of the production of individually distinct
vocalizations be?
The presence of stereotyped individual signals in
Amazonian manatees conforms to the assumption that
recognition between conspecifics is important for both
parent and offspring, due to the high risk of confusion, as
observed in colonial species of birds (Jones et al. 1987).
Intuitively, the absence of a well-developed social organi-
zation (Reynolds 1981), occurrence in turbid water
environment, and the presence of nocturnal activity in
manatees (Best & Da Silva 1979) should select for the
development of individual vocal recognition capabilities
in this species. Dozens of animals are found in feeding
aggregations, called ‘comidia’ by Brazilian hunters. The
lack of any strong social organization and the mother’s
feeding strategies would tend to facilitate the occasional
separation of calves from their mothers (Janik 1999). The
evolution of individually distinct vocal signals would
provide a means for the maintenance of proximity
between mother and calf, and the recognition and loca-
tion of lost calves. The probability of correctly assigning
parental care would be greater, thereby increasing an
individual’s fitness.
Animals that associate with each other early in their
lifetime will have plenty of opportunities to learn the
vocal characteristics of the individuals in their group
(Halliday 1983). The longest and most important bond in
manatees is between mother and calf. Other associations
are casual, occurring temporarily, for feeding or reproduc-
tion (Hartman 1979). Kin recognition, made efficient
through individual recognition, can also have adaptive
value by providing a way to avoid inbreeding and
improve fitness through altruistic behaviour (Halliday
1983).
308 ANIMAL BEHAVIOUR, 61, 2
The signature whistles of bottlenose dolphins are
important in establishing vocal or physical contact
between individuals (Caldwell et al. 1990) and maintain-
ing group cohesion (Janik & Slater 1998). Mother and calf
vocalizations in Amazonian manatees also function as
means for maintaining close proximity between individ-
uals (Hartman 1979). Any species in which an association
with particular individuals is advantageous should pos-
sess some mechanism for group cohesion (Janik & Slater
1998). The individual information encoded in manatee
isolation calls improves the efficiency of the cohesion/
proximity mechanism. This would be especially import-
ant for mothers and calves that are frequently separated
(i.e. during foraging). Therefore, the primary function of
individually distinct vocalizations in manatees would
be identification, location and attraction (cohesion)
between specific individuals, markedly between mother
and calf.
Halliday & Slater (1983) state that long-lived animals
that can recognize each other as individuals can also
develop relationships with one another, depending on
their past experience. Individual recognition may be
important in the maintenance of mating pairs (Halliday
1983) and in male–male competition. A clear social func-
tion of interindividual vocal variation has been verified
in males of several species of pinnipeds (Stirling 1971;
Shipley et al. 1981;Stirling et al. 1987). Reby et al. (1998)
indicate that individual vocal characteristics in polygen-
ous fallow deer, Dama dama, probably facilitate the estab-
lishment of relationships between specific males and
females. Females may exercise some preference to mate
with particular males that are recognized by their individ-
ual vocal pattern. The presence of preferential associ-
ations between male and female Antillean manatees,
T. manatus manatus, in captivity (V. M. Rosa, unpublished
data) could also be based on the recognition of individual
vocal differences.
The results presented in this study provide evidence for
the potential use of vocal parameters, especially those
related to the fundamental frequency, in the individual
recognition of Amazonian manatees. Further research is
needed to demonstrate conclusively whether individual
recognition capabilities exist. The best approach will be to
perform playback experiments of natural and modified
sounds, isolating the most significant parameters, and
verifying how much it can be modified until recognition
is no longer possible.
Given the precarious status of most Amazonian mana-
tee populations (considered vulnerable by the IUCN;
Baille & Groombridge 1996), acoustic surveys can prove
to be especially useful in determining population trends
in habitats where other methods are difficult or even
impossible to use due to poor visibility, dense vegetation,
or cryptic habits of the subjects (Baptista & Gaunt 1997).
