Classification of Captive North American River Otters (Lontra canadensis) Vocal Repertoires: Individual Variations, and Age Class Comparisons

Abstract

This is the first study to examine in detail the vocal behaviors of North American river otters (Lontra canadensis), and the results suggest that river otters have complex vocal repertoires comprised of four distinct vocal types and seven sub-call types. The vocalizations and behaviors of ten captive North American river otter adults, one litter of newborn pups and one litter of pups at eight weeks old were recorded using a SONY Handheld DV camera and an infrared surveillance system. A quantitative analysis of 2726 calls on the adults and 299 calls for the pups was conducted for acoustic parameters that included frequencies, powers, and duration. Whine, chirp and chatter call types were the main vocal elements of the vocal repertoire and were present at birth. Pups vocals were structurally underdeveloped versions of the adult vocals and adults call types showed individual variations. This suggests that vocalizations are likely individually modified as pups enter adulthood. A unique whistle was present in newborn pup vocal repertoires but appeared to be reduced in the repertoire by eight weeks old. However, further research needs to be conducted to determine the function of the whistle.

Full text

Sciknow Publications Ltd. ABC 2014, 1(4)502-517
Animal Behavior and Cognition DOI: 10.12966/abc.11.07.2014
©Attribution 3.0 Unported (CC BY 3.0)
Classification of Captive North American River Otters
(Lontra canadensis) Vocal Repertoires: Individual
Variations, and Age Class Comparisons
Carla Almonte1*
1City University of New York
*Corresponding author (Email: calmonte1215@mac.com)
Citation – Almonte, C. (2014). Classification of captive North American river otters (Lontra canadensis)
vocal repertoires: Individual variations, and age class comparisons. Animal Behavior and Cognition, 1(4),
502-517. doi: 10.12966/abc.11.07.2014
Abstract - This is the first study to examine in detail the vocal behaviors of North American river otters
(Lontra canadensis), and the results suggest that river otters have complex vocal repertoires comprised of
four distinct vocal types and seven sub-call types. The vocalizations and behaviors of ten captive North
American river otter adults, one litter of newborn pups and one litter of pups at eight weeks old were
recorded using a SONY Handheld DV camera and an infrared surveillance system. A quantitative
analysis of 2726 calls on the adults and 299 calls for the pups was conducted for acoustic parameters that
included frequencies, powers, and duration. Whine, chirp and chatter call types were the main vocal
elements of the vocal repertoire and were present at birth. Pups vocals were structurally underdeveloped
versions of the adult vocals and adults call types showed individual variations. This suggests that
vocalizations are likely individually modified as pups enter adulthood. A unique whistle was present in
newborn pup vocal repertoires but appeared to be reduced in the repertoire by eight weeks old. However,
further research needs to be conducted to determine the function of the whistle.
Keywords - River Otters, Vocal repertoire, Vocalizations, Lontra canadensis, Otter pups, Communication.
North American river otters (Lontra canadensis) have relatively fluid social structures that can
adapt to a wide range of ecological conditions with variable group composition (Beckel, 1991). Because
the degree of sociality varies between and within habitat types, it is likely that communication in river
otters is also flexible and serves several functions (Rostain, Ben-David, Groves, & Randall, 2004).
Research shows Prince William Sound nonsocial otters deposit spraints at more latrines sites than social
otters, and this serves to facilitate mutual avoidance (Ben-David, Blundell, Kern, Brown, & Jewitt, 2005).
As well, documented observations suggest that during social interactions vocalizations are flexible. Wild
otters produce ‘screams’ and ‘wickers’ when engaged in physical altercations (Kruuk, 2006, p. 87) and
captive otters ‘grunt’ and ‘chirp’ during play (Beckel, 1991). Research implies that river otters use a
variety of vocalizations to serve to maintain group cohesiveness, signal alarm or danger, to express
apparent fear or anger and avoid aggressive interactions (Melquist & Hornocker, 1983).
Vocal studies conducted on members of the subfamily Lutrinae have partially explained the vocal
repertoire of the giant otter (Pteronura brasiliensis) (Bezzera, Souto, & Schiel, 2010), sea otter (Enhydra
lutris; McShane, Estes, Reidman, & Staedler, 1995), and Eurasian otter (Lutra lutra; Gnoli & Prigioni,

