VARIATION AMONG FREE-LIVING SPOTTED HYENAS
IN THREE PERSONALITY TRAITS
By
Kathryn Cushing Shaw

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Zoology
Ecology, Evolutionary Biology and Behavior
2012

ABSTRACT
VARIATION AMONG FREE-LIVING SPOTTED HYENAS IN THREE PERSONALITY TRAITS
By
Kathryn Cushing Shaw

Inter-individual differences in behavior, termed “animal personality,” are often
consistent over time and across contexts, and can be significantly related to fitness. Most
studies of animal personality are conducted in the laboratory on captive animals or with
experimental protocols on wild animals; few studies have used observational data from freeliving animals to examine personality. In this dissertation, I use longitudinal data collected
under naturalistic conditions to study personality in a wild population of spotted hyenas
(Crocuta crocuta). This study is the first to examine the stability, the proximate causes, and the
fitness implications of consistent inter-individual differences in a large, free-living carnivore.
I investigated three personality traits that are particularly pertinent to survival and
reproduction in spotted hyenas: boldness, aggressiveness, and sociability. Lions are a major
source of mortality for hyenas, yet hyenas regularly interact with them directly due to the
potential benefits of acquiring food. Because hyenas must balance the risk of injury or death
with the benefits of resource acquisition, lion-hyena interactions offer a promising way to study
boldness. Aggression is frequent and easily observable in hyenas; moreover, it is tightly
coupled with access to food, which is a major determinant of reproductive success in this
species. Finally, hyenas vary in their propensity to engage in social behavior with conspecifics,

and sociability has been shown to affect fitness in species with societies that are remarkably
similar in size and structure to those of spotted hyenas.
I found significant differences among individual hyenas in all three traits. However,
whereas both sexes exhibited consistent inter-individual differences in aggressiveness and
sociability, males were less consistent than females in their boldness; this may be due to their
low social status and reduced access to resources. Attributes of the individual (e.g. social rank)
and situational factors (e.g. seasonal prey availability) affected all three personality traits.
Heritability and maternal effects were significant, but small, for both boldness and
sociability. A much larger proportion of the variation in aggressiveness could be attributed to
genetic and maternal effects, supporting previous research linking hormone exposure in utero
to aggressive behavior later in life. Whereas both boldness and aggressiveness were stable
across age classes, sociability changed across ontogeny and the nature of this change varied
with social rank. The advantages of associating with dominant hyenas, such as increased
feeding tolerance, may be especially important for low-ranking individuals. Interestingly, the
fitness benefits of sociability and aggression also varied with rank; high rates of aggression and
sociability enhanced the reproductive success of low-ranking hyenas more than that of highranking hyenas. For low-ranking hyenas, gaining feeding tolerance via sociability or increasing
resource-holding potential via aggressiveness may significantly enhance reproductive success.
Both boldness and sociability were linked to survival, but these traits affected longevity
in different ways. Highly social hyenas lived longer than those that were less social; meanwhile,
selection on boldness was stabilizing, favoring hyenas with intermediate boldness values that
struck a balance between the benefits of risky behavior and the risks of injury and death.

Copyright by
Kathryn Cushing Shaw
2012

ACKNOWLEDGEMENTS

First and foremost, I thank my advisor, Dr. Kay Holekamp, for her support and
encouragement throughout my time at Michigan State University. She has been incredibly
generous with her knowledge, her time, and her resources, and I consider myself extremely
fortunate to have been part of her lab. Kay expects rigorous and independent work from her
students, but she is always willing to offer help and words of wisdom to make us better
scientists and collaborators. My time in the field working on the Mara Hyena Project was an
unforgettable experience for which I am constantly grateful. Kay has been very accommodating
to my personal life and she has been willing to advise me long-distance, even though it often
made her life more complicated.
The other members of my committee, Dr. Fred Dyer, Dr. Tom Getty, and Dr. Laura
Smale, have also been vital to my success. Laura has been a necessary voice of reason and
rationality, and she has worked very hard to make sure I was set up for success with my
dissertation project. Tom has been incredibly supportive of my chosen career path, and I thank
him for being so encouraging about my non-academic aspirations. I am very grateful to Fred for
his guidance and direction as Chair of the Zoology Department.
This work would not have been possible without all the past and present Mara Hyena
Project lab members and research assistants. Each and every one of them has spent long and
tireless hours in the lab and the field, providing me with the data for this dissertation. Genetic
data from Katy Califf, Russ Van Horn, and Heather Watts were especially helpful for my
analyses. Jenn Smith has been incredibly generous with her data and her time, and I am very
v

grateful for the guidance she has given me along the way. Both Cheryl Vanier and Eli Swanson
have helped me with thorny statistical issues, and I can’t thank them enough for putting up
with my incessant questions.
Several lab members have been not only instrumental to my research, but also great
friends to me, including Sarah Benson-Amram, Leslie Curren, Stephanie Dloniak, Audrey
DeRose-Wilson, Andy Flies, Dave Green, Julia Greenberg, Sarah Jones, Joe Kolowski, Jenn Smith,
Greg Stricker, Eli Swanson, Jaime Tanner, and Marc Wiseman. In particular, Katy Califf has been
a constant source of support and laughter through both great times and difficult ones over the
past 6 years; I can’t imagine having gone through this experience without her.
I also thank my numerous undergraduate volunteers, including Maria Bunchman,
Megan Champoux, Rebecca Champney, J.B. Jones, Angie Lane, Danielle Miles, Chris Mirque,
Mike O’Callahan, Kyle Salvati, Jennifer Seid, and Ashleigh Winkelmann, for the many tedious
hours they spent poring over the notebooks, and for being willing to work so independently
while I was off-campus.
I could not have asked for better companions in the field than Audrey DeRose-Wilson,
Dave Green, and Jeff Smith. We had our share of incredible experiences and complete
catastrophes, but having them around made even our misadventures tolerable and memorable.
My time in Kenya would have been impossible without help from John Keshe, James Kerembe,
Stephen and Lesingo Nairori, and Philomen Naiguran. I am so grateful for the kindness and
generosity they showed me while I was in the field. In particular, Philomen was an incredible
cook, handyman, and friend, and I consider myself very lucky to have met him and his family.

vi

My undergraduate thesis advisor, Dr. Ethan Temeles, was very influential in my decision
to pursue graduate school. He taught me the ropes of field research, and I learned so much
from his meticulous work ethic and his passion for research.
Michigan State University’s College of Natural Sciences, Graduate School, Department of
Zoology, and EEBB Program, as well as Amherst College and the National Science Foundation,
have all generously provided me with funding during the course of my dissertation. I also thank
Kenya Wildlife Service, the Office of the President of Kenya, Narok Country Council, and the
Mara Conservancy for providing permission to conduct this research.
I certainly would not be in this position without the love and encouragement from my
parents, Wendy and Stephen Shaw. In particular, I thank my father for instilling in me a love of
science, and my mother for braving bat-filled caves and sleeping in a tent on the Mara to show
her support. I thank my brother Andy Shaw for always being there when I needed a kind word
or a kick in the pants. Finally, Toshi Yoshida has been absolutely invaluable to me during this
process. When things have been difficult, he has motivated me to continue and inspired me to
succeed. I am incredibly grateful for his support, generosity, and patience throughout this
process, and for being such an important part of my life.

vii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................................................ ix
LIST OF FIGURES .............................................................................................................................. xi
CHAPTER 1
GENERAL INTRODUCTION ............................................................................................................... 1
Overview of chapters .......................................................................................................... 8
Literature cited .................................................................................................................. 12
CHAPTER 2
INDIVIDUAL DIFFERENCES IN BOLDNESS AMONG SPOTTED HYENAS .......................................... 19
Introduction ...................................................................................................................... 19
Methods ............................................................................................................................ 24
Results ............................................................................................................................... 33
Discussion.......................................................................................................................... 41
Literature cited .................................................................................................................. 47
CHAPTER 3
INDIVIDUAL DIFFERENCES IN AGGRESSIVENESS AMONG SPOTTED HYENAS .............................. 55
Introduction ...................................................................................................................... 55
Methods ............................................................................................................................ 60
Results ............................................................................................................................... 73
Discussion.......................................................................................................................... 82
Appendix ........................................................................................................................... 88
Literature cited .................................................................................................................. 94
CHAPTER 4
INDIVIDUAL DIFFERENCES IN SOCIABILITY AMONG SPOTTED HYENAS ..................................... 102
Introduction .................................................................................................................... 102
Methods .......................................................................................................................... 106
Results ............................................................................................................................. 116
Discussion........................................................................................................................ 123
Appendix ......................................................................................................................... 128
Literature cited ................................................................................................................ 131

viii

LIST OF TABLES

Table 2.1. Fixed effects predicting average minimum distances from hyenas to lions in sessions
with at least one lion present and at least three minimum distances for known hyenas ........... 34
Table 2.2. Fixed effects predicting the minimum distance separating each male hyena from
lions ............................................................................................................................................... 35
Table 2.3. Fixed effects predicting the minimum distance separating each female hyena from
lions ............................................................................................................................................... 35
Table 3.1. DIC values yielded by separate MCMCglmm models run with NFAs and FAs emitted
by adults (≥24 months of age) of each sex. DIC values are given for models with only a random
effect of the session identifier, and for models run with both a session identifier and an
aggressor identifier. For each context in each sex, the better model is marked with an
asterisk .......................................................................................................................................... 75
Table 3.A1. Fixed effects predicting the total FA count in each session. Only aggressive acts
directed by adults (≥24 months of age) down the hierarchy toward other adults are included.
Significant fixed effects are marked with asterisks ...................................................................... 89
Table 3.A2. Fixed effects predicting the total NFA count in each session. Only aggressive acts
directed by adults (≥24 months of age) down the hierarchy toward other adults are included.
Significant fixed effects are marked with asterisks ...................................................................... 89
Table 3.A3. Fixed effects predicting the average FA intensity in each session. Only aggressions
directed by adults (≥24 months of age) down the hierarchy to other adults are included.
Significant fixed effects are marked with asterisks ...................................................................... 90
Table 3.A4. Fixed effects predicting the average NFA intensity in each session. Only
aggressions directed by adults (≥24 months of age) down the hierarchy to other adults are
included. Significant fixed effects are marked with asterisks ...................................................... 90
Table 3.A5. Fixed effects predicting the FA count by each hyena in each session. Only
aggressions directed by adults (≥24 months of age) down the hierarchy to other adults are
included, and all sessions that were shorter than 30 minutes or had no aggressions were
omitted. Significant fixed effects are marked with asterisks ....................................................... 91
Table 3.A6. Fixed effects predicting the NFA count by each hyena in each session. Only
aggressions directed by adults (≥24 months of age) down the hierarchy to other adults are
included, and all sessions that were shorter than 30 minutes or had no aggressions were
omitted. Significant fixed effects are marked with asterisks. ..................................................... 91
ix

Table 3.A7. Fixed effects predicting the intensity of FAs emitted by adult (≥24 months of age)
females. Significant effects are marked with asterisks ................................................................ 92
Table 3.A8. Fixed effects predicting the intensity of NFAs emitted by adult (≥24 months of age)
females. Significant effects are marked with asterisks ................................................................ 92
Table 3.A9. Fixed effects predicting the intensity of FAs emitted by adult (≥24 months of age)
males. Significant effects are marked with asterisks ................................................................... 92
Table 3.A10. Fixed effects predicting the intensity of NFAs emitted by adult (≥24 months of
age) males. Significant effects are marked with asterisks ........................................................... 92
Table 3.A11. DIC values yielded by separate MCMCglmm models predicting adult (≥24 months
of age) lifetime aggression rate and aggression intensity for each aggression context. DIC values
are given for the full model with all variance components including V A (additive genetic), VM
(maternal effect), and VI (permanent environment effect), as well as for models omitting each
variance component. Relevant fixed effects were retained in each model, and intensity models
also include a random term representing the session identifier ................................................. 93
Table 4.1. DIC values yielded by MCMCglmm models predicting the number of greetings
solicited in each observation session by adult (≥24 months of age) female and male hyenas.
DIC values are given for models with only a random effect of the session identifier, and for
models run with both a session identifier and an individual identifier. For each sex, the better
model is marked with an asterisk ............................................................................................... 117
Table 4.A1. Situational variables predicting the average greeting count of adults (≥24 months
of age) in each observation session ............................................................................................ 129
Table 4.A2. Fixed effects predicting the number of greetings solicited by each adult (≥24
months of age) female hyena in each observation session........................................................ 129
Table 4A.3. Fixed effects predicting the number of greetings solicited by each adult (≥24
months of age) male hyena in each observation session ........................................................... 130

x

LIST OF FIGURES

Figure 2.1. The percent of variation in boldness attributable to heritability, permanent
environment effects, maternal effects, and the remaining unexplained variation. Variances
were estimated with a MCMCglmm animal model. Values on the y-axis between 0.3 and 0.8
are omitted ................................................................................................................................... 36
Figures 2.2a and 2.2b. Predicted behavioral reaction norms for boldness measures across (a)
two age classes and (b) two categories of seasonal prey availability, as predicted by lmer
models. Each line represents how the behavior of an individual changes across categories.
Smaller relative boldness measures represent bolder individuals, since these measures are
based on minimum distances to lions .......................................................................................... 38
Figure 2.3. The relationship between lifetime relative boldness and longevity. Actual values
are displayed as dots, and the values predicted by the second-order regression model are
shown as a line, with 95% confidence intervals indicated with grey dotted lines ....................... 40
Figure 3.1. The percent of variation in lifetime food-related and non-food aggression rates
attributable to heritability and maternal effects, and the remaining unexplained variation.
Variances were estimated with MCMCglmm animal models ...................................................... 77
Figures 3.2a and 3.2b. Predicted behavioral reaction norms for aggression intensity across (a)
two age classes and (b) two aggression contexts. Each line represents how the behavior of an
individual changes across categories ............................................................................................ 79
Figure 4.1. For two measures of sociability, the percent of variation attributable to heritability,
maternal effects, and the remaining unexplained variation. Variances were estimated with
separate MCMCglmm animal models ........................................................................................ 118
Figure 4.2. Predicted behavioral reaction norms for the likelihood of greeting across two age
classes, as predicted by lmer models. Each line represents how the behavior of a female
changes acrss ontogeny. Smaller likelihoods of greeting represent less sociable individuals, as
these individuals are less likely to greet with other hyenas in a session ................................... 120
Figure 4.3. The relationship between group joining rate and longevity. Group joining rate was
calculated as the hourly rate at which adult females joined groups between 1996 and 2002.
Actual values are displayed as dots, and the values predicted by generalized linear model are
shown as a line ............................................................................................................................ 122

xi

CHAPTER 1
GENERAL INTRODUCTION

Understanding animal behavior has been a major component of biological research for
centuries; naturalists and scientists have long been interested in how and why animals behave
the way they do (e.g. Darwin 1872; Aristotle 1929; Hinde 1970). Until recently, much of the
focus of behavioral research on non-human animals has been at the level of species, rather
than at the level of the individual (Jolly 1966; Geist 1974; Yasukawa & Searcy 1982; Thierry
1985). However, in recent years researchers have begun to document significant differences in
behavior among individuals within a single population (e.g. Wilson et al. 1994; Reale et al. 2000;
Dall et al. 2004; Carere & Eens 2005). Moreover, these differences are often consistent over
time and across contexts, and can have significant implications for fitness (Duckworth 2006;
Johnson & Sih 2007; Silk et al. 2010; Vainikka et al. 2010). This type of variation has been
referred to as “individual differences,” “temperament,” “coping styles,” “behavioral
syndromes,” and “animal personality” (e.g. Macdonald 1983; Capitanio 1999; Koolhaas et al.
1999; Reale & Festa-Bianchet 2003; Johnson & Sih 2005).
Evidence for personality has been found in species as diverse as squid, sunfish, great
tits, marmots, and chimpanzees (Armitage 1986a; Coleman & Wilson 1998; Dingemanse et al.
2003; Sinn & Moltschaniwskyj 2005; Anestis et al. 2006). The literature on personality in
humans and their companion animals is also extensive (e.g. Eysenck 1953; Wilsson & Sundgren
1997; Ebstein et al. 2002). In most species in which personality has been studied, individuals
tend to vary across similar dimensions, such as aggressiveness, boldness, activity, sociability,
1

and exploration (for reviews, see Sih et al. 2004, Reale et al. 2007). However, in species where
consistent individual differences have been found, males and females may differ in
repeatability. For instance, male guppies are less consistent in their boldness than are female
guppies (Harris et al. 2010), yet theory predicts that males should be more consistent in
aggressive behavior than females due to high testosterone levels (Wingfield 1994). In a metaanalysis, Bell et al. (2009) found that differences between males and females in behavioral
repeatability vary based on the species and the traits being studied; this suggests that, in terms
of consistency, the optimal strategy for each sex may be determined separately in each species
based on specific aspects of its environment, life history, and morphology.
One important question addressed by personality research is how stable traits are, both
over time and across different contexts. For instance, whereas Sinn et al. (2008) found that
juvenile boldness predicted adult boldness in squid, studies in other species suggest that
personality may change across ontogeny via experience, physiological development, or
changing environmental conditions (Bell & Stamps 2004; Dingemanse et al. 2009; Stamps &
Groothuis 2010). There is also great debate as to the effect of context on personality; for
instance, are individuals that are aggressive in one context also aggressive in another (i.e.
domain-generality), or is aggressiveness in different contexts unrelated (i.e. contextspecificity)? Some studies have suggested that personality traits in humans and other animals
are context-specific (Kagan 2003; Sinn & Moltschaniwskyj 2005; Tanner & Adler 2009), but
there are also several well-established examples of domain-generality in personality (e.g.
Johnson & Sih 2007).

2

Artificial selection for personality traits has been very successful; researchers have been
able to breed strains of very social mice and highly exploratory great tits (Carere et al. 2005;
Sankoorikal et al. 2006), suggesting that heritability plays a significant role in some personality
traits. A heritability estimate is the proportion of the observed phenotypic variance in a trait
that can be explained by genetics; heritability has been estimated to be as high as 0.61 for
aggression in vervet monkeys (Fairbanks et al. 2004) and 0.89 for reactivity in dumpling squid
(Sinn et al. 2006). However, not all traits appear to be equally heritable in all species; for
instance, less than one percent of the variation in aggressiveness observed in a population of
sticklebacks could be explained by genetics (Bell 2005).
Another major focus of research on animal personality in recent years has been the
relationship between fitness and inter-individual differences in behavior. Many studies suggest
that personality has significant implications for reproductive success and survival. Among some
primates, females with strong social ties to other group members have greater longevity than
less social females (Silk et al. 2010). However, extreme trait values may also have negatively
correlated with fitness; Packer et al. (1995) found that aggressive, high-ranking female baboons
had reduced fertility and a higher incidence of miscarriages than did less dominant females. In
other species, the consequences of personality may vary with fluctuating environmental
conditions. For instance, in bighorn sheep, shy females suffer higher mortality than do bold
females in years when predation pressure is high, but not in years of reduced predation (Reale
& Festa-Bianchet 2003).
Here, I present a study of personality in the spotted hyena (Crocuta crocuta). Since
1988, the Mara Hyena Project has continuously observed a group of free-living hyenas in the
3

Masai Mara National Reserve in Kenya. Collaborators on this project have collected a variety of
behavioral, demographic, morphological, genetic, and physiological data on these hyenas. This
rich longitudinal dataset has enabled me to examine the heritability, stability, and fitness
consequences of behavioral traits under naturalistic conditions. To date, personality has most
often been examined in captive individuals in the laboratory, or with experimental protocols on
wild animals (Armitage 1986a; Wilson et al. 1993; Brown et al. 2007; Fox et al. 2009; Taylor et
al. 2012). However, Wilson et al. (1994) suggest that the behavior of an individual in the
laboratory may not accurately reflect its behavior in the wild. Whereas lab-based and
experimental studies have the advantage of enhanced control, direct observational data
documenting personality traits under strictly natural conditions offers a means to assess
simultaneously how such traits vary among free-living individuals, and how they affect fitness.
Only a small number of studies to date have been able to assess personality with the naturallyoccurring behavior of free-living animals (e.g. Armitage 1986b; McPhee & Quinn 1998; Silk et
al. 2010).
In this dissertation, I examine three personality traits in spotted hyenas: boldness,
aggressiveness, and sociability. These traits are all characterized by frequent and easily
observable behavior in this species, and each has direct implications for fitness. Other
personality traits that have been identified in various species, such as activity and docility, are
worthy of future study, but their potential consequences for survival and reproduction in
hyenas are less clear.
Boldness is a particularly important personality trait to understand, because predation is
one of the most significant selection pressures in determining behavior (Lima & Dill 1990; Lima
4

1998). Although lions are a major source of hyena mortality, hyenas do not actively avoid lions
(Watts & Holekamp 2009). Unlike most other species, hyenas have significant incentives to
interact directly with their predators; they are able to steal food from lions in some situations,
and are often forced to defend their own kills from lions (Kruuk 1972; Mills 1990; Watts &
Holekamp 2009). Hyenas therefore face a particularly interesting tradeoff with respect to lions,
balancing the risk of injury or death with the potential benefits of acquiring or retaining
possession of food. In a larger framework, understanding individual differences in boldness
among large carnivores is important in several ways. Shy individuals are prone to being
undersampled (Wilson et al. 1994), so it is important to understand in what types of species
shy-bold continuums exist, and what sex and age classes are most likely to exhibit consistent
differences in boldness. It is also hypothesized that bolder carnivores are more likely to
become “problem individuals” that kill livestock and pets (Linnell et al. 1999; Darrow & Shivik
2009). Studies of boldness in large carnivores are necessary to clarify the processes by which
individuals become “problems,” and to identify which individuals in each species are
particularly prone to livestock and pet depredation.
Aggressiveness among animals is a particularly important trait to understand, since in
many species, aggression facilitates access to resources. Consistent inter-individual variation in
aggressiveness has been documented in several species, including three-spined sticklebacks
(Bakker 1986), red squirrels (Taylor et al. 2012), and vervet monkeys (Fairbanks et al. 2004).
However, aggressiveness has not been examined in any species with sex role reversals, where
females are socially dominant to, and often more aggressive than, males. In these species, the
proximate causes and the fitness consequences of aggressive behavior may be very different
5

than those in species where males are dominant. Spotted hyena societies are femaledominated, and female hyenas exhibit more frequent and intense aggressive behavior than
males do (Kruuk 1972; Van Meter 2009). Among hyenas, aggressive behavior occurs in several
contexts, including food-related and non-food related situations; the occurrence of aggressive
behavior in these different contexts permits a study of context-specificity in aggression.
Additionally, since access to resources drives reproductive success in hyenas (Holekamp &
Smale 1996), aggression may be significantly related to fitness in this species.
Whereas sociability has been well-studied in humans and other primates, far fewer
studies have examined this personality trait in other taxa. Previous research suggests that
sociability is significantly related to fitness; individuals with stronger social bonds and those
that engage in social behavior more often tend to have better reproductive success and
increased longevity, compared to less social individuals (Silk et al. 2003; Cote et al. 2008;
Cameron et al. 2009; Silk et al. 2010; Schulke et al. 2010). Spotted hyenas are highly gregarious
and live in large fission-fusion groups, where subgroups change in size and composition on
short time scales (Holekamp et al. 2006). These social dynamics, in addition to the frequent
affiliative behaviors demonstrated by hyenas, offer a promising way to study inter-individual
variation in sociability. Previous research on hyenas has demonstrated significant advantages
to engaging in social behavior (Holekamp et al. 1997; Boydston et al. 2001; Smith et al. 2007;
Smith et al. 2011), suggesting that sociability may be positively correlated with survival and
reproductive success in this species.
Mine is not the first study of inter-individual differences in the behavior of hyenas;
Gosling (1998) studied personality in a group of captive spotted hyenas. In this study, observers
6

who were familiar with the hyenas were asked to rate them on several personality traits. The
ratings for many traits, including “affiliative,” “bold,” “aggressive,” “nervous,” and “nurturant,”
were highly reliable, with inter-observer reliabilities averaging 0.71. These results suggest that
hyenas vary along several behavioral axes, and that these differences can be reliably quantified
by different observers. However, the study does have some shortcomings. For instance, the
trait ratings were not independent because the polled observers worked together closely, and
had likely discussed these hyenas in detail before the polling commenced. Additionally, the
hyenas were rated on their past behavior, so it is possible that particularly salient memories
affected observers’ ratings. Finally, although Gosling went to great lengths to define each trait,
the setup of the study made it impossible to eliminate subjectivity.
Since this early research, other studies have documented inter-individual differences in
the behavior of hyenas. Via playback experiments, Watts et al. (2010) determined that
individual hyenas varied in the way they responded to lion roars. The authors also found a
correlation between boldness and vigilance behavior; hyenas that took more risks than other
individuals in response to the playback also tended to be more vigilant than other hyenas.
These differences were context-specific, seen only in the context of interactions with lions, not
at dens or in situations where lions were absent. Pangle (2008) reported evidence of consistent
differences in vigilance behavior among adult hyenas, but not among juveniles. Finally, both
Dloniak et al. (2006) and Van Meter (2009) found correlations between circulating maternal
fecal androgen levels during gestation and the aggression rates of offspring, suggesting that
hormone-mediated maternal effects play a significant role in shaping individual differences in
aggressive behavior among hyenas.
7

While this research has been instrumental in laying the groundwork for personality
research in spotted hyenas, no previous study has synthesized genetic, behavioral, and fitness
data to understand inter-individual variation in this species in a broader sense. My dissertation
work is the first to examine the stability, the proximate causes, and the fitness implications of
consistent inter-individual differences in a large, free-living carnivore.

