WI

\

L

 
     
   
   
     

“WNW“

MW“

5

\lM‘lWlml‘Mll

141

'—1_.
100
‘moo

 

“I“ LIBRARY
wg Michigan State
IV University ~

This is to certify that the
thesis entitled

SIGNATURE SIGNALS IN THE TERRITORIAL VOCALIZATIONS
OF RED SQUIRRELS (TAMIASCIURUS HUDSONICUS) AND
THEIR USE IN KIN RECOGNITION.

presented by
Adam Reed Goble

has been accepted towards fulfillment
of the requirements for the

MS. degree in Zoology

 

 

  
 

é/ /./ W

’C/ Major-ER ssonflignature

/’

MSU is an affinnative-action, equal-opportunity employer

A- ...—.—--a-IJ9'-°-‘-

.5...~ ._---c-I-n-0-.-a-I-u--.-.!._g--o-n-n-o-o-o-o-o-o-¢-I-o-o-I-n-n-I‘-- A -4

nqnu-o-c-.-._- —--.-.—-—-—-—.-._.-..-

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K lProVAcoEIPres/CIRC/DateDue mdd

SIGNATURE SIGNALS IN THE TERRITORIAL VOCALIZATIONS OF RED
SQUIRRELS (TAMIASCIUR US HUDSONICUS) AND THEIR USE IN KIN
RECOGNITION.

By

Adam Reed Goble

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree
MASTER OF SCIENCE
ZOOLOGY

2008

ABSTRACT
SIGNATURE SIGNALS IN THE TERRITORIALVOCALIZATIONS OF RED
SQUIRRELS (TAMIASCI UR US HUDSONIC US) AND THEIR USE IN KIN
RECOGNITION.
By
Adam Reed Goble
Current work on kin recognition in rodents has shown that the dominant modality
for recognition is through olfactory mechanisms. In a territorial rodent, such as the red
squirrel (Tamiasciurus hudsonicus), olfactory mechanisms are may not be the prime
modality Of recognition compared to vocal mechanisms. The focus of this thesis was to
examine the territorial vocalizations of red squirrels for the presence of signature signals
and to determine if they are used in kin recognition. Time and frequency elements of the
fundamental frequency were extracted from 319 rattles of 149 unique individuals. These
data were analyzed using first principle component analysis and then general linear
models of the first four principle components, revealing that individual identity and age
significantly influenced the time and frequency components of rattles. Juvenile rattles had
lower fundamental frequencies and faster syllable rates than adults. It was also found that
context did not influence the structure of rattles. Recognition through rattle vocalization
was investigated using playback experiments. When rattles were played on the territories
of individuals, the territory owners were more likely to rattle in response when they were
unrelated to the caller; focal individuals were also more likely to bark in response to
neighbors showing neighbor recognition. This study is the first to demonstrate the
presence of signature signals in red squirrel rattles and Shows acoustical kin and neighbor

recognition in a territorial rodent.

ACKNOWLEDGEMENTS

I’d like to think my primary advisor for helping me to develop myself as a
scientist and complete this Master’s Thesis. Dr. Andrew McAdam offered me a chance to
learn what science was about and what was needed to succeed in the field, and I’m glad I
took the opportunity. His patience and understanding got me on the right track and in the
right frame of mind to do research and his knowledge has helped me expand my
expertise to many fields of science I had little experience in before I came to MSU. When
I started my degree, my goal was to learn what it meant to be a scientist and Dr. McAdam
has certainly helped me achieve that.

The other members of my committee have been very valuable in my development
as a scientist as well. Dr. Kay Holekamp pushed me to learn more about animal behavior
early in my graduate career and that has driven me since that day. I also like to thank Dr.
Holekamp for standing in as my primary advisor toward the end of my graduate degree,
giving me advice and suggestions on my thesis and most importantly being there when I
needed someone to talk to, even if it was not about research or animal behavior. I also
like to thank Dr. Brian Maurer, his class on multivariate statistics my first year of
graduate school provided the basis for my analyses in Chapter Two and his advice and
comments on the analyses helped me finish them quickly and efficiently.

My lab mates, Elizabeth Ball, Andrea Yeager, Ben Dantzer, and Ryan Taylor
have been very supportive and I appreciate their willingness to proof read and comment
on my thesis as a moments notice and for field assistance from Ben and Ryan. Their
advice and comments were invaluable when I was editing and revising my manuscript

this past few months.

iii

I am grateful to the First Nations and Yukon Territorial Government for allowing
me to conduct my research on their lands and I also like to thank Dr. Stan Boutin for
allowing me to work and stay at his field site in the Yukon. Dr. Boutin also provided me
with advice and instilled a great amount of confidence into me when I first started this
project. Without his support early on, I may not have stuck to this project. While I was at
the field station I was also assisted by several of the 2005 and 2006 squirrel camp field
technicians. They were vital in helping me locate my study animals and keeping track of
them.

Finally, I’d like to think my parents, Micheal and Robin Goble, and my
grandparents for their care and support throughout the past three years. They were always
there for me, were interested in my research even if they did not know what I was talking
about, and supported me research anyway they could. My grandparents provided with the
money I needed for my field equipment when funding was tight, without them I would
have never been able to conduct my playback trials for my third chapter. I am grateful I
had the chance to work with and learn from everyone at Michigan State University and I
feel that everyone I have interacted with has contributed to my development as a

scientist.

iv

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................................. vi
LIST OF FIGURES ........................................................................................................... vii
CHAPTER ONE: GENERAL INTRODUCTION .............................................................. 1
1.1 Recognition Systems ......................................................................................... 5
1.2 Vocalizations Of Red Squirrels .......................................................................... 8
1.3 “Rattle” Vocalizations ....................................................................................... 9
1.4 Hypotheses and Predictions ............................................................................. 12
1.5 References ........................................................................................................ 14
CHAPTER TWO: SIGNATURE SIGNALS IN THE VOCALIZATIONS OF RED
SQUIRRELS (TAMIASCI UR US H UDSONICUS) ............................................................ 18
2.1 Introduction ...................................................................................................... 18
2.2 Methods ............................................................................................................ 23
2.3 Results .............................................................................................................. 27
2.4 Discussion ........................................................................................................ 29
2.5 References ........................................................................................................ 35
CHAPTER THREE: KIN RECOGNITION IN RED SQUIRRELS (TAMIASCIUR US
HUDSONIC US) ................................................................................................................. 46
3.1 Introduction ...................................................................................................... 46
3.2 Methods ............................................................................................................ 50
3.3 Results .............................................................................................................. 55
3.4 Discussion ........................................................................................................ 56
3.5 References ........................................................................................................ 61
CHAPTER FOUR: GENERAL CONCLUSION .............................................................. 69
4.1 References ........................................................................................................ 72

LIST OF TABLES

2.1 Summary statistics for the fundamental frequency (F0) measurements of the 319
recorded rattles of red squirrels (Tamiasciurus hudsom'cus). The data were extracted
using Raven 1.2.1 ............................................................................................................... 38

2.2 Principle component eigenvectors and the proportion of the variance explained by
each component for red squirrel rattles. PCl represents bandwidth, PC2 represents
frequency elements of the fundamental frequency, PC3 represents the time elements of
the fundamental frequency and PC4 represents the overall length of the rattles ............... 38

2.3 Principal component eigenvectors and the proportion of the variance explained by
each component for red squirrel rattles using adult and juvenile datasets. PCl represents
bandwidth, PC2 represents frequency elements of the fundamental frequency, PC3
represents the time elements of the fundamental frequency and PC4 represents the overall
length of the rattles ............................................................................................................. 39

vi

LIST OF FIGURES

Figure 2.1: Sonogram of the rattle vocalization of a North American red squirrel
(Tamiasciurus hudsonicus). The dark vertical lines represent individual syllables. The
total length of the rattle portion of this call is four seconds ............................................... 40

2.2 Polygons indicate hypothetical territories of red squirrels, the shaded circle indicates
the focal squirrel and the large circle indicates the 130m radius within which a rattle can
be heard (Smith, 1978) ....................................................................................................... 41

2.3 Spectrograph of measurements taken on the fundamental frequency of a rattle
syllable. The fundamental frequency of the syllable is indicated by the shaded dark bar in
the center of the spectrograph. Seconds are represented on the x-axis while frequency
(Hz) is represented on the y-axis. The relative power of the signal (db) is indicated by
darker shading. This particular syllable has a maximum fundamental frequency of 1500
Hz, a minimum fundamental frequency of 600 Hz and is 0.015 seconds long ................. 42

2.4 The effects of age of an individual on principle component two of red squirrel rattles.
Larger PC scores are indicative of higher fundamental frequencies, low bandwidth of the
fundamental frequency and low amounts of frequency modulation in the rattle. Data are
represented by mean +/- SE. Rattles recorded from adults are represented by “A” and
rattle recorded from juveniles (< 1 year Old) are represented by “J.” ................................ 43

2.5 The effects of the age of an individual on principle component three of red squirrel
rattles. Larger PC scores are indicative of increased rate in the syllables per second and
shorter syllable length in the rattle. Dare are represented by mean +/- SE. Rattles recorded
from adults are represented by “A” and rattle recorded from juveniles (< 1 year old) are
represented by “J.” ............................................................................................................. 44

2.6 The effects of behavioral context on principle component four for red squirrel rattles.
Larger PC scores are indicative of shorter overall call durations. Data are represented by
mean +/- SE. Behavioral contexts were classified as unelicited (1), responsive (2) and
provoked (3) ....................................................................................................................... 45

3.1 Spatial layout of the playback study. The speaker was placed 10m from the center of
the focal individual’s midden directed toward the individual and the observer maintained
between a 60° and 90° angle between the individual and the speaker ............................... 66

3.2 Probability of the focal individual rattling during the pre and post playback
observational periods of a playback trial. Responses are grouped based on the
relationship between the focal individual and the individual used in the playback
recording being neighbors (N) or non-neighbors (NN) in the pre-observational phase, and
grouped based on being neighbors or non-neighbors and kin (K) or non-kin (NK) in the
post-observational phase. Data are presented as mean ...................................................... 67

vii

3.3 Probability of the focal individual barking during the pre and post playback
observational periods of a playback trial. Responses are grouped based on the
relationship between the focal individual and the individual used in the playback
recording being neighbors (N) or non-neighbors (NN). The data are presented as mean
............................................................................................................................................ (SS?

viii

CHAPTER ONE: GENERAL INTRODUCTION

Communication occurs when a signal from one individual influences the behavior
of another individual that received the signal (Wiley, 1981; Endler, 1993; Johnstone,
1996). These signals are conveyed in a variety of modalities including visual displays,
scents, and vocalizations (Marler, 1967). In order for these signals to be effective they
must be detectable by a recipient (Johnstone, 1996), and each of these three modalities
allows for effective communication in different situations and habitats.

Visual displays function most effectively when two individuals are in close
proximity of one another, occupying the same space and time. The physical environment
impacts the effectiveness of visual displays by limiting their range through light
scattering. An example of physical influences on visual signals occurs in the comparison
of air and water environments; light cannot travel as far in water as it does in air, so the
range of the signal is much shorter (Lythgoe, 1988; Wehner, 1997). Visual displays are
most commonly found in diurnal social animals, such as many Species of birds that live in
open environments containing little vegetation or other obstructions (Marler, 1967).

Olfactory communication requires some degree of spatial overlap, but allows for
temporal differences in the interactions. Organisms that use olfactory signals, such as
beavers (Castor Canadensis; Hodgdon and Lancia, 1983) can use scent marks on
territories to orient themselves (Muller-Schwarze and Heckman, 1980), assist in finding
mates or for identifying kin (Butler and Butler, 1979). Beavers utilize these olfactory
cues during times where there is low light, or when communicating with other individuals
that have a temporal difference in Space use. In addition to being used when individuals

visit the same location at different times, scent also plays a role in communication during

episodes of physical contact (Marler, 1967). However, the key feature of scent that sets it
apart from visual and vocal modalities is its ability to persist in the environment.
Olfactory signals are well suited for communication among organisms that occupy stable
spatial locations over time, such as carnivores, rodents and primates because scent marks
can persist for extended periods of time (Marler, 1967).

Vocal communication, like visual communication, relies on spatial and temporal
overlap between the signaler and the receiver to be effective. Vocal signals exist for short
periods of time compared to olfactory signals and dissipate into the environment quickly
after they have been emitted. This attribute of vocal recognition is both a boon and a
hindrance: if an individual hears a vocal signal, it can be aware of the caller’s presence in
the area immediately, and respond accordingly. Because vocal signals exist for such a
short time, temporal differences between presence of the caller and recipient would mean
the Signal goes unheard. The effectiveness of a vocal signal depends on a variety of
factors including habitat (Morton, 1975; Wehner, 1997), background noise (Schwartz and
Wells, 1983) and the recipient’s ability to interpret the signal (Marler, 1967). Vegetation
can scatter vocal signals making it harder for a recipient to extract information from
them, or make the signaler hard to localize because of reflection and echo (Wehner,
1997). The broadcast potential Of a signal based on habitat has been documented by
Morton (1975) who found that different habitats allowed certain frequencies to propagate
stronger and farther. For example, frequencies for vocalizations used in grasslands were
in the 800 — 1585 Hz range and the frequencies best suited to forest habitats were in the
1585 — 2500 Hz range. Morton (1975) compared the results of his study to a selection of

avian vocalizations, and concluded that possessing calls in these frequency ranges boosts

the effectiveness and audible distance of a call. Other environmental components such as
air pressure, temperature, background noise or even altitude influence can also affect call
transmission. While the effectiveness of vocal communication is dependant upon many
physical, temporal, and spatial variables, it allows a type of interaction that is not
permitted by either visual or scent communication — immediate, direct long-range
communication. Because of their quick production and short persistence in the
environment, vocal signals are most suited to species that are mobile and depend on
quick, accurate exchanges of information during interactions (Marler, 1967).

A typical vocalization includes a fundamental frequency and its corresponding
harmonics broken into syllables. The fundamental frequency is the lowest frequency in a
set of harmonics. The harmonics are regularly Spaced multiples of the fundamental
frequency (Halliday et al. 1997). Individual syllables in a vocalization can be
distinguished by discontinuities, or breaks, in the fundamental frequency. Common
measurements of the fundamental frequency of a vocalization can be described by the
minimal and maximum frequencies of each syllable, the modulation of a syllable, the
length of the syllables, and the spacing between syllables (e.g., Cheney and Seyfarth,
1982; Rukstalis et al., 2003).