Almonte 503
1995) and the research suggest they have complex vocal systems. The giant otter is highly social in
comparison to the sea otter and Eurasian otter that engage in limited social interactions. These studies
showed that despite the differences in social behavior, similarities exist in call types exhibited and the
manner in which they are used. They all exhibit a “scream,” which is a high-pitched call and the Eurasian
Otter and giant otter exhibit a “blow” that is described as exhalation of air.
To date, no formal studies have been conducted on river otter vocal systems, and based on the
literature it is likely that the repertoire is similar to other otter species. The purpose of this study was to
expand on the documented observations of river otter vocals by delineating and quantifying the vocal
repertoire of captive river otter adults and pups. This was done to investigate the existence of individual
variation among adults that exhibit the same call types and to compare the vocal repertories of adults and
pups. In addition, the relationship between call types exhibited and the arousal state of the individual
producing the call was examined. If call types are associated with particular arousal states then vocals
might relay information about the temperament of an individual at a given point in time.
In nature elusive, adult river otters can be difficult to observe. As well, pups do not emerge from
the natal den until approximately 6-8 weeks (Gorman, Erb, McMillan, Martin, & Homyack, 2005), which
makes it difficult to observe them during the newborn phase. The captive environment allowed for otters
to be observed at any time in a 24 hr period using non-invasive techniques. Furthermore, footage of
newborn pups vocal behaviors could be captured without disturbing the birthing process or natal den.
Although, captive behaviors are not a direct reflection of wild behaviors, the information gained from this
study about river otter vocal behaviors could provide the groundwork for investigations of wild river otter
vocal behaviors.
Method
Study Sites
The study consisted of ten adult otters, one litter of five newborn pups, and one litter of three
pups at eight weeks old housed in five zoological facilities throughout the Greater New York City Area.
Turtle Back Zoo located in West Orange, NJ (outdoor dimensions: 750 sq. ft; indoor dimensions: 102 ft2)
housed three adults (two female, one male). Beardsley Zoo located in Bridgeport, CT (outdoor
dimensions: 900 ft2; indoor dimensions: 150 ft2) housed a male/female adult mating pair and the litter of
newborn pups. Palisades Park Conservancy situated in Bear Mountain, NY (outdoor dimensions: 150 ft2;
indoor dimensions: 48 ft2) housed a solitary adult male. Atlantis Aquarium located in Riverhead, NY
(outdoor dimensions: 726 ft2; indoor dimensions: 80 ft2) housed a male/female adult mating pair and eight
weeks old pups. Stamford Farm and Museum situated in Stamford, CT (outdoor dimensions: 2040 ft2;
indoor dimensions: 85 ft2) housed a non-mating male/female adult pair.
Data Acquisition
Video and audio recordings were collected from November 2007 – November 2010. Sessions
were randomly conducted during the day and overnight throughout the three years of data collection. Day
sessions gave insight into how otters behaved vocally when in the presence of humans, while overnight
sessions gave insight to the vocals exhibited when the otters were devoid of human presence. Day
sessions occurred when the study sites were open to the public (10:00 a.m. - 4:00 p.m.
1
) and consisted of
manually recording the otters for 30 consecutive minutes. A SONY DV hand-held camera and tripod
were used to capture the recordings. One or two sessions were conducted per visit to a study site, and
study sites were visited three times a week (Monday – Friday). The duration of the day sessions and
number of visits a week to study sites were kept to a minimum to ensure the daily routines of the river
otters were not disrupted. Fifteen sessions were conducted for each facility, however sessions continued if
1
Hours of operation varied per study site.

Almonte 504
thirty usable vocals were not collected from each individual. Night sessions occurred when study sites
were closed to the public (4:00 p.m. - 10:00 a.m.
2
) for 15 - 18 continuous hours (the time recordings
ended were based on when the otters received their morning feedings). A temporary surveillance system
was installed in the overnight sleeping quarter(s) of a given study site. It consisted of one - three infrared
camera(s) (the number of cameras installed was based on the number of indoor dens at the facility), one
LOUROE microphone (audio pick-up of 30 ft diameter) and one digital video recorder (500GB). The
digital video recorder was put on a timer to record for three consecutive nights and each night was one
session. Nine sessions were conducted per facility, but would be increased if thirty vocals were not
obtained from each otter, or to capture a birth (litters were born at Beardsley Zoo and Atlantis Aquarium).
Once data collection was completed at one study site the surveillance system was uninstalled and
reinstalled in the next study site.
Video Analysis
The data collection consisted of 2129 hrs of footage (day session: 55 hrs; overnight sessions:
2074 hrs). The footage from each session was viewed in real time to locate the times vocals were
produced. All sessions that contained at least one vocal were imported into FINAL CUT PRO to edit out
the video footage that contained no vocal activity leaving shorter video clips that contained only vocals.
The shorter video clips were imported into RAVEN PRO 1.3 to obtain a spectrogram (Window Type:
Hann: 256; 270 Hz; overlap 50%; discrete Fourier Transform (dft): 256) of the vocal. The spectrogram
was a visual picture of the sound that was based on frequency and temporal patterns.
Individual Identification & Spectral Analysis of Vocalizations
Spectrograms of each edited file were played back to seek out the times usable calls (isolated,
void of background noise) were produced. The time a usable call was produced in the spectrogram was
corresponded to the timestamp on the edited video file to make an identification of the individual
producing the call. Both physical features (unique markings, fur color, tail shape, tail width and body
size) and behavioral features (mouth opening; up and down movement of shoulders) were used to identify
the individual producing the call. For adults when identification was not possible the call was not
included in the dataset. The pups could not be individually identified because they did not have individual
identities at the time sessions were conducted. The final dataset consisted of 2726 adult usable calls where
the identification of the individual producing the call was known, and 299 calls from the two litters of
pups, collectively. RAVEN PRO 1.3 quantitatively analyzed each call for two acoustic categories and
nine acoustic parameters: Frequency (High, Low, Range, Mean, Max, and Center) and Power (Max and
Average). Duration was determined by subtracting the start time from the end time. When harmonic
bands were present in the spectral image, the number of bands were counted and recorded.
Classification of Call Types for Adult Vocalizations
To develop a classification system all 2726 adult and 299 pup vocals usable calls were assigned a
number as well as a name based solely on the sound it produced to the human ear. Adults produced 11
distinctive sounds (call types): whine (1), chirp (2), chatter (3), creek (4), squeak (5), scream (6), grunt
(7), swish (8), hiss (9), blow (10) and hiccup (11), and the pups produced four distinctive sounds: whine
(1), chirp (2), chatter (3) and whistle (12). For the adults, the spectral analysis showed structural
similarities in calls that were aurally different. The analysis of visual spectrograms showed that the creek,
squeak, scream, hiss, swish, were structurally comparable to the whine, and the hiccup was structurally
comparable to the chirp. Furthermore, the chatter spectrograms revealed it was a series of whines or
2
Hours of closures varied per study site.