Overview of chapters
Each subsequent chapter in this dissertation is a systematic study of a different
personality trait in a group of free-living spotted hyenas. In each chapter, I first investigate the
situational and individual variables that contribute to the trait, and determine whether there is
evidence of consistent inter-individual differences in the trait. Then, I determine whether
genetics or maternal effects contribute significantly to phenotypic variation in the trait, and
whether the trait is stable across ontogeny. Finally, I assess whether there is a relationship
between the trait and various measures of fitness.
In Chapter Two, I investigate boldness in hyenas, using the distances maintained
between hyenas and lions as a measure of boldness. I found that, whereas female hyenas
exhibited consistent inter-individual differences in boldness, males were more plastic in their
behavior around lions. Although this bears further research, sex differences in plasticity may be
related to social rank and access to resources in this species. Situational variables were very
important in determining how closely hyenas approached lions, and genetic and maternal
effects only played a small role in boldness. Boldness in juveniles was a good predictor of
boldness during adulthood, indicating that for hyenas, this trait is determined relatively early in
8

life. Although I found no correlation between boldness and reproductive success, I did find that
it was significantly related to longevity; hyenas with particularly low and high boldness
measures died at younger ages than did those with intermediate boldness values. Whereas the
boldest animals may be especially prone to injury and death due to their propensity to take lifethreatening risks, the shyest animals may fail to reap the benefits of risky behavior.
In the third chapter, I study the frequency and intensity of aggressive behavior in two
contexts: food-related aggressions and non-food-related aggressions. Both males and females
showed consistent inter-individual differences in the intensity of aggression, but hyenas were
not consistent across repeated observations in the frequency of either type of aggression.
However, when averaged over the lifetime of individuals, there were clear differences among
hyenas in both food-related and non-food-related aggressive behavior. These differences were
highly heritable and subject to large maternal effects, echoing previous research indicating that
hormone exposure in utero influences the aggressive behavior of hyenas later in life (Dloniak et
al. 2006; Van Meter 2009). This early determination of aggressive behavior is also supported
by my finding that the intensity of aggressive behavior early in life predicts aggressive behavior
during adulthood. Interestingly, aggressive behavior appears to be context-specific; hyenas
that emitted high-intensity aggressive acts over food did not necessarily emit high-intensity
aggressive acts in other contexts, suggesting different functions, and mediating mechanisms
for, different types of aggressive behavior. Whereas aggressiveness in some species decreases
some aspects of fitness (e.g. Packer et al. 1995), aggressiveness in hyenas is related to
enhanced reproductive success. I found that both food-related and non-food related
aggressiveness are positively correlated with reproductive success, although the relationship
9

between food-related aggressiveness and reproductive success depended on maternal rank. I
also found that the offspring of females with high food-related aggression rates were able to
feed longer at kills among conspecifics than the offspring of mothers that aggressed less
frequently over food. Because access to resources drives variation in reproductive success in
female hyenas (Holekamp & Smale 1996), aggressiveness may benefit females and their
offspring by allowing them to more successfully monopolize food resources.
In Chapter Four I investigate sociability, using several different trait measures such as
the frequency at which hyenas solicit greetings and the rate at which they join subgroups. Both
males and females were consistent in their greeting behavior across repeated observations.
Greeting rate and joining rate were both significantly heritable, but only a small portion of the
variation in each of these behaviors could be explained by maternal effects. Interestingly,
whereas I found in previous chapters that both boldness and aggressiveness are stable across
ontogeny, sociability changes in different ways across age classes in various individuals.
Changes in sociability across ontogeny depend on social rank; as they reach reproductive
maturity, low-ranking hyenas, but not high-ranking hyenas, tend to increase in their tendency
to solicit greetings. The advantages of associating with dominant hyenas, such as increased
feeding tolerance (Smith et al. 2007), may be especially important for low-ranking hyenas.
Although social rank is the major driver of reproductive success among hyenas (Holekamp &
Smale 1996; Swanson et al. 2011), inter-individual differences in sociability also contribute to
variation in reproductive success. I found that the rate at which females joined groups was
positively correlated with various fitness measures, and the benefits for reproductive success
were especially strong among low-ranking hyenas. Sociability has similar fitness benefits in
10

other species (e.g. Silk et al. 2003; Silk et al. 2010), and these positive correlations may be due
to advantages conferred by strong cooperative and coalitionary partnerships (Smith et al.
2007).
The research in this dissertation was made possible by collaboration, both in terms of
data collection over the last two decades, and also in support provided to me more recently.
Because these chapters were prepared in manuscript form and this project was a work of
collaboration, I use first person plural in the remainder of this dissertation rather than first
person singular.

11

LITERATURE CITED

12

LITERATURE CITED
Anestis, S. F., Bribiescas, R. G., & Hasselschwert, D. L. 2006. Age, rank, and personality effects
on the cortisol sedation stress response in young chimpanzees. Physiology & Behavior, 89,
287-294.
Aristotle. 1929. The physics. London: Heinemann.
Armitage, K. B. 1986a. Individual differences in the behavior of juvenile yellow-bellied
marmots. Behavioral Ecology and Sociobiology, 18, 419-424.
Armitage, K. B. 1986b. Individuality, social behavior, and reproductive success in yellow-bellied
marmots. Ecology, 67, 1186-1193.
Bakker, T. C. M. 1986. Aggressiveness in sticklebacks (Gasterosteus aculeatus): a behaviorgenetic study. Behaviour, 98, 1-144.
Bell, A. M. 2005. Behavioural differences between individuals and two populations of
stickleback (Gasterosteus aculeatus). Journal of Evolutionary Biology, 18, 464-473.
Bell, A. M., Hankison, S. J., & Laskowski, K. L. 2009. The repeatability of behaviour: a metaanalysis. Animal Behaviour, 77, 771-783.
Bell, A. M., & Stamps, J. A. 2004. Development of behavioral differences between individuals
and populations of sticklebacks, Gasterosteus aculeatus. Animal Behaviour, 68, 1339-1348.
Boydston, E. E., Morelli, T. L., & Holekamp, K. E. 2001. Sex differences in territorial behavior
exhibited by the spotted hyena (Hyaenidae, Crocuta crocuta). Ethology, 107, 369-385.
Brown, C., Burgess, F., & Braithwaite, V. 2007. Heritable and experiential effects on boldness in
a tropical poeciliid. Behavioral Ecology and Sociobiology, 62, 237-243.
Cameron, E. Z., Setsaas, T. H., & Linklater, W. L. 2009. Social bonds between unrelated females
increase reproductive success in feral horses . Proceedings of the National Academy of
Sciences , 106 , 13850-13853.
Capitanio, J. P. 1999. Personality dimensions in adult male rhesus macaques: Prediction of
behaviors across time and situation. American Journal of Primatology, 47, 299-320.
Carere, C., Drent, P. J., Privitera, L., Koolhaas, J. M., & Groothuis, T. G. G. 2005. Personalities in
great tits, Parus major: stability and consistency. Animal Behaviour, 70, 795-805.
Carere, C., & Eens, M. 2005. Unravelling animal personalities: How and why individuals
consistently differ. Behaviour, 142, 1149-1157.
13

Coleman, K., & Wilson, D. S. 1998. Shyness and boldness in pumpkinseed sunfish: individual
differences are context-specific. Animal Behaviour, 56, 927-936.
Cote, J., Dreiss, A., & Clobert, J. 2008. Social personality trait and fitness. Proceedings of the
Royal Society B: Biological Sciences, 275, 2851-2858.
Dall, S. R. X., Houston, A. I., & McNamara, J. M. 2004. The behavioural ecology of personality:
consistent individual differences from an adaptive perspective. Ecology Letters, 7, 734-739.
Darrow, P. A., & Shivik, J. A. 2009. Bold, shy, and persistent: Variable coyote response to light
and sound stimuli. Applied Animal Behaviour Science, 116, 82-87.
Darwin, C. 1872. The Expression of the Emotions in Man and Animals. Chicago, IL: The
University of Chicago Press.
Dingemanse, N. J., Both, C., van Noordwijk, A. J., Rutten, A. L., & Drent, P. J. 2003. Natal
dispersal and personalities in great tits (Parus major). Proceedings of the Royal Society,
Series B, 270, 741-747.
Dingemanse, N. J., Van der Plas, F., Wright, J., Reale, D., Schrama, M., Roff, D. A., Van der Zee,
E., & Barber, I. 2009. Individual experience and evolutionary history of predation affect
expression of heritable variation in fish personality and morphology. Proceedings of the
Royal Society B: Biological Sciences, 276, 1285-1293.
Dloniak, S. M., French, J. A., & Holekamp, K. E. 2006. Rank-related maternal effects of
androgens on behaviour in wild spotted hyaenas. Nature, 440, 1190-1193.
Duckworth, R. A. 2006. Behavioral correlations across breeding contexts provide a mechanism
for a cost of aggression. Behavioral Ecology, 17, 1011-1019.
Ebstein, R. P., Zohar, A. H., Benjamin, J., & Belmaker, R. H. 2002. An update on molecular
genetic studies of human personality traits. Applied Bioinformatics, 1, 57-68.
Eysenck, H. J. 1953. The structure of human personality. New York: Methuen.
Fairbanks, L. A., Newman, T. K., Bailey, J. N., Jorgensen, M. J., Breidenthal, S. E., Ophoff, R. A.,
Comuzzie, A. G., Martin, L. J., & Rogers, J. 2004. Genetic contributions to social impulsivity
and aggressiveness in vervet monkeys. Biological Psychiatry, 55, 642-647.
Fox, R. A., Ladage, L. D., Roth II, T. C., & Pravosudov, V. V. 2009. Behavioural profile predicts
dominance status in mountain chickadees, Poecile gambeli. Animal Behaviour, 77, 14411448.

14

Geist, V. 1974. On the Relationship of Social Evolution and Ecology in Ungulates. American
Zoologist, 14, 205-220.
Gosling, S. D. 1998. Personality dimensions in spotted hyenas (Crocuta crocuta). Journal of
Comparative Psychology, 112, 107-118.
Harris, S., Ramnarine, I. W., Smith, H. G., & Pettersson, L. B. 2010. Picking personalities apart:
estimating the influence of predation, sex and body size on boldness in the guppy Poecilia
reticulata. Oikos, 119, 1711-1718.
Hinde, R. A. 1970. Animal behaviour: A synthesis of ethology and comparative psychology. 2nd
edn. New York: McGraw Hill.
Holekamp, K. E., Sakai, S. T., & Lundrigan, B. L. 2006. Social intelligence in the spotted hyena
(Crocuta crocuta). Philosophical Transations of the Royal Society (B), 362, 523-538.
Holekamp, K., & Smale, L. 1996. Rank and reproduction in the female spotted hyaena. Journal
of Reproduction, 108, 229-237.
Holekamp, K. E., Smale, L., Berg, R., & Cooper, S. M. 1997. Hunting rates and hunting success
in the spotted hyena (Crocuta crocuta). Journal of Zoology, 242, 1-15.
Johnson, J. C., & Sih, A. 2005. Precopulatory sexual cannibalism in fishing spiders (Dolomedes
triton): a role for behavioral syndromes. Behavioral Ecology and Sociobiology, 58, 390-396.
Johnson, J. C., & Sih, A. 2007. Fear, food, sex and parental care: a syndrome of boldness in the
fishing spider, Dolomedes triton. Animal Behaviour, 74, 1131-1138.
Jolly, A. 1966. Lemur Social Behavior and Primate Intelligence . Science , 153 , 501-506.
Kagan, J. 2003. Biology, Context, and Developmental Inquiry. Annual Review of Psychology, 54,
1-23.
Koolhaas, J. M., Korte, S. M., De Boer, S. F., Van Der Vegt, B. J., Van Reenen, C. G., Hopster,
H., De Jong, I. C., Ruis, M. A. W., & Blokhuis, H. J. 1999. Coping styles in animals: current
status in behavior and stress-physiology. Neuroscience and Biobehavioral Reviews, 23, 925935.
Kruuk, H. 1972. The spotted hyaena: a study of predation and social behavior. Chicago: Chicago
University Press.
Lima, S. L. 1998. Stress and Decision Making under the Risk of Predation: Recent Developments
from Behavioral, Reproductive, and Ecological Perspectives. In: Advances in the Study of

15

Behavior: Stress and Hormones, Vol Volume 27 (Ed. by A. Møller, M. Milinski, & P. Slater),
pp. 215-290. Academic Press.
Lima, S. L., & Dill, L. M. 1990. Behavioral decisions made under the risk of predation: a review
and prospectus. Canadian Journal of Zoology, 68, 619-640.
Linnell, J. D., Odden, J., Smith, M. E., Aanes, R., & Sweonson, J. E. 1999. Large carnivores that
kill livestock: do“ problem individuals” really exist. Wildlife Society Bulletin, 27, 698.
Macdonald, K. 1983. Stability of individual differences in behavior in a litter of wolf cubs (Canis
Lupus). Journal of Camparative Psychology, 97, 99-106.
McPhee, M. V., & Quinn, T. P. 1998. Factors affecting the duration of nest defense and
reproductive lifespan of female sockeye salmon, Oncorhynchus nerka. Environmental
Biology of Fishes, 51, 369-375.
Mills, M. G. L. 1990. Kalahari hyaenas: Comparative behavioral ecology of two species. London:
Unwin Hyman.
Packer, C., Collins, D. A., Sindimwo, A., & Goodall, J. 1995. Reproductive constraints on
aggressive competition in female baboons. Nature, 373, 60-63.
Pangle, W. M. 2008. Threat-sensitive behavior and its ontogenetic development in top
mammalian carnivores. Michigan State University.
Reale, D., & Festa-Bianchet, M. 2003. Predator-induced natural selection on temperament in
bighorn ewes. Animal Behaviour, 65, 463-470.
Reale, D., Gallant, B. Y., Leblanc, M., & Festa-Bianchet, M. 2000. Consistency of temperament
in bighorn ewes and correlates with behaviour and life history. Animal Behaviour, 60, 589597.
Reale, D., Reader, S. M., Sol, D., McDougall, P. T., & Dingemanse, N. J. 2007. Integrating
animal temperament within ecology and evolution. Biological Reviews, 82, 291-318.
Sankoorikal, G. M. V., Kaercher, K. A., Boon, C. J., Lee, J. K., & Brodkin, E. S. 2006. A mouse
model system for genetic analysis of sociability: C57BL/6J versus BALB/cJ inbred mouse
strains. Biological Psychiatry, 59, 415-423.
Schulke, O., Bhagavatula, J., Vigilant, L., & Ostner, J. 2010. Social Bonds Enhance Reproductive
Success in Male Macaques. Current biology, 20, 2207-2210.
Sih, A., Bell, A., & Johnson, J. C. 2004. Behavioral syndromes: an ecological and evolutionary
overview. Trends in Ecology & Evolution, 19, 372-378.
16

Silk, J. B., Alberts, S. C., & Altmann, J. 2003. Social Bonds of Female Baboons Enhance Infant
Survival. Science, 302, 1231-1234.
Silk, J. B., Beehner, J. C., Bergman, T. J., Crockford, C., Engh, A. L., Moscovice, L. R., Wittig, R.
M., Seyfarth, R. M., & Cheney, D. L. 2010. Strong and Consistent Social Bonds Enhance the
Longevity of Female Baboons. Current Biology, 20, 1359-1361.
Sinn, D. L., Apiolaza, L. A., & Moltschaniwskyj, N. A. 2006. Heritability and fitness-related
consequences of squid personality traits. Journal of Evolutionary Biology, 19, 1437-1447.
Sinn, D. L., Gosling, S. D., & Moltschaniwskyj, N. a. 2008. Development of shy/bold behaviour
in squid: context-specific phenotypes associated with developmental plasticity. Animal
Behaviour, 75, 433-442.
Sinn, D. L., & Moltschaniwskyj, N. A. 2005. Personality traits in dumpling squid (Euprymna
tasmanica): context-specific traits and their correlation with biological characteristics.
Journal of Comparative Psychology, 119, 99-110.
Smith, J., Memenis, S., & Holekamp, K. 2007. Rank-related partner choice in the fission–fusion
society of the spotted hyena (Crocuta crocuta). Behavioral Ecology and Sociobiology, 61,
753-765.
Smith, J. E., Powning, K. S., Dawes, S. E., Estrada, J. R., Hopper, A. L., Piotrowski, S. L., &
Holekamp, K. E. 2011. Greetings promote cooperation and reinforce social bonds among
spotted hyaenas. Animal Behaviour, 81, 401-415.
Stamps, J. A., & Groothuis, T. G. G. 2010. Developmental perspectives on personality:
implications for ecological and evolutionary studies of individual differences. Philosophical
Transactions of the Royal Society B: Biological Sciences, 365, 4029-4041.
Swanson, E. M., Dworkin, I., & Holekamp, K. E. 2011. Lifetime selection on a hypoallometric
size trait in the spotted hyena. Proceedings of the Royal Society B: Biological Sciences,
Tanner, C. J., & Adler, F. R. 2009. To fight or not to fight : context-dependent interspecific
aggression in competing ants. Animal Behaviour, 77, 297-305.
Taylor, R. W., Boon, A. K., Dantzer, B., Reale, D., Humphries, M. M., Boutin, S., Gorrell, J. C.,
Coltman, D. W., & McAdam, A. G. 2012. Low heritabilities, but genetic and maternal
correlations between red squirrel behaviours. Journal of Evolutionary Biology, 25, 614-624.
Thierry, B. 1985. Patterns of agonistic interactions in three species of macaque (Macaca
mulatta, M fascicularis, M tonkeana). Aggressive Behavior, 11, 223-233.

17

Vainikka, A., Rantala, M. J., Niemelä, P., Hirvonen, H., & Kortet, R. 2010. Boldness as a
consistent personality trait in the noble crayfish, Astacus astacus. Acta Ethologica, 14, 1725.
Van Meter, P. 2009. Hormones, stress and aggression in the spotted hyena (Crocuta crocuta).
Michigan State University.
Watts, H. E., Blankenship, L. M., Dawes, S. E., & Holekamp, K. E. 2010. Responses of spotted
hyenas to lions reflect individual differences in behavior. Ethology, 116, 1199-1209.
Watts, H. E., & Holekamp, K. E. 2009. Ecological Determinants of Survival and Reproduction in
the Spotted Hyena. Journal of Mammalogy, 90, 461-471.
Wilson, D. S., Clark, A. B., Coleman, K., & Dearstyne, T. 1994. Shyness and boldness in humans
and other animals. Trends in Ecology and Evolution, 9, 442-446.
Wilson, D. S., Coleman, K., Clark, A. B., & Biederman, L. 1993. Shy-bold continuum in
pumpkinseed sunfish (Lepomis gibbosus): An ecological study of a psychological trait.
Journal of Comparative Psychology, 107, 250-260.
Wilsson, E., & Sundgren, P.-E. 1997 October 1. The use of a behaviour test for selection of dogs
for service and breeding. II. Heritability for tested parameters and effect of selection based
on service dog characteristics. Applied Animal Behaviour Science, 54, 235-241.
Wingfield, J. C. 1994. Control of territorial aggression in a changing environment.
Psychoneuroendocrinology, 19, 709-721.
Yasukawa, K., & Searcy, W. A. 1982. Aggression in female red-winged blackbirds: A strategy to
ensure male parental investment. Behavioral Ecology and Sociobiology, 11, 13-17.

18

CHAPTER 2
INDIVIDUAL DIFFERENCES IN BOLDNESS AMONG SPOTTED HYENAS

INTRODUCTION
In recent years, there has been accumulating evidence of consistent differences in
behavior among individuals of a single species, or “animal personality.” Studies have found that
many traits, including aggressiveness, activity, boldness, sociability, and exploratory behavior,
are consistent over time and across contexts in some species (see Sih et al. 2004 and Reale et
al. 2007 for comprehensive reviews). Although it has received considerable attention (e.g.
Wilson et al. 1994; Brown et al. 2007; Frost et al. 2007; Chapman et al. 2011), boldness has
been one of the more complicated personality traits to study, and it has also proven difficult to
understand its implications, due to the difficulties inherent in studying this trait and its
consequences under naturalistic conditions.
Riesch et al. (2009) define the boldness of an individual as “the extent to which it is
willing to trade off potentially increased predation risks for possible gains in resources.”
Because of its direct effects on fitness, predation is one of the most important selection
pressures determining behavior (Lima & Dill 1990; Lima 1998). The tremendous significance of
predation suggests that boldness may be one of the most important traits to understand, both
in terms of how variation in boldness arises and how this variation affects fitness. Boldness has
been studied in a variety of species, including fishing spiders (Johnson & Sih 2005), dumpling
squid (Sinn et al. 2006), sand fiddler crabs (Pratt et al. 2005), several species of fish (Wilson et
al. 1993; Coleman & Wilson 1998; Bell & Stamps 2004; Biro et al. 2009), crayfish (Vainikka et al.
19

2010), great tits (Carere & van Oers 2004), red squirrels (Boon et al. 2008), kangaroo rats
(Dochtermann & Jenkins 2007), and bighorn sheep (Reale et al. 2009). However, whereas these
animals all have multiple predators, individual variation in boldness has never been
systematically studied in species that experience very little risk of predation in nature, such as
large mammalian carnivores. Variation in boldness, as well as the fitness consequences of
these differences, may be very different in predator than prey species. Here we evaluate
boldness in the second-largest carnivore in Africa, the spotted hyena (Crocuta crocuta), for
which lions are the only natural predators.
Most, but not all, studies have found boldness to be consistent within individuals and
repeatable over time (Johnson & Sih 2005; López et al. 2005; Sinn & Moltschaniwskyj 2005;
Niemelä et al. 2012). However, consistent inter-individual differences are not necessarily
observable in all subsets of a population. For example, Harris et al. (2010) found that boldness
in guppies was repeatable in females but not males, and research by Hedrick and Kortet (2012)
showed the same trend in field crickets. Dzieweczynski and Crovo (2011) showed that some,
but not all, measures of boldness are repeatable in juvenile sticklebacks.
With few exceptions (e.g. Riesch et al. 2009), most previous studies have found that
boldness is significantly heritable, with heritability estimates ranging from 0.002 (Bell 2005) to
more than 0.5 (Drent et al. 2003). To date, very few studies have estimated maternal effects on
boldness. Reale et al. (2009) found no evidence of maternal effects on boldness in bighorn
sheep, and the maternal effect estimate in a study of boldness in great tits was significant, but
small, at 0.05 (van Oers et al. 2004).

20

Findings are inconsistent regarding the stability of boldness over time, particularly
before and after the onset of sexual maturity. Magnhagen and Borcherding (2008) found that
boldness in perch was age-specific; individuals that were particularly bold as juveniles were not
necessarily bold as adults. However, Sinn et al. (2008) found a positive association between
feeding-related boldness among individual squid as juveniles and adults, suggesting that
phenotypic boldness is determined well before adulthood.
In terms of fitness, theoretical analyses by Wolf et al. (2007) suggest that bolder animals
may have shorter lifespans and may invest more in current, rather than future, reproduction. A
meta-analysis by Smith and Blumstein (2008) supported this theory, finding that boldness
decreased longevity but increased reproductive success in captive animals. Reale et al. (2009)
tested this tradeoff with a population of male bighorn sheep and found that bolder rams
actually survived longer than shy rams, and that bolder rams also had greater reproductive
success later in life than did shy rams. Variation in the relationships between boldness and
fitness may be due to species-level differences and exposure to varying intensities of predation
pressure.
Most studies of boldness are either performed in the laboratory or use an experimental
setup to quantify boldness in a free-living population; novel object tests (e.g. Wilson et al. 1993;
Sundström et al. 2004; Frost et al. 2007), handling tests (e.g. Carere and van Oers 2004; Taylor
et al. 2012) and open-field tests (e.g. Brown, Burgess, and Braithwaite 2007; Boon, Reale, and
Boutin 2008) are commonly-used tests of boldness. Whereas these studies have the advantage
of enhanced control, direct observational data documenting boldness under strictly natural
conditions offers a means to assess simultaneously how this personality trait varies among free21

living individuals, and how it affects their fitness. But because boldness is difficult to quantify in
naturalistic settings, few studies have used observational data to examine boldness (e.g. Reale
et al. 2000), and only a small number of longitudinal studies of boldness have ever been
conducted with free-living animals (e.g. Boon, Reale, and Boutin 2008; Reale et al. 2009). Here
we conduct a longitudinal analysis of boldness among free-living spotted hyenas (Crocuta
crocuta) in Kenya. The riskiest situations for spotted hyenas occur during encounters with lions.
Hyenas and lions have a high degree of dietary overlap and compete for the same resources
(Kruuk 1972; Hayward 2006). Due to their greater size and power, lions regularly steal kills
from hyenas (Kruuk 1972; Mills 1990; Frank et al. 1995). Generally, interspecific competition
between carnivores is particularly intense because of the specialized adaptations for killing
possessed by these competitors (Creel et al. 2001). Lions represent a major source of hyena
mortality; in studies of four separate populations of hyenas, lions were the largest sources of
known hyena deaths in each population (Kruuk 1972; Trinkel et al. 2006; Watts & Holekamp
2009). In the Masai Mara, lion density is particularly high (Ogutu & Dublin 2002), suggesting
that hyenas there may experience unusually frequent and intense competition with lions.
Despite the risks that lions pose to hyenas, hyenas do not necessarily avoid encounters
with lions, and in fact they are often attracted to lions with kills, presumably due to the
potential benefits of scavenging (Watts et al. 2010). Depending on the situation, such as the
relative numbers of lions and hyenas present and whether or not a male lion is present, hyenas
are sometimes able to steal parts of a carcass or even drive lions off a kill altogether (Kruuk
1972; Mills 1990). For hyenas, the possible benefits of acquiring food may outweigh the risk of
injury or death from lions. Because proximity to lions represents both a risk and a potential
22

benefit to hyenas, measuring the behavior of hyenas during lion-hyena encounters offers a
promising way to study boldness.
Previous research has suggested that individual spotted hyenas may vary in boldness
and other related traits. In a study by Gosling (1998), observers were asked to rate captive
hyenas on several personality traits; the study found that boldness was rated with very high
reliability, but also found no evidence for a shy-bold continuum in the population using a
principal components analysis. However, subsequent studies of free-living hyenas have
suggested that consistent inter-individual differences in boldness do exist in spotted hyenas. A
study by Watts et al. (2010) showed that there were significant individual differences in the way
that hyenas responded to playbacks of lion roars. In this study, hyenas varied both regarding
whether or not they avoided the roars, and also in the strength of their behavioral response to
the playbacks. Watts et al. (2010) also demonstrated a correlation between boldness and
vigilance behavior; hyenas that took more risks in response to the playback also tended to be
more vigilant. These differences were context-specific, seen only in the context of interactions
with lions, not at dens or in the “baseline” context in the absence of lions. Pangle (2008)
reported evidence of consistent differences in vigilance behavior among adult hyenas, but not
among juveniles.
Here, we examine the social, situational, and individual factors that contribute to
boldness in spotted hyenas. Next we determine whether there is evidence for consistent
individual differences in this trait, and determine how much of the phenotypic variation in
boldness is due to heredity, non-genetic maternal effects, and permanent environmental

23

effects. We then examine the stability of boldness across ontogeny and across different
contexts, and finally we evaluate the relationship between boldness and fitness.