Variation in the structure of fundamental frequencies and syllables of
vocalizations is extensively documented and has been shown to contain a variety of
information about the caller, including the caller’s current behavioral context in many
birds and mammals (Norcross and Newman 1993; Rendall et al. 1999; Crockford and
Boesch 2003; Wich et al. 2003). Morton (1977) proposed the motivation-structural rule

that aggressive vocalizations would have lower fundamental frequencies (low pitch),

decreased time between syllables, and wider frequency (noisy) bandwidths in their
vocalizations. In contrast to aggressive vocalizations, submissive vocalizations would
have higher fundamental frequencies (higher pitch), increased time between syllables,
and narrower bandwidths (tonal). Age effects on vocalizations tend to follow the same
trend, such that calls of juveniles are of higher pitch, with a tonal sound to the calls,
whereas adults tend to have calls of lower fundamental frequency and higher modulation
(Morton, 1977). Researchers dating back to Darwin (1872) have studied and discussed
the similarity of the vocalizations of submissive adults and juveniles.

Morton (1977) performed an experimental test showing that the calls Of several
avian species followed the MS-Rule. The MS-Rule has also been Shown to apply to the
vocalizations of two different mammalian species: white-nosed coatis (Nasua narica;
Compton et al., 2001) and elk (Cervus Canadensis; Feighny et al. 2006). Morton (1977)
initially proposed and provided evidence for the MS-Rule in close range vocalizations —
those that are typically used in hostile/submissive interactions, but mentioned that the
MS-Rule could possibly hold true for long range communications as well. In a study
involving bull elk bugles, Feighny et al. (2006) showed that bugles emitted under hostile
circumstances showed all the characteristics of an aggressive call as predicted by the
motivation-structural rule. This indicated that the MS-Rule can be applied to long-range
communication, but to date, there have been few tests of the motivation-structural rule
and how it pertains to long distance vocalizations.

In addition to contextual information about the caller, vocalizations can also
contain non-contextual information, such as size (e.g. Davis and Halliday 1978; Clutton-

Brock and Albon, 1979), sex (Tomaszycki et al., 2001), age (e.g., Cheney and Seyfarth

1990; Nestrova 1990; Blumstein and Daniels 2004; Blumstein and Munos 2005), or
individual identity (e. g. Tooze et al, 1990; Rukstalis et al 2003; Puglisi and Adamo 2004;
Palacios et al 2007). The identity of a caller can be established when vocalizations
contain individual-specific information that is not influenced by the size, age, sex, and
behavioral context of the caller. These individual-specific signal components can exist in
the frequency, temporal, or rate characteristics of a vocalization, and are collectively
called signature signals. Signature signals have been Shown to exist in the vocalizations
of several mammalian and avian species (e.g. Tooze et al. 1990, Rukstalis et a1 2003;
Puglisi and Adamo 2004; Palacios et al 2007). These signature signals exist in calls that
maintain social order and spacing, such as contact-calls (howls) Of wolves (Tooze et al.
1990) and “phee” calls of marmosets (Rukstalis, et al., 2003) and also in anti-predator
calls, such as those produced by vervet monkeys (Cheney and Seyfarth, 1988) or
Belding’s ground squirrels (Leger et al. 1984). The presence of signature signals is a
necessary requirement for individual recognition; without individual variation in a signal,
there would be no possibility for distinguishing specific individuals, and thus impossible
to distinguish specific kin, neighbors, mates, or rivals.
1.1 Recoggition Systems

Recognition is the ability of an individual to differentiate between kin, mates,
rivals, allies, predators, or prey (Sherman et al., 1997). Individual recognition occurs
when an individual-specific trait or signature signal is used to identify an individual
during an interaction (Tibbits and Dale, 2007). Individual recognition has been shown to
exist in many species of birds (e. g., Leonard et a1, 1997; Aubin et al, 2000) and mammals

(Johnston, 2003; Mateo, 2003; Searby and Jouventin 2003). Three contexts in which

individual recognition has been extensively studied include territoriality, competition,
and parent/offspring recognition (reviewed in Tibbits and Dale, 2007). Recognition of
neighbors in a territorial system can allow owners to direct their aggressiveness toward
potentially threatening strangers instead of their established neighbors. Being able to
identify ones established neighbors allows a territory owner to focus energy on showing
new individuals where territory boundaries exist instead of reinforcing established
boundaries with long-time neighbors (Telemes, 1994). The use and function of individual
recognition in territoriality has been documented in both avian (e.g., Falls, 1982;
Whitfield, 1986) and mammalian Species (e.g., Johnston, 2003; Mateo 2003).

When individual recognition allows for the identification of genetically related
individuals it is called kin recognition (Hepper, 1991). Kin recognition can lead to kin
discrimination - differential behavioral responses to kin, based on an acoustical,
olfactory, or visual cue (Beecher, 1982; Hepper 1991). Three different mechanisms of kin
recognition have been proposed: site-specific spatial reference, prior association, and
phenotype matching. Site-specific kin recognition occurs when individuals provide
preferential treatment to individuals with whom they share a common space, such as a
nest or den. (e.g., Holmes and Sherman, 1982; Waldman, 1987; Waldman, 1988; Mateo,
2003). Kin recognition via prior association occurs when individuals interact with one
another in situations exclusive to kin, and in doing so learn the unique traits of each other.
The most common modality of kin recognition through prior association in mammals is
scent, but visual and vocal cues can play an important role as well (Waldman, 1987;
Waldman, 1988; Mateo, 2003). Kin recognition through phenotype matching occurs

when an individual can determine whether an individual is related to them based on the

similarity of the interacting individual’s signal to their own or that of their established kin
(Waldman, 1987; Waldman, 1988). This uniquely allows for the identification of
previously unknown kin (Waldman, 1988). Common modalities of phenotype matching
recognition are through olfactory and vocal mechanisms (Waldman, 1987; Waldman,
1988; Mateo, 2003). Kin recognition can influence many behaviors of animals including
responses to anti-predator calls (e.g., Hepper 1991, Hare, 1994), cooperation between
groups of individuals (e.g., Cheney and Seyfarth, 1982; Hepper 1991), and the location
and identification of juveniles by adults (e.g., Cheney and Seyfarth, 1980; Waldman,
1988; Balcombe 1990; Hepper, 1991; Holekamp et al. 1999; Insley 2000; Searby and
Jouventin, 2003). In all these examples, the responses Of recipients were strongly dictated
by the relation of the recipient to the caller. For example, the probability of an individual
Columbian ground squirrel (Spermophilus columbianus) fleeing in response to the anti-
predator call of a related individual was significantly greater than a call from an unrelated
individual (Hare, 1994). In vervet monkeys (Chlorocebus pygerythrus), the interactions
between troops of individuals are influenced by the patterns of dispersal and recognition
of vocalizations from dispersed individuals (Cheney and Seyfarth 1982). Troops that
shared a genetic link via dispersed individuals were more likely to engage in cooperative
behaviors with one another than troops that did not.

Kin recognition is a commonly studied behavior in rodents. In a review, Mateo
(2003) found that many rodents for which kin recognition has been extensively studied
exhibited kin recognition through olfactory mechanisms. Mateo identified 33 species in
which olfactory kin recognition occurred and 12 of these species were capable of

phenotype matching. Even though the majority of rodents studied use olfactory

mechanisms for kin recognition, olfactory means may not be the most efficient modality
for every rodent species: highly territorial diurnal rodents that minimize direct contact
could utilize vocal signals for recognition more easily than olfactory modalities.

1.2 Vocalizations of Red Sguirrels

One such species in which Olfactory kin recognition would be inefficient
compared to that of acoustical kin recognition is the North American red squirrel
(Tamiasciurus hudsonicus). Red squirrels are territorial tree squirrels whose social
structure consists of solitary males and females defending exclusive territories year-round
(Smith, 1968). These exclusive territories are based around middens, large piles of cone
bracts in which new cones are cached each fall (Smith, 1968; Gurnell, 1984).

Red squirrels use a variety of vocalizations during defense of their territories and
interactions with conspecifics. The screech, also referred to as the squeak (Embry, 1970)
is a loud, high pitched squeal that is used as an offensive threat call against territory
intruders (Nodler, 1973) and sounds like a “tsew” (Lair, 1990). The growl vocalization
used by red squirrels is a low-pitched, defensively oriented call that has a Short audible
range (Smith, 1978). The screech and growl vocalizations are used in direct interactions
between conspecifics and typically indicate forthcoming aggression from a territory
owner to an intruder (Lair, 1990).

Another vocalization used by red squirrels is the buzz, or appeasing call, that are
typically emitted as a submissive attempt to prevent aggression from the territory owner
(Smith, 1978; Lair, 1990). Buzz vocalizations are also emitted during interactions

between young individuals and their mothers (Lair, 1990) and during mating chases

during which males often buzz when attempting to mate with females (Smith, 1968; Lair,
1990)

The bark, or alarm call, was once thought to be a singular generic vocalization
that signaled forthcoming aggressive behavior, such as charging from a territory owner to
an intruder, or the presence of a predator (Smith, 1968; Searing, 1977, Smith, 1978, Lair,
1990). However more recent work by Green and Meagher (1998) showed that the
classical “bark” vocalization was potentially the lumping of three distinct anti-predator
calls. These calls were the bark, the sect, and sect-bark. Bark calls are audibly different
than a screech and seets are sometimes considered squeaks. While bark vocalizations
sound similar, they produce distinctly different sonograms when digitally analyzed, and
they are used in response to different predators. Barks were found to be used in the
presence Of terrestrial predators while the seets and sect-barks were used in the presence
of aerial predators (Greene and Meagher, 1998).

1.3 Rattle Vocalizations

The rattle is the territorial call of red squirrels. It is used to claim ownership of a
territory and declare the owner’s presence on the territory (Smith 1968; Smith 1978).
Rattles are composed of a series of individual syllables emitted in rapid succession in a
bout that can last from a fraction of a second to more than five seconds. The fundamental
frequency of a rattle is between 0.5 and 2.0 kHz (Smith, 1978) and is associated with a
variable number of harmonics (Figure 2.1, also in Smith, 1978). The numerOus harmonics
typically found in rattles result in a wide total bandwidth of the vocalizations that can
exceed 20 kHz (Figure 2.1, also see Smith, 1978). Smith (1978) documented that the

rattle was an easily localizable call that could be heard from up to 130m away. Rattles are

easily localizable due to the repetitive use of similar syllables and the wide range of
frequencies the call contains (Marler, 1955; Marler 1967) The rattle vocalization is
commonly used in response to a neighboring individual’s calls (Smith, 1978) or a
territory intrusion (Smith, 1978). The rattle has also been documented to be broadcasted
in what appears to simply be an unelicited broadcast (Smith, 1978; Lair, 1990). Finally,
rattles are also emitted during mating chases (Smith, 1968; Smith, 1978; Lair, 1990), an
event that occurs once per breeding season per female, where males pursue an estrous
female and attempt to mate with her (Smith, 1968). Both the intruding males and estrous
females rattle frequently during this temporary breakdown of the territorial boundaries
during the event (Smith, 1968). Rattle vocalizations might allow individuals to know who
is present during mating chases and where the female is located during the chase, even if
she moves off of her territory and vocalizes, but this has not yet been investigated.

Even though barks, squeaks, growls, and buzzes are used frequently in different
interactions, they all contain a common trait: they are short range vocalizations that are
associated with direct interactions with other individuals, intruders, or predators. The
rattle, while used in some close-range interactions, such as mating chases, is most often
used in long-range communication where additional visual or scent cues would be
minimized. Based on the variety of situations and contexts in which rattles are emitted,
their use in defense of territories, and the ability to be heard over long distances, the rattle
is the most likely vocalization to exhibit signature signals and thus allowing for vocal kin
recognition in red squirrels.

Kin recognition in red squirrels could mediate several common interactions,

including their territoriality. If the behaviors of red squirrels are considered within the

10

scope of kin-selection theory (Hamiltion, 1964) a territory owner may consider a related
neighbor no threat to their ownership of their midden, or could incline an individual to
attempt to usurp an unrelated neighbor from their territory. Hamilton (1964) stated that an
altruistic act would occur when rB > C where r is the relatedness of two individuals, B is
the benefit of kin altruism (altruistic acts to benefit a genetically similar individual) and C
is the cost of the altruistic act. For red squirrels, benefits of having larger territories could
potentially increase survivability or the survivability of offspring, while costs could
include finding a new territory, loss of offspring, or death. In the scope of territorially, an
altruistic act would be considered not attempting to take a territory from a related
individual, and cost would be the penalty one could incur from not acquiring the
resources on a neighbors territory. Based on Hamilton’s Rule, an individual could lower
its inclusive fitness by taking all or part of a territory away from a relative; however, it is
also possible that usurping the territory would benefit an individual’s inclusive fitness
because of the increase in direct fitness (how many offspring they produce). If an
V individual’s inclusive fitness would be lowered more by usurping a related neighbor’s
territory than the direct fitness gains that might occur from occupying the territory there
would be no net fitness gain associated with taking a related individual’s territory.
Additionally, it would benefit an individual’s inclusive fitness to defend the territory
boundaries shared with unrelated individuals and to direct any potential territory
expansion in the direction of these unrelated neighbors. An example of kin mediated
territorially is exhibited in Belding’s ground squirrels (Spermophilus beldingi), which are
more likely to chase unrelated individuals than related individuals away from potential

nesting sites (Sherman, 1980; Sherman, 1981).

ll

1.4 Hypotheses and Predictions

Through the analysis of digitally recorded vocalizations, I examined the calls of
individual red squirrels in the study population. I then used these recorded vocalizations
to perform a second study involving vocal playback experiments to determine if red
squirrels were capable of kin recognition through vocalizations. The following chapters
document my analysis of the recorded call components and playback experiment.

For my first study I hypothesized that the vocalizations of red squirrels contain
contextual and non-contextual information. If the rattles of red squirrels contain signature
signals, I predicted that the effects of identity on the acoustical structure would account
for a significant amount of the variation in call components that cannot be explained by
physical or environmental variables. If red squirrels follow the motivation-structure rule
proposed by Morton (1977), I predicted that rattles elicited in more aggressive contexts
would have lower fundamental frequencies and higher modulations. I also predicted
juvenile squirrels would have vocalizations with higher fundamental frequencies and
lower modulation compared to adult squirrels. The results of this study are documented in
chapter two of my thesis.

In my second study, I hypothesized that red squirrels are capable of recognizing
kin based on rattle vocalizations. If red squirrels are capable of recognizing kin, I
predicted that squirrels would respond to rattles from related individuals with less rattles
and barks, as well as exhibit a lower tendency to approach the speakers, than if the rattle
came from an unrelated individual. I also examined the potential for individual
recognition in red squirrels by looking at the differential response in probability of

rattling, barking, and approaching the speaker toward neighbors and non-neighbors. The

12

results of this playback study are documented in chapter three of my thesis. By
examining the differences in the probability of barking in response to playbacks, I sought
to determine the function of the bark call. If barks are a precursor to aggression (Lair
1990) then I would expect playbacks to elicit barks in focal animals. In contrast, if barks
serve only as alarm calls (Greene and Meagher, 1998) then on-territory rattle playbacks

Should not influence the probability of an individual barking.