Almonte 505
chirps that occurred in rapid succession. A secondary classification system was developed considering
these visual similarities. The secondary classification consisted of four fundamentally distinct call types:
whine (1), chirp (2), grunt (7), and blow (10). The whine was a fundamental call type because every
individual exhibited it (including pups). The chirp, grunt, and blow produced very distinct sounds and the
spectral images were visually unlike the whine. Chatters were sub-divided into whine-chatter sub-types
that were grouped with whines and chirp-chatter sub-types that were grouped with chirps. The creek,
squeak, scream, swish, hiss, were grouped with the whine, and the hiccup was grouped with the chirp.
Because whistles were unique to the pups they were not included in the discriminant function analysis.
Vocal Usage and Individual Behavior
An ethogram (Table 1) was developed using both video footage and first hand observations in
order to have an inventory of all behaviors (and their definitions) that were observed. The behaviors were
divided into three states of arousal and assigned a number: non-aggressive (0), moderately aggressive (1),
and highly aggressive (2) (Table 1). An arousal state was defined as a particular state of mind at the time
a vocalization is produced that was in regards to the individual’s level of agitation. Physically expressed
behaviors defined in the ethogram were used to determine an individual’s arousal state and assign a code
for each usable call.
Inter-Observer Reliability
To determine that the methods used to assign each usable call a name, individual, and level of
arousal were repeatable a secondary observer checked for inter‐observer reliability. The second observer
viewed 50 video clips (10 clips from each study site) and 301 calls (11%) of the data set. The unbiased
observations of the second observer were based solely on the 11 distinct sounds, identifying techniques
and behavioral definitions that were used in the methods.
Statistical Analysis
STATISTICA v7.1 was used to conduct a discriminant analysis on the adult dataset to search for
statistical meaningful groups in the call classification system. If calls that sound differently are
structurally distinct calls then call types should discriminate under the eleven distinctive sounds heard by
ear. STATISTICA v7.1 was used to conduct a Kruskal-Wallis ANOVA to determine the presence of
individual variations in adults. The ANOVA included all the variables that were determined in the
quantitative analysis, but the squeak and hiccup were not included in this analysis because they were only
heard from one individual. If individual variations exist, then there should be significant differences
across acoustic parameters among individuals that share the same call type. The chi-square test was used
to assess how the frequencies of sounds were distributed over levels of aggression. If there is an
association between the arousal state of the individual and the call type that is exhibited then the calls may
relay information about the temperament of the individual producing it. Morphological variations exist
between adults and pups therefore to test the hypotheses that there are significant differences in
frequencies, harmonics, power and duration of shared call types among age classes a Mann‐Whitney Test
was conducted. The Mann‐Whitney Test was used because it is a non‐parametric test that does not
assume a normal distribution.

Almonte 506
Table 1
Ethogram of Captive River Otter Behaviors and Number of Individuals that Exhibited the Behavior.
Behavior
Sub-Behavior
Definition
Arousal State Code
Number of otters
that exhibited
behavior
Sleeping
Indoor
Period of rest in den for 30
minutes or more
Non-aggressive
0
10**
Outdoor
Period of rest for 30
minutes or more; occurs
mainly by trees
Non-aggressive
0
10*
Social/Group
Sleeping with another
otter(s) (within an otters
distance away) for 30
minutes or more
Non-aggressive
0
9**
Grooming
Self
Licking and biting of own
skin, fur, or paws
Non-aggressive
0
10
Others
Licking and biting of skin,
fur, or paws of another
otter
Non-aggressive
0
9
Massaging
Rubbing body or back
against a substrate (i.e.
rock, log, wall)
Non-aggressive
0
10
Swimming
Submerged
Otter completely
submerged in water or in
pool and maneuvering
through the water for more
than 30 seconds
Non-aggressive
0
10*
Acrobatic
Underwater swimming
involving acrobatic
movements (i.e. spinning,
back flips)
Non-aggressive
0
9
Floating
Swimming on the back
during floating
Non-aggressive
0
1
Play
Toys
Grasping, manipulating, or
playing with toys
Non-aggressive
0
10*
Climbing
Scaling a gate or wall
Non-aggressive
0
3
Plopping
Running forward and then
laying body and head flat
and spread out onto the
ground
Non-aggressive
0
1
Water
Splashing water out of the
pool with paws
Non-aggressive
0
1
Self
Playing with body parts,
tail or feet
Non-aggressive
0
1
Otter
Physical contact with
otter(s) that are docile, and
playful; or more than one
otter manipulating a toy(s)
Non-aggressive
0
9*
Enrichment
Painting
Painting stimulated by the
trainers
Non-aggressive
0
2