In addition to their shy behaviour on the surface,
Amazonian manatees occur in areas with characteristics
that make it impossible to observe the animals directly
(turbid waters, difficult access, thick vegetation).
By achieving a better understanding of the character-
istics and functions of manatee vocalizations, it will be
possible to determine the applicability of bioacoustics in
the management and preservation of these sirenians
throughout the Amazon basin.
Acknowledgments
We thank the directors and the staff of INPA, CPPMA and
MPEG for the opportunity to work with their animals and
for their support during this project, especially Vera M. F.
da Silva, J. Anselmo d’Affonseca Neto, Stella M. Lazzarini,
Eliete Carvalho and Paulo H. G. Castro. We thank Luiz
Pedreira Gonzaga and Maria Nazareth F. da Silva for their
help and support. Anthony B. Rylands, Jeff Podos,
Dwaine Santee, Jason A. Mobley, Karen Paglia, Jacques
Vielliard, Daniel W. Leger and the anonymous referees
provided most valuable comments. This research was
funded by MacArthur Foundation, Fundac¸a˜o O Botica´rio
de Protec¸a˜o a` Natureza, Conservation International,
Financiadora de Estudos e Projetos (FINEP), Ministe´rio de
Cieˆncia e Tecnologia/Programa Piloto para Protec¸a˜o das
Florestas Tropicais do Brasil (MCT/PP-G7), Conselho
Nacional de Desenvolvimento Cienti´fico e Tecnolo´ gico
(CNPq) and the U.S. Fish and Wildlife Service. The
research presented here was evaluated and approved by
the Animal Behavior Society’s Animal Care Committee
on 1 February 2001.
References
Abs, M. & Jeismann, R. 1988. Do courtship songs differ individually
in the domestic pigeon Columbia livia domestica?Bioacoustics,1,
151–157.
Anderson, P. K. & Barclay, R. M. R. 1995. Acoustic signals of
solitary dugongs: physical characteristics and behavioral corre-
lates. Journal of Mammalogy,76, 1226–1237.
Baille, J. & Groombridge, B. (Eds) 1996. Red List of Threatened
Animals. Gland, Switzerland: IUCN.
Baptista, L. F. & Gaunt, S. L. L. 1997. Bioacustics as a tool in
conservation studies. In: Behavioral Approaches to Conservation in
the Wild (Ed. by J. R. Clemmons & R. Buchholtz), pp. 212–242.
Cambridge: Cambridge University Press.
Balcombe, J. P. 1990. Vocal recognition of pups by mother Mexican
free-tailed bats, Tadarida brasiliensis mexicana.Animal Behaviour,
39, 960–966.
Bengston, J. L. & Fitzgerald, S. M. 1985. Potential role of vocaliz-
ations in West Indian manatees. Journal of Mammalogy,66,
816–819.
Best, R. C. 1982. Seasonal breeding in the Amazonian manatee,
Trichechus inunguis (Mammalia: Sirenia). Biotropica,14, 76–78.
Best, R. C. & Da Silva, V. M. 1979. Peixe-Boi. Uma Sereia na
represa? Cespaulista,16, 26–29.
Caldwell, M. C. & Caldwell, D. K. 1965. Individualized whistle
contours in bottlenose dolphins (Tursiops truncatus). Nature,207,
434–435.
Caldwell, M. C. & Caldwell, D. K. 1968. Vocalization of naive
captive dolphins in small groups. Science,159, 1121–1123.
Caldwell, M. C. & Caldwell, D. K. 1979. The whistle of the Atlantic
bottlenosed dolphin (Tursiops truncatus): ontogeny. In: Behavior of
Marine Animals: Current Pespectives in Research. Vol. 3 (Ed.byH.E.
Winn & B. L. Olla), pp. 369–401. New York: Plenum.
Caldwell, M. C., Caldwell, D. K. & Tyack, P. L. 1990. Review of the
signature-whistle hypothesis for the Atlantic bottlenose dolphin.