13

1.5 References

Aubin, T. P. Jouventin, and C. Hildebrand. 2000. Penguins use the two-voice system to
recognize each other. Proceedings of the Royal Society of London B Biological
Science. 267, 1081-1087.

Balcombe, J. P. 1990. Vocal recognition of pups by mother Mexican-free tailed bats.
Animal Behaviour, 39, 960-966.

Blumstein, D. T., and O. Munos. 2005. Individual, age, and sex-specific information is
contained in yellow-bellied marmot alarm calls. Animal Behaviour. 69, 353-361.

Blumstein, D. T., L. Vemeyre and J. C. Daniel. 2004. Reliability and the adaptive utility
of discrimination among alarm callers. Proceedings of the Royal Society of
London, Series B. 271, 1851-1857.

Boutin, S. and S. Schweiger. 1988. Manipulation of intruder pressure in the red squirrel
(T amiasciurus hudsonicus): Effects on territory size and acquisition. Canadian
Journal of Zoology. 66, 252-262.

Butler, R. G. and L. A. Butler. 1979. Toward a functional interpretation of scent marking
in the beaver (Castor canadensis). Behavioral and Neural Biology, 26, 442-454.

Charif, RA, CW Clark, and KM Fristrup. 2004. Raven 1.2 User ’3 Manual. Ithaca:
Cornell Laboratory of Ornithology.

Cheney, D. L., and R. M Seyfarth. 1982. Recognition of individuals within and between
groups of free-ranging vervet monkeys. American Zoologist, 22, 519-529.

Cheney, D. L. and R. M. Seyfarth. 1990. How Monkeys see the World. University of
Chicago Press. Chicago.

Clutton-Brock, T. H. and S. D. Albon. 1979. The roaring Of red deer and the evolution of
honest advertisement. Behaviour, 69, 131-143.

Compton, L. A., J. A. Clarke, J. Seidenstricker, and D. R. Ingrisano. 2001. Acoustic
characteristics of white-nosed coati vocalizations: a test of the motivation-

structural rules. Journal of Mammalogy, 82(4), 1054-1058.

Connor, J. K. and D. L. Hart]. 2004. A Primer of Ecological Genetic. Sunderland: Sinauer
Associates, Inc.

Darwin, C. 1872. The expressions of the emotions in man and animal. John Murray.
London.

14

Davis, N. B. and T. R. Halliday 1978. Deep croaks and fighting assessment in toads Bufo
bufo. Nature. 274, 683-685.

Endler, J. A. 1993. Some general comments on the evolution and design of animal
Signaling systems. Phil Trans Royal Society of London, 340, 215-225.

Falls, J. B. and L. G. D’Agincourt. 1981. A comparison of neighbor-stranger
discrimination in eastern and western meadowlarks. Canadian Journal of
Zoology. 59, 2380-2382.

Falls, J. 1982. Individual recognition by sound in birds. In Song and Learning and Its
Consequences (Kroodsma, D. E. and Miller, E. H. eds) Academic Press 237-278.

Feighny, J. A., K. E. Williamson, and J. A. Clarke. 2006. North American elk bugle
vocalizations: male and female bugle call structure and context. Journal of
Mammalogy, 87(6), 1072-1077.

Gurnell, J. 1984. Home range, territoriality, caching behaviour and food supply of the
red squirrel (T amiasciurus hudsonicus) in a sub alpine lodgepole pine forest.
Animal Behaviour. 32, 1119-1131.

Greene, E. and T. Meaher. 1998. Red Squirrels, Tamiasciurus hudsonicus, produce
predator-class specific alarm calls. Animal Behavior, 55, 511-518.

Halliday, D; P, Resnik; J. Walden. 1997. Fundamentals of Physics, 5th ed. John Wiley
and Son Inc. New York.

Hamilton, W. D. 1964. The genetical evolution of social behavior, I and 11. Journal of
Theoretical Biology, 7, 1—52.

Hare, J. F. 1994. Group member discrimination via familiarity with substrate -
biochemical cues by Columbian ground squirrels. Animal Behavior, 47, 803—813.

Hauser, M. D. 1993. The evolution of nonhuman primate vocalization: Effects of
phylogeny, body weight, and social context. American Naturalist, 142, 528-542.

Holekamp, K. E., E. E. Boydston, M. Szykman, 1. Graham, K. J. Nutt, S. Birch, A.
Piskel, and M. Singh. 1999. Vocal recognition in the spotted hyena and its possible
implications regarding the evolution of intelligence. Animal Behaviour, 58, 383-395.

Holmes, W. G., and P. W. Sherman. 1982. The ontogeny of kin recognition in two
species of ground squirrels. American Zoologist 22, 491—517

Insley, S. J. 2000. Long—term vocal recognition in the northern fur seal. Nature, 406, 404-
405.

15

Johnstone, RA. 1997. The evolution of animal signals. In Behavioural Ecology (eds.
Krebs, J. R. and Davies, N. B) 155-178. Oxford: Blackwell Science.

Johnston, R. E. 2003. Chemical Communication in Rodents: From Pheromones to
Individual Recognition. Journal of Mammalogy, 84, 1141-1162.

Lair, H. 1990. The calls of the red squirrel: a contextual analysis of functions. Behaviour.
115 (3-4) 254-281.

Larsen, K. W. and S. Boutin. 1993. Does food availability affect the growth and survival

of males and females differently in a promiscuous small mammals, T amiasciurus
hudsonicus? Journal of Animal Ecology, 62, 364-370.

Lythgoe, J. N. 1988, Light and vision in the aquatic environment. In: Sensory Biology of
Aquatic Animals (ed. J. Atema, R. R. Fay, A. N. Popper, and N. Tavoga),
Springer-Verlag, New York, 57-82.

Leger, D. W., D. H. Owings, and L. M. Boa]. 1984. Contextual information and
differential response to alarm whistles in California ground squirrels — Z.
Teirpsycol. Journal of Comparative Ethology, 49, 142-155.

Leonard, M. L., A. G. Horn, C. R. Brown, and N. J. Fernandez. Parent-offspring
recognition in tree swallows, T achycineta bicolor. Animal Behavior, 54, 1107-
1116.

Marler, P. 1955. Characteristics of some animal calls. Nature, 176, 6-8.

Marler, P. 1967. Animal communication signals. Science. 157, 769-774.

Mateo, J. M. 2003. Kin recognition in ground squirrels and other rodents. Journal of
Mammalogy. 84(4), 1163-1181.

McAdam, A. G., S. Boutin, A. K. Sykes, and M. M. Humphries. 2007. Life histories of
female red squirrels and their contributions to population growth and lifetime

fitness. Ecoscience, 14(3), 362-369.

Morton, E. S. 1975. Ecological sources of selection on avian sounds. American
Naturalist, 109, 17-34

Morton, E. S. 1977. Occurrence and significance of motivational-structural rules in some
birds and mammal sounds. American Naturalist. 111. 855-869.

Miiller-Schwarze, D. and S. Heckman. 1980. The social role of scent marking in beaver
(Castor canadensis). Journal of Chemical Ecology, 6, 81-95.

16

Nesterova, N. L. 1996. Age-dependant alarm behaviour and response in alarm call in
bobac marmots (Marmota bobac Mull). In: Biodiversity in Mammals (eds. M.
LeBerre, R. Rmousse, and L. LeGuelte), 181-186. Moscow-Lyon: International
Network on Marmots.

Nodler, F. A. 1973. Food habits, vocalizations, and territoriality of Alaskan red squirrels
(Genus Tamiasciurus). M.Sc. thesis, University of Alaska.

Palacios, V, E. Font, and R Marquez. 2007. Iberian wolf howls: Acoustic structure,
individual variation, and a comparison with North American populations. Journal
of Mammlogy, 88(3), 606-613.

Puglisi, L. and C. Adamo. 2004. Discrimination of individual voices in male great
bitterns (Botaurus stellaris) in Italy. The Auk, 121, 541-547.

Price, K., K. Broughton, S. Boutin, and A. R. E. Sinclair. 1986. Territory size and

ownership in red squirrels: response to removals. Canadian Journal of
Zoology. 64, 1 144-1147.

Rukstalis, M. J. E. Fite and J. French. 2003. Social change affects vocal change in
Callitrichid primates (Callithrix kuhlii). Ethology, 109. 327-340.

Schwartz, J. J. and K. D. Wells. 1983. The influence of background noise on a
neotropical treefrog, Hyla ebraccata. Herpatologica, 39(2). 121-129.

Searby, A., and P. Jouventin. 2003. Mother-lamb acoustic recognition in sheep: a
frequency coding. Proceedings of the Royal Society of London B Biological
Science, 270, 1765-1771.

Searing, G. F. 1977. The function Of the bark call of the red squirrel. Canadian Field
Naturalist, 91, 187-189.

Sherman, P. W. 1980. The limits of ground squirrel nepotism. In Sociobiology: Beyond
Nature/Nurture? (eds. G. W. Barlow and J. Silverrnan) Westview, Boulder, 505-
544.

Sherman, P. W. 1981. Kinship, demography, and Belding’s ground squirrel nepotism.
Behavioral Ecology Sociobiology, 21, 229-235.

Sherman, P. W., H. K. Reeve, and D. W. Pfennig. 1997. Recognition systems. In
Behavioural ecology: an evolutionary approach (J. R. Krebs and N. B. Davies,
eds.). Blackwell Scientific Publications, Oxford, United Kingdom. 69—96

Smith, C. C. 1968. The adaptive nature of social organization in the genus of tree
squirrels Tamiasciurus hudsonicus. Ecological Monographs, 38, 31-63.

17

Smith, C. C. 1978. Structure and function of the vocalizations of tree squirrels
(T amiasciurus). Journal of Mammalogy. 59, 793-808.

Sousa-Lima, R. R., A. P. Pagalia, G. A. B. F Da Fonseca. 2002. Signature information
and individual recognition in the isolation calls of Amazonian manatees,
Trichechus inunguis (Mammalia: Sirenia). Animal Behavior. 63, 715-724.

Telemes, E. J. 1994. The role of neighbors in territorial systems — when are they dear
enemies. Animal Behavior, 47, 339-350.

Tibbits, E. A. and J. Dale. 2007. Individual recognition: it is good to be different. Trends
in Ecology and Evolution, 22 529-537.

Theis, K. R., K. M. Greene, S. R. Benson-Amram, and K. E. Holekamp. 2007. Sources of
variation in the long-distance vocalizations of Spotted hyenas. Beahviour.ll4,
557-584.

Tomszyacki, M. L., J. E. Davis, H. Gouzoules, and K. Wallen. 2001. Sex differences in
infant rhesus macaque separation-rejection vocalizations and effects of prenatal
androgens. Hormones and Behavior. 39, 267-276.

Tooze, Z. J., F. H. Harrington, and J. C. Fentress. 1990. Individually distinct
vocalizations in timber wolves, Canis lupus. Animal Behaviour 40, 723-730

Waldman, B. 1987. Mechanisms of Kin Recognition. Journal of Theoretical Biology,
128, 159-185.

Waldman, B. 1988. The ecology Of kin recognition. Annual Reviews Ecological
Systems, 19, pp 543- 571.

Wehner, R. 1997. The ant’s celestial compass system: spectral and polarization channels.
In Orientation and Communication in Arthropods (ed. M. Lehrer), 145—185.
Basel: Birkhauser Verlag.

Wich, S. A., S. Koski, H. de Varies, C. P. van Schaik. 2003. Individual and contextual
variation in the Thomas langur male loud calls. Ethology, 109, 1-13.

Wiley, R. H. 1983. Associations of song properties with habitats for territorial oscine
birds of eastern North America. American Naturalist, 138, 973-993.

Whitfield, D. P. 1986. Plumage variability and territoriality in breeding tumstone

Aernaria interpres — status signaling or individual recognition. Animal Behavior,
34, 1471-1482.

18

CHAPTER TWO
SIGNATURE SIGNALS IN THE RATTLES OF RED SQUIRRELS
(TAMIASCI UR US H UDSONICUS).
2.1 Introduction:

Vocalizations are one of the modalities by which animals communicate with one
another. Vocal communication allows animals to communicate over long distances
without engaging in direct contact with another individual (Marler, 1967). Variation in
the structure of vocalizations has been Shown to convey information about the caller’s
size (Davis and Halliday 1978; Clutton-Brock and Albon 1979) sex (Tomaszycki et al
2001), age (Cheney and Seyfarth 1990; Nestrova 1990; Blumstein and Daniels 2004;
Blumstein and Munos 2005) current circumstances (Norcross and Newman 1993;
Rendall et al. 1999; Crockford and Boesch 2003; Wich et al. 2003, Slocombe and
Zuberbuhler, 2005a,b) and identity (Holekamp et al. 1999; Rukstalis et a1 2003; Puglisi
and Adamo 2004; Palacios et al 2007).

Vocalizations of animals can also vary based on the contextual state of the caller
and can be explained using the MS-Rule (Morton, 1977). Morton (1977) proposed that
aggressive vocalizations would have lower fundamental frequencies (low pitch),
decreased time between syllables, and wider frequency (noisy) bandwidths in their
vocalizations because larger animals would be able to produce lower frequencies due to
the increased size of their vocal cords. These larger animals also tend to win aggressive
interactions due to their size, so the lower frequencies are a reliable signal of which
individuals to avoid (Morton, 1977). In contrast to aggressive vocalizations, submissive

vocalizations tend to have higher fundamental frequencies (higher pitch), increased time

between syllables, and narrower bandwidths (tonal; see Compton et al., 2001, Feighney et

19

al., 2006, and Theis et al., 2007). Age effects on vocalizations were first described by
Darwin (1872) and many researchers have confirmed his ideas on the similarity of the
vocalizations of submissive adults and juveniles. Darwin concluded that adults who have
been defeated or are in a submissive role use vocalizations structurally similar to juvenile
vocalizations to prevent or stop aggression from other individuals. Morton (1977)
included an empirical test of the motivation-structural rule in birds in which he showed
that the calls of several avian species followed the MS-Rule. In addition to the initial
work in avian species, the MS-Rule has been Shown to apply to the vocalizations of two
different mammalian species: white-nosed coatis (Nasua narica; Compton et al., 2001)
and elk (Cervus canadensis; F eighny et al. 2006).