Almonte 507
Behavior
Sub-Behavior
Definition
Arousal State Code
Number of otters
that exhibited
behavior
Training
Other activities stimulated
by trainers (tricks)
Non-aggressive
0
5
Seeking
Searching behavior that
results from enrichment
(i.e. stashing food in toys)
Non-aggressive
0
2
Aggressive
Vocal
Vocalizations that occur
within the 2 minutes
before, during, or 2
minutes after a hostile
physical confrontation or
vocal quarrel
Moderately Aggressive
1
10
Physical
Physical attack on another
otter (biting, scratching), or
violent wrestling
Highly Aggressive
2
9
Defensive/Non-physical
Crouching head down, tail
positioned down
Moderately Aggressive
1
4
Non-aggressive
Vocal
Vocalizations that occur
within the 2 minute before,
during, or 2 minute after a
non-aggressive physical
behavior
Non-aggressive
0
10
Submissive
Physical
Subordinate behavior (i.e.
cowering, running away,
giving up after a quarrel or
confrontation)
Moderately Aggressive
1
5
Stereotypical
Scratching
Scratching at gate or door
with paws repeatedly more
than 5 times in a row
Moderately Aggressive
1
2
Pacing
Walking back and forth
repeatedly in the same one
area
Moderately Aggressive
1
5
Swimming
Swimming in circles
Moderately Aggressive
1
2
Chewing
Chewing on inanimate
body part
Moderately Aggressive
1
1
Note: Arousal state code assigned to each behavior is also shown in parenthesis “()”.
**Behaviors exhibited by newborn pups and 2-months old pups.
* Behaviors exhibited by 2-months old pups.
Results
Call Classification
The results of the discriminant analysis on the classification based on sound suggested that 72%
of the calls were accurately identified based on the variables that were used in this study. However, there
Table 1 (cont.)

Almonte 508
was extensive overlapping in the distribution of the discriminant scores (Figure 1A). The discriminant
analysis conducted on the four fundamentally distinct call classifications reached an overall accuracy of
89.1%. The canonical discriminant functions showed distinctive groups (Whine, Chirp, Grunt, and Blow)
in the distribution of the discriminant scores (Figure 1B).
A) B)
Figure 1. A) Discriminant function analysis results for the eleven sound classification systems. Each square represents a centroid
for each call category and is marked with a number that represents the sound. Images were obtained using SPSS for Mac, version
20.0, SPSS Inc. B) Discriminant function analysis results for the four sound classification systems. Each square represents a
centroid for each call category and is marked with a number that represents the sound. Images were obtained using SPSS for
Mac, version 20.0, SPSS Inc.
Fundamental Call Types
Whine. The whine was universal to all otter’s vocal repertoires. The whine call type had the
greatest variation in sound. It was moderately pitched, and showed a mean duration of 1.4 s (Table 2).
Spectrally the whine could be static or harmonic (Figure 2A, B).
Chirp. The chirp had a high frequency with a short mean duration of 0.2 s (Table 2), but occurred
in intermittent succession. It was aurally comparable to a bird’s chirp and spectrally produced a series of
harmonic bands. The shape of the bands were either linear shaped or had an upside down “v” shape
(Figure 2C, D).
Grunt. The grunt sounded comparable to the sound produced when a human clears their throat. It
was low-pitched, with a mean duration of 0.01 s (Table 2). It produced a very distinct spectrogram with
little variation in the images within or among the individuals that exhibited it (Figure 2E).
Blow. The blow sounds like air being blown out from the nose. Three otters exhibited this call
type: two males and one female. The males produced similar spectral images, but the female produced a
different spectrogram. The blow had a high frequency of 10514 kHz, and a mean duration of 0.9 s (Figure
2F, Table 2).

Almonte 509
Table 2
Descriptive Statistics for all Call Types Across the Variables (mean ± SD)
Note: Call (n) = Call type and sample size; HF = high frequency; LF = low frequency; RF = range of frequency; MNF = mean frequency; MXF = max frequency; CTF = center
frequency; MXP = max power; AVP = average power; DRT = duration; HRM = harmonics.
Call
(n)
HF
(kHZ)
LF
(kHz)
RF
(kHZ)
MNF
(kHZ)
MXF
(kHZ)
CTF
(kHZ)
MXP
(dB)
AVP
(dB)
DRT
(s)
HRM
(#)
Whine
(959)
ADULTS
5970+/-1735
609+/-501
5471+/-4153
3289+/-922
2587+/-913
2573+/-817
88+/-11
69+/-10
1.4+/-1.1
1.1+/-2
Whine
(170)
PUPS
4852+/-754
1165+/-450
3695+/-923
3009+/-478
1993+/-667
1980+/-539
70+/-4
54+/-3
0.5+/-0.2
1.9+/-0.8
Chirp
(1024)
ADULTS
8764+/-3395
1407+/-792
7480+/-4951
5080+/-1752
2478+/-1077
2525+/-905
91+/-8
73+/-7
0.2+/-0.4
3.3+/-1.7
Chirp
(100)
PUPS
4209+/-1120
1890+/-684
2706+/-4354
3049+/-530
2700 +/-572
2681+/-484
81+/-6
66+/-5
0.2+/-0.1
1.6+/-0.8
Grunt
(154)
1996+/-3103
300+/-921
1697+/-2266
1146+/-1990
583+/-1083
621+/-1151
75+/-7
62+/-8
0.7+/-0.4
0.01+/-0.16
Blow
(78)
10514+/-5296
708+/-418
9807+/-5331
5611+/-2646
1897+/-815
2317+/-646
86+/-8
69+/-6
0.3+/-0.1
0.9+/-0.2
Chatter
(290)
ADULTS
6571+/-1073
661+/-472
5900+/-1233
3616+/-577
3028+/-754
3076+/-544
94+/-9
74+/-8
1.8+/-1.3
0.4+/-1.5
Chatter
(5)
PUPS
5453+/-122
2342 +/-136
2982 +/- 103
3897 +/- 119
3188 +/- 265
3263 +/-168
58 +/-1.9
45 +/-1.5
0.6 +/- 0.1
2.4 +/-2.2
Creek
(124)
6481+/-1216
982+/-582
5500+/-1346
3669+/-823
2945+/-653
2896+/-527
86+/-10
67+/-10
1.1+/-0.9
0.4+/1.28
Squeak
(14)
9644+/-2433
447+/-335
9197+/-2374
5045+/-1268
3295+/-1210
3830+/-689
102+/-2
83+/-1
2.1+/-1.5
2.4+/-3.9
Scream
(20)
6619+/-306
449+/-388
6170+/-605
3534+/-175
2897+/-575
3000+/-304
101+/-2
83+/-3
1.5+/-1.2
0.8+/-1.5
Swish
(20)
5428+/-901
881+/-300
4548+/-1052
3155+/-418
2269+/-881
2475+/-862
74+/-7
58+/-5.6
1.4+/-1
0.9+/-1
Hiss
(39)
5502+/-986
1713+/-576
3788+/-1072
3608+/-605
3034+/-668
3067+/-620
73+/-7
54+/-11
1+/-0.4
0.3+/-0.6
Hiccup
(4)
5698+/-774
543+/-445
5155+/-1092
3121+/-317
2344+/-1285
2344+/-1151
88+/-9
67+/-7
0.2+/-0.1
1.8+/-2.1
Whistle
(24)
4947+/-868
998+/-274
3949+/-924
2973+/-448
1898+/-545
1789+/-354
71+/-3
55+/-3
0.4+/-0.1
2+/-1