METHODS
Study Subjects
Between 1988 and 2009, personnel from the Mara Hyena Project collected behavioral
data on a group of free-living spotted hyenas, called the Talek Clan, in the Masai Mara National
Reserve in Kenya (Boydston et al. 2001). All hyenas in this clan were recognized individually by
observers based on their unique spot patterns, and were of known sex, age, and social rank.
Hyenas were sexed based on the morphology of the glans of the erect phallus (Frank et al.
1990). We assigned ages to all natal animals (to ±7 days) based on their appearance when they
are first seen above ground at dens, and we were able to assign immigrant males’ ages reliably
(to ± 6 monts) with an age-estimation model (Van Horn et al. 2003).
The social rank of each hyena in the clan was assigned using a dominance matrix based
on observations of dyadic agonistic interactions (Holekamp & Smale 1993; Smale et al. 1993).
Rank was reassessed each time a new animal reached adulthood, entered or left the clan, or
died. Cubs were assigned the rank of their mother until they reached reproductive maturity at
24 months of age (Frank 1986; Glickman et al. 1992; Holekamp et al. 1997). We also calculated
standardized rank for each hyena by dividing their numerical rank by the total number of
ranked animals in the clan at the time.
We established maternity based on genotyping and observations of cubs nursing
(Holekamp & Smale 1993). Paternity was assigned based exclusively on genotyping, as
24

previously described (Van Horn et al. 2004; Watts et al. 2011). Methods used to immobilize
animals and collect blood samples for DNA extraction are described in detail in Engh et al.
(2002). All sampling procedures were approved by the Institutional Animal Care and Use
Committee at Michigan State University (AUF 07/08-099-00), complied with Kenyan law, and
conformed to guidelines approved by the American Society of Mammalogists (Sikes & Gannon
2011).

Data collection
We generally observed hyenas daily from vehicles between 0530 and 0930 hours and
between 1700 and 2000 hours. We initiated observation sessions whenever we saw a hyena or
group of hyenas separated from others by at least 200 meters. Upon initiating an observation
session, we recorded the GPS location of the session and conducted a behavioral scan that
identified each hyena, noted its current behavior, and described its position in space; these
scans were repeated every 20 minutes. During each session, we used critical incident sampling
(Altmann 1974) to record specific behaviors such as aggressive acts, appeasements,
vocalizations, and greetings. We also recorded whenever a hyena entered or left a session in
progress. The presence, species, number, and sex of other carnivores present were also noted
when possible, as was the presence of food in the possession of hyenas or other carnivores.
Observational data for this chapter were extracted from all sessions in which at least
one known hyena and at least one lion were present simultaneously. During these sessions, we
noted distances between each hyena and the lion or lions, both in regular scans at 20-minute
intervals and during critical incident sampling. We also included distance data when it could be
25

reliably extrapolated from other data; for instance, if a hyena was recorded as “biting” a lion, a
distance of zero meters was inferred. Distance data involving unidentified hyenas were not
included in these analyses.
To assess boldness in hyenas, we measured two aspects of the behavior of each
individual present during lion-hyena interactions. The first was the closest distance to which
each hyena approached a lion (called the “minimum distance”) during a session. We also
calculated the minimum standardized distance rank of each hyena; this was determined by
ranking the minimum distances of all hyenas in a session from the closest to the farthest from
lions, and dividing the distance rank of each hyena by the total number of hyenas with recorded
distances in that session. This standardized measure allowed us to compare the behavior of
individuals across very different circumstances. Hyenas that had smaller minimum distances
and smaller minimum standardized distance ranks were considered bolder, because they were
more prone to interactions with and attacks from lions than were other hyenas.
Other situational data were also collected from these sessions, including the maximum
number of hyenas and lions present in the session, the time of day (either from midnight to
noon, or from noon until midnight), whether or not food was available in the session, whether
or not at least one male lion was present, and the current seasonal prey availability (high for
June through October, and low from November through May).

Statistical Analysis
Situational factors influencing boldness

26

We assessed boldness based on two different levels of analysis; one at the level of the
observation session, and another at the level of individual hyenas. First, we performed an
analysis using the individual observation session as the unit of analysis to determine which
social and environmental factors influence hyenas’ distance from lions. In order to generate an
accurate and unbiased dataset, we included only sessions in which at least three hyenas had
known distances from the lion or lions. As the response variable for this session, we used an
average of the minimum distances for all hyenas present in each session. This measure,
calculated for each session, represented the average minimum distance that hyenas kept from
lions during that session. Using this composite measure of boldness allowed us to assess the
importance of situational variables that were common to all hyenas present in the session.
Using a Bayesian approach, we fitted a model with the Markov Chain Monte Carlo for
Generalized Linear Mixed Models (MCMCglmm) analysis tool in the R statistical package
(Hadfield 2010; R Core Development Team 2011), with the average minimum distance for all
hyenas present in each session as the response variable, and various social and environmental
aspects of the session as fixed effects. These fixed effects included the time of day, the ratio of
the maximum number of hyenas present to the maximum number of lions present in the
session, whether or not food was available in the session, whether or not at least one male lion
was present, and whether seasonal prey availability was high or low. We included year and
month as random effects. The posterior distribution of the model was sampled every 500
iterations after a burn-in period of 20,000 iterations for a total of 2,000 samples.
For all MCMCglmm models, the estimates are reported as modes of the posterior
distribution, and we tested fixed effects for significance by determining whether or not the 95%
27

credible interval for the fixed effect included zero; variables for which the credible interval
included zero were not significant predictors of boldness, whereas those that did not include
zero were considered significant. We tested random effects for significance by comparing the
deviance information criterion (DIC) values of models fitted with and without each random
effect of interest; if the inclusion of a random effect lowered the DIC value of the model, the
random effect was considered significant in predicting boldness.
For these models, we used a relatively uninformative prior that partitioned variance
equally among all random effects (Hadfield 2010); using a more informative prior did not affect
reported results. In these and the other MCMCglmm models described below, we assessed
convergence by examining time series of the model parameters. We also inspected
autocorrelation values for the models, all of which were less than 0.1 for reported values.

Consistency in boldness within individuals
We then performed a similar Bayesian analysis using individual hyenas as the units of
analysis to (1) determine which characteristics of individuals affected the distances maintained
by hyenas between themselves and lions, and (2) determine whether there was consistency
across sessions within individuals in this measure of boldness.
In the individual-level analysis, we used hyenas’ minimum standardized distance ranks
from lions, rather than their absolute minimum distances from lions, as the measure of
boldness. Here we only included sessions in which at least two hyenas had recorded distances,
and we only included hyenas that had recorded distances from lions in at least four different
observation sessions. We analyzed the boldness of males and females in separate models.
28

For each sex, we created a repeated measures MCMCglmm model with minimum
standardized distance rank as the response variable. The standardized social rank and age of
the hyena on the date of the session were included as fixed effects, and were tested for
significance as described above. A unique session identifier was included as a random effect in
both analyses to control for pseudoreplication. To determine whether individual hyenas were
consistent in the minimum distances they maintained from lions, we tested whether the
inclusion of a random effect representing the identity of the hyena improved the model by
comparing DIC values. If this term significantly improved the fit of the model, individuals’
behavior was considered consistent across sessions.
The posterior distribution of the model was sampled every 500 iterations after a burn-in
period of 20,000 iterations for a total of 2,000 samples. For these models, we used an
uninformative prior that partitioned variance equally among all random effects (Hadfield 2010);
the results did not change when we used a more informative prior.

Heritability of boldness
To determine whether or not boldness is heritable in spotted hyenas, we created a
restricted-likelihood Bayesian animal model (Kruuk 2004). Animal models use pedigree
information to partition the heritable and non-heritable components of phenotypic variance
(Kruuk 2004; Wilson et al. 2009). Phenotypic, additive genetic, maternal, and permanent
environmental variances were estimated with a univariate-trait animal model predicting the
boldness of each hyena. Permanent environment effects are created by each individual’s
environment and are consistent over its lifetime; maternal effects are non-genetic effects that a
29

mother has on all her offspring. We used the previously described model with individual
hyenas as the units of analysis, but included boldness data from both sexes. We included all
significant individual-level fixed effects as well as random effects representing the identity of
the hyena, the session identifier, the identity of the hyena’s mother, and a term representing
the additive genetic effect. Because female hyenas mate with several males across their
reproductive history, maternal effects can be distinguished from heritability via the effect of
these different sires.
Because MCMCglmm requires values for all random effects, we generated unique
maternal identities for immigrant males and natal animals with unknown mothers, following
Taylor et al. (2012) This technique allows for the estimation of maternal effects, even when
pedigree information is missing; however, it assumes that all hyenas with unknown maternity
are unrelated.
We calculated heritability, maternal effects, and permanent environment effects by
dividing the relevant variance component by the total phenotypic variance. Following Lessels
and Boag (1987), we calculated repeatability by dividing the among-individual variance by the
total among- and within-individual variance.
The posterior distribution of the animal model was sampled every 500 iterations after a
burn-in period of 20,000 iterations for a total of 5,000 samples. For this model, we used an
uninformative prior that partitioned variance equally among all random effects (Hadfield 2010);
the results did not change when we used a more informative prior.

30

Stability over time and across situations
In order to determine whether the boldness of individual hyenas remained stable over
time, we created generalized linear mixed models using the lme4 function in R. In each session,
we determined whether each hyena with a recorded minimum distance was less or more than
24 months of age; hyenas that were less than 24 months of age were considered to be
juveniles, and those over 24 months of age were considered to be adults. Only hyenas with at
least two recorded distances in each age class were included in this analysis.
In these models, we used minimum standardized distance rank as the response variable
and age class as the main fixed effect of interest. Other fixed effects described previously,
including the hyena’s sex and standardized rank, were also included in the analysis but only
fixed effects that lowered the Akaike information criterion (AIC) value of the model were
retained. Here again, a unique session identifier was included as a random effect in all models
to control for pseudoreplication.
We included individual identity as a random intercept to allow for the variation in
boldness between hyenas suggested by earlier analyses. We also added a random slope to the
model, which described the extent to which the boldness of individual hyenas varied over time.
If this term significantly improved the fit of the model, we concluded that boldness in hyenas
was not a trait that changed consistently in individuals across ontogeny. To test whether the
random slope significantly improved the fit of the model, we created two models that were
identical except for the inclusion of the random slope. We then compared these two models
with a log-likelihood ratio test, in which the log-likelihood ratio is calculated as 2[log-likelihood
of model B - log-likelihood of model A], and tested as a chi-squared distribution (Pinheiro &
31

Bates 2000; Martin & Réale 2008). The approach of using random regression to investigate
behavioral reaction norms is described in more detail in Dingemanse et al. (2010).
Using the same methodology, we ran a set of similar models to test whether the
boldness exhibited by individual hyenas varied with prey availability. In these models, we used
prey availability as the main fixed effect of interest, and defined the random slope to allow
hyenas’ boldness to change across sessions with high and low prey availability. Only hyenas
present in at least two sessions in each prey availability category were included in this analysis.

Relationship between boldness and fitness
Finally we tested whether a female’s boldness was related to fitness, as indicated by
various measures of reproductive success and longevity. For these analyses, we used two
lifetime measures of boldness: (1) a measure of relative boldness calculated by averaging all a
hyena’s minimum standardized distance ranks from lions during lion-hyena interactions, and (2)
a measure of absolute boldness calculated by averaging all a hyena’s actual minimum distances
from lions during lion-hyena interactions.
As indicators of reproductive success, we used the total number of cubs borne by each
female over her lifetime, as well as the number and proportion of her cubs that survived to
three life-history milestones: den independence, weaning, and reproductive maturity. We
created a set of generalized linear and second-order polynomial models with these fitness
components as the response variables, and the lifetime boldness measure, maternal rank, and
an interaction term as predictors. Only female hyenas that were born after our study began,

32

died before our study ended, and had recorded distances in at least two sessions with lions
were included in this analysis.
We also used hyenas’ longevity as a response variable to determine whether boldness is
related to survival. For this dataset, we used the same two lifetime measures of boldness, but
included only females for which we had known death dates and had recorded distances in at
least two sessions with lions. We created both generalized linear and second order polynomials
with longevity (in months) as the response variable. As fixed effects, we included our lifetime
boldness measures and average lifetime standardized rank, as well as the interaction terms
between these variables.
For longevity and all measures of reproductive success, we used likelihood ratio tests to
determine whether the second-order polynomial models were able to predict these fitness
measures better than linear models. All variables in these analyses were Z-transformed.

RESULTS
Situational factors influencing boldness
Between 1988 and 2009, there were 128 sessions with recorded minimum distances
that met our criteria for inclusion in this analysis. In these sessions, the average hyena-to-lion
ratio was 6.13±0.634, and male lions were present in 34.4% of these sessions. The average
minimum distance between hyenas and the lion or lions present was 24.51±3.24 meters. On
average, there were 8.50±0.61 hyenas with recorded minimum distances from lions in each of
these sessions. In general, hyenas approached lions more closely as the hyena-to-lion ratio
increased and when food was present in a session (Table 2.1). Hyenas maintained greater
33

distances from lions when at least one male lion was present (Table 2.1). Time of day and
seasonal food availability were not significant predictors of the average minimum distance in
these sessions, although there was a non-significant trend for larger minimum distances from
noon to midnight than from midnight to noon (Table 2.1).

Table 2.1
Fixed effects predicting average minimum distances from hyenas to lions in sessions with at
least one lion present and at least three minimum distances for known hyenas.
pMCMC
Coefficients
Posterior mode
95% credible interval
67.9058
(33.5133 to 98.7528)
<0.001*
(Intercept)
14.8892
( -2.6071 to 30.7114)
0.08
Time of day (PM)
-1.4121
(-2.5021 to -0.3243)
0.016*
Hyena-to-lion ratio
16.5883
(2.1574 to 32.4285)
0.034*
Male lion present
-28.9227
(-55.9679 to -3.9985)
0.034*
Food available
-3.1148
( -17.8168 to 11.5887)
0.664
Prey availability (low)

Including a random effect of year improved the fit of the model (DIC=938.24) compared
to models that included random effects of both year and month (DIC=939.54), only month
(DIC=941.69), or neither random effect (DIC=939.21), indicating that the year affects the
general boldness of hyenas in some way.

Consistency in boldness within individuals
In our repeated measures analysis, there were 43 females that ranged in age from 9.87
to 198.80 months, averaging 66.32±2.00 months, and 32 males that ranged in age from 10.13
to 165.90 months, averaging 77.37±2.32 months.
For males, none of the individual-level predictor variables were significant; neither age
nor standardized rank affected how closely male hyenas approached lions (Table 2.2). For
34

females, age was not a significant predictor of boldness, but higher-ranking hyenas tended to
approach lions more closely than did lower-ranking hyenas (Table 2.3).

Table 2.2
Fixed effects predicting the minimum distance separating each male hyena from lions.
pMCMC
posterior mean
95% credible interval
(Intercept)
Standardized rank
Age

0.4242757
0.11127
-0.0004126

(0.3198692 to 0.5258341)
(-0.0467974 to 0.2601023)
(-0.0013439 to 0.0004587)

<0.001*
0.160
0.376

Table 2.3
Fixed effects predicting the minimum distance separating each female hyena from lions.
pMCMC
posterior mean
95% credible interval
(Intercept)
Standardized rank
Age

0.3798397
0.2244043
0.000239

(0.2953735 to 0.4660331)
(0.0235000 to 0.4879079)
(-0.0006693 to 0.0010733)

<0.001*
0.044*
0.594

For females, including a random effect of an individual identifier improved the fit of the
model (DIC=147.87), compared to a model that included only the session identifier as a random
effect (DIC=160.84), indicating that individual females have consistent boldness measures
across sessions. For males, the inclusion of a random effect of an individual identifier did not
improve the model (DIC=120.44), compared to a model without this term (DIC=119.78),
suggesting that males are not consistent across sessions in their boldness.

Heritability of boldness
Genetic effects, maternal effects, and permanent environment effects were all significant for
boldness, as indicated by a lower DIC value when these effects were included

35

0.6
1.0

Proportion of variance explained

0.5
0.9

0.4
0.8

0.3
0.3

0.2
0.2

0.1
0.1

00
Heritability (4.16%)

Maternal effect
(2.75%)

Permanent
Unexplained (89.67%)
environment effect
(3.42%)

Figure 2.1. The percent of variation in boldness attributable to heritability, permanent
environment effects, maternal effects, and the remaining unexplained variation. Variances
were estimated with a MCMCglmm animal model. Values on the y-axis between 0.3 and 0.8
are omitted.

36

in a model (DIC= 296.912), compared to when they were omitted (genetic effect omitted,
DIC=304.211; maternal effect omitted, DIC=-302.640, permanent environment effect omitted,
DIC=302.332). The heritability estimate was 0.0416 (95% CI=0.0116 to 0.0770, Figure 2.1), the
permanent environment estimate was 0.0342 (95% CI=0.0141 to 0.0744, Figure 2.1), and the
maternal effect estimate was 0.0275 (95%CI=0.0115 to 0.0665, Figure 2.1). The estimate for
repeatability was 0.0682 (95% CI=0.0417 to 0.1271).

Stability over time and across situations
There were 17 hyenas present with lions in at least two sessions during each age
interval; 12 (70.6%) were female and 5 (29.4%) were male. The model including a random
slope of age class in addition to the random intercept was not significantly better at predicting
the boldness of individuals than the model with only the random intercept (LRT=0.301, df=2,
p=0.860). The inclusion of a random slope was not significant, indicating that the boldness
measures of individual hyenas generally change in a consistent way with age (Figure 2.2a).
There were 49 hyenas that met our criteria for the prey availability analysis. Of these
hyenas, 27 (55.1%) were female and 22 (44.9%) were male. For this analysis, including a
random slope of prey availability did not improve the fit of the model (LRT=-0.757, df=2, p=1).
The inclusion of a random slope was not significant, indicating that the boldness measures of
individual hyenas generally change in a consistent way across varying levels of prey availability
(Figure 2.2b).

37

a

b

Figures 2.2a and 2.2b. Predicted behavioral reaction norms for boldness measures across (a)
two age classes and (b) two categories of seasonal prey availability, as predicted by lmer
models. Each line represents how the behavior of an individual changes across categories.
Smaller relative boldness measures represent bolder individuals, since these measures are
based on minimum distances to lions.
38

Relationship between boldness and fitness
Thirty females satisfied the criteria for inclusion in our analysis of reproductive success.
Using generalized linear models, neither the lifetime measure of relative boldness nor the
lifetime measure of absolute boldness was significant in predicting any measure of reproductive
success (p>0.2). No interaction terms containing either boldness measure improved the fit of
the reproductive success models, and second-order regression models were not better at
predicting reproductive success than linear models (p>0.2).
There were 40 female hyenas that satisfied the criteria for our analysis of longevity. The
age at death of hyenas ranged from 41.23 to 212.73 months, with an average of 106.82±6.75
months. Using generalized linear models, neither the lifetime measure of relative boldness nor
the lifetime measure of absolute boldness was significant in predicting longevity (p>0.2).
Standardized rank was not a significant predictor of age at death (p=0.684). No interaction
terms containing either boldness measure improved the fit of the generalized linear longevity
models. Interestingly, however, a second-order polynomial model with lifetime relative
boldness was significantly better at predicting age at death than a corresponding linear model
(LRT=4.488, df=1, p=0.034, Figure 2.3). In this model, the second-order boldness term was
significant in predicting females’ age at death (-467.402±222.995, t=-2.096, p=0.043).

39

Figure 2.3. The relationship between lifetime relative boldness and longevity. Actual values
are displayed as dots, and the values predicted by the second-order regression model are
shown as a line, with 95% confidence intervals indicated with grey dotted lines.

40

DISCUSSION
Overall, we found that there are consistent inter-individual differences in lion-related
boldness in female spotted hyenas, but not in males. This finding is consistent with previous
research on other species (guppies, Harris et al. (2010); field crickets, Hedrick and Kortet
(2012)), where males also failed to show consistency in boldness across repeated observations.
Female and male hyenas may employ different strategies in terms of boldness, with females
being more consistent and males maintaining more situational plasticity. Because immigrant
male hyenas hold lower social ranks, and therefore have reduced access to resources, a less
rigid and more adaptable strategy might be advantageous for males. Males have less reliable
access to food, smaller body sizes, and higher movement and activity rates than females, all of
which may make them more sensitive to environmental conditions than females (Kruuk 1972;
Mills 1990; Kolowski et al. 2007).
Several situational aspects of lion-hyena interactions, such as the time of day and the
availability of food were important determinants of boldness among hyenas. We found that
the presence of at least one male lion and the hyena-to-lion ratio both affected the behavior of
hyenas, consistent with previous findings on lion-hyena interactions (Kruuk 1972; Cooper
1991). Not surprisingly, hyenas were bolder when food was available in a session, supporting
the hypothesis that access to food is a major incentive for hyenas to interact with lions, despite
the risks involved (Watts et al. 2010).
Social rank had no effect on boldness among males, but we found that higher-ranking
females tended to approach lions more closely than did lower-ranking females. This is
somewhat surprising in terms of the relationship between boldness and resource availability, as
41

females with restricted access to food might be expected to take more risks than those with
better access to food (Riesch et al. 2009). However, high-ranking females would gain
possession of any scraps or carcasses stolen from lions, so they may actually benefit the most
from approaching and interacting with lions. Even if food is successfully scavenged from lions,
low-ranking hyenas might be prohibited from feeding by higher-ranking individuals.
Our population-wide repeatability estimate was significant but small, compared to
estimates for other personality traits (Bell et al. 2009). Although this estimate may be accurate,
the low value may also be partially due to statistical artifacts. Generally, when estimating
repeatability from repeated measures, the experimenter assumes that the conditions remain
constant over each trial. In laboratory experiments this is often a reasonable assumption, but
in the field there are innumerable conditions that change between trials for which we could not
account in our study. Additionally, the low repeatability measure may be due to the inclusion
of males in the animal model. Data from immigrant males are necessary to estimate
heritability, but we found no evidence of consistent differences in boldness among males in an
earlier analysis, and it is possible that the inclusion of these values decreased the repeatability
estimate here.
The heritability estimate from our animal model was also low but significant. This
estimate is much lower than the average heritability for behavioral traits (0.30±0.02)
(Mousseau & Roff 1987), but is within the range of reported values in the literature. Sinn,
Apiolaza, and Moltschaniwskyj (2006) reported a boldness heritability estimate of 0.05, and
estimates from Bell (2005) were as low as 0.002. However, repeatability generally sets an
upper limit for heritability (Falconer 1981), and in light of our small repeatability estimate, our
42

heritability estimate is relatively substantial. Statistically, traits with low repeatabilities tend to
give underestimates of heritability and other important quantitative genetic parameters
(Whitlock 1998). Estimates for both maternal effects and permanent environment effects for
boldness in our study were both small, but significant. As noted previously, differences
between heritability and maternal effect estimates were due to sire effects within dams.
While we were able to explain a small amount of the observed phenotypic variation in
boldness via heritability, maternal effects, and permanent environment effects, more than 89%
of the observed variation in boldness was unexplained by the model. None of the three sources
of inter-individual variation that we assessed appear to play a large role in boldness in hyenas.
Lion-hyena interactions are complex and dynamic, and there are many environmental and
social conditions that we could not include in our models or even measure accurately in the
field, such as motivational states and events preceding the lion-hyena interaction. It is likely
that these factors are responsible for much of the observed variation for which we could not
account statistically in our analyses.
Random regressions showed that the distances hyenas maintain from lions are relatively
stable over time and varying circumstances. Boldness was consistent across age classes, as
indicated by the fact that the rank order of the boldness of individual hyenas remained
relatively stable over time. In other words, hyenas that were bold as juveniles were also
generally bold as adults, suggesting that the boldness of an individual is determined relatively
early in life. Research on other species has also found evidence of ontogentic consistency in
boldness (great tits, Carere et al. 2005; fishing spiders, Johnson & Sih 2005; dumpling squid,
Sinn et al. 2008).
43

A similar trend was found across varying levels of seasonal prey availability; hyenas that
were relatively bold when prey was abundant were also bolder than their peers when prey was
scarce. This finding, along with the earlier finding that prey availability does not affect general
boldness levels in hyenas, was somewhat surprising, since boldness behavior is predicted to
change with hunger levels (Riesch et al. 2009). When food is hard to come by, the potential
benefits of approaching and interacting with lions are expected to increase; however, our
results showed no difference in boldness between seasons of high and low prey availability, and
no difference between individuals in how they respond to these changes. It is possible that our
measure of seasonal prey availability did not accurately reflect the amount of food available to
hyenas on a fine enough time scale here. Boldness may be also constrained in some way, and
the hyenas’ behavior may not be sufficiently labile to adjust to differing levels of prey
abundance. The genetic, cognitive, or physical constraints underlying personality traits such as
boldness may prohibit individuals from behaving optimally in any given situation (Sih et al.
2004b; Sih et al. 2004c; Dingemanse & Reale 2005; Dochtermann & Roff 2010). Due to one or
more of these constraints, behavior in one environment might be optimal, whereas the
behavior may seem suboptimal, or even maladaptive, in a different context. While none of
these factors were tested in our analyses, future research that investigated constraints on
boldness among hyenas might be beneficial.
We found that, although boldness was not correlated with the reproductive success of
female hyenas, it was significantly related to longevity. We found evidence of stabilizing
selection on boldness; females that were particularly bold and those that were particularly shy
had shorter lifespans than did those in the middle of the shy-bold continuum. Hyenas that
44

consistently approach lions closely may be more prone to injury and death than those that
maintain greater distances from lions. Meanwhile, hyenas that stay relatively far from lions
may suffer because they do not reap the same benefits that bolder hyenas are able to procure.
Because we only measured boldness in hyenas in one context in this study, it is not
possible to draw conclusions about this trait in contexts other than lion-hyena interactions. It
would be beneficial for future researchers to examine other types of boldness in hyenas, such
as boldness related to mating, dispersal, or anthropogenic influences. Future studies should
also test boldness in hyenas with experiments in the field; it would then be possible to compare
experimental measures of boldness with the reliable observational estimates measured here,
to determine whether there is consistency between these two measures.
This study confirms the existence of individual differences in boldness among female
spotted hyenas that are consistent over ontogeny and across contexts. We found that a small
but significant proportion of the variation in boldness behavior was due to heritability; this
effect may be smaller than effects in other species because the spotted hyena is a predator,
and thus faces a very different set of tradeoffs than do prey species. Whereas prey species
rarely have anything to gain from approaching and interacting with their predators, hyenas can
potentially benefit greatly from interactions with lions by acquiring food, despite the risk of
injury or death. Differences among hyenas in boldness may reflect plastic individual strategies
aimed at resource acquisition, rather than direct genetic effects. However, stabilizing selection
may limit extreme boldness or shyness in this species, because hyenas with intermediate
phenotypes survive longer than those on either extreme end of the boldness spectrum.