In addition to being influenced by context, age, sex, or environmental factors calls
can also contain individual-specific components. These individual-specific signal
components can exist in the frequency, time, or rate aspects of a vocalization and are
collectively called signature signals. Signature signals have been shown to exist in the
vocalizations of several mammalian and avian species (e.g. Tooze et al. 1990, Rukstalis
et al 2003; Puglisi and Adamo 2004; Palacios et al 2007). These signature signals are
present in vocalizations that maintain social order and spacing, such as contact-calls
(howls) of wolves (Canis lupus; Tooze et al. 1990) and “phee” calls of marmosets
(Callithrix jacchus; Rukstalis, et al., 2003) and also in anti-predator calls, such as those
produced by vervet monkeys (Cheney and Seyfarth, 1988) or Belding’s ground squirrels
(Leger et al. 1984).

The presence of signature signals in vocalizations provides the potential for

individual kin recognition. Without individual variation it is impossible to identify

20

specific individuals, and thus identify specific kin. Most previous investigations of kin
recognition in small mammal systems, however, have been based on olfactory cues
(Mateo 2003), which may not be the primary modality for communication when there is
limited or no direct contact between individuals such as in territorial species.

One such system where acoustical cues may be used in place of olfactory cues for
kin recognition is the North American red squirrel (T amiasciurus hudsonicus). North
American red squirrels are solitary, territorial animals that communicate primarily
through vocalizations (Smith, 1968, Smith, 1978). Territories of red squirrels in white
Spruce (Picea glauca) forests are approximately 0.28 to 0.46 hectares (Price et al., 1986),
and are centered on a midden, or cache, where spruce cones are stored as a food source
for the winter and subsequent years (Smith, 1968; Price et al. 1986). Both male and
female red squirrels are highly territorial, defending year round territories from all
conspecifics with the exception of the one day per breeding season when a female is in
estrus and allows males onto her territory to mate (Smith, 1968; Gurnell, 1984; Price et
al. 1986; Boutin and Schweiger, 1988). This strict territorialin and limited direct
interactions in red squirrels suggests that acoustical cues may be used for kin recognition.

In the defense and maintenance of their territories, red squirrels use a vocalization
known as a rattle to proclaim ownership of a territory. This vocalization is the most
common vocalization used by red squirrels (Smith, 1968, Smith 1978, Lair, 1990). The
rattle is used as an advertisement by a territory owner to proclaim its presence on and
willingness to defend a territory (Smith, 1968). The rattle vocalization is composed of a
series of rapidly repeating syllables that are emitted in calling events. One calling event

can be composed of a short rattle lasting approximately one second to a series of

21

successive rattles lasting over five minutes (Smith, 1978). These syllables include a well
defined fundamental frequency and the corresponding harmonics. The harmonics are
tightly spaced and the individual syllables appear as dark lines on a sonogram (Figure 1;
see also Smith 1978). Rattles can be emitted in response to another call, a territory
intrusion, or used spontaneously to proclaim ownership of a territory (Smith, 1968;
Embry, 1970; Lair, 1990). Rattles are always broadcast from within an individual’s
territory with one exception; males and females rattle outside Of their territories during a
mating chase (Smith, 1978; Price et al. 1986), when the territorial system temporarily
breaks down. Initial studies of the rattle by Smith (1978) were inconclusive regarding
whether the rattles of red squirrels carried information related to the identity of the caller.
Smith was unable tO detect a Significant difference in his small number Of replicated
samples, but believed there were sufficient differences in the calls to allow for limited
identification of individuals.

An individual’s rattle has been reported to be heard up to 130m away in western
hemlock (T suga heterophylla) and lodgepole pine (Pinus contorta) forests (Smith, 1978),
which would correspond to an audible area of 5.3 hectares. Densities of red squirrels in
the southwest Yukon Territory of Canada varied from 1.3 individuals per hectare to 2.3
individuals per hectare during the 2005 and 2006 summers (Boutin, Humphries and
McAdam, unpublished data). Therefore, a single red squirrel could potentially hear
vocalizations from 7 to 13 surrounding conspecifics (Figure 2). The direction of a
vocalization alone might, therefore, be insufficient to determine identity along a single
bearing. Because directionality alone would not allow for accurate identification of

neighboring individuals through rattles, I hypothesized that red squirrel rattles contain

22

unique signature signals. I predicted that the identity of the rattling individual would
explain a significant amount of variance in the rattles from different individuals. I also
hypothesized that the rattles of red squirrels would contain information regarding
physical and contextual states of the caller. I predicted that the vocalizations recorded in
aggressive contexts would be of lower frequency and higher modulation than the rattles
recorded in non-aggressive contexts based on the motivation-structural rule (Morton,
1977). I predicted juvenile calls would be of lower fundamental frequencies and have
higher rates, than adults. Finally, because there is no physical sexual dimorphism (Larsen
and Boutin, 1993) or behavioral sexual dimorphism (males and females both defend
exclusive territories and use the same vocalizations) of red squirrels (Smith, 1968); I
predicted that male and female squirrels would no exhibit sexual dimorphism in their
rattles
2.2 Methods
2.2.1 Study Site and Location

Recordings of red squirrel rattles were made in a natural population in the
southwestern Yukon Territory, Canada (61°N, 138°W) during the months of June, July,
and August in 2005 and 2006. A detailed description of the area can be found in
McAdam et a1. (2007). A total of 291 adult recordings and 21 juvenile recordings were
made from 149 unique individuals. All individuals sampled were part of a long-term
study. Individuals were identified using numbered metal ear tags, but for behavioral
observation identification was established using unique combinations of colored wires
passed through the ear tags of the individuals.

2.2.2 Recording of Rattles

23

Rattles were recorded between 0600 and 1100 hours by waiting at the midden of
the focal individual until a call was recorded opportunistically or by eliciting a call via a
playback. Opportunistic recordings were made by placing the recording equipment on or
near the focal squirrel’s midden until that individual emitted a call. If an individual did
not rattle within approximately five minutes of my arrival at the midden, a playback of a
conspecific’s rattle was used in an attempt to elicit a response from the focal individual.
Playbacks were performed using a portable CD player with built-in speakers that was
placed on the focal individual’s midden. A conspecific’s rattle was arbitrarily chosen as
the stimulus from a catalogue of rattles recorded from the same study area. During a
playback, the recording equipment was placed underneath or near the focal individual to
record any responses. Rattles were recorded using a Marantz Professional Solid State
Recorder PMD660 that was able to record signals in the 0.01 kHz to 22.5 kHz range. The
sound file was stored as an uncompressed 16 bit .wav file to preserve call characteristics.
A foam windshield was employed to reduce interference and a low-noise cord connected
the multi-directional microphone to the recorder

During each rattle recording, the identity of the focal individual was confirmed
and the squirrel’s location in the tree and behavior were noted. The location of the
individual was classified as being within 3m of the ground or greater than 3m from the
ground.

Recorded rattles were grouped into three categories based on the behavioral
context of the call: unelicited, responsive, and provoked. Unelicited and responsive rattles
were recorded by waiting at the midden of the focal individual until they emitted a

spontaneous rattle. Unelicited rattles were rattles emitted with no apparent stimulus and

24

responsive rattles were rattles emitted in response to a vocalization Of a nearby neighbor.
A provoked call was a rattle recorded in response to a playback on the focal individual’s
territory or a rattle recorded during a chase or other direct confrontation between two
individuals. A provoked call emitted in response to a playback was defined to be any call
that occurred within three minutes of the start of the playback call. In most cases (>90%)
the provoked rattle occurred within thirty seconds of the playback rattle ending.
2.2.4 Data Extraction

The first data extracted were for total call length, which was measured as the
length of time from the first rattle syllable to last rattle syllable. Secondly, a subset of 10
syllables was used to characterize each recorded call. Because the calls’ fundamental
frequencies tended to change rapidly in the beginning and end of the calls, a sub-sample
of syllables was taken between the 11th and 21St syllable, which is approximately the
middle of most calls. If the call was shorter than 21 syllables, the mid-rattle syllable was
used as the midpoint for the sub-sample and the four syllables before and after it were
used to create the sub-sample. All calls with less then 11 syllables (n = 2) were omitted
from this study. Only the fundamental frequency of this subset of syllables was analyzed
and the following frequency measurements were taken: maximum fundamental
frequency, minimum fundamental frequency, bandwidth (maximum fundamental
frequency — minimum fundamental frequency), and modulation (bandwidth/syllable
length). Time measurements taken from a vocalization included the average syllable
length, and call rate (syllables per second). The measurements are shown in Figure 3.

Calls were analyzed using RAVEN 1.2.1 with a 256 sample fast fourier transformation

25

using a Hanning Window 4096 sample DF T with a 10.8 hertz grid resolution (Charif et
al. 2004).
2.2.5 Analyses

Two principal component analyses (PCA) were used to reduce the multivariate
dataset into a smaller set Of variables. The first PCA analysis incorporated only rattles of
adult squirrels while the second PCA included both adult and juvenile squirrels. The
second analysis was preformed to determine the effects of age on the structure of rattles.
Principal components were calculated using an eigenvector analysis on the correlation
matrix of the extracted data variables. The principal component scores (PC scores) for
each call were calculated by multiplying the vector of standardized values of the eight
acoustical variables by the eigenvector for each principal component for each call.

General linear models and linear mixed-effect models were used to analyze
variation in the four principal component scores for both the adult PCA and the combined
adult and juvenile PCA. Each principal component was modeled using two individual
variables (sex and age), three environmental variables (time of day, Julian date, location:
< 3m or > 3m from the ground), and the behavioral context (unelicited, elicited, or
provoked). Individual identity was also included as a random effect in mixed-effects
models for the PCA incorporating both adults and juveniles. Mass was not included in the
final models because mass data were available for only 200 of 320 recordings; models
based on a smaller data set that included lowest seasonal mass of individuals gave similar
results as the models that excluded mass. Significance of the variables was assessed using
the complete model. Analyses were performed using R (R Development Core Team,

2007). The presence of signature signals was assessed by comparing the linear mixed-

26

effects models and their corresponding nested general linear model with a likelihood ratio
test to assess the significance of the identity random effect in each principal component
(Edwards, 1972).
2.3 Results

The seven measurements taken on the recorded rattles (Table 2.1) were reduced
into four principal components for both principal component analyses. For the data set
including only adults, the first four principal components account 91% of the variation
(Table 2.2) while the four principal components from the combined adult and juvenile
PCA accounted for 94% of the variance in rattle vocalizations (Table 2.3). Principal
component one described the frequency and time elements in the fundamental frequency
of calls. Larger scores of principal component one corresponded to rattles with narrow
bandwidths, lower modulation and shorter syllable lengths. The second principal
component also described frequency aspects and syllable length of the. rattles. Larger
scores for principal component two corresponded to rattles with higher fundamental
frequencies, shorter syllable lengths and higher modulations. Principal component three
described mostly the time elements of the fundamental frequencies of the rattles. Larger
scores for principal component three corresponded to calls with higher rates
(syllables/second) and Shorter syllable lengths. Principal component four scores were
heavily based on the length of the rattles. Since the call length loaded in negatively with
the principal component score, larger scores were indicative of Shorter calls.

The results of the models for the PCA including only adults showed there were no
differences between male and females (ltlm S 1.1, p > 0.3 for all models), and there were

no differences in the structure of rattles in regards to context (i.e., unelicited, responsive,

27

or provoked call; NM 5 1.8, p > 0.07 for all models) for any of the principal components.
Environmental factors (time of day, Julian date, and location) were Shown to not
influence the structure of calls (III277 S 0.9 p > 0.5 for all models) except for principal
component two for which recordings made toward the end of the field season tended to
have higher scores for principal component two than those recorded earlier in the season
([3 :t SE, 0.009 d: 0.005, t155 = 2.05, p = 0.04). Higher scores of PC2 were associated with
generally higher fundamental frequencies.

The results of the models Of the combined adult and juvenile principal
components also showed no differences between male and female squirrel rattles in any
of the four principal components (|t|155 < 1.45, df = 155 p > 0.1 for all models). Juveniles
had lower values for principal component two ([3 :t SE, -1.05 :l: 0.29, t|55 = 3.63, p =
0.011; Figure 2.4) and principal component three ([3 i SE, -0.83 i 0.25, t155 = 3.267, p =
0.015; Figure 2.5). On average, the fundamental frequency of a rattle syllable produced
by adults (age > 1 year) started at 1280 Hz and ended at 870 Hz while rattle syllables
produced by juveniles (age < 1 year) started at 1180 Hz and ended at 783 Hz. The
average rate of syllables for juveniles was 25 syllables per second with a total average
call length of 3.4 seconds while adults produced on average 23 syllables per second with
an overall average call length of 3.0 seconds (Table 1.1)

The analysis of principal component four showed no differences between
unelicited calls and responsive calls ([3 i SE, 0.04 i 0.16 t155 = 0.08, p > 0.5), however,
provoked rattles had significantly lower scores for principal component four than
unelicited calls (0 3:: SE, -0.28 :I: 0.14, t155 = 2.11, p = 0.036; Figure 6). Provoked calls

were on average 3.42 seconds long (n= 88) whereas unprovoked calls were 2.91 seconds

28

long (n = 231). Provoked calls were on average 10 Hz lower in frequency and contained
1 additional syllable per second than unprovoked calls. No environmental factors were
shown to influence the structure of rattles: Julian date, time of day, and location showed
no significant influence on the structure of the rattles (|t|155 < 1.87, p > 0.05 for all
models).

The influence of identity on rattles structure was assessed using the adult and
juvenile data set. The random effect for individual identity was shown to Significantly
improve the models relative to models of PC scores that included only fixed effects (i.e.,
sex, age, time of day, Julian day, location, and context) for components one (x2 =15.16,
df= 1, p < 0.0001), two (x2 = 43.02, df= 1, p < 0.0001), three (x2 = 71.44, df= 1, p <
0.0001), and four (x 2 = 39.52, df = l, p < 0.0001), which is indicative of Signature
signals in each principal component. The random effect for identity was calculated prior
to fixed effects and explained 38.8% of the variance for PCI, 50.6% of the variance for
PC2, 54.7% of the variance for PC3, and 48.3% of the variance for PC4.

2.4 Discussion

The presence of a signature signal in a vocalization is a fundamental requirement
for individual recognition through vocalizations (Sherman et al., 1997). Individual
recognition allows an organism to identify their kin, neighbors, rivals, friends, or mates
based on prior interactions (Sherman et al., 1997). Individual recognition has been shown
to exist in many avian (e.g. Leonard et al, 1997; Aubin et a1, 2000) and mammalian
species (e.g. Johnston, 2003; Mateo, 2003; Searby and Jouventin 2003) and this study has
shown another mammalian system with signature signals, the North American red

squirrel (T amiasciurus hudsonicus). Individual variation accounted for between

29

approximately 39% and 55% of variation in the acoustical components of rattles,
meaning that approximately half of the rattle structure is a signature signal based on the
identity of the caller.