Almonte 510
Sub-Call Types
Chatter. The chatter was aurally comparable to teeth chattering. The spectral appearance of the
chatter showed it was a chirp or a whine that occurred in a rapid succession. It had a high frequency with
an average duration of 1.8 s (Figure 2G, H). Chatters have been described as a staccato cry in Eurasian
Otters (Gnoli & Prigioni, 1995).
Creek. The creek was heard in five otters, and predominately females exhibited creeks. Creeks
sounded like an old wooden door opening, with some slight variations based on the individual producing
the call. The creek had an average power of 67dB, and an average duration of 1.1 s (Figure 2I).
Squeak. The squeak sounded like a shrieking whine with an average duration of 2.1 s. Squeaks
were spectrally graded and appeared to be a whine and a chirp used in conjunction (Figure 2J). The
squeak has been described in sea otters (McShane, Estes, Riedman, & Staedler, 1995) although, is not
clear if these authors description of a squeak is analogous to the description of the chirp presented in this
study.
Scream. Screams were difficult to distinguish from the whine because they looked spectrally
identical and sounded very similar aurally. The main difference was the scream became increasingly
louder the longer expressed. The scream was high-pitched with a mean duration of 1.5 s (Figure 2K,
Table 2). Female sea otters have been observed screaming when separated from pups (McShane, Estes,
Riedman, & Staedler, 1995).
Hiss. The hiss sounds comparable to the hiss of a snake. The hiss is a static low frequency call
(Figure 2L) that averaged 1 s. The hiss had an average power of 54dB.
Swish. The swish sounded like water swirling around in a container. Its spectral image produces a
static appearance (Figure 2M), and was similar to the hiss. It was a low frequency call with an average
power of 58dB.
Hiccup. The hiccup was heard in a young female while engaged in play with a toy. It aurally
sounded like a human hiccup and spectrally appeared similar to the chirp (Figure 2N).
Whistle. The whistle was unique to newborn pups, it was harmonic and appeared similar to the
chirp, but has a down sweep of the harmonic bands at the start of the vocal (Figure 4G). The whistle had
an average of two harmonic bands.
Individual Variation
The results of the Kruskal-Wallis ANOVA (Table 3) showed highly significant differences (p <
0.01) for all call types except the scream and the hiss. The hiss showed no individual variations likely
because one adult female exhibited 36 of the 39 hiss vocals recorded.
Vocalization Usage and Behavior
The results of the chi-square test suggested that vocal usage was highly associated with the state
of arousal (X2 = 1357, df = 20, p < 0.01). The bar graph (Figure 3) showed that chirps were the dominant
call type exhibited when the otter was in a non-aggressive state, and when the otter was in a highly
aggressive state chatters were the dominant call types. Otters expressed various call types when exhibiting
moderately aggressive behaviors.

Almonte 511
A) B) C)
D) E) F)
G) H) I)
J) K) L)
M) N)
Figure 2. Spectrogram images for the call types: A) non-harmonic whine, B) harmonic whine, C) “v” shaped harmonic chirp, D)
linear shaped harmonic chirp, E) grunt, F) blow, G) whine in succession chatter, H) chirp in succession chatter, I) creek, J)
squeak, K) scream, L) hiss, M) swish, N) hiccup. Y-axis represents frequency (kHz) and X-axis represents time (s).