45

Understanding boldness in hyenas and other large carnivores can potentially inform our
understanding and treatment of "problem individuals" (Linnell et al. 1999). Carnivores are
especially prone to causing problems for herders, ranchers, and other humans living in or near
undeveloped land; hyenas, wolves, lions, and other predators are known to prey on livestock
and pets (Patterson et al. 2004; Kolowski & Holekamp 2006; Darrow & Shivik 2009). It has been
suggested that that boldness may play a role in the likelihood of depredation (Linnell et al.
1999; Darrow & Shivik 2009). This study represents only the first step in understanding
patterns of boldness among hyenas, but future studies of boldness in large carnivores might be
able to determine which individuals in each species are more likely to cause problems
associated with livestock or pet depredation, and may thus be able to suggest new ways to
mitigate and manage these problems in ways consistent with conservation goals.

46

LITERATURE CITED

47

LITERATURE CITED
Altmann, S. E. 1974. Observational study of behavior: sampling methods. Behavior, 49, 227265.
Bell, A. M. 2005. Behavioural differences between individuals and two populations of
stickleback (Gasterosteus aculeatus). Journal of Evolutionary Biology, 18, 464-473.
Bell, A. M., Hankison, S. J., & Laskowski, K. L. 2009. The repeatability of behaviour: a metaanalysis. Animal Behaviour, 77, 771-783.
Bell, A. M., & Stamps, J. A. 2004. Development of behavioral differences between individuals
and populations of sticklebacks, Gasterosteus aculeatus. Animal Behaviour, 68, 1339-1348.
Biro, P. A., Beckmann, C., & Stamps, J. A. 2009. Small within-day increases in temperature
affects boldness and alters personality in coral reef fish. Proceedings of the Royal Society B:
Biological Sciences, 277, 71-77.
Boon, A. K., Reale, D., & Boutin, S. 2008. Personality, habitat use, and their consequences for
survival in North American red squirrels Tamiasciurus hudsonicus. Oikos, 117, 1321-1328.
Boydston, E. E., Morelli, T. L., & Holekamp, K. E. 2001. Sex differences in territorial behavior
exhibited by the spotted hyena (Hyaenidae, Crocuta crocuta). Ethology, 107, 369-385.
Brown, C., Burgess, F., & Braithwaite, V. 2007. Heritable and experiential effects on boldness in
a tropical poeciliid. Behavioral Ecology and Sociobiology, 62, 237-243.
Carere, C., Drent, P. J., Privitera, L., Koolhaas, J. M., & Groothuis, T. G. G. 2005. Personalities in
great tits, Parus major: stability and consistency. Animal Behaviour, 70, 795-805.
Carere, C., & van Oers, K. 2004. Shy and bold great tits (Parus major): body temperature and
breath rate in response to handling stress. Physiology &amp; Behavior, 82, 905-912.
Chapman, B. B., Hulthén, K., Blomqvist, D. R., Hansson, L.-A., Nilsson, J.-Å., Brodersen, J.,
Anders Nilsson, P., Skov, C., & Brönmark, C. 2011. To boldly go: individual differences in
boldness influence migratory tendency. Ecology Letters, 14, 871-876.
Coleman, K., & Wilson, D. S. 1998. Shyness and boldness in pumpkinseed sunfish: individual
differences are context-specific. Animal Behaviour, 56, 927-936.
Cooper, S. 1991. Optimal hunting group size: the need for lions to defend their kills against loss
to spotted hyaenas. African Journal of Ecology, 29, 130-136.

48

Creel, S., Spong, G., & Creel, N. 2001. Interspecific competition and the population biology of
extinction-prone carnivores. In: Carnivore Conservation, (Ed. by J. L. Gittleman, S. M. Funk,
D. Macdonald, & R. K. Wayne), pp. 35-60. Cambridge: Cambridge University Press.
Darrow, P. A., & Shivik, J. A. 2009. Bold, shy, and persistent: Variable coyote response to light
and sound stimuli. Applied Animal Behaviour Science, 116, 82-87.
Dingemanse, N. J., Kazem, A. J. N., Reale, D., & Wright, J. 2010. Behavioural reaction norms:
animal personality meets individual plasticity. Trends in Ecology & Evolution, 25, 81-89.
Dingemanse, N., & Reale, D. 2005. Natural selection and animal personality. Behaviour, 142,
1159-1184.
Dochtermann, N. A., & Jenkins, S. H. 2007. Behavioural syndromes in Merriam’s kangaroo rats
(Dipodomys merriami): a test of competing hypotheses. Proceedings of the Royal Society B:
Biological Sciences, 274, 2343-2349.
Dochtermann, N. A., & Roff, D. A. 2010. Applying a quantitative genetics framework to
behavioural syndrome research. Philosophical Transactions of the Royal Society B:
Biological Sciences, 365, 4013-4020.
Drent, P. J., van Oers, K., & van Noordwijk, A. J. 2003. Realized heritability of personalities in
the great tit (Parus major). Proceedings of the Royal Society, Series B, 270, 45-51.
Dzieweczynski, T. L., & Crovo, J. A. 2011. Shyness and boldness differences across contexts in
juvenile three-spined stickleback Gasterosteus aculeatus from an anadromous population.
Journal of Fish Biology, 79, 776-788.
Engh, A. L., Funk, S. M., Van Horn, R. C., Scribner, K. T., Bruford, M. W., Libants, S., Szykman,
M., Smale, L., & Holekamp, K. E. 2002. Reproductive skew among males in a femaledominated mammalian society. Behav. Ecol., 13, 193-200.
Falconer, D. S. 1981. An introduction to quantitative genetics. 2nd edn. New York: Longman,
Inc.
Frank, L. G. 1986. Social organization of the spotted hyaena (Crocuta crocuta). I. Demography.
Animal Behaviour, 34, 1500-1509.
Frank, L. G., Glickman, S. E., & Powch, I. 1990. Sexual dimorphism in the spotted hyaena
(Crocuta crocuta). Journal of Zoology, 221, 308-313.
Frank, L. G., Holekamp, K. E., & Smale, L. 1995. Dominance, demography, and reproductive
success of female spotted hyenas. In: Serengeti II: Conservation, Research, and
Management, (Ed. by A. R. E. Sinclair), pp. 364-384. Chicago: University of Chicago Press.
49

Frost, A. J., Winrow-Giffen, A., Ashley, P. J., & Sneddon, L. U. 2007. Plasticity in animal
personality traits: does prior experience alter the degree of boldness? Proceedings of the
Royal Society B: Biological Sciences, 274, 333-339.
Glickman, S. E., Frank, L. G., Pavgi, S., & Licht, P. 1992. Hormonal correlates of
“masculinization” in female spotted hyaenas (Crocuta crocuta). 1. Infancy to sexual
maturity . Journal of Reproduction and Fertility , 95 , 451-462.
Gosling, S. D. 1998. Personality dimensions in spotted hyenas (Crocuta crocuta). Journal of
Comparative Psychology, 112, 107-118.
Hadfield, J. D. 2010. MCMC methods for multi-response generalized linear mixed models: the
MCMCglmm R package. Journal of Statistical Software, 33, 1-22.
Harris, S., Ramnarine, I. W., Smith, H. G., & Pettersson, L. B. 2010. Picking personalities apart:
estimating the influence of predation, sex and body size on boldness in the guppy Poecilia
reticulata. Oikos, 119, 1711-1718.
Hayward, M. W. 2006. Prey preferences of the spotted hyaena (Crocuta crocuta) and degree of
dietary overlap with the lion (Panthera leo). Journal of Zoology, 270, 606-614.
Hedrick, A., & Kortet, R. 2012. Sex differences in the repeatability of boldness over
metamorphosis. Behavioral Ecology and Sociobiology, 66, 407-412.
Holekamp, K. E., Cooper, S. M., Katona, C. I., Berry, N. A., Frank, L. G., & Smale, L. 1997.
Patterns of association among female spotted hyenas (Crocuta crocuta). Journal of
Mammalogy, 78, 55-64.
Holekamp, K. E., & Smale, L. 1993. Ontogeny of dominance in free-living spotted hyenas:
juvenile rank relations with other immature individuals. Animal Behaviour, 46, 451-466.
Johnson, J. C., & Sih, A. 2005. Precopulatory sexual cannibalism in fishing spiders (Dolomedes
triton): a role for behavioral syndromes. Behavioral Ecology and Sociobiology, 58, 390-396.
Kolowski, J. M., & Holekamp, K. E. 2006. Spatial, temporal, and physical characteristics of
livestock depredations by large carnivores along a Kenyan reserve border. Biological
Conservation, 128, 529-541.
Kolowski, J. M., Katan, D., Theis, K. R., & Holekamp, K. E. 2007. Daily patterns of activity in the
spotted hyena . Journal of Mammalogy, 88, 1017-1028.
Kruuk, H. 1972. The spotted hyaena: a study of predation and social behavior. Chicago: Chicago
University Press.

50

Kruuk, L. E. B. 2004. Estimating genetic parameters in natural populations using the “animal
model”. Philosophical Transations of the Royal Society (B), 359, 873-890.
Lessells, C. M., & Boag, P. T. 1987. Unrepeatable repeatabilities—a common mistake. The Auk,
2, 116-121.
Lima, S. L. 1998. Stress and Decision Making under the Risk of Predation: Recent Developments
from Behavioral, Reproductive, and Ecological Perspectives. In: Advances in the Study of
Behavior: Stress and Hormones, Vol Volume 27 (Ed. by A. Møller, M. Milinski, & P. Slater),
pp. 215-290. Academic Press.
Lima, S. L., & Dill, L. M. 1990. Behavioral decisions made under the risk of predation: a review
and prospectus. Canadian Journal of Zoology, 68, 619-640.
Linnell, J. D., Odden, J., Smith, M. E., Aanes, R., & Sweonson, J. E. 1999. Large carnivores that
kill livestock: do“ problem individuals” really exist. Wildlife Society Bulletin, 27, 698.
López, P., Hawlena, D., Polo, V., Amo, L., & Martín, J. 2005. Sources of individual shy-bold
variations in antipredator behaviour of male Iberian rock lizards. Animal Behaviour, 69, 19.
Magnhagen, C., & Borcherding, J. 2008. Risk-taking behaviour in foraging perch: does
predation pressure influence age-specific boldness? Animal Behaviour, 75, 509-517.
Martin, J. G., & Réale, D. 2008. Temperament, risk assessment and habituation to novelty in
eastern chipmunks, Tamias striatus. Animal Behaviour, 75, 309-318.
Mills, M. G. L. 1990. Kalahari hyaenas: Comparative behavioral ecology of two species. London:
Unwin Hyman.
Mousseau, T. A., & Roff, D. A. 1987. Natural selection and the heritability of fitness
components. Heredity, 59, 181-197.
Niemelä, P. T., DiRienzo, N., & Hedrick, A. V. 2012. Predator-induced changes in the boldness
of naïve field crickets, Gryllus integer, depends on behavioural type. Animal Behaviour, -.
van Oers, K., Drent, P. J., de Jong, G., & van Noordwijk, A. 2004. Additive and nonadditive
genetic variation in avian personality traits. Heredity, 93, 496-503.
Ogutu, J. O., & Dublin, H. T. 2002. Demography of lions in relation to prey and habitat in the
Maasai Mara National Reserve, Kenya. African Journal of Ecology, 40, 120-129.
Pangle, W. M. 2008. Threat-sensitive behavior and its ontogenetic development in top
mammalian carnivores. Michigan State University.
51

Patterson, B. D., Kasiki, S. M., Selempo, E., & Kays, R. W. 2004. Livestock predation by lions
(Panthera leo) and other carnivores on ranches neighboring Tsavo National ParkS, Kenya.
Biological Conservation, 119, 507-516.
Pinheiro, J. C., & Bates, D. M. 2000. Mixed-effects Models in S and S-Plus. New York: SpringerVerlag.
Pratt, A. E., McLain, D. K., & Berry, A. S. 2005. Variation in the boldness of courting sand fiddler
crabs (Uca pugilator). Ethology, 111, 63-76.
Reale, D., Gallant, B. Y., Leblanc, M., & Festa-Bianchet, M. 2000. Consistency of temperament
in bighorn ewes and correlates with behaviour and life history. Animal Behaviour, 60, 589597.
Reale, D., Martin, J., Coltman, D. W., Poissant, J., & Festa-Bianchet, M. 2009. Male personality,
life-history strategies and reproductive success in a promiscuous mammal. Journal of
Evolutionary Biology, 22, 1599-1607.
Reale, D., Reader, S. M., Sol, D., McDougall, P. T., & Dingemanse, N. J. 2007. Integrating
animal temperament within ecology and evolution. Biological Reviews, 82, 291-318.
Riesch, R., Duwe, V., Herrmann, N., Padur, L., Ramm, A., Scharnweber, K., Schulte, M., SchulzMirbach, T., Ziege, M., & Plath, M. 2009. Variation along the shy–bold continuum in
extremophile fishes (Poecilia mexicana, Poecilia sulphuraria). Behavioral Ecology and
Sociobiology, 63, 1515-1526.
Sih, A., Bell, A. M., & Johnson, J. C. 2004a. Reply to Neff and Sherman. Behavioral syndromes
versus darwinian algorithms. Trends in Ecology & Evolution, 19, 622-623.
Sih, A., Bell, A., & Johnson, J. C. 2004b. Behavioral syndromes: an ecological and evolutionary
overview. Trends in Ecology & Evolution, 19, 372-378.
Sih, A., Bell, A., Johnson, Jc., & Ziemba, R. 2004c. Behavioral Syndromes: An Integrative
Overview. The Quarterly Review of Biology, 79, 241-277.
Sikes, R. S., & Gannon, W. L. 2011. Guidelines of the American Society of Mammalogists for the
use of wild mammals in research. Journal of Mammalogy, 92, 235-253.
Sinn, D. L., Apiolaza, L. A., & Moltschaniwskyj, N. A. 2006. Heritability and fitness-related
consequences of squid personality traits. Journal of Evolutionary Biology, 19, 1437-1447.
Sinn, D. L., Gosling, S. D., & Moltschaniwskyj, N. a. 2008. Development of shy/bold behaviour
in squid: context-specific phenotypes associated with developmental plasticity. Animal
Behaviour, 75, 433-442.
52

Sinn, D. L., & Moltschaniwskyj, N. A. 2005. Personality traits in dumpling squid (Euprymna
tasmanica): context-specific traits and their correlation with biological characteristics.
Journal of Comparative Psychology, 119, 99-110.
Smale, L., Frank, L. G., & Holekamp, K. E. 1993. Ontogeny of dominance in free-living spotted
hyaenas: juvenile rank relations with adult females and immigrant males. Animal
Behaviour, 46, 467-477.
Smith, B. R., & Blumstein, D. T. 2008. Fitness consequences of personality: a meta-analysis.
Behavioral Ecology, 19, 448-455.
Sundström, L. F., Petersson, E., Höjesjö, J., Johnsson, J. I., & Järvi, T. 2004. Hatchery selection
promotes boldness in newly hatched brown trout (Salmo trutta): implications for
dominance. Behavioral Ecology, 15, 192-198.
Taylor, R. W., Boon, A. K., Dantzer, B., Reale, D., Humphries, M. M., Boutin, S., Gorrell, J. C.,
Coltman, D. W., & McAdam, A. G. 2012. Low heritabilities, but genetic and maternal
correlations between red squirrel behaviours. Journal of Evolutionary Biology, 25, 614-624.
R Development Core Team, 2011. R: A Language and Environment for Statistical Computing.
Trinkel, M., Fleischmann, P. H., & Kastberger, G. 2006. Comparison of land-use strategies of
spotted hyenas (Crocuta crocuta, Erxleben) in different ecosystems. African Journal of
Ecology, 44, 537-539.
Vainikka, A., Rantala, M. J., Niemelä, P., Hirvonen, H., & Kortet, R. 2010. Boldness as a
consistent personality trait in the noble crayfish, Astacus astacus. Acta Ethologica, 14, 1725.
Van Horn, R. C., Engh, A. L., Scribner, K. T., Funk, S. M., & Holekamp, K. E. 2004. Behavioural
structuring of relatedness in the spotted hyena (Crocuta crocuta) suggests direct fitness
benefits of clan-level cooperation. Molecular Ecology, 13, 449-458.
Van Horn, R. C., McElhinny, T. L., & Holekamp, K. E. 2003. Age estimation and dispersal in the
spotted hyena (Crocuta crocuta). Journal of Mammalogy, 84, 1019-1030.
Watts, H. E., Blankenship, L. M., Dawes, S. E., & Holekamp, K. E. 2010. Responses of spotted
hyenas to lions reflect individual differences in behavior. Ethology, 116, 1199-1209.
Watts, H. E., & Holekamp, K. E. 2009. Ecological determinants of survival and reproduction in
the spotted hyena. Journal of Mammalogy, 90, 461-471.

53

Watts, H. E., Scribner, K. T., Garcia, H. A., & Holekamp, K. E. 2011. Genetic diversity and
structure in two spotted hyena populations reflects social organization and male dispersal.
Journal of Zoology, 285, 281-291.
Whitlock, M. 1998. The repeatability of fluctuating asymmetry: a revision and extension.
Proceedings of the Royal Society of London. Series B: Biological Sciences, 265, 1429-1431.
Wilson, D. S., Clark, A. B., Coleman, K., & Dearstyne, T. 1994. Shyness and boldness in humans
and other animals. Trends in Ecology and Evolution, 9, 442-446.
Wilson, D. S., Coleman, K., Clark, A. B., & Biederman, L. 1993. Shy-bold continuum in
pumpkinseed sunfish (Lepomis gibbosus): An ecological study of a psychological trait.
Journal of Comparative Psychology, 107, 250-260.
Wilson, A. J., Réale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A., Kruuk, L.
E. B., & Nussey, D. H. 2009. An ecologist’s guide to the animal model. Journal of Animal
Ecology, 79, 13-26.
Wolf, M., van Doorn, G. S., Leimar, O., & Weissing, F. J. 2007. Life-history trade-offs favour the
evolution of animal personalities. Nature, 447, 581-584.

54

CHAPTER 3
INDIVIDUAL DIFFERENCES IN AGGRESSIVENESS AMONG SPOTTED HYENAS

INTRODUCTION
Aggression is one of the most noticeable and fundamental aspects of behavior;
extensive research has been conducted on when, why, and how animals behave aggressively
toward one another (e.g. Smith and Price 1973; van den Berghe 1974; Francis 1988; Hemelrijk
2000; Panksepp and Zellner 2004). However, until recently, much of the focus of aggression
research was at the level of species (e.g. Grant & Ulmer 1974; Yasukawa & Searcy 1982; Thierry
1985; Digby 1999), while far fewer studies examined behavioral differences among individuals
within a population. In the last several years, research on animal personality has begun to
address consistent inter-individual differences in aggressive behavior, revealing that these
differences may be stable across time and contexts and can have a significant impact on fitness
(Dingemanse & de Goede 2004; Johnson & Sih 2005; Bell & Sih 2007; Boon et al. 2008; Sinn et
al. 2008).
Male animals are generally considered to be more aggressive than females, because in
many species, competition for mates selects for large, aggressive males (Geist 1974; Walters &
Seyfarth 1987; Archer 1988; Kunc et al. 2006). Thus, many studies of inter-individual
differences in aggressiveness have focused on species in which either males are the more
aggressive sex, or males and females emit aggressive behaviors at similar rates (e.g. Leshner
and Moyer 1975; Armitage 1986; Verbeek, Boon, and Drent 1996; Taylor et al. 2012). However,
females also often fight with one another over resources and in defense of their young, and the
55

aggressiveness of females may be generally underestimated (Sandell 1998; Bebié & McElligott
2006; Clutton-Brock 2007; Van Meter 2009). Extreme cases of aggressive behavior in females
present themselves in species with sex role reversals, in which females are socially dominant to
males (e.g. Kappeler 1990; Gwynne 1991; Delehanty et al. 1998; Berglund and Rosenqvist
2003). In these species, individual differences in aggressive behavior may have very different
patterns, and the fitness consequences of this variation may be very different from those
documented in species in which males dominate females. To date, we know of no studies
investigating the personality trait of aggression in a female-dominated mammal. Here we
examine inter-individual differences in aggression in a highly aggressive and female-dominated
carnivore, the spotted hyena (Crocuta crocuta). Hyenas offer an excellent model system in
which to study aggressive behavior because aggression in this species is very frequent, is easily
observable, varies in intensity, and occurs in several different contexts. Additionally, because
the hyenas in our study population have been monitored continuously since 1988, we have
genetic and fitness data with which we can assess the causes and consequences of
aggressiveness in this species.
Research on sex differences in the consistency of aggression has been inconclusive in
both humans and other animals. Whereas some studies (e.g. While et al. 2010) have found that
behavioral consistency does not differ between the sexes, a meta-analysis by Bell et al. (2009)
found that differences between males and females in behavioral repeatability vary based on
the species and the traits being studied. This suggests that, in terms of consistency, the optimal
strategy for each sex may be determined separately for each species based on specific aspects
of its environment, life history, and morphology.
56

Although there is a general consensus that inter-individual differences in aggressiveness
exist in many species (e.g. marmots: Armitage 1986; sticklebacks: Bakker 1986; great tits:
Verbeek, Boon, and Drent 1996; red squirrels: Boon, Reale, and Boutin 2008), there is much
debate as to whether aggressiveness is domain-general or context-dependent. In other words,
are individuals that are aggressive in one context also aggressive in another (i.e. domaingenerality), or is aggressiveness in different contexts unrelated (i.e. context-specificity)? In
female fishing spiders, aggression is stable across contexts; individuals that are highly
aggressive in a food-related context are also more likely to engage in precopulatory cannibalism
(Johnson 2001; Johnson & Sih 2005). However, research suggests that aggression in some
insect species is context-specific based on the identity of the opponent (Pfennig & Reeve 1989;
Tanner & Adler 2009). Contexts of aggression have also received a significant amount of
attention in humans; most psychologists agree that aggression in humans is context-specific
(for reviews, see Buss and Shackelford 1997 and Kagan 2003). However, there are few studies
addressing the effect of context on aggressive behavior in non-human mammals.
Artificial selection for short or long attack latency in house mice has been very
successful, indicating that there is a strong genetic component to aggressiveness (Benus et al.
1991; Miczek et al. 2001). A study of aggressiveness in captive vervet monkeys estimated the
heritability of aggressiveness in that species to be 0.64 (Fairbanks et al. 2004). However, other
studies, especially those with free-living subjects, have generally yielded lower estimates of
heritability. For instance, Taylor et al. (2012) estimated that only 12 percent of the phenotypic
variation in aggressive behavior observed in wild red squirrels could be explained by genetics,

57

and Bell (2005) estimated that the heritability of aggressiveness in some populations of
sticklebacks may be as low as 0.01.
Studies of the fitness consequences of aggressiveness have also produced conflicting
results; whereas aggression enhances reproductive success in some species, it decreases some
aspects of fitness in others. For instance, a study of aggressiveness in wild skinks showed that
the offspring of particularly aggressive females had enhanced survival, compared to the
offspring of less aggressive females (Sinn et al. 2008). By contrast, in western bluebirds,
aggressiveness in males was linked to lower reproductive success; males that defended their
nests most intensely fledged fewer offspring than other males, due to a tradeoff between
aggression and parental care (Duckworth 2006). Finally, Packer et al. (1995) found that despite
having shorter interbirth intervals and increased infant survival, aggressive female baboons of
high social rank had reduced fertility and a higher incidence of miscarriages.
Among hyenas, adult females are larger than adult males by about 10 percent, and are
socially dominant to immigrant males (Kruuk 1972). Within a clan, there is a rigid linear
dominance hierarchy, and social status is reinforced through frequent aggressive behaviors
(Kruuk 1972; Frank 1986; Mills 1990). Rank determines access to resources, and competition
among clan members for food is intense; aggression over food occurs frequently, as highranking females frequently direct food-related aggression toward lower-ranking clan members
(Kruuk 1972; Mills 1990; Frank et al. 1995). Aggressive behavior also commonly occurs in
contexts unrelated to food; in fact much aggression appears to be entirely unprovoked,
apparently functioning mainly to maintain the status quo in the social hierarchy (Van Meter
2009). Another common form of non-food-related aggression occurs during maternal
58

interventions, in which a mother directs aggression toward conspecifics in defense of her
offspring (Holekamp & Smale 1991; Engh et al. 2000).
The aggressive behavior of female hyenas differs markedly from that of male hyenas.
Female hyenas direct aggression toward conspecifics almost three times more often than do
males, although the rate at which female hyenas emit aggressions is not higher than that of
most female primates (Van Meter 2009). Additionally, although male hyenas exhibit intrasexual aggression at a low level (Van Meter 2009), with the exception of “baiting” behavior,
males very rarely direct aggressions toward females (see Szykman et al. 2003). Thus, although
we know that female hyenas are more aggressive than males, we lack information on variation
in aggressiveness within each sex; we do not know how aggressive behavior varies among
either females or males.
Previous research has shown that hormones influence aggressive behavior in hyenas via
maternal effects. Dloniak et al. (2006) found that cubs born to mothers with higher androgen
concentrations during pregnancy exhibited more aggressive behavior than those born to
pregnant females with lower androgen levels. Van Meter (2009) also found a positive
correlation between maternal fecal androgen levels during pregnancy and rates at which
offspring emitted aggressive acts as adults, suggesting an organizational role for hormones in
shaping aggressiveness in hyenas. However, beyond this maternal effect, we know very little
about the causes of inter-individual variation in aggression in this species.
Much of the variation in reproductive success among female spotted hyenas is driven by
social rank. High-ranking females begin breeding at younger ages and have shorter interbirth
intervals, longer reproductive life-spans, and more surviving offspring than do lower-ranking
59

females (Holekamp & Smale 1996; Swanson et al. 2011). However, low-ranking individuals
benefit more than high-ranking individuals do from high prey availability in terms of
reproductive success. Whereas high-ranking hyenas had enhanced reproductive success at all
times, low-ranking hyenas enjoyed better reproductive success when prey abundance was high
than when prey were scarce. Together, these results suggest that increased access to
resources drives reproductive performance in hyenas, so it is possible that more aggressive
hyenas experience better reproductive success than less aggressive ones due to a superior
ability to monopolize resources.
In this chapter, we examine two types of aggressive acts: food-related aggressions and
non-food-related aggressions. We first determine which situational and individual factors
influence aggressive behavior among hyenas. Then, we inquire whether there is evidence of
consistent differences among individuals in two measures of aggressiveness, and whether these
differences are heritable and subject to maternal effects. We also assess the stability of
aggressiveness over time and across contexts, and we conclude with an evaluation of the
relationship between aggressiveness and fitness.