The initial studies of the rattle by Smith (1978) were inconclusive as to whether
the vocalizations of red squirrels carried information related to identity of the caller.
Smith was unable to detect a significant difference in his replicated samples, but believed
there were sufficient differences in the calls to allow for limited identification of
individuals. The normal response to a territory owner’s rattle is for the intruder to leave
the caller’s territory (Smith, 1968), but because of the use of rattles in mating chases,
where territoriality breaks down, Lair (1990) suggested that the rattle might also serve as
a Signature signal for individual recognition in red squirrels. The results of the current
study confirm that there are signature signals in the rattles of red squirrels and that they
possess the fundamental requirements for individual recognition. The rattle vocalization
is used extensively in mating chases (Smith, 1968; Smith 1978; Lair 1990) and could be
used to assist males trying to find estrus females during the chases, but I found no
differences between the structure of male and female rattles. It is likely, therefore, that
red squirrels involved in mating chases identify the female based on her unique call
components and not on a generalized “female” call.

Smith (1978) and Lair (1990) had differing views on the purpose of the rattles.
Smith (1978) considered the rattle a “keep-out” call while Lair (1990) proposed that the
rattle vocalization was used to convey multiple meanings. Lair’s ideas would be
supported if there was a difference in the frequency and time elements of rattles under

different contexts. The results of the analysis of principal component four shows there

30

were slight differences between the unprovoked and provoked rattles when I included
juveniles in the data set — the provoked rattles were ~10 Hz lower in frequency on
average and ~0.5 seconds longer. Even though there was only a Slight frequency
difference, we did see a significant difference in the length of rattles. This trend was not
present when I examined the data set excluding juveniles. This is most likely due to the
juvenile recordings being taken in a provoked state (11 = 4). The eigenvectors for principal
component four showed that call length was the dominant component in PC4, so
removing the juveniles from the data set did not radically change the eigenvectors for
PC4. Because the effect of context on PC4 disappears when we remove juveniles from
the data set I can conclude that behavioral context does not influence the structure of
adult rattles.

My results Show that red squirrels do not adhere to the motivation-structural rule
(Morton, 1977). If red squirrels followed the MS-Rule I would have expected the
fundamental frequency of the provoked rattles to be Significantly lower than those of
unprovoked rattles, which was not the case in this study as we saw only a Slight
difference. In retrospect, it is perhaps unreasonable to expect the MS-Rule to apply to red
squirrel territorial vocalizations. If an individual emitted passive or submissive rattles,
they would identify themselves as a prime target for territory intrusion or cone theft. The
purpose of emitting a submissive, long range call is to avoid aggression from other
individuals. However, in a system where displacement from a territory reduces
survivability dramatically (Larsen and Boutin, 1993), the benefit of being submissive
towards potential territory usurpers is minimal. While the motivation-structural rule

exists in close contact vocalizations (Compton et al., 2001) and some long range

31

vocalizations (F eighney et al., 2006), the motivation-structural rule is not followed in the
rattles of red squirrels

Differences in vocalizations based on age have been documented in many
different studies (e.g. Darwin, 1872; Morton, 1977; Sousa-Lima et al. 2002, Theis et a1.
2007). To date, many papers have discussed the relationship between lower fundamental
frequencies and larger body size associated with adulthood (Darwin, 1872; Morton, 1977;
Hauser, 1993; Thesis et al. 2007). This study found the opposite of previous literature.
Principal component two was explained by fundamental frequency and modulation of the
calls in the opposite direction to what would be predicted based on previous literature.
Larger scores for principal component two were indicative of calls with higher
modulation and higher frequencies. While principal component three is not as heavily
influenced by the frequencies of the rattles as it is the rate of the rattles, the eigenvectors
associate larger scores in principal component three with lower frequencies, reinforcing
the difference between adults and juveniles found for principal component two. Higher
scores for principal component three, which are indicative of higher rattle rates
(syllables/sec), indicate that juveniles vocalized more rapidly than adults when they
rattle.

I propose two possible explanations for the occurrence of lower fundamental
frequencies in juvenile red squirrel rattles compared to adult red squirrels. The first
possibility is that the emphasis harmonics of the juvenile rattles are higher than those of
adults. A harmonic series is comprised of a fundamental frequency and its associated
harmonics that are linearly spaced frequency intervals above the fundamental frequency;

the emphasis harmonic is the harmonic with the highest amplitude in the series (Halliday

32

et al. 1997). If the emphasis harmonics of juveniles are higher than those of adults, then
the juvenile vocalizations would generally sound higher pitched than adults, as has been
documented previously (e.g. Darwin, 1872; Morton, 1977; Hauser, 1993; Sousa-Lima et
al. 2002; Thesis et al. 2007). This study did not empirically examine the differences in the
harmonic structure of red squirrel rattles, only the differences in the structure of the
fundamental frequency.

The second possibility is that the variance we saw was due to low sample size for
individual juveniles. Due to the large amounts of variation based on identity we found in
the initial modeling, it is possible that these juveniles sampled have lower frequency calls
in general, which would cause the juvenile data set to be skewed toward lower
frequencies. Only seven juveniles were sampled and the recordings (n = 21) were made
after these individuals had dispersed and claimed their own territories. Collecting
recordings of juveniles is difficult due to high mortality rates and difficulty in tracking
their dispersal. During the summer in which juvenile recordings were made, few vacant
territories in the monitored research area were available. While both of these ideas are
potential answers tO the differences we see between adults and juveniles in principal
components two and three, none can fully explain them.

The results of this study provide the opportunity for much future research on a
system that goes against the grain when compared to classic communication literature.
Future work could focus on differences between adult and juvenile rattles to understand
the ontogeny of rattle development. Finally, my study has shown that identity has only
provided us with an explanation for 39% of the variation we see in principal component

one; the other 61% was not thoroughly explained by any of my included physical or

33

environmental variables. An expansion of physical (reproductive status, actual age) and
environmental predictors (weather, temperature, density, habitat, and seasons) could yield

additional information on what influences the acoustical structure of rattles.

34

2.5 References

Aubin, T. P. Jouventin, and C. Hildebrand. 2000. Penguins use the two-voice system to
recognize each other. Proceedings of the Royal Society of London B Biological
Science. 267, 1081-1087.

Blumstein, D. T., and O. Munos. 2005. Individual, age, and sex-specific information is
contained in yellow-bellied marmot alarm calls. Animal Behaviour. 69, 353-361.

Blumstein, D. T., L. Vemeyre and J. C. Daniel. 2004. Reliability and the adaptive utility

of discrimination among alarm callers. Proceedings of the Royal Society of
London, Series B. 271, 1851-1857.

Boutin, S. and S. Schweiger. 1988. Manipulation of intruder pressure in the red squirrel

(Tamiasciurus hudsonicus): Effects on territory size and acquisition. Canadian
Journal of Zoology. 66, 252-262.

Charif, RA, CW Clark, and KM Fristrup. 2004. Raven 1.2 User’s Manual. Cornell
Laboratory of Ornithology, Ithaca, NY.

Cheney, D. L. and R. M. Seyfarth. 1990. How Monkeys see the World. University of
Chicago Press. Chicago.

Clutton-Brock, T. H. and S. D. Albon. 1979. The roaring of red deer and the evolution of
honest advertisement. Behaviour. 69, 131-143.

Compton, L. A., J. A. Clarke, J. Seidenstricker, and D. R. Ingrisano. 2001. Acoustic
characteristics of white-nosed coati vocalizations: a test of the motivation-
structural rules. Journal of Mammalogy, 82(4), 1054-1058.

Darwin, C. 1872. The expressions of the emotions in man and animal. John Murray,
London.

Davis, N. B. and T. R. Halliday 1978. Deep croaks and fighting assessment in toads Bufo
bufo. Nature. 274, 683-685.

Edwards, Anthony W.F. 1972. Likelihood, lst edition, Cambridge: Cambridge University
Press.

Falls, J. B. and L. G. D’Agincourt. 1981. A comparison of neighbor-stranger

discrimination in eastern and western meadowlarks. Canadian Journal of
Zoology. 59, 2380-2382.

Feighny, J. A., K. E. Williamson, and J. A. Clarke. 2006. North American elk bugle

vocalizations: male and female bugle call structure and context. Journal of
Mammalogy, 87(6), 1072-1077.

35

Gurnell, J. 1984. Home range, territoriality, caching behaviour and food supply of the red

squirrel (Tamiasciurus hudsonicus) in a sub alpine lodgepole pine forest. Animal
Bahviour.32, 1119-1131.

Halliday, D., P. Resnik, and J. Walden. 1997. Fundamentals of Physics, 5th ed. John
Wiley and Son Inc. New York.

Lair, H. 1990. The calls of the red squirrel: a contextual analysis of functions. Behaviour.
115 (3-4) 254-281.

Larsen, K. W. and S. Boutin. 1993. Does food availability affect the growth and survival
of males and females differently in a promiscuous small mammal, T amiasciurus
hudsonicus? Journal of Animal Ecology, 62, 364-370.

Leonard, M. L., A. G. Horn, C. R. Brown, and N. J. Fernandez. Parent-offspring

recognition in tree swallows, Tachycineta bicolor. Animal Behavior, 54, 1107-
1116.

Marler, P. 1967. Animal communication signals. Science. 157, 769-774.

Mateo, J. M. 2003. Kin recognition in ground squirrels and other rodents. Journal of
Mammalogy. 84(4), 1 163-1 181.

McAdam, A. G., S. Boutin, A. K. Sykes, and M. M. Humphries. 2007. Life histories of
female red squirrels and their contributions to population growth and lifetime
fitness. Ecoscience,14(3), 362-369.

Morton, E. S. 1977. Occurance and significance of motivational-structural rules in some
birds and mammal sounds. American Naturalist. 111. 855-869.

Nesterova, N. L. 1996. Age-dependant alarm behaviour and response in alarm call in
bobac marmots (Marmota bobac Mull). In: biodiversity in Mammals (Ed. By M.
LeBerre, R. Rmousse, and L. LeGuelte), 181-186. Moscow-Lyon: International
Network on Marmots.

Palacios, V, E. Font, and R Marquez. 2007. Iberian wolf howls: Acoustic structure,
individual variation, and a comparison with North American populations. Journal
of Mammlogy, 88(3), 606-613.

Puglisi, L. and C. Adamo. 2004. Discrimination of individual voices in male great
bitterns (Botaurus stellaris) in Italy. The Auk, 121, 541-547.

Price, K., K. Broughton, S. Boutin, and A. R. E. Sinclair. 1986. Territory size and

ownership in red squirrels: response to removals. Canadian Journal of
Zoology. 64, 1 144-1147.

36

Rukstalis, M. J. E. Fite and J. French. 2003. Social change affects vocal change in
Callitrichid primates (Callithrix kuhlii). Ethology. 109. 327-340.

Searby, A., and P. Jouventin. 2003. Mother-lamb acoustic recognition in sheep: a
frequency coding. Proceedings of the Royal Society of London B Biological
Science, 270, 1765-1771.

Sherman, P. W., H. K. Reeve, and D. W. Pfennig. 1997. Recognition systems. In
Behavioural ecology: an evolutionary approach (J. R. Krebs and N. B. Davies,
eds.). Blackwell Scientific Publications, Oxford, United Kingdom. 69—96

Smith, C. C. 1968. The adaptive nature of social organization in the genus of tree
squirrels Tamiasciurus hudsonicus .Ecological Monographs, 38, 31-63.

Smith, C. C. 1978. Structure and function of the vocalizations of tree squirrels
(Tamiasciurus). Journal of Mammalogy. 59, 793-808.

Sousa-Lima, R. R., A. P. Pagalia, G. A. B. F Da Fonseca. 2002. Signature information
and individual recognition in the isolation calls of Amazonian manatees,
T richechus inunguis (Mammalia: Sirenia). Animal Behaviour, 63, 715-724.

Theis, K. R., K. M. Greene, S. R. Benson-Amram, and K. E. Holekamp. 2007. Sources of
variation in the long-distance vocalizations of spotted hyenas. Behaviour,114,
557-584.

Tibbits, E. A. and J. Dale. 2007. Individual recognition: it is good to be different. Trends
in Ecology and Evolution, 22 529-537.

Tomszyacki, M. L., J. E. Davis, H. Gouzoules, and K. Wallen. 2001. Sex differences in
infant rhesus macaque separation-rejection vocalizations and effects of prenatal

androgens. Hormones and Behavior. 39, 267-276.

Tooze, Z. J., F. H. Harrington, and J. C. Fentress. 1990. Individually distinct
vocalizations in timber wolves, Canis lupus. Animal Behaviour 40, 723-730

37

Table 1. Summary statistics for the fundamental frequency (F0) measurements of the 319
recorded rattles of red squirrels (T amiasciurus hudsonicus). The data were extracted
using Raven 1.2.1.

 

 

Adults Juveniles

Rattle Component Mean :I:SE Mean :tSE
Peak F0 (Hz) 1287.9 6.37 1181.0 38.29
Minimum F0 (Hz) 869.3 6.59 783.5 29.50
Bandwidth (Hz) 418.5 6.73 397.1 27.26
Modulation (kHz/second) 37.8 0.05 36.0 2.39
Syllable Length (seconds) 0.011 0.01 0.011 0.003
Call Length (seconds) 2.96 0.09 3.47 0.46
Syllables/second 22.83 0.16 24.90 0.57

 

Table 2.2: Principal component eigenvectors and the proportion of the variance explained

by each component for red squirrel rattles using the adult dataset. PC1 represents

bandwidth, PC2 represents frequency elements of the fundamental frequency, PC3
represents the time elements of the fundamental frequency and PC4 represents the
fundamental frequency, modulation, and length of the rattles.

 

 

Call Component PCl PC2 PC3 PC4
Peak Fundamental -0.32 0.53 -0.15 -0.22
Min Fundamental 0.33 0.60 -0.26 -0.20
Bandwidth -0.63 -0.01 0.1 1 -0.01
Syllable Length -0.44 -0.33 0.34 -0.26
Call Length 0.02 -0.09 -0.34 -0.88
Modulation -0.40 0.35 -0.46 0.25
Syllables/ Sec 0.21 0.00 -0.67 -0.01
Proportion of the

Variance Explained 0.35 0.23 0.19 0.14

 

38

Table 2.3: Principal component eigenvectors and the proportion of the variance explained
by each component for red squirrel rattles using adult and juvenile datasets. PCl
represents bandwidth, PC2 represents frequency elements of the fundamental frequency,
PC3 represents the time elements of the fundamental frequency and PC4 represents

fundamental frequency, modulation, and length of the rattles.