In: The Bottlenose Dolphin (Ed. by S. Leatherwood & R. R. Reeves),
pp. 199–234. San Diego, California: Academic Press.
309SOUSA-LIMA ET AL.: IDENTITY IN MANATEE CALLS
Cheney, D. L. & Seyfarth, R. M. 1982. How vervet monkeys
perceive their grunts: field playback experiments. Animal Behav-
iour,30, 739–751.
Evans, W. E. & Herald, E. S. 1970. Underwater calls of a captive
amazon manatee, Trichechus inunguis.Journal of Mammalogy,51,
820–823.
Everitt, B. S. & Dunn, G. 1992. Applied Multivariate Data Analysis.
New York: Oxford University Press.
Falls, J. B. 1982. Individual recognition by sounds in birds. In:
Acoustic Communication in Birds. Vol. 2 (Ed. by D. H. Kroodsma &
E. H. Miller), pp. 237–278. New York: Academic Press.
Frommolt, K.-H., Kruchenkova, E. P. & Russig, H. 1997. Individu-
ality of territorial barking in Arctic foxes, Alopex lagopus (L., 1758).
Zeitschrift fu¨r Sa¨ ugetierkunde, suppl. II,62, 66–70.
Gerhardt, H. C. 1983. Communication and the environment. In:
Animal Behaviour Communication. Vol. 2 (Ed. by T. R. Halliday &
P. J. B. Slater), pp. 82–113. Oxford: Blackwell Scientific.
Gelfand, D. L. & McCracken, G. F. 1986. Individual variation in the
isolation calls of Mexican free-tailed bat pups (Tadarida brasiliensis
mexicana). Animal Behaviour,34, 1078–1086.
Halliday, T. 1983. Information and communication. In: Animal
Behaviour Communication. Vol. 2 (Ed. by T. R. Halliday & P. J. B.
Slater), pp. 43–81. Oxford: Blackwell Scientific.
Halliday, T. R. & Slater, P. J. B. 1983. Introduction. In: Animal
Behaviour Communication. Vol. 2 (Ed. by T. R. Halliday & P. J. B.
Slater), pp. 1–8. Oxford: Blackwell Scientific.
Hammerschmidt, K. & Todt, D. 1995. Individual differences in
vocalizations of young barbary macaques (Macaca sylvanus): a
multi-parametric analysis to identify critical cues in acoustic sig-
nalling. Behaviour,132, 381–399.
Hartman, D. S. 1979. Ecology and behavior of the manatee,
Trichechus manatus, in Florida. Special Publication of the American
Society of Mammalogy,5, 1–153.
Insley, S. J. 1992. Mother–offspring separation and acoustic
stereotypy: a comparison of call morphology in two species of
pinnipeds. Behaviour,120, 103–122.
Janik, V. M. 1999. Origins and implications of vocal learning in
bottlenose dolphins. In: Mammalian Social Learning: Comparative
and Ecological Perspectives (Ed.byH.O.Box&K.R.Gibson),
pp. 308–326. Cambridge: Cambridge University Press.
Janik, V. M. & Slater, P. J. B. 1998. Context-specific use suggests
that bottlenose dolphin signature whistles are cohesion calls.
Animal Behaviour,56, 829–838.
Janik, V. M. & Slater, P. J. B. 2000. The different roles of social
learning in vocal communication. Animal Behaviour,60, 1–11. doi:
10.1006/anbe.2000.1410.
Janik, V. M., Dehnhardt, G. & Todt, D. 1994. Signature whistle
variations in a bottlenose dolphin, Tursiops truncatus.Behaviour
Ecology and Sociobiology,35, 243–248.
Jones, G. & Ransome, R. D. 1993. Echolocation calls of bats are
influenced by maternal effects and change over a lifetime. Pro-
ceedings of the Royal Society of London, Series B,252, 125–128.
Jones, I. L., Falls, J. B. & Gaston, A. J. 1987. Vocal recognition
between parents and young of ancient merrelets, Synthliboram-
phus antiquus (Aves: Alcidae). Animal Behaviour,35, 1405–1415.