METHODS
Study Subjects
Between 1988 and 2009, personnel from the Mara Hyena Project collected behavioral
data on a large group of free-living spotted hyenas, called the Talek Clan, in the Masai Mara
National Reserve, Kenya (Boydston et al. 2001). All hyenas in this clan were recognized
individually by observers based on their unique spot pattern and were of known sex, age, and
60

social rank. Hyenas were sexed based on the morphology of the glans of the erect phallus
(Frank et al. 1990). We assigned ages to all natal animals (to ±7 days) based on their
appearance when they were first seen above ground at dens, and we were able to assign the
ages of immigrant males reliably (to ± 6 months) with an age-estimation model (Van Horn et al.
2003).
The social rank of each hyena in the clan was assigned using a dominance matrix based
on observations of dyadic agonistic interactions (Holekamp & Smale 1993; Smale et al. 1993).
Rank was reassessed each time a new animal reached adulthood, entered or left the clan, or
died. Each cub was assigned the rank of its mother until it reached reproductive maturity at 24
months of age (Frank 1986b; Glickman et al. 1992; Holekamp et al. 1997); at that time young
females were assigned their own ranks in the adult hierarchy. We calculated standardized rank
by dividing each hyena’s numerical rank by the total number of ranked animals in the clan at
the time.
We established maternity based on genotyping and observations of cubs nursing
(Holekamp & Smale 1993). Paternity was assigned based exclusively on genotyping, as
previously described (Van Horn et al. 2004; Watts et al. 2011). Methods used to immobilize
animals and collect blood samples used for DNA extraction are described in detail in Engh et al.
(2002). All sampling procedures were approved by the Institutional Animal Care and Use
Committee at Michigan State University (AUF 07/08-099-00), complied with Kenyan law, and
met guidelines approved by the American Society of Mammalogists (Sikes & Gannon 2011).

61

Data collection
We generally observed hyenas daily from vehicles between 0530 and 0930 hours and
between 1700 and 2000 hours. We initiated an observation session whenever we saw one or
more hyenas separated from others by at least 200 meters. Upon initiating an observation
session, we recorded the GPS location of the session and conducted a behavioral scan that
identified each hyena and noted its current behavior; these scans were repeated every 20
minutes. During each session, we used critical incident sampling (Altmann 1974) to record
specific behaviors such as aggressive acts, appeasements, vocalizations, and greetings. We also
recorded whenever a hyena entered or left a session in progress.
We used three different measures of aggressiveness: (1) the intensity of particular
aggressive acts within observation sessions, (2) the aggression ‘count’ for each individual within
each observation session, and (3) the lifetime hourly aggression rate calculated over all
observations sessions in which an individual was seen during its adulthood. Aggression
intensity was a measure of the severity of an aggressive act, measured on a scale of 1 to 3.
Aggressive acts with intensities of 1 were low-intensity threats characterized by an intention
movement to bite. Aggressive acts with intensities of 2 were intermediate-intensity threats
involving snapping or lunging, but no actual contact. Aggressive acts with intensities of 3 were
high-intensity threats involving a biting attack. An aggression count measured how many times
a hyena emitted aggressive acts during a session. Following Van Meter (2009), we also
calculated an hourly adjusted aggression rate for each individual present in each session as:
number of aggressive acts ÷ number of potential targets
number of hours the session lasted
62

where the number of potential targets was a count of all lower-ranking hyenas present. This
measure controlled for variation in the number of opportunities each hyena had to emit
aggressive acts during each session. To calculate a lifetime aggression rate for each hyena, we
averaged its adjusted aggression rates in all its sessions after reaching reproductive maturity at
24 months of age.
Aggressions were categorized as either food-related or non-food-related, and
aggression intensities, counts, and rates for these two contexts were treated separately. Foodrelated aggressions included all aggressive behaviors that took place in the presence of food.
All other aggressions were considered non-food aggressions. In this study, we used the
abbreviation NFA (for “non-food aggression”) to refer to aggressive acts occurring under
circumstances not associated with food, and FA (for “food-related aggression”) to refer to
aggressive acts occurring in the presence of food at dens, fresh kills, and at older carcasses.
Only dyadic aggressive interactions, and only aggressive acts directed down the
hierarchy (i.e. at lower-ranking hyenas), were included in our analyses. When individuals had
no lower-ranking targets present in a session, we did not include aggression counts or rates for
them in that session, as they could not emit any aggressions down the hierarchy. Additionally,
for adult females, we only included sessions in which an individual was lactating, because
aggression levels vary with reproductive state among female hyenas (Van Meter 2009).
Females were considered to be lactating between the date of birth of their last litter and the
first of three milestones: (1) the date of conception of the next litter, (2) the date of weaning of
the last litter, or (3) the disappearance of the last litter (Holekamp & Smale 1996).

63

Specific situational data were collected from all observation sessions, including the
location of the session (a fresh kill, an old carcass, the communal den, a mating event, or
“other” location), the time of day (either from midnight to noon, or from noon until midnight),
the duration of the session, the maximum number of hyenas present, and the seasonal prey
availability (high for June through October, and low from November through June).
We calculated the age (in months) and standardized rank of each hyena present in each
session. Additionally, for each hyena in each session, we divided the number of lower-ranking
hyenas present by the total number of hyenas present in the session; this measure is referred
to as the proportion of hyenas present that were potential targets of aggression.

Statistical Analysis
Bayesian analyses
For many of our analyses, we adopted a Bayesian approach and created models with the
Markov Chain Monte Carlo for Generalized Linear Mixed Models (MCMCglmm) analysis tool in
the R statistical package (Hadfield 2010; R Core Development Team 2011). For each of these
analyses, we tested fixed effects for significance by determining whether or not the 95%
credible interval for the fixed effect included zero; variables for which the credible interval
included zero were not significant predictors of the response variable, whereas those that did
not include zero were considered significant. We tested random effects for significance by
comparing the deviance information criterion (DIC) values of models fitted with and without
the random effect of interest; if the inclusion of a random effect lowered the DIC value of the
model, the random effect was considered significant in predicting the response variable.
64

For zero-inflated MCMCglmm models, we began by using a saturated model with fixed
effects contributing to both the poisson and zero-inflated portions of the model, but only
retained zero-inflation terms that lowered the DIC value of the model. In the zero-inflated
portion of the model, the estimates for fixed effects were interpreted as the log(e) of the ratio
of the odds of the response variable not occurring compared to the odds of the response
variable occurring.
For all MCMCglmm models, we used a relatively uninformative prior that partitioned
variance equally among all random effects (Hadfield 2010); using a more informative prior did
not affect reported results. We assessed convergence of these models by examining time
series of the model parameters. The estimates are reported as posterior means, with 95%
credible intervals. We also inspected autocorrelation values for these models, all of which were
less than 0.1 for reported values.

Situational factors influencing aggressiveness
We assessed aggressiveness based on two different levels of analysis; one at the level of
the observation session, and another at the level of individual hyenas. First, we performed
analyses using the observation session as the unit of analysis to determine which social and
environmental factors influenced hyenas’ aggressiveness.
Aggression count
For each session, we counted the total number of NFAs emitted. In sessions where food
was present, we also counted the total number of FAs emitted during the session. These
measures represent the general food-related and non-food-related aggressiveness of hyenas
65

present in each session. We used these two measures as response variables in two zeroinflated MCMCglmm models. We ran separate session-level models for FAs and NFAs, because
particular situational factors may influence these two types of aggressions in different ways.
We included various situational aspects of the session as fixed effects, including the location
and duration of the session, the time of day, the maximum number of hyenas present, and the
seasonal prey availability.
Data for these analyses included only aggressive acts emitted by adult hyenas. To
reduce the number of zeros and encourage model convergence, in these analyses we did not
include sessions that were less than ten minutes long. The posterior distribution of the model
was sampled every 500 iterations after a burn-in period of 20,000 iterations for a total of 2,000
samples. We used a one-way Kruskal Wallis ANOVA test to determine whether the mean FA
count and mean NFA count differed significantly.
Aggression intensity
We also ran separate session-level models using the average FA intensity and the
average NFA intensity exhibited by all hyenas in the session as the response variables, with the
same fixed effects described above. Only aggressive acts emitted by adults were included in this
analysis, and sessions of all lengths during which aggression occurred were included. The
posterior distribution of the model was sampled every 500 iterations after a burn-in period of
15,000 iterations for a total of 2,000 samples. We used a one-way Kruskal Wallis ANOVA test to
determine whether the mean FA intensity and mean NFA intensity differed significantly.

66

Consistency in aggressiveness within individuals
We then performed similar Bayesian analyses using individual hyenas as the units of
analysis to (1) determine which characteristics of individuals affected aggressiveness, and (2)
determine whether there was consistency within individuals in aggressiveness across repeated
observations.
Aggression count
In the first two repeated measures analyses, we used the FA and NFA counts for each
hyena in each session as the response variables in zero-inflated poisson MCMCglmm models.
We included the sex of the hyena, its standardized social rank, its age at the time of the session,
and the proportion of targets in the session as fixed effects in these individual-level models. In
these models, we only included adults appearing in at least 20 sessions with potential targets
for either FAs or NFAs. Because of the large number of zeros, we only included aggression
counts from sessions that were at least 30 minutes long, and we omitted data from sessions in
which no aggressive interactions occurred.
A unique session identifier was included as a random effect to control for
pseudoreplication. To determine whether individual hyenas were consistent in their
aggressiveness, we also included a random effect representing the identity of the aggressor. If
this random effect of individual identity significantly improved the fit of the model, individuals’
behavior was considered consistent across sessions. The posterior distribution of the model
was sampled every 500 iterations after a burn-in period of 20,000 iterations for a total of 2,000
samples.

67

Aggression intensity
We then created similar individual-level models for FAs and NFAs emitted by adults,
using the intensity of each type of aggressive act as the response variables, with the same
individual-level fixed effects as described above. We carried out separate analyses for males
and females. The posterior distribution of the model was sampled every 100 iterations after a
burn-in period of 10,000 iterations for a total of 1,000 samples.

Heritability of aggressiveness
To determine whether or not aggressiveness is heritable in hyenas, we created
restricted-likelihood Bayesian animal models for several measures of aggressiveness (Kruuk
2004). Animal models use pedigree information to partition the heritable and non-heritable
components of phenotypic variance (Kruuk 2004; Wilson et al. 2009). Phenotypic, additive
genetic, maternal, and permanent environmental variances were estimated with univariate
trait animal models predicting NFA and FA measures for each hyena. Permanent environment
effects are created by each individual’s environment and are consistent over its lifetime;
maternal effects are non-genetic effects that a mother has on all her offspring. Because female
hyenas mate with several males across their reproductive history, maternal effects can be
distinguished from heritability via the effect of these different sires.
We carried out separate analyses for FA intensity, NFA intensity, lifetime FA rate, and
lifetime NFA rate. In the intensity models, we used the same fixed effects as in the individuallevel models described above. In the lifetime aggression rate models, we included only sex and
lifetime standardized rank as fixed effects. We retained only fixed effects that contributed
68

significantly to the model. In all animal models, we included random effects representing the
identity of the hyena, the identity of the hyena’s mother, and a term representing the additive
genetic effect in order to estimate the desired parameters of each model. For the aggression
intensity models, we also included a random effect of the session identifier.
Because MCMCglmm requires values for all random effects, we generated unique
maternal identities for immigrant males and natal animals born before our study began,
following Taylor et al. (2012). This technique allows for the estimation of maternal effects,
even when pedigree information is missing; however, it assumes that all hyenas with unknown
maternity are unrelated.
For each measure of aggressiveness, we calculated heritability, maternal effects, and
permanent environment effects by dividing the relevant variance component by the total
phenotypic variance. Following Lessels and Boag (1987), we calculated repeatability by dividing
the among-individual variance by the total among- and within-individual variance.
The posterior distribution of the animal model was sampled every 500 iterations after a
burn-in period of 20,000 iterations for a total of 5,000 samples.

Stability over time and across aggression contexts
To assess whether the aggressiveness of individual hyenas remained stable over time,
we created generalized linear mixed models using the lme4 function in R. To compare
aggressive behavior early in life to that later in life, we included only aggressive acts emitted
when the aggressor was either a cub (less than 12 months of age) or an adult (more than 24
months of age). For cubs, we included all aggressive acts, and for adults we included only

69

aggressive acts directed at adults but not in the presence of food; this way, we could assess
whether the general level of aggressiveness early in life predicted a more specialized type of
aggressive behavior during adulthood. Only hyenas observed to emit at least two aggressive
acts during each age interval (cub and adult) were included in this analysis.
In these models, we used aggression intensity as the response variable and age class as
the main fixed effect of interest. The sex, standardized social rank and age of the aggressor and
the proportion of targets in the session were also included as fixed effects. Only fixed effects
that lowered the Akaike information criterion (AIC) value of the model were retained in the
final model. A unique session identifier was included as a random effect in all models to
control for pseudoreplication.
We included individual as a random intercept to allow for the variation in
aggressiveness between hyenas, as suggested by earlier analyses. We also added a random
slope to the model, which described the extent to which the aggressiveness of individual
hyenas changed across ontogeny. If this term significantly improved the fit of the model, we
concluded that aggressiveness in hyenas was not a trait that changed consistently among
individuals across age classes. To test the significance of this term, we created two models that
were identical except for the inclusion of the random slope. We then compared these two
models with a log-likelihood ratio test, in which the log-likelihood ratio was calculated as 2[loglikelihood of model B - log-likelihood of model A], and tested as a chi-squared distribution
(Pinheiro & Bates 2000; Martin & Réale 2008). The approach of using random regression to
investigate behavioral reaction norms is described in more detail in Dingemanse et al. (2010).

70

Using the same methodology, we ran a set of similar models to test whether the
aggressiveness exhibited by individual hyenas was context-specific or domain-general. If
aggressiveness in hyenas is context-specific, hyenas that are relatively aggressive over food
should also be more aggressive than other hyenas in non-food contexts. In these models, we
used aggression context (either “food” or “non-food”) as the main fixed effect of interest, and
defined the random slope to allow the aggression intensity of an individual to differ across
contexts. Only adult female hyenas with at least 50 FAs and 50 NFAs were included in this
analysis.

Relationship between aggressiveness and fitness
Finally, we tested whether a hyena’s aggressiveness was related to its fitness, as
indicated by various measures of reproductive success. For these analyses, we used NFA and
FA lifetime aggression rates as our measures of aggressiveness. The only hyenas included in
this analysis were females that were born after our study began, died before our study ended,
and were present in at least 50 sessions with potential targets for FAs and at least 50 sessions
with potential targets for NFAs.
As indicators of reproductive success, we used the total number of cubs borne by each
female over her lifetime, as well as the proportion of her cubs that survived to three life-history
milestones: den independence, weaning, and reproductive maturity (24 months of age). We
created two sets of linear and second-order polynomial models with these fitness components
as the response variables, and maternal rank and either lifetime FA rate or lifetime NFA rate as

71

predictors, as well as an interaction term between the lifetime aggression rate and maternal
rank. All variables in these analyses were Z-transformed.
We also assessed whether lifetime aggression rates of females were related to the
amount of time their offspring spent feeding at kills. For this analysis, we included only
offspring younger than 36 months of age, because until this age, hyenas are at a developmental
feeding disadvantage (Watts & Holekamp 2009; Tanner et al. 2010). For all sessions in which
hyenas less than 36 months of age with known mothers were observed at carcasses or kills with
conspecifics, we calculated the number of minutes these young hyenas fed, based on
behavioral scan data. Scans were conducted, when possible, at regular 15-minute intervals.
We created an MCMCglmm model predicting the number of minutes the hyena fed, and
included as fixed effects the total number of scans (i.e. the potential minutes during which the
hyena could have been observed feeding), the sex and age of the hyena, its standardized rank
in the session, the total number of hyenas present, the proportion of higher-ranking hyenas
present, whether or not the cub’s mother was present in the session, and the estimated size
category of the prey (less than 100 kg, between 100 kg and 200 kg, between 200 kg and 500 kg,
or more than 500 kg). A random effect representing the session identifier was included to
control for pseudoreplication.
First, we added, and tested the significance of, a random effect representing the identity
of the mother to determine whether her identity affected the feeding success of her offspring
in some way. Then, we removed the random effect of maternal identity and added a fixed
effect representing the lifetime aggression rate of the mother. We created two similar models,
one with her lifetime NFA rate and one with her lifetime FA rate. Significance of either of these
72

terms would suggest that the aggressiveness of female hyenas can affect the feeding success of
their offspring. In this analysis, we only included the offspring of mothers that were present in
at least 50 sessions with potential targets for FAs, and in at least 50 sessions with potential
targets for NFAs.

RESULTS
Social and environmental factors influencing aggressiveness
Aggression count
There were 8126 sessions that met our criteria for the FA count analysis. On average,
there were 0.529±0.02 FAs per session and in 6738 (82.9%) of these sessions, hyenas did not
emit aggressions in the context of food. Hyenas emitted more FAs in longer sessions than
shorter sessions, in sessions with fewer hyenas present than those with more hyenas present,
and in sessions between noon and midnight, compared to sessions earlier in the day (Appendix,
Table 3.A1). The FA count in each session was higher during seasons of low prey availability
than during seasons of high prey availability, and there were more FAs at fresh kills than dens,
but fewer FAs at old carcasses than at dens (Appendix, Table 3.A1).
In the same time frame, there were 14735 sessions that met our criteria for inclusion in
the NFA count analysis. On average, there were 0.637±0.016 NFAs per session; hyenas did not
emit any NFAs in 10766 (73.1%) of these sessions. The mean NFA count was significantly
greater than the mean FA count (Kruskal-Wallis: H=238.561, df=1, p<0.001). There were more
NFAs when seasonal prey availability was low than when it was high, and more NFAs at kills,
dens, and at mating events than at other locations (Appendix, Table 3.A2). NFA counts were
73

higher in longer than shorter sessions, and higher in sessions with more hyenas present than in
those with fewer hyenas (Appendix, Table 3.A2).
Aggression intensity
Between 1988 and 2009, there were 859 sessions with FAs that met our criteria for
inclusion in the intensity analysis; the average intensity of these FAs was 1.84±0.02. During this
time, there were 3038 sessions with NFAs that met our criteria for inclusion in this analysis.
The intensity of the NFAs in these sessions averaged 1.59±0.01. Average FA intensities were
significantly greater than mean NFA intensities (Kruskal-Wallis: H=133.818, df=1, p<0.001).
Intensities of both FAs and NFAs were greater between noon and midnight than earlier
in the day (Appendix, Tables 3.A3 and 3.A4). NFAs emitted at dens and during mating events
were more intense than those emitted at other locations (Appendix, Table 3.A4). The
intensities of FAs did not differ significantly between dens and fresh kills, but FAs emitted at old
carcasses were less intense than those emitted at dens (Appendix, Table 3.A3). NFA intensity
increased as the number of hyenas present increased, and there was a similar, but nonsignificant, trend for FA intensity to increase as the number of hyenas present increased
(Appendix, Tables 3.A3 and 3.A4).

Consistency in aggressiveness within individuals
Aggression count
There was no evidence for consistency within individuals in either FA or NFA counts;
including a random effect of the identity of the aggressor did not improve model fit (food:

74

DIC=14528.05; non-food, DIC=26035.88) over models that did not include this random effect
(food: DIC=12172.34; non-food, DIC=27319.11).
Both FA and NFA counts increased as the proportion of lower-ranking hyenas in the
session increased (Appendix, Tables 3.A5 and 3.A6). Older females emitted more NFAs than did
younger females, higher-ranking hyenas emitted more NFAs than did lower-ranking hyenas,
and females emitted more NFAs than did males (Appendix, Table 3.A6).
Aggression intensity
For aggression intensity, including a random effect of the identity of the aggressor
significantly improved the model for each aggression context in both sexes (Table 3.1),
suggesting that there are consistent inter-individual differences in the intensity of FAs and NFAs
emitted by both male and female hyenas.

Table 3.1
DIC values yielded by separate MCMCglmm models run with NFAs and FAs emitted by adults
(≥24 months of age) of each sex. DIC values are given for models with only a random effect of
the session identifier, and for models run with both a session identifier and an aggressor
identifier. For each context in each sex, the better model is marked with an asterisk.
Random effects included
Sex and aggression context
Session
Aggressor and Session
Females, non-food
8311.299
8248.891*
Females, food
5274.435
5242.795*
Males, non-food
1091.263
1065.257*
Males, food
1232.201
1223.305*

Younger females emitted higher-intensity NFAs than did older females, but age had no
effect on the intensity of FAs emitted by females (Appendix, Tables 3.A7 and 3.A8). FAs emitted
by females were of higher intensity when there was a greater proportion of lower-ranking

75

hyenas present than when there were fewer lower-ranking hyenas present (Appendix, Table
3.A7). No fixed effects significantly predicted the intensity of either FAs or NFAs emitted by
males; however, in each context, there was a non-significant trend for intensity to increase as
the proportion of lower-ranking hyenas increased (Appendix, Tables 3.A9 and 3.A10)

Heritability of aggressiveness
Lifetime aggression rate
For lifetime FA rates, both heritability and maternal effects were significant, as
indicated by a lower DIC value when these effects were included in a model, compared to when
they were omitted (Appendix, Table 3.A11). The heritability estimate from the MCMCglmm
animal model for lifetime FA rates was 0.260 (95% CI=0.093 to 0.593, Figure 3.1) and the
maternal effect estimate was 0.128 (95%CI=0.059 to 0.277, Figure 3.1).
In terms of lifetime NFA rates, both heritability and maternal effects were significant,
as indicated by a lower DIC value when these effects were included in a model, compared to
when they were omitted (Appendix, Table 3.A11). The heritability estimate from the
MCMCglmm animal model for lifetime NFA rate was 0.320 (95% CI=0.065 to 0.590, Figure 3.1),
and the maternal effect estimate was 0.291 (95%CI=0.079 to 0.543, Figure 3.1).
Aggression intensity
Animal model estimates for aggression intensities were much lower than animal model
estimates for lifetime aggression rates, and in some cases were non-significant. In terms of FA
intensity, heritability, maternal effects, and permanent environment effects were significant, as
indicated by a lower DIC value when these effects were included in a model, compared to when
76

0.70
0.61

Proportion of variance explained

0.60

0.50

0.39

0.40

Food-related
aggression rate

0.32
0.30

0.29
Non-food related
aggression rate

0.26

0.20
0.13
0.10

0.00
Heritability

Maternal effect

Unexplained

Figure 3.1. The percent of variation in lifetime food-related and non-food aggression rates
attributable to heritability and maternal effects, and the remaining unexplained variation.
Variances were estimated with MCMCglmm animal models.

77

they were omitted (Appendix, Table 3.A11). The heritability estimate from the MCMCglmm
-5

animal model for FA intensity was 0.042 (95% CI=2.506x10 to 0.102), the permanent
-4

environment effect was 1.181x10 (95% CI=7.011x10
-5

estimate was 6.969x10 (95%CI=1.587x10

-10

-11

to 0.065), and the maternal effect

to 0.0602).

For NFA intensity, heritability and maternal effects were significant, as indicated by a
lower DIC value when these effects were included in a model, compared to when they were
omitted (Appendix, Table 3.A11). Permanent environment effects for this trait were not
significant (Appendix, Table 3.A11). For NFA intensity, the heritability estimate was 0.026 (95%
-4

CI=0.003 to 0.080) and the maternal effect estimate was 0.044 (95%CI=5.299x10 to 0.116).
The estimate for the repeatability of FA intensity was 0.067 (95% CI=0.013 to 0.126),
and the estimate for the repeatability of NFA intensity was 0.0263 (95% CI=0.003 to 0.080).
These repeatability values suggest that aggression intensity in each context is significantly
repeatable, but there is still a great amount of within-individual variation in this measure.

Stability over time and across contexts
Fifteen female hyenas met our criteria for inclusion in the age class analysis. The model
including a random slope of age class in addition to the random intercept was not significantly
better at predicting the aggressiveness of individuals than the model with only the random
intercept (LRT=2.199, df=2, p=0.333, Figure 3.2a). The inclusion of a random slope in this
model was not significant, indicating that the intensities of aggressions emitted by individual
hyenas generally change in a consistent way across ontogeny.

78

a

b

Figures 3.2a and 3.2b. Predicted behavioral reaction norms for aggression intensity across (a)
two age classes and (b) two aggression contexts. Each line represents how the behavior of an
individual changes across categories.
79

Thirty one females met our criteria for inclusion in the context analysis. For this
analysis, including a random slope of aggression context did improve the fit of the model (LRT=9.076, df=2, p=0.011, Figure 3.2b). The inclusion of a random slope was significant, indicating
that the aggression intensity in hyenas is context-specific, rather than domain-general.

Relationship between aggressiveness and fitness
Aggression intensity
Thirty-one females met the criteria for our analysis of the relationship between
aggression intensity and fitness. Generalized linear models and second-order polynomial
models showed no relationship between either FA or NFA intensity and any measure of
reproductive success, and there were no significant interactions between reproductive success
and aggression intensity (p>0.1).
Lifetime aggression rate
Thirty-two females satisfied the criteria for inclusion in our analysis of the relationship
between aggression rate and reproductive success. We found that the lifetime NFA rate of a
female is significantly related to the proportion of her cubs that survive to 24 months of age;
the offspring of females with higher lifetime NFA rates are more likely to survive to
reproductive maturity than the offspring of other females (0.719±0.300, t=2.401, p=0.025).
There was a similar, although non-significant, relationship between a female’s lifetime NFA rate
and the proportion of her cubs that survive to weaning (0.512±0.296, t=1.733, p=0.096).
We did not find any direct relationship between any measure of reproductive success
and lifetime FA rate (p>0.1). However, we found several significant interactions between
80

lifetime FA rate and maternal rank. Whereas high-ranking hyenas had consistently high
reproductive success regardless of their lifetime aggression rates, low-ranking females with
high lifetime FA rates had better reproductive success than low-ranking females that were less
aggressive in feeding contexts. This interaction was significant for the proportion of cubs
surviving to weaning (0.690±0.253, t=2.725, p=0.012) and the proportion of cubs surviving to
reproductive maturity (0.651 ±0.269, t=2.416, p=0.024). No second-order polynomial terms
were significant predictors of reproductive success, and second-order polynomial models were
not significantly better than linear models at predicting the reproductive success of females
(p>0.1).
In our analysis of time spent feeding at kills, we found that maternal identity was an
important predictor of offspring feeding time at kills, even after controlling for social rank and
the presence of higher-ranking hyenas. Including a random effect of maternal identity
improved the model (DIC=6713.464), compared to one without maternal identity
(DIC=6735.044). Furthermore, after removing the maternal identity term from the model, the
lifetime FA rate of the mother was a significant predictor of offspring feeding time; the
offspring of mothers with higher lifetime FA rates fed longer among conspecifics at kills than
did the offspring of mothers with lower lifetime FA rates (0.846, 95% CI=-0.028 to 1.571,
p=0.05). The lifetime NFA rate of the mother was not a significant predictor of offspring
feeding time (p=0.775).