 

 

Call (3011330116111 PCl PC2 PC3 PC4
Peak Fundamental -0.36 0.58 -0.16 -0.16
Min Fundamental 0.27 0.62 -0.28 -0.14
Bandwidth -0.64 -0.03 0.1 1 -0.02
Syllable Length -0.39 -0.40 -0.37 -0.19
Call Length 0.02 -0.05 0.20 -0.94
Modulation -0.43 0.31 0.45 0.16
Syllables/ Sec 0.19 -0.01 0.69 -0.03
Proportion of the

Variance Explained 0.33 0.25 0.19 0.16

 

39

Rattle Syllables

     

I l

    
  

  
  

1; I

’:l. . I'II'V‘

    
  

   

. K‘,

I I’II'II i'

   

   

. I

        

I.I
I
I .fi’.;iyl:‘i“l :2: ”II!“ ”1‘

h. . 1' '~
. I I" ' II
.I.
I .
14‘ ‘ . ..1 .
4

Frequency (kHz)
.2:
l

 

-4'.

Time (sec)

Figure 2.1: Sonogram of the rattle vocalization of a North American red squirrel
(Tamiasciurus hudsonicus). The dark vertical lines represent individual syllables. The
total length of the rattle portion of this call is four seconds.

40

 

 

 

 

 

 

 

 

_“
Q
4?
@

 

 

 

 

 

7

Figure 2.2: Polygons indicate hypothetical territories of red squirrels, the shaded circle
indicates the focal squirrel and the large circle indicates the 130m radius that a rattle can

be heard from (Smith, 1978).

41

2.000
1.800

1.600 ' .

0.000 _
0.400 "

0.200

0 000
kHz

.v,

80.432 0.44 0.45 0.46 0.47

 

 

Figure 2.3. Spectrograph of measurements taken on the fimdamental frequency of a rattle
syllable. The fundamental frequency of the syllable is indicated by the shaded dark bar in
the center of the spectrograph. Seconds are represented on the x-axis while frequency
(Hz) is represented on the y-axis. The relative power of the signal (db) is indicated by
darker shading. This particular syllable has a maximum fundamental frequency of 1500
Hz, a minimum fundamental frequency of 600 Hz and is 0.015 seconds long.

42

 

0.5

-0.5 0.0

 

P02

 

-1.0

 

-‘l .5
I
I

 

 

 

-2.0
I

Figure 2.4: The effects of age of an individual on principal component two of red squirrel
rattles. Larger PC scores are indicative of higher fundamental frequencies, low bandwidth
of the fundamental frequency and low amounts of frequency modulation in the rattle.
Data are represented by mean +/- SE. Rattles recorded from adults are represented by
“A” and rattle recorded from juveniles (< 1 year old) are represented by “J.”

43

 

1.0

 

 

 

 

 

 

fl
'0._
O
('0
O
0 d—I ‘1—
[L .._L
'0.
O—
I
O.
‘_—l
l

 

Figure 5: The effects of the age of an individual on principal component three of red
squirrel rattles. Larger PC scores are indicative of increased rate in the syllables per
second and shorter syllable length in the rattle. Dare are represented by mean +/- SE.
Rattles recorded from adults are represented by “A” and rattle recorded from juveniles (<
1 year old) are represented by “J.”

44

 

0.4

0.2
I

 

 

 

 

PC4

 

-0.2 0.0

 

-0 .4

 

 

 

I I I

1 2 3
Behavioral Context

Figure 6: The effects of behavioral context on principal component four for red squirrel
rattles. Larger PC scores are indicative of shorter overall call durations. Data are
represented by mean +/- SE. Behavioral contexts were classified as unelicited (1),
responsive (2) and provoked (3). These data were generated from the adult and juvenile
combined data set.

45

Chapter Three
Kin recognition in the Territorial Vocalizations of Red Squirrels ( T amiasciurus
hudsonicus)
3.1 Introduction:

Kin recognition is the ability of an individual to assess the genetic relatedness of
another individual based on acoustical, olfactory, or visual cues (Hepper, 1991; Beecher,
1992). The study of kin recognition can provide insights into many aspects of animal
behavior, including dispersal, mate choice, social cooperation and even territoriality
(Blaustein et al. 1987; Hepper 1991; Pusey and Wolf 1996; Sherman et al. 1997, Tibbits
and Dale, 2007). Kin recognition has been shown to mediate territoriality (e.g. Falls,
1982; Whitfield, 1986; Johnston, 2003; Mateo 2003) by helping establish territory
borders and focus energy on defending territories from new neighbors rather than kin or
established neighbors (Telemes, 1994).

The ability to differentiate kin from non-kin allows the individual to alter its
behavior based on how closely it is related to the individual it is interacting with.
Recognition of kin has been Shown to influence responses to anti-predator calls (e.g.,
Hepper 1991, Hare, 1994), cooperation between groups of individuals (e.g., Cheney and
Seyfarth, 1982; Hepper 1991), and the location and identification of juveniles by adults
(Waldman, 1988; Balcombe 1990; Hepper, 1991; Holekamp et al. 1999; Insley 2000;
Searby and Jouventin, 2003). In all the examples given above, the response of the
recipient was strongly dictated by the recipient’s relation to the signaler. For example, in
Columbian ground squirrels (Spermophilus columbianus), the probability Of an individual
fleeing in response to the anti-predator call of a related individual was significantly

greater than a call from an unrelated individual (Hare, 1994). In vervet monkeys,

(Cercopithecus aethiops), interactions between troops are influenced by the patterns of

46

dispersal and prior knowledge of related individuals’ vocalizations (Cheney and Seyfarth
1982). The behavior of the monkey troops documented by Cheney and Seyfarth (1982) is
an example of kin selection theory (Hamilton, 1964). Kin selection (Hamilton, 1964)
predicts that genetic relatedness plays an important part in how animals will interact.
Individuals should be more inclined to associate with and assist related individuals
because it will increase their inclusive fitness and kin could potentially be lower threat
than non-kin because usurping one’s kin can negatively impact an individual’s inclusive
fitness.

Possible mechanisms Of kin recognition include site-specific spatial reference
(Beecher, 1981; Holmes and Sherman, 1982), prior association (Holmes and Sherman,
1982; Waldman, 1988), and phenotype matching (Holmes and Sherman, 1982, Waldman,
1988). Site-specific kin recognition is considered an indirect form of kin recognition
where individuals assume that conspecifics they share a common space with, such as a
nest or den, are kin. Kin recognition by prior association occurs when individuals interact
with conspecifics in a fashion only kin would, such as juvenile play fighting or nursing,
and in doing so learn the unique traits of that individual. The unique traits learned
through these interactions are commonly olfactory in mammals, but the possibility for
learning visual and acoustical traits of relatives exists as well (Waldman, 1988; Mateo,
2003). The third mechanism is phenotype matching, which occurs when an individual
receives a signal from an unknown individual and can determine if the caller is related to
them based on the Signal similarity to the recipient’s own signal or a signal from their

established kin (Waldman, 1988).

47

Recognition through these mechanisms does not allow for positive identification
in every situation and can be misleading. For example, cross-fostered offspring exposed
to foster parents calling during a key point in vocal development might begin to structure
their calls after the foster parents (Sharp et al., 2005). By structuring their vocalizations
after the foster parents, the juveniles gained a vocal signal that would allow for
misrecognition by related individuals. Misrecognition of foster offspring has been shown
experimentally (e.g., Sharp et al., 2005) and naturally (e.g., Brown, 1984; Lyon, 1993) in
birds. In order for phenotype matching to be a reliable mechanism for kin recognition
some portion of the signals must also be genetically influenced.

Kin recognition is a commonly studied phenomenon in rodents. In a review,
Mateo (2003) found that many rodent Species recognize kin through olfactory
mechanisms. Based on a comprehensive review of rodent recognition literature, Mateo
(2003) identified 33 species in which olfactory kin recognition occurred. Twelve of the
species reviewed were capable of phenotype matching based on these Olfactory cues
because individuals were able to recognize kin without prior association. Even though the
majority of rodents studied use olfactory mechanisms as a primary modality for kin
recognition, the potential for recognition through other modalities exists. In highly
territorial rodents that minimize direct contact, the primary modality for kin recognition
might instead be vocalizations.

One such species in which vocalizations could provide a basis for kin recognition
is the North American red squirrel (Tamiasciurus hudsonicus). North American red
squirrels are solitary, territorial animals that communicate territoriality primarily through

a vocalization called a rattle (Smith, 1968, Smith, 1978; Lair 1990). The rattle is used as

48

an advertisement by a territory owner to proclaim its presence on, and willingness to
defend a territory (Smith, 1968). The rattle call is a series of rapidly repeating syllables
that are emitted in bouts that can be as short as a single rattle lasting one second to a
series of rattles used in succession, which can exceed five minutes (Smith, 1978). These
syllables include a well defined fundamental frequency and the corresponding harmonics.
The harmonics are tightly Spaced and the individual syllables appear as dark lines on a
sonogram (Figure 2.1; see also Smith 1978). This call is emitted as a response to a
neighboring individual’s calls, a territory intrusion, or as an unelicited broadcast. Red
squirrels have also been thought to use a “bark” alarm call, in response to different
predators (Green and Meagher, 1998) and in interactions with conspecifics (Lair, 1990).

Red squirrel rattles have been shown to possess individual-specific components
(i.e., signature signals; Chapter Two). The presence of signature signals provides the
Opportunity for identification of Specific individuals in the population by other
individuals. Territories Of red squirrels in white spruce (Picea glauca) forests are
approximately 0.28 to 0.46 hectares (Price et al., 1986) and are centered on a midden, or
cache, where Spruce cones are stored for future consumption (Smith, 1968; Price et al.
1986). Both male and female red squirrels are highly territorial, defending year-round
territories from all conspecifics with the only exceptions being juveniles on their natal
territory prior to dispersal and during mating chases when the territorial system
temporarily breaks down (Smith, 1968; Gurnell, 1984; Price et al. 1986; Boutin and
Schweiger, 1988).

Given the territoriality of red squirrels and the established presence of signature

signals in their territorial rattle vocalization, I hypothesized that red squirrels are capable

49

of recognizing kin and other conspecifics through rattle vocalizations. I further predict
that if red squirrels are capable of individual recognition, I would see a difference in the
probabilities of receiving rattles, barks, and differences in the probability of individuals
approaching the speaker by individuals involved in interactions with familiar individuals
(neighbors) and unfamiliar individuals (non-neighbors). The dear-enemy hypothesis
(Telemes, 1994) predicts that an individual will devote more energy toward defending
territories from previously unknown individuals than established neighbors; therefore, I
predicted focal individuals would have a higher probability of rattling at on-neighbors to
affirm their status as territory owners

Finally, I also examined whether or not red squirrels barked in response to
playbacks of rattles. If barks serve only as an alarm call (sensu Green and Meagher,
1998), then I would expect no effect of the playback on the probability of barking.
However, if barks are used in place of direct (physical) aggressive interactions between
individuals (Lair, 1990) then I predicted that focal individuals would have higher
probabilities of barking in response to the playback of a rattle compared to a pre-playback
observational period.
3.2 Methods:
3.2.1 Study Location

This study was conducted on a natural population of North American red squirrels
(T amiasciurus hudsonicus) in the southwestern Yukon Territory, Canada (61°N, 138°W)
during the months of June, July, and August in 2006. A detailed description of the area
can be found in McAdam et al. (2007). All individuals sampled were part of a long-term

study and were uniquely identified using numbered metal ear tags. For behavioral

50

observation, identification was established using unique combinations of colored wires,
pipe cleaners, or disks passed through the ear tags of the individuals that allowed for
identification from a distance. Females were marked with only colored wires, males were
marked with a combination of wires and pipe cleaners, and young-of-the-year (juveniles)
were marked with a combination of plastic discs and wires. The individuals involved in
this study were pre-tagged for a prior study (McAdam et a1. 2007) and were not captured
or handled in any manner for this study.
3.2.2 Playback Design

Acoustical playbacks of red squirrel rattles were conducted on focal individuals
by placing a GPX portable stereo approximately 10 meters from the focal squirrel with
the speakers directed toward the individual and camouflaged with foliage. Each playback
trial was comprised of a three-minute pre-observational period, a playback of a recorded
rattle, and then a three-minute post-observational period. The post-rattle playback
Observation period began with the recorded rattle being emitted from the speaker. Only
one rattle was played per playback trial and these rattles ranged from two to five seconds
in length. In order to differentiate between a response to the speaker and the observer, the
observer stood perpendicular to the Speaker and the focal squirrel. There were no trials
during this study in which the focal individual faced the Observer and then vocalized. If
the focal individual moved during the playback trial, the observer moved to attempt to
keep the focal individual in sight. If visual contact was lost for more then several seconds
or the focal individual moved more than 20 meters from the speaker, the trial was
restarted. The spatial layout of the speaker, midden, and observer for playback trials is

depicted in Figure 3.1.

51

Rattles used for the playbacks were recorded from individuals in the study
population. These recordings were made opportunistically by recording the Spontaneous
and responsive rattles of red squirrels (Chapter two). The recorded calls were imported
into Raven 1.2.1 and were amplified by a factor of five. Each amplified call was then
burned onto a blank CD as an mp3 file and played from a portable stereo. The rattle used
for each playback trial was selected based on two conditions: kinship and proximity
(neighbor relationship) that were randomly applied in a fully factorial design. For the
purpose of this study, proximity was assessed as being neighboring or non-neighboring.
Neighbors were considered to be pairs of individuals whose primary middens were less
than 150m apart; non-neighbors were considered to be pairs of individuals whose primary
middens were more then 150m from one another. Middens closer than 150m were
typically found on adjacent territories. The 150m distance was chosen based on the 130m
audible distance documented by Smith (1978) with 20m additional buffer zone added to
compensate for territory size. Because territory size ranges from 0.28 to 0.43 hectares
(~15m radius) it is possible that while the primary middens may be greater then 130m
apart, the edge of the territories could be as close as 100m. The distance of 150m was
selected to minimize the possibility that the two individuals involved in the non-
neighboring playbacks interacted regularly. Kin were classified as individuals with a
relatedness value (r-value) Z 0.25. Relatedness was calculated using a maternal and
paternal pedigree for the individuals in the long term study area. The maternal pedigree
was constructed using behavioral observations of females and natal young at nests (Lane

et al. 2007) and the paternal side of the pedigree was established using microsatellite

52

analysis of tissue samples collected from individuals in the study population (Gunn et al.
2005; Lane et al. 2007).

An r-value of 0.25 could represent grandmother-granddaughter relationships or
half-siblings. Sibling pairings were assigned an r-value of 0.25 even if paternity was not
known since both full and half—sibs have r-values of at least 0.25. Of the 30 trials
conducted involving kin, 20 trials were conducted using mother-offspring pairings, seven
involving sibling pairings, and three involving sire-offspring pairings. Non-kin were
classified as any individuals with an r < 0.25.