Almonte 512
Table 3
Results of the Kruskal-Wallis ANOVA Test for Individual Differences in Call Type.
Call Type
Source Of Variation
df
F
p value
Whine
Between Groups
Within Groups
9
949
26.28837
< 0.01*
Chirp
Between Groups
Within Groups
9
1017
8.83566
< 0.01*
Grunt
Between Groups
Within Groups
4
149
166.951
< 0.01*
Blow
Between Groups
Within Groups
2
75
222.92068
< 0.01*
Chatter
Between Groups
Within Groups
9
280
36.70185
< 0.01*
Creek
Between Groups
Within Groups
4
119
7.25704
< 0.01*
Scream
Between Groups
Within Groups
2
17
11.52863
> 0.05
Swish
Between Groups
Within Groups
12
8
0.31311
< 0.01*
Hiss
Between Groups
Within Groups
3
35
1.06097
> 0.05
Note. Significant differences are highlighted in bold and marked with an asterisk (*).
Figure 3. Bar graph representing the number of vocals heard from each call under the three states of arousal (non-aggressive;
moderately aggressive; highly aggressive)
Age Class Comparison
The spectral images of the whines (Figure 4A, B) showed no distinct differences between age
classes and could appear static or harmonic for both adult and pups. The Mann‐Whitney results (Table 4)

Almonte 513
showed a significant difference in all the variables. The adults were significantly higher across variables
except in the low frequency and the harmonics. The pups had a higher low frequency but a lower high
frequency therefore pups had an overall lower range of frequency. The chirp’s spectral images (Figure 4
D) showed that the range of frequency were different in pups and adults. As well the pups chirp produced
linear harmonics, and the adults shape could either be linear or upside down “v” shaped. The
Mann‐Whitney showed (Table 4) significant differences across all variables. The pups had a higher low
frequency and the adults had higher powers. Pups chatters are spectrally similar to the chatter-chirp
spectrogram of adults because they appeared to be chirps in rapid succession (Figure 4E). The pups had a
higher low frequency and the adults had higher powers and longer durations. The Mann-Whitney (Table
4) showed that adult and pup chatters were significantly different in all variables except the max and
center frequency.
Table 4
Table of Hypotheses Proposed on the Variations of Adults and Pups Share Call Types
Call Type
Ha
Ho
Mann-Whitney p-value
Whine
HFAD≠ HFPP
HFAD=HFPP
<0.01
LFAD≠ LFPP
LFAD=LFPP
<0.01
RGFAD≠ RGFPP
RGFAD=RGFPP
<0.01
MNFAD≠ MNFPP
MNFAD=MNFPP
<0.01
MXFAD ≠ MXFPP
MXFAD =MXFPP
<0.01
CTFAD≠CTFPP
CTFAD=CTFPP
<0.01
MXPWRAD ≠ MXPWRPP
MXPWRAD = MXPWRPP
<0.01
AVGPWRAD≠ AVGPWRPP
AVGPWRAD=AVGPWRPP
<0.01
DRTAD ≠ DRTPP
DRTAD = DRTPP
<0.01
HRMAD≠ HRMPP
HRMAD= HRMPP
<0.01
Chirp
HFAD≠ HFPP
HFAD=HFPP
<0.01
LFAD≠ LFPP
LFAD=LFPP
<0.01
RGFAD≠ RGFPP
RGFAD=RGFPP
<0.01
MNFAD≠ MNFPP
MNFAD=MNFPP
<0.01
MXFAD ≠ MXFPP
MXFAD =MXFPP
<0.01
CTFAD≠CTFPP
CTFAD=CTFPP
<0.01
MXPWRAD ≠ MXPWRPP
MXPWRAD = MXPWRPP
<0.01
AVGPWRAD≠ AVGPWRPP
AVGPWRAD=AVGPWRPP
<0.01
DRTAD ≠ DRTPP
DRTAD = DRTPP
<0.01
HRMAD≠ HRMPP
HRMAD= HRMPP
<0.01
Chatter
HFAD≠ HFPP
HFAD=HFPP
<0.01
LFAD≠ LFPP
LFAD=LFPP
<0.01
RGFAD≠ RGFPP
RGFAD=RGFPP
<0.01
MNFAD≠ MNFPP
MNFAD=MNFPP
<0.01
MXFAD ≠ MXFPP
MXFAD =MXFPP
>0.05
CTFAD≠CTFPP
CTFAD=CTFPP
>0.05
MXPWRAD ≠ MXPWRPP
MXPWRAD = MXPWRPP
<0.01
AVGPWRAD≠ AVGPWRPP
AVGPWRAD=AVGPWRPP
<0.01
DRTAD ≠ DRTPP
DRTAD = DRTPP
<0.01
HRMAD≠ HRMPP
HRMAD= HRMPP
<0.01
Note: (HF = high frequency; LF = low frequency, MXF = maximum frequency; MXPWR = maximum power; DRT = duration;
HRM = harmonics). Hypotheses, the results of the Mann-Whitney and interpretation of results are shown. Significant p-values
are highlighted in bold and marked with an asterisk (*).