81

DISCUSSION
Overall, our analyses show that there are consistent inter-individual differences among
spotted hyenas in some measures of aggressiveness, and that aggressiveness has significant
implications for the reproductive success of females. We found that males and females are
consistent in the intensity of their aggressive behavior, both in the context of food and in
situations unrelated to food. However, we did not find that the number of aggressive acts
emitted by hyenas in either context were consistent across repeated observations.
For both food and non-food-related aggression contexts, the location of the observation
session and several other situational variables were highly significant predictors of both the
number and intensity of aggressive acts observed. The number of FAs and the number and
intensity of NFAs increased with the number of hyenas present in the session. Hyenas tended
to emit more food-related and non-food-related aggressions during seasons when prey
availability was low, compared to seasons where food was more abundant. However,
aggression intensity was not related to seasonal prey availability.
At the level of individual, adults emitted more food-related and non-food-related
aggressions as the proportion of lower-ranking hyenas in a session increased, indicating that
hyenas generally become more aggressive when more targets of aggression are present. Males
emitted fewer aggressions than females did in non-food contexts, but not in food-related
contexts, which supports a similar finding by Van Meter (2009) that the rates of intra-sexual
aggression over food did not differ between the sexes. None of our individual-level predictor
variables were significant predictors of the intensity of aggressive acts emitted by males,
indicating that male aggressiveness may be more strongly driven by the situational variables
82

and personality traits than by individual attributes at the time of the session, such as age or
social rank.
Among spotted hyenas, aggression intensity appears to be context-specific; that is,
hyenas that emitted particularly intense aggressions over food did not necessarily also emit
intense aggressions in contexts away from food. This resembles findings in humans (e.g. Kagan
2003), and suggests that food-related and non-food-related aggression emitted by hyenas likely
have very different functions, and that the mechanisms mediating these two types of
aggressions may differ.
Unlike aggression intensity, which was consistent for both males and females across
repeated observations, we found that aggression counts for individual hyenas were not
consistent across sessions. The number of aggressions exhibited by a hyena in a session may be
strongly driven by the situational aspects of the session included in our analyses, as well as
others that we could not account for. For instance, group composition is an important
determinant of whether or not hyenas attack each other (Smith et al. 2010); whereas we were
able to control for the proportion of lower-ranking hyenas present in our analyses, we could
not control for the presence of specific hyenas that may have incited or suppressed aggression,
or for combinations of individuals that may have been particularly volatile. Moreover,
especially in regard to food-related aggression, motivational state is likely to vary between
sessions and drive aggressive behavior (Stocker & Huber 2001; Hodge et al. 2009).
When aggression counts were averaged over time in our lifetime aggression rate
measure, there were significant heritable differences among individuals, and aggressive
behavior was subject to significant maternal effects. As noted previously, differences between
83

heritability and maternal effect estimates were due to sire effects within dams. The average
heritability of behavioral traits is 0.30±0.02 (Mousseau & Roff 1987), and our heritability
estimates for aggressiveness in both contexts were close to this average. Our maternal effect
estimate for lifetime non-food aggression rate was quite substantial; roughly a third of the
variation in this type of aggressive behavior could be explained by maternal effects. The
maternal effect estimate for food-related aggression rate was lower, but still larger than
maternal effect estimates for aggressive behavior in some other free-living mammals (e.g.
Taylor et al., 2012), indicating that maternal effects play a substantial role in multiple types of
aggressive behavior in hyenas. This is consistent with earlier findings by Dloniak et al. (2006)
and Van Meter (2009), who found significant correlations between circulating maternal fecal
androgen levels during gestation and the aggression rates exhibited by offspring. Our significant
heritability and maternal effect estimates for lifetime aggression rates, along with our finding
that aggression intensity was stable across age classes, suggest that some aspects of aggressive
behavior are determined very early in life.
Interestingly, for aggression intensity, our estimates for heritability, maternal effects,
and permanent environment effects were extremely small, despite evidence of consistent
differences across observation sessions. Again, subtle or unmeasurable differences between
sessions may have contributed to these very low estimates. Our low repeatability estimates for
these measures suggest that, although individual differences in aggression intensity are
statistically consistent across observation sessions, there is still much within-individual variation
across repeated observations in this measure of aggressiveness. Differences in aggressiveness
may be difficult to observe in hyenas on a session-by-session basis in the wild. However, over
84

time, inter-individual differences in aggression in this species become evident and have
significant implications for fitness.
When we used lifetime aggression rates to predict the reproductive success of females,
we found that the offspring of females with high non-food aggression rates enjoyed better
survival than did offspring of females that are less aggressive in non-food situations. Mortality
rates are high before 2 years of age (Watts & Holekamp 2009), and maternal interventions and
other aggressive behaviors that are emitted outside the context of food likely play an important
role in defending and protecting offspring during vulnerable life history stages. Furthermore,
we found a significant interaction between lifetime food-related aggression rate and maternal
rank, indicating that frequent food-related aggressions emitted by an adult female may
increase the likelihood that her cubs will survive to weaning and reproductive maturity, and
that this beneficial effect of aggressiveness increased as rank decreased. Hyenas that are
particularly aggressive over food are likely to displace conspecifics from kills more often,
thereby increasing their own ability to feed. Reproductive success is determined by access to
resources among female hyenas (Holekamp & Smale 1996), and it appears that females can
enhance their reproductive success by increasing their food intake. The advantages accruing
from food-related aggression increased as rank decreased, indicating that better access to food
may be particularly beneficial for lower-ranking individuals. Holekamp and Smale (1996)
suggested that high-ranking female hyenas were buffered from the consequences of fluctuating
prey availability due to their priority access to resources; a similar effect may be at work here,
as increased aggression over food may not benefit higher-ranking females as much as it does
lower-ranking females.
85

The results of our offspring feeding time analysis suggest that increased tolerance of a
female’s offspring at kills may be a benefit of high rates of food-related aggression emitted by
mothers. We found that the offspring of females that were particularly aggressive over food
were able to feed longer among conspecifics than could the offspring of females that were less
aggressive over food. Our results suggest that hyenas up to 36 months of age may benefit in
this way from having more aggressive mothers. Until about three years of age, young hyenas
are at a competitive disadvantage during group feeding events because their skulls and feeding
apparatuses have not yet finished developing (Watts et al. 2009; Tanner et al. 2010). It is
thought that this protracted developmental period, in tandem with the intense feeding
competition seen in hyena clans, favored the evolution of large, aggressive females (Watts et al.
2009). Our results suggest that female aggressiveness in food-related contexts might mitigate
this competitive disadvantage for offspring. The rather unusual set of constraints imposed by
slow morphological development and intense feeding competition may be the reason that
aggressiveness has a positive effect on reproductive success in hyenas, whereas high levels of
aggression decrease some aspects of fitness in other species (e.g. Packer et al. 1995;
Duckworth 2006).
Further studies investigating individual differences in both food-related and non-food
aggression should inform our understanding of the effects of aggressiveness on fitness, as well
as the interaction between rank and aggressiveness. Although we have shown here that rankindependent differences in aggressiveness among hyenas drive some variation in reproductive
success, social rank is still the most important predictor of reproductive success (Holekamp &
Smale 1996; Swanson et al. 2011). Interestingly, we see definite differences in the fitness
86

benefits of aggressiveness for low- and high-ranking hyenas. We hypothesize that these
differences are due to the effect of rank on resource acquisition; whereas high-ranking females
are always able to displace conspecifics from kills, increased aggressiveness may improve the
resource holding potential of lower-ranking females to a greater degree. Future studies
focusing on the behavior of individuals at group feeding events may be able to clarify the
relationship between aggression rate, food intake, and social rank. For instance, it is currently
unknown whether a high aggression rate at a kill directly increases the time an individual
spends feeding in that session, and whether this effect varies with social rank. Alternatively, it
is possible that a “history” of frequent aggressions over food is enough to increase an
individual’s food consumption at a kill. Additionally, although we did not find a negative
relationship between aggressiveness and reproductive success, it is possible that high rates of
aggression may have associated costs in other areas. For example, in some species, individuals
prone to aggressive behavior experience increased metabolic demands (e.g. Matsumasa &
Murai 2005) and injury rates (e.g. Wilson & Boelkins 1970) compared to less aggressive
individuals; it is not yet clear whether high levels of aggression have similar consequences for
hyenas.

87

APPENDIX

88

Table 3.A1
Fixed effects predicting the total FA count in each session. Only aggressive acts directed by
adults (≥24 months of age) down the hierarchy toward other adults are included. Significant
fixed effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
Poisson
(Intercept)
Location (carcass)
Location (kill)
Time of day (PM)
Number of hyenas present
Session length
Prey availability (low)
Zero-inflated
(Intercept)
Session length
Prey availability (low)

2.2977
-4.37379
3.11857
0.47438
0.03666
0.74779
0.19324

(2.01705 to 2.58161)
(-4.87854 to -3.78269)
(2.85094 to 3.36689)
(0.31709 to 0.65134)
(0.02439 to 0.04805)
(0.59287 to 0.91887)
(0.01666 to 0.38226)

<0.001*
<0.001*
<0.001*
<0.001*
<0.001*
<0.001*
0.0389*

1.21186
-0.54217
0.26127

(0.96698 to 1.45397)
(-0.67857 to -0.3971)
(-0.04531 to 0.65901)

<0.001*
<0.001*
0.1306

Table 3.A2
Fixed effects predicting the total NFA count in each session. Only aggressive acts directed by
adults (≥24 months of age) down the hierarchy toward other adults are included. Significant
fixed effects are marked with asterisks.
Posterior mode 95% credible interval
pMCMC
Poisson
(Intercept)
Location (den)
Location (kill)
Location (mating)
Location (carcass)
Time of day (PM)
Number of hyenas present
Session length
Prey availability (low)
Zero-inflated
(Intercept)
Session length

-2.693988
0.503852
0.209515
1.338414
0.181835
-0.004432
0.056039
0.277183
0.142071

(-2.911986 to -2.48687)
(0.385957 to 0.621126)
(0.054158 to 0.361203)
(1.100465 to 1.568188)
(-0.132920 to 0.491380)
(-0.095446 to 0.090116)
(0.048147 to 0.064742)
(0.190224 to 0.360318)
(0.049520 to 0.240325)

<0.001*
<0.001*
0.008*
<0.001*
0.261
0.903
<0.001*
<0.001*
0.002*

3.478454
-8.504293

(2.769206 to 3.978756)
(-9.676932 to -5.820411)

<0.001*
<0.001*

89

Table 3.A3
Fixed effects predicting the average FA intensity in each session. Only aggressions directed by
adults (≥24 months of age) down the hierarchy to other adults are included. Significant fixed
effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
(Intercept)
Location (carcass)
Location (kill)
Number of hyenas present
Time of day (PM)
Prey availability (low)

1.816063
-0.2376
0.059606
0.004783
0.113481
0.031494

(1.692739 to 1.999028)
(-0.476201 to -0.059231)
(-0.162098 to 0.038545)
(-0.001109 to 0.008981)
(0.031890 to 0.212463)
(-0.050913 to 0.102473)

<0.01*
<0.01 *
0.38
0.08
0.04*
0.38

Table 3.A4
Fixed effects predicting the average NFA intensity in each session. Only aggressions directed by
adults (≥24 months of age) down the hierarchy to other adults are included. Significant fixed
effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
(Intercept)
Location (den)
Location (kill)
Location (mating)
Location(carcass)
Number of hyenas present
Time of day (PM)
Prey availability (low)

1.4610220
0.126464
0.081427
0.155056
0.065156
0.005805
0.061963
-0.034516

90

(1.343075 to 1.545208)
(0.069829 to 0.177375)
(0.022554 to 0.158401)
(0.076016 to 0.238748)
(-0.035453 to 0.165568)
(0.002295 to 0.009367)
(0.019212 to 0.101909)
(-0.073343 to 0.002779)

<0.01*
<0.01*
0.06
<0.01*
0.16
<0.01*
<0.01*
0.16

Table 3.A5
Fixed effects predicting the FA count by each hyena in each session. Only aggressions directed
by adults (≥24 months of age) down the hierarchy to other adults are included, and all sessions
that were shorter than 30 minutes or had no aggressions were omitted. Significant fixed effects
are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
Poisson
(Intercept)
Sex (male)
Age
Standardized rank
Proportion of targets
Zero-inflated
(Intercept)
Age

-8.3263814
-0.1566183
0.0007395
0.9155038
5.2782115

(-9.7106529 to -6.3727372)
(-0.4255145 to 0.1454277)
(-0.0017825 to 0.0032776)
(-0.3683651 to 2.1918252)
(3.5488414 to 6.9030409)

<0.001*
0.292
0.58
0.148
<0.001*

-6.1499061
0.0084016

(-7.6089937 to -4.0470538)
(0.0032528 to 0.0142680)

<0.001*
<0.001*

Table 3.A6
Fixed effects predicting the NFA count by each hyena in each session. Only aggressions
directed by adults (≥24 months of age) down the hierarchy to other adults are included, and all
sessions that were shorter than 30 minutes or had no aggressions were omitted. Significant
fixed effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
Poisson
(Intercept)
-3.136161
(-3.614926 to -2.712158)
<0.001*
Sex (male)
-0.904921
(-1.157348 to -0.672857)
<0.001*
Age
0.005109
(0.002837 to 0.007353)
<0.001*
Standardized rank
-0.936564
(-1.490512 to -0.344600)
<0.001*
Proportion of targets
1.615436
(1.117032 to 2.200340)
<0.001*
Zero-inflated
(Intercept)
-3.312539
(-3.921171 to -2.83293)
<0.001*
Age
0.014052
(0.010708 to 0.017718)
<0.001*

91

Table 3.A7
Fixed effects predicting the intensity of FAs emitted by adult (≥24 months of age) females.
Significant effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
(Intercept)
Age
Standardized rank
Proportion of targets

0.5053389
0.0014929
-0.5000345
0.7582205

(-0.0403970 to 1.0255745)
(-0.0037310 to 0.0005699)
(-1.1307486 to 0.0232113)
(0.1616710 to 1.3726560)

0.06*
0.19
0.08
0.005*

Table 3.A8
Fixed effects predicting the intensity of NFAs emitted by adult (≥24 months of age) females.
Significant effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
(Intercept)
Age
Standardized rank
Proportion of targets

0.5386560
-0.0035970
-0.1021480
-0.3603050

(0.129524 to 0.978908)
(-0.006204 to -0.001508)
(-0.585966 to 0.420753)
(-0.798148 to 0.125683)

0.005*
<0.003*
0.68
0.115

Table 3.A9
Fixed effects predicting the intensity of FAs emitted by adult (≥24 months of age) males.
Significant effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
(Intercept)
Age
Standardized rank
Proportion of targets

-0.12736
0.004635
-1.111081
1.140719

(-1.008385 to 0.960142)
(-0.002327 to 0.010014)
(-2.596731 to 0.143540)
(-0.095912 to 2.161934)

0.77
0.16
0.1
0.06

Table 3.A10
Fixed effects predicting the intensity of NFAs emitted by adult (≥24 months of age) males.
Significant effects are marked with asterisks.
Posterior mode
95% credible interval
pMCMC
(Intercept)
Age
Standardized rank
Proportion of targets

-0.9926710
0.0013020
0.8906690
1.2727010

(-1.991511 to -0.007707)
(-0.004247 to 0.006771)
(-0.427330 to 2.260357)
(-0.159216 to 2.508053)

92

0.05*
0.665
0.16
0.065

Table 3.A11
DIC values yielded by separate MCMCglmm models predicting adult (≥24 months of age)
lifetime aggression rate and aggression intensity for each aggression context. DIC values are
given for the full model with all variance components including VA (additive genetic), VM
(maternal effect), and VI (permanent environment effect), as well as for models omitting each
variance component. Relevant fixed effects were retained in each model, and intensity models
also include a random term representing the session identifier.
Models with omitted effects
Context
Full model VA omitted VM omitted
VI omitted
Lifetime aggression rate
Lifetime aggression rate
Aggression intensity
Aggression intensity

Food
Non-food
Food
Non-food

-403.452
-435.156
6101.237
13613.51

93

-382.919
-420.296
6104.711
13619.19

-400.942
-407.336
6110.397
13623.31

NA
NA
6103.556
13605.48

LITERATURE CITED

94

LITERATURE CITED
Altmann, S. E. 1974. Observational study of behavior: sampling methods. Behavior, 49, 227265.
Archer, J. 1988. The Behavioural Biology of Aggression. New York: Cambridge University Press.
Armitage, K. B. 1986. Individual differences in the behavior of juvenile yellow-bellied marmots.
Behavioral Ecology and Sociobiology, 18, 419-424.
Bakker, T. C. M. 1986. Aggressiveness in sticklebacks (Gasterosteus aculeatus): a behaviorgenetic study. Behaviour, 98, 1-144.
Bebié, N., & McElligott, A. G. 2006. Female aggression in red deer: Does it indicate competition
for mates? Mammalian Biology - Zeitschrift für Säugetierkunde, 71, 347-355.
Bell, A. M. 2005. Behavioural differences between individuals and two populations of
stickleback (Gasterosteus aculeatus). Journal of Evolutionary Biology, 18, 464-473.
Bell, A. M., Hankison, S. J., & Laskowski, K. L. 2009. The repeatability of behaviour: a metaanalysis. Animal Behaviour, 77, 771-783.
Bell, A. M., & Sih, A. 2007. Exposure to predation generates personality in threespined
sticklebacks (Gasterosteus aculeatus). Ecology Letters, 10, 828-834.
Benus, R. F., Bohus, B., Koolhaas, J. M., & van Oortmerssen, G. A. 1991. Heritable variation for
aggression as a reflection of individual coping strategies. Cellular and Molecular Life
Sciences, 47, 1008-1019.
van den Berghe, P. 1974. Bringing beasts back in: toward a biosocial theory of aggression.
American Sociological Review, 39, 777-788.
Berglund, A., & Rosenqvist, G. 2003. Sex role reversal in pipefish. In: Vol 32 pp. 131-167.
Academic Press.
Boon, A. K., Reale, D., & Boutin, S. 2008. Personality, habitat use, and their consequences for
survival in North American red squirrels Tamiasciurus hudsonicus. Oikos, 117, 1321-1328.
Boydston, E. E., Morelli, T. L., & Holekamp, K. E. 2001. Sex differences in territorial behavior
exhibited by the spotted hyena (Hyaenidae, Crocuta crocuta). Ethology, 107, 369-385.
Buss, D. M., & Shackelford, T. K. 1997. Human aggression in evolutionary psychological
perspective. Clinical Psychology Review, 17, 605-619.

95

Clutton-Brock, T. 2007. Sexual Selection in Males and Females . Science , 318 , 1882-1885.
Delehanty, D. J., Fleischer, R. C., Colwell, M. A., & Oring, L. W. 1998. Sex-role reversal and the
absence of extra-pair fertilization in Wilson’s phalaropes. Animal Behaviour, 55, 995-1002.
Digby, L. 1999. Targeting aggression in blue-eyed black lemurs (Eulemur macaco flavifrons).
Primates, 40, 613-617.
Dingemanse, N. J., & de Goede, P. 2004. The relation between dominance and exploratory
behavior is context-dependent in wild great tits. Behavioral Ecology, 15, 1023-1030.
Dingemanse, N. J., Kazem, A. J. N., Reale, D., & Wright, J. 2010. Behavioural reaction norms:
animal personality meets individual plasticity. Trends in Ecology &amp; Evolution, 25, 8189.
Dloniak, S. M., French, J. A., & Holekamp, K. E. 2006. Rank-related maternal effects of
androgens on behaviour in wild spotted hyaenas. Nature, 440, 1190-1193.
Duckworth, R. A. 2006. Behavioral correlations across breeding contexts provide a mechanism
for a cost of aggression. Behavioral Ecology, 17, 1011-1019.
Engh, A. L., Esch, K., Smale, L., & Holekamp, K. E. 2000. Mechanisms of maternal rank
“inheritance” in the spotted hyaena, Crocuta crocuta. Animal Behaviour, 60, 323-332.
Engh, A. L., Funk, S. M., Van Horn, R. C., Scribner, K. T., Bruford, M. W., Libants, S., Szykman,
M., Smale, L., & Holekamp, K. E. 2002. Reproductive skew among males in a femaledominated mammalian society. Behav. Ecol., 13, 193-200.
Fairbanks, L. A., Newman, T. K., Bailey, J. N., Jorgensen, M. J., Breidenthal, S. E., Ophoff, R. A.,
Comuzzie, A. G., Martin, L. J., & Rogers, J. 2004. Genetic contributions to social impulsivity
and aggressiveness in vervet monkeys. Biological Psychiatry, 55, 642-647.
Francis, R. C. 1988. On the relationship between aggression and social dominance. Ethology, 78,
223-237.
Frank, L. G. 1986a. Social organization of the spotted hyaena (Crocuta crocuta). II. Dominance
and reproduction. Animal Behaviour, 35, 1510-1527.
Frank, L. G. 1986b. Social organization of the spotted hyaena (Crocuta crocuta). I. Demography.
Animal Behaviour, 34, 1500-1509.
Frank, L. G., Glickman, S. E., & Powch, I. 1990. Sexual dimorphism in the spotted hyaena
(Crocuta crocuta). Journal of Zoology, 221, 308-313.

96

Frank, L. G., Holekamp, K. E., & Smale, L. 1995. Dominance, demography, and reproductive
success of female spotted hyenas. In: Serengeti II: Conservation, Research, and
Management, (Ed. by A. R. E. Sinclair), pp. 364-384. Chicago: University of Chicago Press.
Geist, V. 1974. On the relationship of social evolution and ecology in ungulates. American
Zoologist, 14, 205-220.
Glickman, S. E., Frank, L. G., Pavgi, S., & Licht, P. 1992. Hormonal correlates of
“masculinization” in female spotted hyaenas (Crocuta crocuta). 1. Infancy to sexual
maturity . Journal of Reproduction and Fertility , 95 , 451-462.
Grant, W. C., & Ulmer, K. M. 1974. Shell selection and aggressive behavior in two sympatric
species of hermit crabs. The Biological Bulletin, 146, 32-43.
Gwynne, D. T. 1991 April 1. Sexual competition among females: What causes courtship-role
reversal? Trends in Ecology & Evolution, 6, 118-121.
Hadfield, J. D. 2010. MCMC methods for multi-response generalized linear mixed models: the
MCMCglmm R package. Journal of Statistical Software, 33, 1-22.
Hemelrijk, C. K. 2000. Towards the integration of social dominance and spatial structure.
Animal Behaviour, 59, 1035-1048.
Hodge, S. J., Thornton, A., Flower, T. P., & Clutton-Brock, T. H. 2009. Food limitation increases
aggression in juvenile meerkats. Behavioral Ecology, 20, 930-935.
Holekamp, K. E., Cooper, S. M., Katona, C. I., Berry, N. A., Frank, L. G., & Smale, L. 1997.
Patterns of association among female spotted hyenas (Crocuta crocuta). Journal of
Mammalogy, 78, 55-64.
Holekamp, K., & Smale, L. 1991. Dominance acquisition during mammalian social development:
the “inheritance” of maternal rank. American Zoologist, 31, 306-317.
Holekamp, K. E., & Smale, L. 1993. Ontogeny of dominance in free-living spotted hyenas:
juvenile rank relations with other immature individuals. Animal Behaviour, 46, 451-466.
Holekamp, K., & Smale, L. 1996. Rank and reproduction in the female spotted hyaena. Journal
of Reproduction, 108, 229-237.
Van Horn, R. C., Engh, A. L., Scribner, K. T., Funk, S. M., & Holekamp, K. E. 2004. Behavioural
structuring of relatedness in the spotted hyena (Crocuta crocuta) suggests direct fitness
benefits of clan-level cooperation. Molecular Ecology, 13, 449-458.

97

Van Horn, R. C., McElhinny, T. L., & Holekamp, K. E. 2003. Age estimation and dispersal in the
spotted hyena (Crocuta crocuta). Journal of Mammalogy, 84, 1019-1030.
Johnson, J. C. 2001. Sexual cannibalism in fishing spiders (Dolomedes triton): an evaluation of
two explanations for female aggression towards potential mates. Animal Behaviour, 61,
905-914.
Johnson, J. C., & Sih, A. 2005. Precopulatory sexual cannibalism in fishing spiders (Dolomedes
triton): a role for behavioral syndromes. Behavioral Ecology and Sociobiology, 58, 390-396.
Kagan, J. 2003. Biology, Context, and Developmental Inquiry. Annual Review of Psychology, 54,
1-23.
Kappeler, P. M. 1990. Female dominance in Lemur catta: more than just female feeding
priority? Folia Primatologica, 55, 92-95.
Kruuk, H. 1972. The spotted hyaena: a study of predation and social behavior. Chicago: Chicago
University Press.
Kruuk, L. E. B. 2004. Estimating genetic parameters in natural populations using the “animal
model”. Philosophical Transations of the Royal Society (B), 359, 873-890.
Kunc, H. P., Amrhein, V., & Naguib, M. 2006. Vocal interactions in nightingales, Luscinia
megarhynchos: more aggressive males have higher pairing success. Animal Behaviour, 72,
25-30.
Leshner, A. I., & Moyer, J. A. 1975. Androgens and agonistic behavior in mice: relevance to
aggression and irrelevance to avoidance-of-attack. Physiology & Behavior, 15, 659-695.
Lessells, C. M., & Boag, P. T. 1987. Unrepeatable repeatabilities—a common mistake. The Auk,
2, 116-121.
Martin, J. G. a., & Réale, D. 2008. Temperament, risk assessment and habituation to novelty in
eastern chipmunks, Tamias striatus. Animal Behaviour, 75, 309-318.
Matsumasa, M., & Murai, M. 2005. Changes in blood glucose and lactate levels of male fiddler
crabs: effects of aggression and claw waving. Animal Behaviour, 69, 569-577.
Miczek, K. A., Maxson, S. C., Fish, E. W., & Faccidomo, S. 2001. Aggressive behavioral
phenotypes in mice. Behavioural Brain Research, 125, 167-181.
Mills, M. G. L. 1990. Kalahari hyaenas: Comparative behavioral ecology of two species. London:
Unwin Hyman.