NO individual was used for the same treatment twice (e. g., a mother would not be
used in a neighboring-kin treatment twice if she lived adjacent to two daughters). In
addition to not replicating treatments on a focal individual, each recorded rattle was
selected and used for three or four different treatments, but only once for a specific
treatment.

Trials were nullified if an intruding squirrel entered the focal individual’s territory
and a direct interaction occurred (n = 1) or visual contact with the focal individual was
lost (n = 6). Playback trials were rescheduled if the rattles of neighboring individuals
were heard during the pre-observation period (11 = 3) or if the focal individual engaged in
alarm calling at the researcher (n = 5). Any individuals that engaged in alarm calling
behavior as the observer approached their midden were left alone and revisited to at a
later time to conduct the playback trial.

3.2.3 Data Analysis
During the pre and post-observational periods, the number and type of

vocalizations emitted by the focal individual were recorded (rattles, rattles with chews,

53

barks, and squeaks). In addition to the vocalizations being documented in these periods, I
also documented whether the focal individual looked at or approached the speaker and
whether there were any neighboring conspecifics in the area at the time of the playback
trial. The data were analyzed using generalized linear models. Time looking at the
speaker was not recorded because it was not possible to video tape the playback trials.
The probability of rattling, barking, and approaching the Speaker in response to the
playback were modeled using separate generalized linear models with a binary response
for each behavior. The two experimental factors were included as predictors in the model:
proximity (Neighbor, n = 32; Non-Neighbor, n = 32) and kin (Kin, n = 30; Non-Kin, n =
34). Individuals in the pre-observation period were not split based on relatedness while
individuals in the post-observational phase were divided into four categories based on the
kinship and neighbor relationships.

There was some over-dispersion in the rattle model (11) = 1.24; residual deviance =
75.77 on 61 degrees of freedom) and the bark model (it) = 1.36; residual deviance = 83.03
on 61 degrees of freedom) and there were no signs of data points with heavy influence on
the relationships found by the models. The models were initially constructed using
factors for sex of the focal individual, time of day (converted to continuous decimal
format, 0.0 = 0000h, 0.5 = 1200h, etc.), and an interaction between kinship and
proximity, but these factors had no significant influence on the behavioral response
frequencies. These factors were removed and the models reassessed. In the one instance
in which the focal individual did not look at the speaker, he raised his head following the

playback then immediately put it down again. The frequency of rattles and barks in the

54

pre-observational period were compared to the post-observational rates using Fisher’s
Exact tests (Zar, 1996).
3.3 Results:

During the three-minute pre-observational phase, the focal individual rattled in
22% of the trials and barked in 19% of the trials. There was a trend for the focal
individuals chosen for the non-neighboring treatments to rattle more during the pre-
Observation period than those chosen for the neighboring treatments (2 = 1.83, df= 63, p
= 0.066). This trend was also evident in barks (z = 1.94, df = 63, p = 0.052). During no
pre-observational phase did the focal individual look at or approach the speaker. The
probability of the focal individual rattling during the pre-observational phase was not
influenced by the sex of the individual (2 = -1.02, df = 63, p = 0.308) or the time of day (2
= -0.54, df = 63, p = 0.96). The probability of the focal individual barking during the pre-
Observational phase was also not influenced by the sex of the individual (2 = -0.49, df =
63, p = 0.624) or the time of day (2 = -0.588, df= 63, p = 0.556).

Focal individuals were less likely to rattle in response to the playback of a related
individual (2 = 2.95, df = 63, p = 0.003), but there was no influence of neighbor
relationship on the probability of rattling (z = 1.419, df = 63, p = 0.156). The probability
of hearing a rattle in the post-observational period in trials involving kin (24%) was not
significantly different from the frequency of rattles in the pre-observational phase for
neighbors (19%; Fisher’s Exact, p = 0.22) or non-neighbors (32%; Fisher’s Exact, p =
0.34). The frequency of the focal individual responding to a playback of an unrelated
individual during the post-observational phase (62%) was significantly higher than the

frequency of rattles in the pre-observational phase of the neighboring treatments (12.5%;

55

Fisher’s Exact, p < 0.001) and non-neighboring treatments (31.3%; Fisher’s Exact, p =
0.012, Figure 3.2).

There was no difference in the probability of receiving a bark in response to
playbacks based on kinship (z = 0.954, df = 63, p = 0.330) or proximity (z = -1.339, df =
63, p = 0.181). However, focal individuals were three times more likely to emit a bark in
response to a playback of their neighbors rattle (Fisher’s Exact, p = 0.015) when
compared to the pre-observational period (Figure 3.3); however, focal individuals were
not more likely to bark at the play back of a non-neighbor (Fisher’s Exact, p = 0.41;
Figure 3.3). Individuals tended to approach the speaker more frequently in response to a
playback involving a neighboring individual’s call than a non-neighboring individual’s
call (2 = -1.757, d‘f = 63, p = 0.079) but there was no effect of relatedness on the
probability Of approaching the speaker (z = 0.386, df = 63, p = 0.7).

3.4 Discussion:

Evidence from this study confirmed my prediction that red squirrels are capable
of kin recognition — focal individuals were more likely to rattle in response to a playback
of an unrelated individual’s rattle. Focal individuals were no more likely to rattle at a
playback of their own kin than they were to spontaneously rattle when there was no
stimulus involved during the pre-playback observational period. Behaviorally, this shows
that individuals do not alter their vocalization rates in response to simulations of territory
intrusions by kin, but they did toward non-kin. This clear ability to alter behavior in
response to the relatedness of the individual with which they are interacting shows that

red squirrels are capable of kin recognition.

56

One point of interest in the playback data is the trend for individuals chosen for
the non-neighboring treatments to rattle more during the pre-observational phase than
those in the neighboring treatments. The individuals for each treatment were randomly
chosen, and the trend could be due the personality of the squirrels. Recent work by Boon
et al. (2007) showed that red squirrels exhibit different personalities. Given that the
squirrels exhibit different personalities, it is not unreasonable to conjecture that some
individuals rattle more often than other. It is possible that the individuals chosen in the
non-neighboring treatments were generally more vocal than the ones chosen for the
neighboring treatments. Beyond this idea, there is no explanation for the difference we
see between neighboring and non-neighboring rattle probabilities in the pre-observational
phase of the playbacks.

Barks have been Shown to be used in place of aggressive behavior, such as
charging (Lair, 1990) and they are used as anti-predator calls (Green and Meagher, 1998).
If the bark served solely as an anti-predator call we would expect the same probability of
barking in the pre- and post-observational phase and this was not the case in my study. I
found that focal individuals were more likely to bark after hearing a rattle playback from
a neighboring individual. However, the probability Of barking in response to a rattle
playback was independent of relatedness, so I found no evidence for kin discrimination
based on the barking vocalization. If red squirrels bark in response in an aggressive
encounter then it is not clear why they would respond differentially towards kin in terms
of rattles but not barks.

While this study offers little insight into the function of the bark vocalization, the

results of the bark analysis do show that red squirrels are capable Of neighbor-class

57

recognition. When we compare the probabilities of barking in the pre and post
observational phases, we see a significant increase in the probability of barking at
neighboring individuals. These differential responses to neighbors and non-neighbors
shows that red squirrel recognize and respond differently to their neighbors. This is
opposite of what we would expect based on the dear-enemy hypothesis; I expected higher
probabilities of responding to non-neighboring individuals. One discrepancy to the dear-
enemy hypothesis is that the playback studies simulated a territory intrusion by the
neighbor; this would represent a breakdown in the “dear-enemy” by the focal individual’s
neighbor. This could result in the focal individuals taking quick and aggressive action
against the neighbor that violated the “dear-enemy” status.

We have already shown that red squirrels can recognize kin and neighbors.
However, these groups can be recognized without recognizing a specific individual. Kin
can be recognized based on lineage specific traits (Waldman, 1988) and neighbors can be
recognized because the vocalizations are heard regularly and the recipient learns them as
a neighboring call. The current study cannot determine if red squirrels engage in
individual recognition, but signature signals and class recognition are present in the red
squirrel rattles.

This study of the vocalizations of red squirrels is the first to Show vocal
communication as a primary modality for kin recognition in a rodent (Mateo, 2003). All
previous work regarding kin recognition in rodents has been directed at prior association
and phenotype matching via olfactory mechanisms. We can gain some insight into the
possible mechanism of kin interaction by examining the results of this study in

conjunction with the basic behavior of red squirrels. Site-specific kin recognition would

58

be a poor mechanism of recognition in red squirrels because this mechanism of
recognition typically involves social organisms with low rates of dispersal (Sherman and
Holmes, 1982). This is quite the opposite of red squirrels who are asocial (Smith 1968)
and readily disperse (Berteaux and Boutin, 2000). Even though adults most likely do not
use site-specific recognition, there is evidence to support the idea that mother’s use Site-
specific recognition to identify their offspring at early ages. McAdam and Boutin (2003)
conducted a cross fostering experiment where unrelated pups were taken from their nest
and placed in a nest of an unrelated individual. These pups were not rejected by the foster
mother and were raised as her own. Based on this information, we can conclude site-
specific kin recognition most likely occurs in the natal nest, but once the offspring
emerge from the nest, the territoriality and dispersal patterns of red squirrels remove any
possibility for site-specific kin recognition.

Determining whether adult red squirrels are capable of kin recognition based on
prior association, phenotype matching, or a combination Of both is difficult at the present
time. Mothers and juveniles interact on the natal territory for some time after emergence,
and there is the possibility that the mother and her offspring learn to recognize one
another’s rattles in the process. Because most of the playback trials (90%) were done
with mother-offspring or nest mate sibiling pairs, we cannot completely tease apart the
effects of prior association and phenotype matching in red squirrels. Some possible
insight into phenotype matching can be found in the remaining trials that were conducted
on sire-offspring pairs (n = 3). One-third of the sire-Offspring pairs, which had not
previously associated with one another, rattled in response to their relative’s rattle. This is

similar to the rattle probabilities of mothers and offspring rattling at one another (26%)

59

and is approximately half the probability of rattling at a non-kin playback (~62%). The
possibility is strong that phenotype matching between Sires and offspring occurs, but we
do not have the sample size to make a definitive conclusion at this time. More playback
experiments involving related individuals that have never interacted with one another,
such as father-Offspring pairs, would provide us with definitive proof for phenotype

matching should the results be consistent with this study.

60

3.5 References:

Balcombe, J. P. 1990. Vocal recognition of pups by mother Mexican-free tailed bats.
Animal Behaviour, 39, 960-966.

Beecher, M. D., I. M. Beecher, and S. H. Nichols. 1981. Parent-offspring recognition in

bank swallows (Riparia riparia): 11. Development and acoustic basis. Animal
Behavior. 29, 95-101.

Beecher, M. D. 1982. Signature Systems and Kin Recognition. American Zoologist, 22,
477-490.

Berteaux, D., and S. Boutin. 2000. Breeding dispersal in female North American red
squirrels. Ecology, 81, 131 1-1326.

Blaustein, A. R. 1983. Kin recognition mechanisms: phenotypic matching or recognition
alleles? American Naturalist 121, 749—754.

Blumstein, D. T., and O. Munos. 2005. Individual, age, and sex-Specific information is
contained in yellow-bellied marmot alarm calls. Animal Behaviour. 69, 353-361.

Blumstein, D. T., L. Verneyre and J. C. Daniel. 2004. Reliability and the adaptive utility

of discrimination among alarm callers. Proceedings of the Royal Society of
London, Series B. 271, 1851-1857.

Boon, A. K., D. Réale, and S. Boutin. 2007. The interaction between personality,

offspring fitness and food abundance in North American red squirrels. Ecology
Letters. 10, 1094-1104.

Boutin, S. and S. Schweiger. 1988. Manipulation of intruder pressure in the red squirrel

(Tamiasciurus hudsonicus): Effects on territory size and acquisition. Canadian
Journal of Zoology. 66, 252-262.

Brown, C. R. 1984. Laying eggs in a neighbor nest: benefit and cost of colonial nesting in
swallows. Science, 225, 518-519.

Charif, RA, CW Clark, and KM Fristrup. 2004. Raven 1.2 User ’s Manual. Cornell
Laboratory of Ornithology, Ithaca, NY.

Cheney, D. L. and R. M. Seyfarth. 1982. Recognition of individuals within and between
groups of free-ranging vervet monkeys. American Zoologist, 22, 519-529.

Clutton-Brock, T. H. and S. D. Albon. 1979. The roaring of red deer and the evolution of
honest advertisement. Behaviour. 69, 131-143.

61

Compton, L. A., J. A. Clarke, J. Seidenstricker, and D. R. Ingrisano. 2001. Acoustic
characteristics of white-nosed coati vocalizations: a test of the motivation-
structural rules. Journal of Mammalogy, 82(4), 1054-1058.

Crockford, C. and C. Boesch. 2003. Context-specific calls in wild chimpanzees, Pan
troglodyte verus: analysis of barks. Animal Behaviour, 66, 115-125.

Darwin, C. 1872. The expressions of the emotions in man and animal. John Murray,
London. ‘

Davis, N. B. and T. R. Halliday 1978. Deep croaks and fighting assessment in toads Bufo
bufo. Nature. 274, 683-685.

Falls, J. B. and L. G. D’Agincourt. 1981. A comparison of neighbor-stranger
discrimination in eastern and western meadowlarks. Canadian Journal of
Zoology. 59, 2380-2382.

Falls, J. 1982. Individual recognition by sound in birds. In Song and Learning and Its
Consequences (Kroodsma, D. E. and Miller, E. H. eds) Academic Press 237-278.

Feighny, J. A., K. E. Williamson, and J. A. Clarke. 2006. North American elk bugle
vocalizations: male and femal bugle call structure and context. Journal of
Mammalogy, 87(6), 1072-1077.

Gunn, M. R., D. A. Dawson, A. Leviston, K. Hartnup, C. S. Davis, C. S. Strobeck, J.
Slate, and D. W. Coltrnan. 2005. Isolation of 18 polymorphic microsatellite loci
from the North American red squirrel, Tamiasciurus hudsonicus (Sciuridae,

Rodentia), and their cross-utility in other species. Molecular Ecology Notes, 5,
650-653.

Gurnell, J. 1984. Home range,territoriality, caching behaviour and food supply of the red
squirrel (T amiasciurus hudsonicus) in a sub alpine lodgepole pine forest. Animal
Bahviour.32, 1119-1131.

Greene, E. and T. Meaher. 1998. Red Squirrels, T amiasciurus hudsonicus, produce
predator-class specific alarm calls. Animal Behavior, 55, 511-518.

Hamilton, W. D. 1964. The genetical evolution of social behavior, I and II. Journal of
Theoretical Biology, 7, 1—52.