Lengagne, T., Lauga, J. & Jouventin, P. 1997. A method of
independent time and frequency decomposition of bioacoustic
signals: inter-individual recognition in four species of penguins.
Comptes rendus de l’Acade´mie des sciences, Serie III, Sciences de la
Vie/Life Sciences,320, 885–891.
Lessells, C. M., Coulthard, N. D., Hodgson, P. J. & Krebs, J. R.
1991. Chick recognition in European bee-eaters: acoustic play-
back experiments. Animal Behaviour,42, 1031–1033.
Loesche, P., Stoddard, P. K., Higgins, B. J. & Beecher, M. D. 1991.
Signature versus perceptual adaptations for individual vocal
recognition in swallows. Behaviour,118, 15–25.
McArthur, P. D. 1982. Mechanisms and development of parent–
young vocal recognition in the pin˜on jay (Gymnorhinus cyano-
cephalus). Animal Behaviour,30, 62–74.
McCowan, B., Reiss, D. & Gubbins, C. 1998. Social familiarity
influences whistle acoustic structure in adult female bottlenose
dolphins (Tursiops truncatus). Aquatic Mammals,24, 27–40.
McGregor, P. K. & Byle, P. 1992. Individually distinctive bittern
booms: potential as a census tool. Bioacoustics,4, 93–109.
Manly, B. J. F. 1994. Multivariate Statistical Methods: a Primer. 2nd
edn. London: Chapman & Hall.
Reby, D., Joaquim, J., Lauga, J., Lek, S. & Aulagnier, S. 1998.
Individuality in the groams of fallow deer (Dama dama) bucks.
Journal of Zoology,245, 79–84.
Rendall, D., Rodman, P. S. & Emond, R. E. 1996. Vocal recognition
of individuals and kin in free-ranging rhesus monkeys. Animal
Behaviour,51, 1007–1015.
Reynolds III, J. E. 1981. Aspects of the social behaviour and herd
structure of a semi-isolated colony of West Indian manatees
Trichechus manatus.Mammalia,45, 431–451.
Robisson, P. 1992. Roles of pitch and duration in the discrimination
of the mate’s call in the king penguin Aptenodytes patagonicus.
Bioacoustics,4, 25–36.
Sanvito, S. & Galimberti, F. 2000. Bioacoustics of southern
elephant seals. II. Individual and geographical variation in male
aggressive vocalisations. Bioacoustics,10, 287–307.
Shipley, C., Hines, M. & Buchwald, J. S. 1981. Individual differ-
ences in threat calls of northern elephant seal bulls. Animal
Behaviour,29, 12–19.
Sokal, R. R. & Rohlf, F. J. 1995. Biometry. 3rd edn. New York: W. H.
Freeman.
Sonoda, S. & Takemura, A. 1973. Underwater sounds of the
manatees, Trichechus manatus manatus and Trichechus inunguis
(Trichechidae). Reports of the Institute for Breeding Research, Tokyo
University of Agriculture,4, 19–24.
Steel, C. 1982. Vocalization patterns and corresponding behaviour
of the West Indian manatee (Trichechus manatus). Ph.D. thesis,
Florida Institute of Technology.
Stirling, I. 1971. Studies on the behaviour of the South Australian
fur seal, Arctocephalus forsteri (Lesson) I. Annual cycle, posture and
calls, and adult males during the breeding season. Australian
Journal of Zoology,19, 243–266.
Stirling, I., Calvert, W. & Spencer, C. 1987. Evidence of stereo-
typed underwater vocalisations of male Atlantic walruses
(Odobenus rosmarus rosmarus). Canadian Journal of Zoology,65,
2311–2321.
Tooze, Z. J., Harrington, F. H. & Fentress, J. C. 1990. Individually
distinct vocalizations in timber wolves, Canis lupus.Animal Behav-
iour,40, 723–730.
310 ANIMAL BEHAVIOUR, 61, 2