98

Mousseau, T. A., & Roff, D. A. 1987. Natural selection and the heritability of fitness
components. Heredity, 59, 181-197.
Packer, C., Collins, D. A., Sindimwo, A., & Goodall, J. 1995. Reproductive constraints on
aggressive competition in female baboons. Nature, 373, 60-63.
Panksepp, J., & Zellner, M. 2004. Towards a neurobiologically based unified theory of
aggression. Revue internationale de psychologie, 17, 37-61.
Pfennig, D. W., & Reeve, H. K. 1989. Neighbor recognition and context-dependent aggression
in a solitary wasp, Sphecius speciosus (Hymenoptera: Sphecidae). Ethology, 80, 1-18.
Pinheiro, J. C., & Bates, D. M. 2000. Mixed-effects Models in S and S-Plus. New York: SpringerVerlag.
Sandell, M. I. 1998. Female aggression and the maintenance of monogamy: female behaviour
predicts male mating status in European starlings. Proceedings of the Royal Society of
London. Series B: Biological Sciences, 265, 1307-1311.
Sikes, R. S., & Gannon, W. L. 2011. Guidelines of the American Society of Mammalogists for the
use of wild mammals in research. Journal of Mammalogy, 92, 235-253.
Sinn, D. L., While, G. M., & Wapstra, E. 2008. Maternal care in a social lizard: links between
female aggression and offspring fitness. Animal Behaviour, 76, 1249-1257.
Smale, L., Frank, L. G., & Holekamp, K. E. 1993. Ontogeny of dominance in free-living spotted
hyaenas: juvenile rank relations with adult females and immigrant males. Animal
Behaviour, 46, 467-477.
Smith, J. E., Van Horn, R. C., Powning, K. S., Cole, A. R., Graham, K. E., Memenis, S. K., &
Holekamp, K. E. 2010. Evolutionary forces favoring intragroup coalitions among spotted
hyenas and other animals . Behavioral Ecology , 21 , 284-303.
Smith, J. M., & Price, G. R. 1973. The Logic of Animal Conflict. Nature, 246, 15-18.
Stocker, A. M., & Huber, R. 2001. Fighting strategies in crayfish Orconectes rusticus (Decapoda,
Cambaridae) differ with hunger state and the presence of food cues. Ethology, 107, 727736.
Swanson, E. M., Dworkin, I., & Holekamp, K. E. 2011. Lifetime selection on a hypoallometric
size trait in the spotted hyena. Proceedings of the Royal Society B: Biological Sciences,

99

Szykman, M., Engh, A. L., Horn, R. C. V., Boydston, E. E., Scribner, K. T., & Holekamp, K. E.
2003. Rare male aggression directed toward females in a female-dominated society:
Baiting behavior in the spotted hyena. Aggressive Behavior, 29, 457-474.
Tanner, C. J., & Adler, F. R. 2009. To fight or not to fight : context-dependent interspecific
aggression in competing ants. Animal Behaviour, 77, 297-305.
Tanner, J. B., Zelditch, M. L., Lundrigan, B. L., & Holekamp, K. E. 2010. Ontogenetic change in
skull morphology and mechanical advantage in the spotted hyena (Crocuta crocuta).
Journal of Morphology, 271, 353-365.
Taylor, R. W., Boon, A. K., Dantzer, B., Reale, D., Humphries, M. M., Boutin, S., Gorrell, J. C.,
Coltman, D. W., & McAdam, A. G. 2012. Low heritabilities, but genetic and maternal
correlations between red squirrel behaviours. Journal of Evolutionary Biology, 25, 614-624.
R Development Core Team. 2011. R: A Language and Environment for Statistical Computing.
Thierry, B. 1985. Patterns of agonistic interactions in three species of macaque (Macaca
mulatta, M fascicularis, M tonkeana). Aggressive Behavior, 11, 223-233.
Van Meter, P. 2009. Hormones, stress and aggression in the spotted hyena (Crocuta crocuta).
Michigan State University.
Verbeek, M. E., Boon, A., & Drent, P. J. 1996. Exploration, aggressive behaviour and dominance
in pair-wise confrontations of juvenile male great tits. Behaviour, 133, 945-963.
Walters, J. R., & Seyfarth, R. M. 1987. Conflict and cooperation. In: Primate Societies, (Ed. by B.
Smuts, D. Cheney, R. Seyfarth, R. Wrangham, & T. Struhsaker), pp. 306-317. Chicago:
Univerisity of Chicago Press.
Watts, H. E., & Holekamp, K. E. 2009. Ecological determinants of survival and reproduction in
the spotted hyena. Journal of Mammalogy, 90, 461-471.
Watts, H. E., Scribner, K. T., Garcia, H. A., & Holekamp, K. E. 2011. Genetic diversity and
structure in two spotted hyena populations reflects social organization and male dispersal.
Journal of Zoology, 285, 281-291.
Watts, H. E., Tanner, J. B., Lundrigan, B. L., & Holekamp, K. E. 2009. Post-weaning maternal
effects and the evolution of female dominance in the spotted hyena. Proceedings of the
Royal Society B: Biological Sciences, 276, 2291-2298.
While, G. M., Isaksson, C., McEvoy, J., Sinn, D. L., Komdeur, J., Wapstra, E., & Groothuis, T. G.
G. 2010. Repeatable intra-individual variation in plasma testosterone concentration and its
sex-specific link to aggression in a social lizard. Hormones and behavior, 58, 208-13.
100

Wilson, A. P., & Boelkins, R. C. 1970. Evidence for seasonal variation in aggressive behaviour by
Macaca mulatta. Animal Behaviour, 18, Part 4, 719-724.
Wilson, A. J., Réale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A., Kruuk, L.
E. B., & Nussey, D. H. 2009. An ecologist’s guide to the animal model. Journal of Animal
Ecology, 79, 13-26.
Yasukawa, K., & Searcy, W. A. 1982. Aggression in female red-winged blackbirds: A strategy to
ensure male parental investment. Behavioral Ecology and Sociobiology, 11, 13-17.

101

CHAPTER 4
INDIVIDUAL DIFFERENCES IN SOCIABILITY AMONG SPOTTED HYENAS

INTRODUCTION
Sociality has evolved multiple times in species representing very diverse taxa, and
gregariousness can be advantageous in many ways, including cooperative defense of resources,
group acquisition of food, and enhanced vigilance (O’Brien 1991; Baker et al. 1998; Hass &
Valenzuela 2002; Baglione et al. 2002; Majolo et al. 2008). On an individual level, animals living
in groups may also benefit from associations with conspecifics through mechanisms like
coalitionary support, cooperative breeding, and social learning (Coussi-Korbel & Fragaszy 1995;
Clutton-Brock 2002; Silk et al. 2004). In gregarious species, group members may differ from
one another with respect to how often they engage in social behavior, or in the strength of
their associations with conspecifics. Sociability is the extent to which an individual engages in
social behavior; Reale et al. (2007) described sociable individuals as those that “seek the
presence of conspecifics, while unsociable individuals avoid conspecifics.” As studies of animal
personality become more common, researchers are finding that this inter-individual variation in
sociability is often consistent over time, and that it can have significant effects on fitness (e.g.
Wilson et al. 1994; Reale & Festa-Bianchet 2003; Carere et al. 2005; Boon et al. 2008; Smith &
Blumstein 2008). Whereas traits such as aggression, exploratory behavior, and activity have
received a substantial amount of attention, sociability is one of the least-studied personality
traits in non-human animals (however, see Armitage 1986; Capitanio 2002; Silk et al. 2006).

102

Consistent inter-individual variation in sociability has been reported in several species,
including sticklebacks (Dingemanse et al. 2009), yellow-bellied marmots (Barash 1976; Armitage
1986), zebra finches (Figueredo et al. 1995), domestic dogs (Svartberg & Forkman 2002), rhesus
macaques (Capitanio 2002; Maninger et al. 2003), and humans (Goldsmith & Gottesman 1981;
Loehlin 1992). Research on common lizards suggests that sociability is stable across ontogeny;
in these studies, the social behavior of juvenile lizards did not change significantly as they
reached reproductive maturity (Cote & Clobert 2007; Cote et al. 2008). However, in a study of
wolf cubs, Macdonald (1983) found that the ontogenetic development of sociability and other
personality traits varied greatly among individuals. Some cubs changed their behavior in
drastically different ways in early development, and the behavior of others remained relatively
stable. In humans and other primates, early experience, such as interactions with kin or
members of a peer cohort, is hypothesized to shape sociability later in life (Capitanio 1999;
Fries et al. 2005).
Some studies point to genetics as a major source of variability in sociability (e.g. Haas et
al. 2009). Researchers have successfully selected for strains of mice that show particularly high
or low rates of social interaction (Sankoorikal et al. 2006; Brodkin 2007). A heritability estimate
is the proportion of the phenotypic variance that can be accounted for by genetics; for humans,
the heritability of sociability has been estimated at between 0.25 and 0.4 (Plomin 1994).
Similarly, Wilsson and Sundgren (1997) estimated that just over one third of the variation in
social behavior among German Shepherds could be explained by genetic effects. As sociability
is one of the less well-studied personality traits in non-human animals, heritability estimates for
this trait in species other than humans and domestic dogs are rare.
103

The effects of sociability on fitness have received more attention than other aspects of
sociability. Cote et al. (2008) found that, among common lizards, higher levels of sociability
increased the reproductive success of females, but not males. In these lizards, sociability also
influenced dispersal decisions; individuals with greater social tolerance tended to disperse to
higher-density populations, whereas those with less social tolerance dispersed toward lowerdensity populations (Cote & Clobert 2007). Silk et al. (2003) demonstrated that the offspring of
female baboons that rated highly on a composite sociality index were more likely to survive
than the offspring of females that were less social, even when controlling for the effects of
dominance rank. Female baboons that form stronger and more stable bonds also live longer
than do less social females (Silk et al. 2010). It is suggested that sociability enhances fitness via
increased protection from harassment, better cooperative relationships with partners, and
more frequent interactions with potential mates (Armitage 1986; Silk et al. 2003; Cote et al.
2008; Schulke et al. 2010)
Here, we study sociability in a social carnivore, the spotted hyena (Crocuta crocuta).
Hyenas are among the most gregarious large carnivores, living in groups containing up to 90
individuals (Holekamp et al. 2006). These groups are fission-fusion societies, where group
members are seldom, if ever, all together. Instead, smaller subgroups form (“fission”) and
break apart (“fusion”), and these subgroups vary in size and composition over time (Kummer
1971; Smith et al. 2008). Various social and environmental factors contribute to fission-fusion
dynamics; shared resource defense against lions or alien hyenas promotes subgroup fusion, and
intense resource competition generally limits subgroup size and promotes subgroup fission
(Smith et al. 2008). In contrast to many gregarious species in which subgroups are more stable
104

and decisions regarding movement are made communally (Conradt & Roper 2005), spotted
hyenas generally make individual decisions to join and leave subgroups (Gittleman 1989). In
this system, individuals can vary in their propensity to socialize and act in their best interest in
terms of social behavior (Smith et al. 2008). Studies have examined partner choice in hyenas to
determine whether there are advantages to socializing with other group members. Smith et al.
(2007) found that adult females most often associate with higher-ranking females and with
those females occupying adjacent rank positions in the clan’s dominance hierarchy. Both
partners in these associations benefit; dominant hyenas benefit from priority of access to
resources defended by a group, whereas subordinate hyenas receive reduced aggression from
dominants at kills and better access to resources (Holekamp et al. 1997b; Boydston et al. 2001;
Smith et al. 2007).
Hyenas frequently engage in several types of affiliative behaviors, including greeting
behavior, in which two or more hyenas stand parallel to one another and reciprocally sniff the
other’s anogenital region (Kruuk 1972; East et al. 1993; Smith et al. 2011). These greeting
ceremonies function either in the context of aggression to reconcile opponents and prevent
escalation (Wahaj et al. 2001), or as nonconciliatory greetings unrelated to aggression (East et
al. 1993; Colmenares et al. 2000). Smith et al. (2011) tested three hypotheses regarding
nonconciliatory greeting behavior in hyenas and found that these greetings function to
reinforce affiliative relationships, rather than to reduce social tension or to signal submission.
Hyenas tended to solicit more greetings from preferred companions and greet more often with
high-than low-ranking hyenas; greetings also offer opportunities for hyenas to assess the
cooperative potential of a partner (Smith et al. 2011). There is limited evidence for inter105

individual variation in greeting behavior in hyenas; Smith et al. (2011) found that individuals are
consistent across observations in the duration of their greetings.
In light of the advantages of associating with conspecifics, sociability may provide
considerable benefits to hyenas in terms of fitness. Hyenas face intense feeding competition,
because they feed in large groups on ephemeral kills. An individual can eat as much as 15kg of
food in 15 minutes, and a group of hyenas can consume an entire carcass in that amount of
time (Kruuk 1972; Holekamp & Smale 1996). Because food is a limited resource and the
reproductive success of female hyenas is strongly driven by access to resources (Holekamp &
Smale 1996), gaining feeding tolerance via associations with conspecifics might represent a
significant fitness advantage for females.
In this chapter, we first assess the situational factors promoting sociability among
hyenas, then determine whether there is evidence of consistent differences in sociability
among individuals. We assess the stability of sociability across ontogeny, and determine
whether genetics and maternal effects are responsible for inter-individual variation in
sociability. Finally, we investigate the relationships between sociability and reproductive
success and longevity.

METHODS
Study Subjects
Between 1988 and 2009, personnel from the Mara Hyena Project collected behavioral
data on a group of free-living spotted hyenas, called the Talek Clan, in the Masai Mara National
Reserve, Kenya (Boydston et al. 2001). All hyenas in this clan were recognized individually by
106

observers based on their unique spot patterns, and were of known sex, age, and social rank.
Hyenas were sexed based on the morphology of the glans of the erect phallus (Frank et al.
1990). We assigned ages to all natal animals (to ±7 days) based on their appearance when they
were first seen above ground at dens, and we were able to assign immigrant males’ ages
reliably (to ± 6 months) with an age-estimation model (Van Horn et al. 2003).
The social rank of each hyena in the clan was assigned using a dominance matrix based
on observations of dyadic agonistic interactions (Holekamp & Smale 1993; Smale et al. 1993).
Rank was reassessed each time a new animal reached adulthood, entered or left the clan, or
died. Cubs were assigned the rank of their mother until they reached reproductive maturity at
24 months of age (Frank 1986; Glickman et al. 1992; Holekamp et al. 1997a). We calculated
standardized rank for each hyena by dividing their numerical rank by the total number of
ranked animals in the clan at the time.
We established maternity based on genotyping and observations of cubs nursing
(Holekamp & Smale 1993). Paternity was assigned based exclusively on genotyping, as
previously described (Van Horn et al. 2004; Watts et al. 2011). Methods used to immobilize
animals and collect blood samples for DNA extraction are described in detail in Engh et al.
(2002). All sampling procedures were approved by the Institutional Animal Care and Use
Committee at Michigan State University (AUF 07/08-099-00), complied with Kenyan law, and
conformed to guidelines approved by the American Society of Mammalogists (Sikes & Gannon
2011).

107

Data collection
We generally observed hyenas from vehicles daily between 0530 and 0930 hours and
between 1700 and 2000 hours. We initiated observation sessions whenever we saw a hyena or
group of hyenas separated from conspecifics by at least 200 meters. Upon initiating an
observation session, we recorded the GPS location of the session and conducted a behavioral
scan that identified each hyena, noted its current behavior, and described its position in space
relative to landmarks and conspecifics; these scans were repeated every 20 minutes. During
each session, we used critical incident sampling (Altmann 1974) to record all occurrences of
specific behaviors including aggressive acts, appeasements, vocalizations, and greetings. We
also recorded whenever a hyena entered or left a session in progress.
In this study, we used four measures to describe the sociability of each individual: (1) its
tendency to solicit greetings, (2) its greeting count, (3) its lifetime greeting rate, and (4) its
mean group joining rate. First, during each observation session where an individual was
present with at least one other hyena, we determined whether or not the hyena solicited at
least one greeting. “Soliciting” a greeting involved a hyena lifting its leg to initiate a greeting
with another hyena. We also calculated the number of greetings a hyena solicited during each
observation session where at least one other hyena was present; this was called the greeting
count for the hyena in that session. Following Van Meter (2009) and Smith et al. (2011), we
calculated hourly adjusted greeting rates for each individual present in each session as:
number of greetings solicited ÷ number of potential greeting partners
number of hours the session lasted

108

where the number of potential greeting partners was a count of all other hyenas present in the
observation session. This measure controlled for variation in the number of opportunities each
hyena had to solicit greetings during each session. To calculate a lifetime greeting rate for each
hyena, we averaged its greeting rates in all its sessions with conspecifics after it reached 24
months of age. For our final measure of sociability, we calculated the hourly rate at which each
hyena joined other groups of hyenas during the period between 1996 and 2002. A hyena was
determined to have joined another group if it approached and remained within 200 meters of
at least one other hyena. To calculate the hourly group joining rate, we divided the number of
joining events for each individual during the seven-year period by the number of hours that the
individual was observed during this period.
Other situational data were also collected from all observation sessions, including the
maximum number of hyenas present in the session, the time of day (either from midnight to
noon, or from noon until midnight), the location of the session (at a communal den, a kill, a
carcass, a mating event, a natal den, or a session away from these contexts), and the current
seasonal prey availability (high for June to October, and low from November to May).

Statistical Analysis
Bayesian analyses
For many of our analyses, we adopted a Bayesian approach and created models with the
Markov Chain Monte Carlo for Generalized Linear Mixed Models (MCMCglmm) analysis tool in
the R statistical package (Hadfield 2010; R Core Development Team 2011). For each of these
analyses, we tested fixed effects for significance by determining whether or not the 95%
109

credible interval for the fixed effect included zero; variables for which the credible interval
included zero were not significant predictors of the response variable, whereas those that did
not include zero were considered significant. We tested random effects for significance by
comparing the deviance information criterion (DIC) values of models fitted with and without
the random effect of interest; if the inclusion of a random effect lowered the DIC value of the
model, the random effect was considered significant in predicting the response variable.
For zero-inflated MCMCglmm models, we began by using a saturated model with fixed
effects contributing to both the poisson and zero-inflated portions of the model, but we only
retained zero-inflation terms that lowered the DIC value of the model. In the zero-inflated
portion of the model, the estimates for fixed effects were interpreted as the log(e) of the ratio
of the odds of the response variable not occurring, compared to the odds of the response
variable occurring.
For all MCMCglmm models, we used a relatively uninformative prior that partitioned
variance equally among all random effects (Hadfield 2010); using a more informative prior did
not affect reported results. Estimates are reported as posterior means with 95% credible
intervals. We assessed convergence of these models by examining time series of the model
parameters. We also inspected autocorrelation values for these models, all of which were less
than 0.1 for reported values.

Situational factors influencing sociability
We assessed sociability based on two different levels of analysis; one at the level of the
observation session, and another at the level of individual hyenas. First, we performed an
110

analysis using the observation session as the unit of analysis to determine which social and
environmental factors influenced how often hyenas greeted.
As the response variable for this analysis, we summed the greeting count in each
observation session for all adult hyenas present; this measure represented a general measure
of sociability for each observation session. Using this composite measure of sociability allowed
us to assess the importance of situational variables that were common to all hyenas present in
the session. We fitted a MCMCglmm model with the total greeting count in each observation
session as the response variable, and various social and environmental aspects of the session as
fixed effects. These fixed effects included the location of the session, the duration of the
session, the time of day, the maximum number of hyenas present in the session, and whether
seasonal prey availability was high or low.

Consistency in sociability within individuals
We then performed a similar Bayesian analysis using individual hyenas as the units of
analysis to (1) determine which characteristics of individuals affected sociability, and (2)
determine whether there was within-individual consistency over time in this measure of
sociability.
In the individual-level analysis, we used the greeting count of each hyena in each session
as the measure of sociability. For each sex, we created a separate repeated measures
MCMCglmm model with the greeting count of each adult hyena in each session as the response
variable. The standardized social rank and age of the hyena on the date of the session were
included as fixed effects. A unique session identifier was included as a random effect in each
111

analysis to control for pseudoreplication. To determine whether individual hyenas were
consistent in the frequency of greets they solicited across observation sessions, we tested
whether the inclusion of a random effect representing the identity of the hyena improved each
model. If this term significantly improved the fit of these models, individuals’ behavior could be
considered consistent across repeated observations. Only hyenas present during adulthood in
at least 20 observation sessions with potential greeting partners in were included in this
analysis.

Heritability of sociability
To determine whether or not sociability is heritable in spotted hyenas, we created
restricted-likelihood Bayesian animal models (Kruuk 2004). Animal models use pedigree
information to partition the heritable and non-heritable components of phenotypic variance
(Kruuk 2004; Wilson et al. 2009). Phenotypic, additive genetic, and maternal effects were
estimated with univariate-trait animal models predicting the sociability of each hyena. Because
female hyenas mate with several males across their reproductive history, maternal effects can
be distinguished from heritability via the effect of these different sires.
As the response variable of our first animal model, we used the lifetime greeting rate of
each hyena. Only hyenas that were present with potential greeting partners in at least 50
sessions after 24 months of age were included in this analysis.
We included sex and average lifetime standardized rank as fixed effects, but only
retained fixed effects that contributed significantly to the model. We included random effects
representing the additive genetic effect and the identity of the hyena’s mother. We calculated
112

heritability and maternal effects by dividing the relevant variance component by the total
phenotypic variance.
Because MCMCglmm requires values for all random effects, we generated unique
maternal identities for immigrant males and natal animals born before our study began,
following Taylor et al. (2012). This technique allows for the estimation of maternal effects,
even when pedigree information is missing; however, it assumes that all hyenas with unknown
maternity are unrelated.
We also created and ran an animal model that was identical to the one described above,
except that here we used the group joining rate of each hyena as the response variable.
Heritability and maternal effects were calculated as described above. Only hyenas that were
observed for at least 15 hours between 1996 and 2002 were included in this analysis.

Stability over time and across situations
In order to determine whether the sociability of individual hyenas remained stable
across ontogeny, we created generalized linear mixed models using the lme4 function in R. We
determined the tendency to greet in each session for all cubs less than 8 months of age, as well
as the tendency to greet in each session for hyenas that were at least 24 months of age. These
data were used as the response variables in linear models involving repeated measures. Only
female hyenas present with potential greeting partners in at least 50 sessions during each age
class were included in this analysis.

113

We included age class (either cub or adult) as the main fixed effect of interest and
included the hyena’s standardized rank as a fixed effect. A unique session identifier was
included as a random effect to control for pseudoreplication.
We included individual identity as a random intercept to allow for the variation among
hyenas in greeting behavior that had been suggested by earlier analyses. We also added a
random slope to the model, which described the extent to which the sociability of individual
hyenas changed across ontogeny. If this term significantly improved the fit of the model, we
could conclude that the sociability of individual hyenas did not change consistently across age
classes. To test whether the random slope significantly improved the fit of the model, we
created two models that were identical except for the inclusion of the random slope. We then
compared these two models with a log-likelihood ratio test, in which the log-likelihood ratio is
calculated as 2[log-likelihood of model B - log-likelihood of model A], and tested as a chisquared distribution (Pinheiro & Bates 2000; Martin & Réale 2008). Finally, we included an
interaction term between standardized rank and the random slope, to determine whether
social rank determined how the sociability of individuals changed across ontogeny. Using a loglikelihood ratio test, we determined whether the inclusion of this interaction term significantly
improved the fit of the model. The approach of using random regression to investigate
behavioral reaction norms is described in more detail in Dingemanse et al. (2010).

Relationship between sociability and fitness
Finally we tested whether a hyena’s sociability affected its fitness, as indicated by
various measures of reproductive success and longevity. We used the lifetime greeting rate
114

and the overall group joining rate of each female as the measures of sociability in this analysis.
For the greeting rate analysis, we only included females present during adulthood with
potential greeting partners in at least 50 sessions. For the joining rate analysis, we only
included adult females that were observed for at least 15 hours between 1996 and 2002.
As indicators of reproductive success, we used the total number of cubs borne by each
female over her lifetime, as well as the proportion of her cubs that survived to three life-history
milestones: den independence, weaning, and 24 months of age. We created a set of
generalized linear and second-order polynomial models with these fitness components as the
response variables, and the sociability measure of interest, the maternal rank of the female,
and an interaction term as predictors. Separate analyses were carried out for greeting rate and
group joining rate. Only female hyenas that were born after our study began and died before
our study ended were included in these reproductive success analyses.
We also used hyenas’ longevity as a response variable to determine whether sociability
is related to survival. For this dataset, we used the same two measures of sociability and
criteria for inclusion, except we included only females for which we had known death dates.
We created both generalized linear and second-order polynomials with longevity (in months) as
the response variable. As fixed effects, we included the sociability measure of interest and the
average lifetime standardized rank of the female, as well as an interaction term between these
variables. Separate analyses were run for greeting rate and group joining rate.
For longevity and all measures of reproductive success, we used likelihood ratio tests to
determine whether second-order polynomial models were able to predict these fitness
measures better than linear models. All variables in these analyses were Z-transformed.
115

RESULTS
Social and environmental factors influencing sociability
Between 1988 and 2009, there were 10526 observation sessions that met our criteria
for inclusion in the analysis of greeting count. In 8775 (83.4%) of these sessions, hyenas did not
greet at all. Greetings were less common at kills than at sessions away from dens, mating, and
food, and hyenas solicited a greater number of greetings between noon and midnight than
between midnight and noon (Appendix, Table 4.A1). Hyenas solicited more greetings in
sessions with longer durations and in sessions with more hyenas present than in shorter
sessions or those with fewer hyenas present (Appendix, Table 4.A1).