Almonte 514
Inter-Observer Reliability
The inter-observer agreed with the sound identification for 270 calls (89.7% of the data set) and
disagreed for 31 calls (10.3% of the data set). For individual identification the secondary observer agreed
for 276 calls (91.7% of the data set) and disagreed for 24 calls (8% of the data set) The secondary
observer agreed with the arousal state identification for 274 interactions (91% of the data set) and
disagreed for 27 interactions (8.9% of the data set) The inter-observer reliability tests results suggested
that the techniques used to identify the individual, sound and arousal state were accurate and repeatable.
A) B)
C) D)
E) F)
G)
Figure 4. Spectral comparison of adults and pups shared call types: A) adult harmonic whine, B) pup whine, C) adult chirp, D)
pup chirp, E) adult chatter-chirp, F) pup chatter, G) newborn pup’s unique whistle
Discussion
Vocal Repertoire and Call Usage of North American River Otters
This is the first study to delineate and quantify the vocal communication system of North
American river otters and the results showed that river otters have a set of complex vocals aurally similar
to giant otters (Pteronura brasiliensis), sea otters (Enhydra lutris), and Eurasian otters (Lutra lutra)
(Bezzera et al., 2011; Gnoli & Prigioni, 1995; McShane et al., 1995). The vocal repertoire iss composed
of several calls that have been described in other otter species (whine; chirp/squeak; blow; and scream).
Like giant otters that show significant differences in their acoustic structures (Bezzera, Souto, & Schiel,
2011), river otter call structures varied and the variations were a distinguishing factor when classifying

Almonte 515
calls in the repertoire. Captive river otters had four distinct call categories (whine, chirp, grunt, blow) and
seven sub-type call categories (chatter, squeak, creek, scream, swish, hiss, and hiccup). Sub-type calls
sounded different from the distinct calls, but were structurally similar to whines or chirps. Pup vocal
repertoires consisted of whines, chirps, and chatters.
The whine was universally present in the adults and pups repertoires, and the discriminant
function analysis showed that six of the sub-call type categories were structurally categorized with
whines. This suggests that the whine is essential to the vocal repertoire and is the vocal that sub-call types
are modified from. Furthermore, newborn pups exhibited the whine within 45 s of birth, implying the
whine is innately present. Adults showed variations in the vocal parameters of the whine; therefore the
whine might be present at birth and over time becomes individually distinct. Although, individual
variations were present the whine was expressed arbitrarily and occurred when the otter was in various
arousal states (Figure 3), suggesting it likely does not function to relay specific information about the
state of arousal of the individual.
Newborn pups expressed the chirp at birth, and adult chirps showed individual variations.
Therefore, chirps may be innately present and become individually distinct like the whine call type.
Chirps were the dominant call type expressed when otters were in a non-aggressive state of arousal
(Figure 3). This suggests that chirps are likely used to convey friendly messages. In addition, otters
chirped during self-grooming or when massaging their body against a substrate. Thus, chirps might also
be used to express feelings of contentment.
An adult female and male housed in separate facilities exhibited grunts and blows. Individual
variations existed in grunts but they were spectrally and aurally similar. Morphological differences
between the sexes may explain the variations, but further research is needed to confirm this. Despite the
individual variations, grunts were particularly expressed when the otter was in a moderately agitated
arousal state (Figure 3). Both individuals grunted when stereotypical behaviors (Table 1) were exhibited,
suggesting that grunts may serve to express frustration or anxiety. The female grunted when she paced
before her morning feed and the male grunted overnight while he chewed on his tail.
The blow showed individual variation, with females having higher frequencies. Like the grunt,
the variations in the blow likely existed because of morphological differences between sexes. Blows were
mainly exhibited when the otters were moderately agitated by a human approaching the den. Hence, the
blow may be used when an otter feels endangered and serves to ward off potential threats. It has been
suggested that captive Eurasian otters use their blow to ward of potential threats (Gnoli & Prigioni, 1995).
Spectrograms revealed the chatter sound was produced when the otter whined or chirped in rapid
succession rather than in isolation or intermittently. All adults exhibited chatters in their repertoires, but
only the newborn pups produced chatters. Chatters were structurally like a whine or chirp and also
showed individual variations. However, all adults chattered while exhibiting aggressive behaviors. During
physical altercations when otters were highly aggressive chatters were the dominant call type exhibited.
This suggests that chatters serves to relay antagonistic or hostile messages.
Squeaks were graded calls and were produced when the otter used the whine and chirp in
conjunction. The spectral image revealed the squeak was a static whine with a harmonic chirp situated in
the center of the call. Although, squeaks visually appeared to be whine and chirp call types used in
conjunction, squeaks were aurally distinct from either call type. Only a solitary male otter produced
squeaks particularly when a human approached the back entrance of the exhibit, and squeaks continued to
be expressed for the duration of the human presence. It is possible this particular individual developed a
unique call to interact with humans, and he used a squeak to fend off potential threats.
The creek, hiss, and swish vocal types were exhibited by several adult otters and were associated
with moderately aggressive states of arousal (Figure 3). The eldest otter (a female) in the study
predominately used these calls. The female was blind and often retreated to a corner after an altercation.
In the corner she would express one or more of the calls types (creek, hiss, swish) while opening her
mouth. Swish, hiss, and creek call types likely relay aggression or defensiveness.
Two separately housed pregnant females exhibited the scream. The scream did not show
individual variations and both females used screams in the same manner. Approximately one month