Hare, J. F. 1994. Group member discrimination via familiarity with substrate -
bomechemical cues by Columbian ground squirrels. Animal Behavior, 47, 803-
813

Hauser, M. D. 1993. The evolution of nonhuman primate vocalization: Effects of
phylogeny, body weight, and social context. American Naturalist, 142, 528-542.

62

Hepper, P. G. 1991. Kin Recognition. Cambridge, New York, NY.

Holekamp, K. E., E. E. Boydston, M. Szykman, 1. Graham, K. J. Nutt, S. Birch, A.
Piskel, and M. Singh. 1999. Vocal recognition in the Spotted hyena and its possible
implications regarding the evolution of intelligence. Animal Behaviour, 58, 383-395.

Holmes, W. G., and P. W. Sherman. 1982. The ontogeny Of kin recognition in two
species of ground squirrels. American Zoologist 22, 491—517

Insley, S. J. 2000. Long-terrn vocal recognition in the northern fur seal. Nature, 406, 404-
405.

Johnston, R. E. 2003. Chemical Communication in Rodents: From Pheromones to
Individual Recognition. Journal of Mammalogy, 84, 1141-1162.

Lair, H. 1990. The calls of the red squirrel: a contextual analysis of functions. Behaviour.
115 (3-4) 254-281.

Lane, J. E., S. Boutin, M. R. Gunn, J. Slate, and D. W. Coltman. 2007. Genetic
relatedness of mates does not predict patterns of parentage in North American red
squirrels. Animal Behaviour, 74, 611-619.

Larsen, K. W. and S. Boutin. 1993. Does food availability affect the growth and survival

of males and females differently in a promiscuous small mammal, T amiasciurus
hudsonicus? Journal of Animal Ecology, 62, 364-370.

Larsen, K. W., and S. Boutin. 1994. Movements, survival, and settlement of red squirrel
(Tamiasciurus hudsorzicus) offspring. Ecology 75:21-223.

Lyon, B. E. 1993. Tactics of parasitic American coots: host choice and the pattern of egg
dispersion among host nests. Behavioral Ecology and Sociobiology, 33, 87-100.

Marler, P. 1967. Animal communication signals. Science. 157, 769-774.

Mateo, J. M. 2003. Kin recognition in ground squirrels and other rodents. Journal of
Mammalogy. 84(4), 1 163-1 181.

McAdam, A. G., S. Boutin, A. K. Sykes, and M. M. Humphries. 2007. Life histories of
female red squirrels and their contributions to population growth and lifetime
fitness. Ecoscience,14(3), 362-369.

McAdam, A. G., and S. Boutin. 2003. Effects of food abundance on genetic and maternal
variation in the growth rate of juvenile red squirrels. Journal of Evolutionary
Biology 16, 1249-1256.

63

Morton, E. S. 1977. Occurance and significance of motivational-structural rules in some
birds and mammal sounds. American Naturalist. 111. 855-869.

Norcross, J. L. and J. D. Newman. 1993. Context and gender-specific differences in the

acoustic structure of common marmoset (Callithrixjacchus) phee calls. American
Journal of Primatology 30, 37-54.

Palacios, V, E. Font, and R. Marquez. 2007. Iberian wolf howls: Acoustic structure,

individual variation, and a comparison with North American populations. Journal
of Mammlogy, 88(3), 606-613.

Puglisi, L. and C. Adamo. 2004. Discrimination of individual voices in male great
bitterns (Botaurus stellaris) in Italy. The Auk, 121, 541-547.

Price, K.., K. Broughton, S. Boutin, and A. R. E. Sinclair. 1986. Territory size and

ownership in red squirrels: response to removals. Canadian Journal of
Zoology. 64, 1144-1147.

Pusey, A., and M. Wolf. 1996. Inbreeding avoidance n animals. Trends in Ecology and
Evolution 11, 201—206.

R Development Core Team. 2007. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-
900051-07-0, URL: http://www.R-project.org.

Rendall, D., R. M. Seyfarth, D. L. Cheney, and M. J. Owren. 1999. The meaning and
function of grunt variants in baboons. Animal Behavior, 57, 583-592.

Rukstalis, M. J. E. Fite and J. French. 2003. Social change affects vocal change in
Callitrichid primates (Callithrix kuhlii). Ethology. 109. 327-340.

Searby, A., and P. Jouventin. 2003. Mother-lamb acoustic recognition in sheep: a
frequency coding. Proceedings of the Royal Society of London B Biological
Science, 270, 1765-1771.

Sharp, 8., A. McGowan, M. Wood, and B. Hatchwell. 2005. Learned kin recognition in a
social bird. Nature, 434, 1127-1129.

Sherman, P. W., H. K. Reeve, and D. W. Pfennig. 1997. Recognition systems. In
Behavioural ecology: an evolutionary approach (J. R. Krebs and N. B. Davies,
eds.). Blackwell Scientific Publications, Oxford, United Kingdom. 69—96

Smith, C. C. 1968. The adaptive nature of social organization in the genus of tree
squirrels Tamiasciurus hudsonicus .Ecological Monographs, 38, 31-63.

Smith, C. C. 1978. Structure and function of the vocalizations of tree squirrels
(Tamiasciurus). Journal of Mammalogy. 59, 793-808.

64

Sousa-Lima, R. R., A. P. Pagalia, G. A. B. F Da Fonseca. 2002. Signature information
and individual recognition in the isolation calls of Amazonian manatees,
T richechus inunguis (Mammalia: Sirenia). Animal Behavior. 63, 715-724.

Telemes, E. J. 1994. The role of neighbors in territorial systems — when are they dear
enemies. Animal Behavior, 47, 339-350

Tibbits, E. A. and J. Dale. 2007. Individual recognition: it is good to be different. Trends
in Ecology and Evolution, 22 529-537.

Theis, K. R., K. M. Greene, S. R. Benson-Amram, and K. E. Holekamp. 2007. Sources of
variation in the long-distance vocalizations of spotted hyenas. Behaviour.114,
557-584.

Tomszyacki, M. L., J. E. Davis, H. Gouzoules, and K. Wallen. 2001. Sex differences in
infant rhesus macaque separation-rejection vocalizations and effects of prenatal
androgens. Hormones and Behavior. 39, 267-276.

Tooze, Z. J., F. H. Harrington, and J. C. Fentress. 1990. Individually distinct
vocalizations in timber wolves, Canis lupus. Animal Behaviour 40, 723-730

Waldman, B. 1988. The ecology Of kin recognition. Annual Reveiws Ecological
Systems, 19, pp 543- 571.

Whitfield, D. P. 1986. Plumage variability and territoriality in breeding tumstone
Aernaria interpres — status signaling or individual recognition. Animal Behavior,

34, 1471-1482.

Wich, S. A., S. Koski, H. de Vries, and C. P. van Schaik. 2003. Individual and contextual
variation in the Thomas langur male loud calls. Ethology, 109, 1-13.

Zar, J. H. 1996. Biostatistical Analysis. Prentice-Hall, Inc. Upper Saddle River, NJ.

65

/Midden

Speaker

 

 

 

 

I 10m J

 

Focal Squirrel
Observer —' 0

Figure 3.1: Spatial layout Of the playback study. The speaker was placed 10m
from the center of the focal individual’s midden directed toward the individual
and the observer maintained between a 60° and 900 angle between the individual
and the speaker.

66

 

 

 

 

 

 

 

 

 

0 8 -
0.7 4
.".
g 0.6 -
is;
m 0.5 ‘1
I.—
° 04 - +N-K
g .-.-.
'8 0.3 - +N-NK
g ---O---NN-NK
a. 0.2 -
0.1 -
0 . . 4 ,
Pre-Playback Post-Playback
Pefiod Peflod

 

 

 

Figure 3.2. Probability of the focal individual rattling during the pre and post playback
observational periods of a playback trial. Responses are grouped based on the
relationship between the focal individual and the individual used in the playback
recording being neighbors (N) or non-neighbors (NN) in the pre-observational phase, and
grouped based on being neighbors or non-neighbors and kin (K) or non-kin (NK) in the
post-Observational phase. Data are presented as mean.

67

 

0.6 .

0.4 -

 

0.3 a N

-~o-NN

 

 

 

 

Probability of Barking

0.1 -

 

0.0

 

Pre-Playback I Post-Playback
Pefiod Pefiod

 

 

 

Figure 3.3. Probability of the focal individual barking during the pre and post playback
observational periods of a playback trial. Responses are grouped based on the
relationship between the focal individual and the individual used in the playback
recording being neighbors (N) or non-neighbors (NN). The data are presented as mean.

68

CHAPTER FOUR: GENERAL CONCLUSION

Early studies of red squirrel vocalizations led researchers (Smith, 1978; Lair,
1990) to believe that the rattles of red squirrels could have signature signals or be used to
identify one another. The analysis of our principal components Showed that a significant
amount Of the variation in rattle structure is due to the identity of the rattling squirrel.
This result is consistent with my hypothesis that signature Signals exist in the rattles of
red squirrels. This identity effect accounted for between 39% and 54% of the variation I
saw in rattle structure.

The dimensional reduction of the dataset yielded four variables that explained
94% of the variance in rattles. Principal component one explained 33% of the variance,
but the principal component was not significantly influenced by any other variable
besides identity, and even then it was the lowest of the four principal components (39%).
The second and third principal components were significantly influenced by the age of
the calling individual. Juvenile squirrels had rattles with syllables per second and lower
fundamental frequencies, which is opposite of what has been found previously (e.g.
Darwin, 1872; Morton, 1977; Sousa-Lima et al. 2002, Theis et al. 2007). Currently there
is no explanation as to why juveniles would have lower fundamental frequencies.
Possible explanations could include: 1) red squirrels have an unique ontogeny in their
vocalizations or 2) there is more information present in the harmonics of rattles that I did
not examine. Only future work on the ontogeny or harmonics in red squirrels will be able
to determine this.

The fourth principal component was explained by the behavior context in which

the vocalization was emitted in when I looked at the data set including adults and

69

juveniles, but these effects went away when I looked at the adult calls. Even though age
was not a Significant predictor of PC4 in the combined adult and juvenile data set, the
removal of the juvenile rattles altered the model of the principle component score. This
could be because the juvenile vocalizations recorded in the provoked state (n = 4) were
driving principle component four, therefore leading me to believe that all squirrels rattle
longer in a provoked state, when it is possibly only juveniles.

The motivation-structural rule (Morton, 1977) predicts that vocalizations used in
aggressive contexts, such as provoked circumstances, would have lower fundamental
frequencies and modulations, sounding more tonal, but this was not the case with my
data. The current study shows that red squirrels do not follow the MS-Rule. Morton
proposed his motivation-structural rule for short range communications, but conjectured
that long range vocalizations might also contained signature signals too. F eighny et al.
(2006) found that long range vocalizations of elk follows the motivation-structural rule,
but the results of my study Show this is not true for all mammals. While there are
occurrences of the motivation-structural rule in long range mammalian communications,
each species will need to be independently evaluated to determine if the vocalizations
follow the MS-Rule.

Identity plays a large role in explaining the variation in rattles we see in the study
population. Having an individual-specific variation in a signal is a key requirement for
individual recognition (Sherman et al., 1997). Individual recognition allows an individual
to identify specific kin, neighbors, and mates (Sherman et al., 1997; Tibbits and Dale,
2007). In this study I have shown definitively that red squirrels exhibit kin recognition

and that they potentially recognize other classes of individuals (such as neighbors) or

70

even specific individuals. The results of my playback study showed that individuals were
significantly less likely to rattle in response to the rattle of a related individual than they
were an unrelated individual; this different response in behavior based on relatedness is
evidence for kin recognition (Sherman et al., 1997). When I compared our preliminary
Observation period results to the post-playback Observation period results, we can also
conclude that an individual is no more likely to rattle at a related individual than it is to
spontaneously rattle when there is no stimulus involved (Figure 3.2).

One area of kin recognition that this study was not able to completely address was
the mechanism through which red squirrels recognize kin. Based on the success of cross-
fostering experiments (McAdam and Boutin, 2003) we can assume that site-specific
recognition occurs in the natal nest prior to emergence of the pups, but this most likely
breaks down once the pups emerge and disperse. We cannot determine whether prior
association or phenotype matching is the primary mechanism for recognition in red
squirrels because the majority Of my playback trials were done with mother-offspring and
maternal sibling pairs. These groups have the potential to use either mechanism for kin
recognition. Future work involving father-offspring or paternal siblings could provide
insight into kin recognition in red squirrels and will be able to determine if they use prior
association, phenotype matching, or both to identify their kin. Should phenotype
matching occur in red squirrels, the most likely source of this matching will be in the
frequency or harmonic elements of the calls and any future research to investigate the
mechanism of kin recognition of red squirrel should begin their search in the fundamental

frequencies and harmonics of the rattles.

71

4.1 References:

Darwin, C. 1872. The expressions of the emotions in man and animal. John Murray,
London.

Feighny, J. A., K. E. Williamson, and J. A. Clarke. 2006. North American elk bugle
vocalizations: male and female bugle call structure and context. Journal of
Mammalogy, 87(6), 1072-1077.

Lair, H. 1990. The calls Of the red squirrel: a contextual analysis of functions. Behaviour.
115 (3-4), 254-281.

McAdam, A. G., S. Boutin, A. K. Sykes, and M. M. Humphries. 2007. Life histories of
female red squirrels and their contributions to population growth and lifetime
fitness. Ecoscience, 14(3), 362-369.

Morton, E. S. 1977. Occurance and significance Of motivational-structural rules in some
birds and mammal sounds. American Naturalist. 111. 855-869

Sherman, P. W., H. K. Reeve, and D. W. Pfennig. 1997. Recognition systems. In
Behavioural ecology: an evolutionary approach (J. R. Krebs and N. B. Davies,
eds.). Blackwell Scientific Publications, Oxford, United Kingdom. 69—96

Smith, C. C. 1978. Structure and function of the vocalizations of tree squirrels
(T amiasciurus). Journal of Mammalogy. 59, 793-808.

Sousa-Lima, R. R., A. P. Pagalia, G. A. B. F DaFonseca. 2002. Signature information
and individual recognition in the isolation calls Of Amazonian manatees,
T richechus inunguis (Mammalia: Sirenia). Animal Behavior. 63, 715-724.

Tibbits, E. A. and J. Dale. 2007. Individual recognition: it is good to be different. Trends
in Ecology and Evolution, 22 529-537.

Theis, K. R., K. M. Greene, S. R. Benson-Amram, and K. E. Holekamp. 2007. Sources of
variation in the long-distance vocalizations of spotted hyenas. Behaviour. 1 14, 557-584

72

     

IIIIIIIIIIIIIIIIIIII