Almonte 516
before the pups were born the females began to exhibit an extremely loud scream when the male was in
her vicinity. Females continued to exhibit screams until pups left the natal den. Screams were produced
when the male or a human approached the natal den, or when females were separated from the pups. The
temporary use of screams and lack of individual variation implies that in females the scream might serve
specifically to protect newborn pups. Males made several attempts to enter the den but retreated when the
female screamed, suggesting this vocal was useful to deter males from approaching newborn pups. River
otter newborn pups are vulnerable to infanticide (Blundell, Ben-David, & Bowyer, 2001) and predation
(Gorman et al., 2005) therefore in the wild the loud scream may function to ward off potential threats to
pups from long distances.
The whistle call type was unique to newborn pups, and was expressed when the pups appeared to
be locating their mother, or exploring the natal den. The natal den was probably difficult for newborn
pups to maneuver through because the pups were blind at birth, and the den was dark. Therefore, it is
possible that newborn pups echolocate using the whistle call type to help them navigate the natal den.
Furthermore, eight weeks old pups observed after leaving the natal den did not exhibit the whistle. This
implies that the whistle might be apparent in the early stages of life, but as dependency on the mother
decreases the whistle is reduced and is eventually eliminated from the repertoire. However, further
research is needed to confirm this.
Wild otters are adept in avoiding confrontation using scent communication (Kruuk, 2006, p. 78),
and river otter spraints are used for species identification, to maintain territories, and relay social status to
group members (Rostain, Ben-David, Groves, & Randall 2004). But, the results suggest that vocal
communication might also serve several functions. This study suggests that river otter vocalizations are
possibly used for individual identification, to relay friendliness, aggression, frustration, or defensiveness,
and to protect pups from predation. Chirps were highly associated with non-aggression and chatters were
highly associated with aggression, implying that chirps and chatters vocal types are likely the most
effective at expressing the temperament of an otter.
Adult-Pup Vocal Comparisons
The greatest difference between adult and pups for shared call types (whine, chirp, and chatter)
lied in the range of frequencies and powers. Adult calls had greater range of frequencies because their
calls contained more harmonic bands than pup vocals. As well, adult calls had a greater average power in
comparison to pups. The structural differences between adult and pups suggests that whines, chirps and
chatters are inherently present, underdeveloped versions of the respective adult call, and are modified as
pups enter adulthood. These modifications are potentially due to changes in morphology during
development, such as changes in vocal cord morphology or in body size. However, further research is
needed to verify this theory.
Future Research
This study provides insight to the vocal behaviors of captive river otters, and the potential
functions of call types. However, because captive behaviors are not a direct mirror of wild behaviors, it is
important to investigate the vocal systems of wild otters to get a complete understanding of North
American river otter vocal systems. With the groundwork established here questions can now be
expanded to wild studies.
Acknowledgements
I would like to thank all the people involved with allowing me open access to the facilities for data
collection: Atlantis Aquarium: Ann Yauillo (Head Curator) & Candyce Paparo (Head Trainer), Beardsley
Zoo: Don Goff (Curator), Rob Thomas (Associate Curator), Bethany Baldwin (Keeper), Chris Barker
(Keeper) & Gregg Dancho (Director), Palisades Park Conservancy/Bear Mountain Park: Edwin

Almonte 517
McGowan (Director) & Melissa Gilmer (Head Trainer), Stamford Museum and Nature Center/Heckscher
Farm: Lauren Satterfield (Former Director), Victoria Marr (Director) & Mark Mogehsen (Head Keeper)
and Turtle Back Zoo: Dr. Jeremy Goodman D.V.M (Director) Annemarie Ferrie (Trainer/Keeper), Gina
Galanoplos & Tamara Myhal (Trainer/Keeper). Without the support and extensive cooperation from
everyone this project would not have been possible. Thank you to Lainga Tong for taking on the
responsibility of secondary observer and working very hard to complete the work in a short period of
time. A special thank you to my doctoral advisor/mentor Dr. Richard Veit and mentor Dr. Jennifer Basil
for their valuable advice, feedback, and support. Thank you to the National Science Foundation for
funding (NSF/AGEP Award HRD 0450360).
References
Beckel, A. L. (1991). Wrestling play in adult river otters (Lutra canadensis). Journal of Mammalogy, 72, 386-390.
Ben-David, M., Blundell, G. M., Kern, J. W., Maier, J. A. K., Brown, E. D., & Jewitt, S. C. (2005). Communication
in river otters: Creation of variable resource sheds for terrestrial communities. Ecology, 86, 1331-1345.
Bezzera, B. M., Souto, A. S., & Schiel, N. (2011. Notes on vocalizations of giant otters in the flooded Igapo forests
of Jau National Park, Amazonas, Brazil. Journal of Ethology, 29,169-175.
Blundell, G. M., Ben-David, M., & Bowyer, R. T. (2001). Sociality in river otters: Cooperative foraging or
reproductive strategies? Behavioral Ecology, 13, 234-141.
Gnoli, C., & Prigioni, C. (1995). Preliminary study on the acoustic communication of captive otters (Lutra lutra).
Hystrix Italian Journal of Mammalogy, 7, 289-296.
Gorman, T. A., Erb, J. D., McMillan, B. R., Martin, D. J., & Homyack, J. A. (2005). Site characteristics of river
otter (Lontra canadensis) natal dens in Minnesota. The American Midland Naturalist, 156, 109-117.
Kruuk, H. (2006). Otters: Ecology, behaviour, and conservation. Oxford University Press Inc., New York.
Melquist, W. E., & Hornocker, M. G. (1983). Ecology of river otters in west central Idaho. Wildlife Monographs,
83, 1-60.
McShane, L. J., Estes, J. A., Riedman, M. L., & Staedler, M. M. (1995). Repertoire, structure, and individual
variation of vocalization in the sea otter. Journal of Mammalogy, 76, 414-427.
Rostain, R. R., Ben-David, M., Groves, P., & Randall, J. A. (2004). Why do river otters scent-mark? An
experimental test of several hypotheses. Animal Behaviour, 68, 703-711.