Consistency in sociability within individuals
Eighty-five females met our criteria for this analysis. Older females were less likely than
younger females to solicit greetings in a session, and there was a non-significant trend for
lower-ranking hyenas to solicit more greetings than did higher-ranking hyenas (Appendix, Table
4.A2). Ninety-six male hyenas met our criteria for this analysis. Both standardized rank and
age affected the number of greetings solicited by male hyenas; higher-ranking males solicited
more greetings than did lower-ranking males, and older males solicited more greetings than did
younger males (Appendix, Table 4.A3).
For members of both sexes, the inclusion of a random effect of individual improved the
fit of the model (Table 4.1), suggesting that both males and females were consistent over time
with respect to the frequency at which they solicited greetings.

116

Table 4.1
DIC values yielded by MCMCglmm models predicting the number of greetings solicited in each
observation session by adult (≥24 months of age) female and male hyenas. DIC values are given
for models with only a random effect of the session identifier, and for models run with both a
session identifier and an individual identifier. For each sex, the better model is marked with an
asterisk.
Random effects included
Sex
Session
Individual and Session
Females
Males

17789.79
5456.81

15606.06*
4615.105*

Heritability of sociability
For lifetime greeting rate, genetic and maternal effects were both significant, as
indicated by a lower DIC value when these effects were included (-863.521), compared to when
they were omitted (genetic effect omitted, DIC=-834.117; maternal effect omitted, DIC=848.817). The heritability estimate from the MCMCglmm animal model for lifetime greeting
rate was 0.160 (95% CI=0.0305 to 0.605, Figure 4.1) and the maternal effect estimate was 0.068
(95%CI=0.026 to 0.312, Figure 4.1).
For group joining rate, genetic and maternal effects were both significant, as indicated
by a lower DIC value when these effects were included (131.586), compared to when they were
omitted (genetic effect omitted, DIC=139.349; maternal effect omitted, DIC= 136.554). The
heritability estimate from the MCMCglmm animal model for the group joining rate was 0.090
(95% CI=0.024 to 0.397, Figure 4.1) and the maternal effect estimate was 0.088 (95%CI=0.024
to 0.336, Figure 4.1).

117

0.9
0.82
0.8

0.77

Proportion of variance explained

0.7
0.6
0.5

Greeting rate
0.4

Group joining rate

0.3
0.2

0.1

0.16
0.09

0.07

0.09

0
Heritability

Maternal effect

Unexplained

Figure 4.1. For two measures of sociability, the percent of variation attributable to heritability,
maternal effects, and the remaining unexplained variation. Variances were estimated with
separate MCMCglmm animal models.

118

Stability of sociability
There were 42 female hyenas with potential greeting opportunities in at least 50
sessions during each of the two age classes. The model that included a random slope of age
class in addition to the random intercept was significantly better at predicting the likelihood of
greeting in a session than the model with only the random intercept (LRT=58.680, df=2,
p<0.001), indicating that within individuals, sociability does not change in consistent ways
across ontogeny (Figure 4.2).
The model that included an interaction between standardized rank and the random
slope was significantly better at predicting the data than the random slope model without this
interaction term (LRT=15.417, df=7, p=0.031), indicating that social rank affected how
sociability changed across ontogeny. The interaction was positive (29.625±5.443), suggesting
that as social rank decreased, females became more likely to greet as adults, compared to
when they were cubs.

Relationship between sociability and fitness
Thirty-eight females satisfied the criteria for inclusion in our analysis of the relationship
between greeting rate and reproductive success. Using generalized linear models, lifetime
greeting rate was not a significant predictor of any measure of reproductive success as a fixed
effect or in an interaction with maternal rank (p>0.1). Second-order polynomial models failed
to describe the data better than linear models (p>0.1)
There were 31 females that satisfied the criteria for inclusion in our analysis of the
relationship between group joining rate and reproductive success. In generalized mixed
119

Figure 4.2. Predicted behavioral reaction norms for the likelihood of greeting across two age
classes, as predicted by lmer models. Each line represents how the behavior of a female
changes acrss ontogeny. Smaller likelihoods of greeting represent less sociable individuals, as
these individuals are less likely to greet with other hyenas in a session.

120

models, group joining rate was not significant in predicting any measure of reproductive
success. However, we found several significant interactions between group joining rate and
maternal rank. Although high-ranking hyenas had consistently high reproductive success, lowranking females with high group joining rates enjoyed better reproductive success than did lowranking females with low group joining rates. The interaction between maternal rank and
group joining rate was significant for all measures of reproductive success associated with
offspring survivorship, including the percent of cubs surviving to weaning (0.022±0.007,
t=3.292, p=0.003), the percent of cubs surviving to den independence (0.024±0.007, t=3.447,
p=0.002), and the percent of cubs surviving to reproductive maturity (0.018±0.007, t=2.615,
p=0.014).
There were 64 female hyenas that satisfied the criteria for inclusion in our analysis of
the relationship between greeting rate and longevity. The age at death of these females
averaged 108.80±5.77 months. Greeting rate was not significantly related to longevity, using
either generalized linear or second-degree polynomial models (p>0.1). Furthermore, no
interaction terms containing greeting rate significantly predicted longevity (p>0.1).
There were 34 female hyenas that satisfied the criteria for inclusion in our analysis of
the relationship between group joining rate and longevity. The age at death of these females
averaged 112.38 ±1.30 months. Group joining rate significantly predicted age at death; hyenas
that joined groups more often lived longer than did those that joined groups less often
(19.820±7.354, t=2.695, p=0.011, Figure 4.3). No interaction terms containing group joining
rate significantly predicted longevity, and second-order polynomial terms were not significantly
better than linear models at predicting age at death (p>0.1).
121

Longevity (age in months at death)

250

200

150

100

50

0
0

0.5

1

1.5
2
2.5
Hourly group joining rate

3

3.5

Figure 4.3. The relationship between group joining rate and longevity. Group joining rate was
calculated as the hourly rate at which adult females joined groups between 1996 and 2002.
Actual values are displayed as dots, and the values predicted by generalized linear model are
shown as a line.

122

DISCUSSION
Both situational factors and individual attributes at the time of the session affected the
number of greetings solicited by hyenas. Hyenas solicited fewer nonconciliatory greetings at
kills than in other situations; in other words, these greetings occurred less frequently in
situations where tension was high due to resource competition than they did in situations with
less tension. This supports previous research showing that nonconciliatory greetings do not
function to reduce social tension (Smith et al. 2011). Interestingly, there was a non-significant
trend for hyenas to solicit a greater number of greetings when seasonal prey abundance was
low. Social relationships may be especially important to maintain when resource availability is
reduced, due to the increased importance of feeding tolerance during these periods.
High-ranking males solicited more greetings than low-ranking males. This effect may be
due to our inclusion of both immigrant and adult natal males in this analysis; adult natal males
outrank immigrant males, and may be more likely to greet with members of their natal clan
than immigrants are. The relationship between rank and greeting behavior was reversed in
females; we observed a non-significant trend for low-ranking females to solicit more greetings
than higher-ranking females. Adult females prefer to solicit greetings from higher-ranking
individuals (Smith et al. 2011), and low-ranking females have a larger number of dominant
animals with which to forge bonds than do higher-ranking females.
Our results show that adult hyenas were consistent across repeated observations in the
number of greetings they solicited; we found significant inter-individual differences in this
measure of sociability for both males and females. From our heritability analyses, we
estimated that about 16 percent of the observed phenotypic variation in the frequency with
123

which hyenas solicit greetings was due to genetic effects, and for the rate at which hyenas join
groups, about 9 percent of the observed phenotypic variation was due to genetic effects. These
heritability estimates are significantly smaller than those reported for sociability in humans (e.g.
Plomin 1994), and are also lower than the average heritability estimate for behavioral traits in
non-human animals, which is 0.30±0.02 (Mousseau & Roff 1987). However, they show that
genetics play a significant role in shaping sociability, and that sociability is more strongly
heritable than some other traits in hyenas, such as boldness (see Chapter Two). As noted
previously, differences between heritability and maternal effect estimates were due to sire
effects within dams. Despite significant heritability and maternal effect estimates, a large
amount of the phenotypic variance observed in both greeting behavior and group joining
behavior remained unexplained by our models. One potential explanation for this unexplained
variation is the effect of early experience on sociability. There is accumulating evidence that
experience is important in determining sociability in non-human primates; periods of maternal
separation and reduced maternal care have been found to reduce social tendencies in pigtail
macaques (Caine et al. 1983; Capitanio & Reite 1984), and Andrews and Rosenblum (1991)
found that fluctuating resource availability early in life affected social competence in bonnet
macaques. Hyenas show many similarities to cercopithecine primates in terms of both social
structure and cognition (Drea & Frank 2003; Holekamp et al. 2006), and thus it is likely that
early experience contributes to some of the inter-individual variation in sociability observed in
hyenas, as it does in highly social primates.
Interestingly, sociability does not appear to be stable across ontogeny; individuals that
are particularly likely to solicit greetings as cubs are not generally more likely to solicit greetings
124

as adults. This runs contrary to our findings regarding aggressiveness and boldness in hyenas,
each of which appear to be determined relatively early in life (see Chapters Two and Three).
This effect may be partially explained by our finding that as social rank decreases, the tendency
to greet increases across ontogeny. In other words, low-ranking individuals are more likely
than high-ranking individuals to solicit more greetings as adults than they did as cubs. Before
they are 8 months old, cubs are still learning their social ranks in relation to those of their
peers, and do not yet understand their social rank in relation to older hyenas (Holekamp &
Smale 1993; Smale et al. 1993); therefore, rank likely has little or no effect on greeting behavior
during this time. However, after reaching reproductive maturity, social rank potentially has
large ramifications for greeting behavior. As discussed previously, there are more dominant
greeting partners available to females of low social status than to those of high social status.
Furthermore, limited access to resources may drive lower-ranking females to seek out the
advantages conferred by greeting, such as reinforced social bonds with cooperative and
affiliative partners, especially those of high rank (Smith et al. 2011).
In a similar vein, we found that the strength of the relationship between sociability and
reproductive success depended on social rank. Whereas high-ranking females tended to enjoy
high offspring survivorship no matter how social they were, the offspring of lower-ranking
females that joined groups more often were more likely to survive to den independence,
weaning, and reproductive maturity than were the offspring of lower-ranking females that did
not join groups as often. This indicates that the benefits of sociability are particularly large for
lower-ranking hyenas. Previous studies have shown that by associating with higher-ranking
females, subordinate females gain reduced aggression at kills and better access to resources
125

(Holekamp et al. 1997b; Boydston et al. 2001; Smith et al. 2007); these advantages likely
translate to an increase in feeding time for hyenas that are particularly social. Because the
reproductive success of female hyenas is strongly driven by access to resources (Holekamp &
Smale 1996), better access to food may be responsible for the increased survival of the
offspring of more gregarious low-ranking females, compared to low-ranking females that are
less social. It is interesting that both sociability and aggression (see Chapter Three) have more
significant ramifications for the reproductive success of low- than high-ranking hyenas. For
species such as hyenas, in which access to resources is vastly different for dominant and
subordinate individuals, the consequences of personality traits that affect resource holding
power may have especially large effects on low-ranking individuals. Findings regarding this
phenomenon are inconsistent; in contrast to our findings, the benefits of sociability are
independent of dominance in baboons, a species with a similar rank-based social organization
(Silk et al. 2003). The interaction of personality, social rank, and fitness may depend on many
factors, such as rigidity of the dominance hierarchy, the availability of food, and the benefits (or
disadvantages) conferred by the personality trait.
The lifespan of adult female hyenas was positively correlated with the rate at which they
joined subgroups within their fission-fusion society. This resembles earlier results from
baboons, in which females with more social support tended to live longer than those with less
social support (Silk et al. 2010). We know that there are definite benefits associated with social
behavior in both hyenas and baboons (Silk et al. 2006; Smith et al. 2007; Smith et al. 2010), but
it is not yet clear exactly how sociability affects longevity in hyenas. Although starvation is not a
major source of mortality among spotted hyenas (Kruuk 1972; Watts & Holekamp 2009),
126

increased access to resources may play an indirect role in prolonging lifespan. Moreover,
because sociability reinforces bonds and promotes cooperation (Smith et al. 2011), hyenas that
frequently engage in social behavior may have more allies present and thus enjoy better
protection during attacks from lions or neighboring groups of hyenas.
Future researchers should strive to elucidate the mechanism by which sociability is
related to survival in hyenas, as this remains poorly understood. Additionally, because we were
only able to explain less than a quarter of the phenotypic variation in the greeting solicitation
rate and group joining rate of hyenas, it would be fruitful to determine what other factors
contribute to inter-individual variation in sociability in this species. Neuropeptides, oxytocin,
and other hormones are known to influence social behavior in other species (Fries et al. 2005;
Lim & Young 2006), but whether hormonal mediation of sociability occurs in hyenas is currently
unknown. Other studies suggest that resource availability or periods of reduced maternal care
may affect sociability via early experience (Capitanio & Reite 1984; Andrews & Rosenblum
1991). It is also possible that litter or cohort effects may play a role in sociability in hyenas, as
these have been found to shape personality traits in other species (e.g. Taylor et al. 2012).
Finally, it would be particularly interesting to study the ramifications of sociability, as well as
other personality traits, in other species where access to resources is extremely unequal. This
type of study would help clarify the circumstances in which personality and social rank interact
to affect the fitness of dominant and subordinate animals in different ways.

127

APPENDIX

128

Table 4.A1
Situational variables predicting the average greeting count in each observation session.
Posterior mode
(95% credible interval)
pMCMC
Poisson
(Intercept)
-2.49937
(-3.27394 to -0.6501)
<0.002*
Location (carcass)
0.19204
( -0.19259 to 0.50616)
0.32
Location (den)
0.0752
(-0.12020 to 0.27166)
0.456
Location (kill)
-0.41461
(-0.65320 to -0.10078)
0.004*
Location (mating event)
0.37107
(-0.01014 to 0.78210)
0.052
Location (natal den)
0.08447
( -0.32860 to 0.43825)
0.68
Time of day (PM)
0.52476
(0.11271 to 0.75499)
<0.002*
Number of hyenas present
0.10805
( 0.05401 to 0.13304)
<0.002*
Session length
0.25974
(0.12581 to 0.42382)
<0.002*
Prey availability (low)
0.1246
(-0.01150 to 0.29387)
0.076
Zero-inflated
(Intercept)
2.50302
( 2.03461 to 2.85827)
<0.002*
Session length
-3.27528
(-3.83302 to -2.26259)
<0.002*

Table 4.A2
Fixed effects predicting the number of greetings solicited by each adult (> 24 months of age)
female hyena in each observation session.
Posterior mode
95% credible interval
pMCMC
Poisson
(Intercept)
Age
Standardized rank
Zero-inflated
(Intercept)
Age
Standardized rank

-2.552586
-0.001007
0.960474

(-2.996015 to -1.882476)
(-0.003943 to 0.001459)
(-0.066043 to 2.014638)

0.003*
0.4686
0.0743

-0.107573
0.012174
-0.31696

(-0.640567 to 0.578720)
(0.008033 to 0.015028)
(-1.254729 to 0.592191)

0.5943
<0.003*
0.601

129

Table 4.A3
Fixed effects predicting the number of greetings solicited by each adult (> 24 months of age)
male hyena in each observation session.
Posterior mode
95% credible interval
pMCMC
Poisson
(Intercept)
Age
Standardized rank
Zero-inflated
(Intercept)
Age
Standardized rank

-2.012691
0.008833
-1.722163

(-2.76507 to -1.318199)
(0.002756 to 0.016745)
(-2.905801 to -0.046431)

<0.003*
<0.003*
0.041*

1.123227
-0.019016
-2.071601

(0.331993 to 1.816341)
(-0.029105 to -0.010240)
(-3.530891 to 0.432572)

<0.003*
<0.003*
0.137

130

LITERATURE CITED

131

LITERATURE CITED
Altmann, S. E. 1974. Observational study of behavior: sampling methods. Behavior, 49, 227265.
Andrews, M., & Rosenblum, L. 1991. Dominance and social competence in differentially reared
bonnet macaques. In: Primatology Today, (Ed. by A. Ehara, T. Kimura, O. Takenaka, & M.
Iwamoto), pp. 347 - 350. Amsterdam: Elsevier.
Armitage, K. B. 1986. Individuality, social behavior, and reproductive success in yellow-bellied
marmots. Ecology, 67, 1186-1193.
Baglione, V., Marcos, J. M., Canestrari, D., & Ekman, J. 2002. Direct fitness benefits of group
living in a complex cooperative society of carrion crows, Corvus corone corone. Animal
Behaviour, 64, 887-893.
Baker, P. J., Robertson, C. P. J., Funk, S. M., & Harris, S. 1998. Potential fitness benefits of
group living in the red fox, Vulpes vulpes. Animal Behaviour, 56, 1411-1424.
Barash, D. P. 1976. Social behaviour and individual differences in free-living Alpine marmots
(Marmota marmota). Animal Behaviour, 24, 27-35.
Boon, A. K., Reale, D., & Boutin, S. 2008. Personality, habitat use, and their consequences for
survival in North American red squirrels Tamiasciurus hudsonicus. Oikos, 117, 1321-1328.
Boydston, E. E., Morelli, T. L., & Holekamp, K. E. 2001. Sex differences in territorial behavior
exhibited by the spotted hyena (Hyaenidae, Crocuta crocuta). Ethology, 107, 369-385.
Brodkin, E. S. 2007. BALB/c mice: Low sociability and other phenotypes that may be relevant to
autism. Behavioural Brain Research, 176, 53-65.
Caine, N. G., Earle, H., & Reite, M. 1983. Personality traits of adolescent pig-tailed monkeys
(Macaca nemestrina): An analysis of social rank and early separation experience. American
Journal of Primatology, 4, 253-260.
Capitanio, J. P. 1999. Personality dimensions in adult male rhesus macaques: Prediction of
behaviors across time and situation. American Journal of Primatology, 47, 299-320.
Capitanio, J. 2002. Sociability and responses to video playbacks in adult male rhesus monkeys
(Macaca mulatta). Primates, 43, 169-177.
Capitanio, J., & Reite, M. 1984. The roles of early separation experience and prior familiarity in
the social relations of pigtail macaques: A descriptive multivariate study. Primates, 25,
475-484.
132

Carere, C., Drent, P. J., Privitera, L., Koolhaas, J. M., & Groothuis, T. G. G. 2005. Personalities in
great tits, Parus major: stability and consistency. Animal Behaviour, 70, 795-805.
Clutton-Brock, T. 2002. Breeding Together: Kin Selection and Mutualism in Cooperative
Vertebrates. Science, 296 , 69-72.
Colmenares, F., Hofer, H., & East, M. L. 2000. Greeting ceremonies in baboons and hyenas. In:
Natural Conflict Resolution, (Ed. by F. Aureli & F. B. . de Waal), pp. 94-96. Berkeley:
University of California Press.
Conradt, L., & Roper, T. J. 2005. Consensus decision making in animals. Trends in Ecology &
Evolution, 20, 449-56.
Cote, J., & Clobert, J. 2007. Social personalities influence natal dispersal in a lizard. Proceedings
of the Royal Society B: Biological Sciences, 274, 383-390.
Cote, J., Dreiss, A., & Clobert, J. 2008. Social personality trait and fitness. Proceedings of the
Royal Society B: Biological Sciences, 275, 2851-2858.
Coussi-Korbel, S., & Fragaszy, D. M. 1995. On the relation between social dynamics and social
learning. Animal Behaviour, 50, 1441-1453.
Dingemanse, N. J., Kazem, A. J. N., Reale, D., & Wright, J. 2010. Behavioural reaction norms:
animal personality meets individual plasticity. Trends in Ecology &amp; Evolution, 25, 8189.
Dingemanse, N. J., Van der Plas, F., Wright, J., Réale, D., Schrama, M., Roff, D. A., Van der
Zee, E., & Barber, I. 2009. Individual experience and evolutionary history of predation
affect expression of heritable variation in fish personality and morphology. Proceedings of
the Royal Society B: Biological Sciences, 276, 1285-1293.
Drea, C. M., & Frank, L. G. 2003. The social complexity of spotted hyenas. In: Animal Social
Complexity, (Ed. by F. B. M. de Waal & P. L. Tyack), pp. 121–148. Harvard University Press.
East, M., Hofer, H., & Wickler, W. 1993. The erect “penis” is a flag of submission in a femaledominated society: greetings in Serengeti spotted hyenas. Behavioral Ecology and
Sociobiology, 33, 355-370.
Engh, A. L., Funk, S. M., Van Horn, R. C., Scribner, K. T., Bruford, M. W., Libants, S., Szykman,
M., Smale, L., & Holekamp, K. E. 2002. Reproductive skew among males in a femaledominated mammalian society. Behav. Ecol., 13, 193-200.

133

Figueredo, A. J., Cox, R. L., & Rhine, R. J. 1995. A generalizability analysis of subjective
personality assessments in the stumptail macaque and the zebra finch. Multivariate
Behavioral Research, 30, 167-197.
Frank, L. G. 1986. Social organization of the spotted hyaena (Crocuta crocuta). I. Demography.
Animal Behaviour, 34, 1500-1509.
Frank, L. G., Glickman, S. E., & Powch, I. 1990. Sexual dimorphism in the spotted hyaena
(Crocuta crocuta). Journal of Zoology, 221, 308-313.
Fries, A. B. W., Ziegler, T. E., Kurian, J. R., Jacoris, S., & Pollak, S. D. 2005. Early experience in
humans is associated with changes in neuropeptides critical for regulating social behavior .
Proceedings of the National Academy of Sciences of the United States of America , 102 ,
17237-17240.
Gittleman, J. 1989. Carnivore Behavior, Ecology and Evolution. Ithaca, New York: Cornell
University Press.
Glickman, S. E., Frank, L. G., Pavgi, S., & Licht, P. 1992. Hormonal correlates of
“masculinization” in female spotted hyaenas (Crocuta crocuta). 1. Infancy to sexual
maturity . Journal of Reproduction and Fertility , 95 , 451-462.
Goldsmith, H., & Gottesman, I. I. 1981. Origins of variation in behavioral style: a longitudinal
study of temperament in young twins. Child Development, 52, 91-103.
Haas, B. W., Mills, D., Yam, A., Hoeft, F., Bellugi, U., & Reiss, A. 2009. Genetic influences on
sociability: heightened amygdala reactivity and event-related responses to positive social
stimuli in Williams syndrome. The Journal of Neuroscience, 29, 1132-9.
Hadfield, J. D. 2010. MCMC methods for multi-response generalized linear mixed models: the
MCMCglmm R package. Journal of Statistical Software, 33, 1-22.
Hass, C., & Valenzuela, D. 2002. Anti-predator benefits of group living in white-nosed coatis
(Nasua narica). Behavioral Ecology and Sociobiology, 51, 570-578.
Holekamp, K. E., Cooper, S. M., Katona, C. I., Berry, N. A., Frank, L. G., & Smale, L. 1997a.
Patterns of association among female spotted hyenas (Crocuta crocuta). Journal of
Mammalogy, 78, 55-64.
Holekamp, K. E., Sakai, S. T., & Lundrigan, B. L. 2006. Social intelligence in the spotted hyena
(Crocuta crocuta). Philosophical Transactions of the Royal Society (B), 362, 523-538.
Holekamp, K. E., & Smale, L. 1993. Ontogeny of dominance in free-living spotted hyenas:
juvenile rank relations with other immature individuals. Animal Behaviour, 46, 451-466.
134

Holekamp, K., & Smale, L. 1996. Rank and reproduction in the female spotted hyaena. Journal
of Reproduction, 108, 229-237.
Holekamp, K. E., Smale, L., Berg, R., & Cooper, S. M. 1997b. Hunting rates and hunting success
in the spotted hyena (Crocuta crocuta). Journal of Zoology, 242, 1-15.
Kruuk, H. 1972. The spotted hyaena: a study of predation and social behavior. Chicago: Chicago
University Press.
Kruuk, L. E. B. 2004. Estimating genetic parameters in natural populations using the “animal
model”. Philosophical Transations of the Royal Society (B), 359, 873-890.
Kummer, H. 1971. Primate Societies: Group Techniques of Ecological Adaptation. Chicago:
Aldine.
Lim, M. M., & Young, L. J. 2006. Neuropeptidergic regulation of affiliative behavior and social
bonding in animals. Hormones and Behavior, 50, 506-517.
Loehlin, J. C. 1992. Genes and Environment in Personality Development. Thousand Oaks, CA:
Sage Publications.
Macdonald, K. 1983. Stability of individual differences in behavior in a litter of wolf cubs (Canis
Lupus). Journal of Comparative Psychology, 97, 99-106.
Majolo, B., de Bortoli Vizioli, A., & Schino, G. 2008. Costs and benefits of group living in
primates: group size effects on behaviour and demography. Animal Behaviour, 76, 12351247.
Maninger, N., Capitanio, J. P., Mendoza, S. P., & Mason, W. A. 2003. Personality influences
tetanus-specific antibody response in adult male rhesus macaques after removal from
natal group and housing relocation. American Journal of Primatology, 61, 73-83.
Martin, J. G. a., & Réale, D. 2008. Temperament, risk assessment and habituation to novelty in
eastern chipmunks, Tamias striatus. Animal Behaviour, 75, 309-318.
Mousseau, T. A., & Roff, D. A. 1987. Natural selection and the heritability of fitness
components. Heredity, 59, 181-197.
O’Brien, T. G. 1991. Female-male social interactions in wedge-capped capuchin monkeys:
benefits and costs of group living. Animal Behaviour, 41, 555-567.
Pinheiro, J. C., & Bates, D. M. 2000. Mixed-effects Models in S and S-Plus. New York: SpringerVerlag.

135

Plomin, R. 1994. Genetics and experience. London: Sage Publications.
Reale, D., & Festa-Bianchet, M. 2003. Predator-induced natural selection on temperament in
bighorn ewes. Animal Behaviour, 65, 463-470.
Reale, D., Reader, S. M., Sol, D., McDougall, P. T., & Dingemanse, N. J. 2007. Integrating
animal temperament within ecology and evolution. Biological Reviews, 82, 291-318.
Sankoorikal, G. M. V., Kaercher, K. A., Boon, C. J., Lee, J. K., & Brodkin, E. S. 2006. A mouse
model system for genetic analysis of sociability: C57BL/6J versus BALB/cJ inbred mouse
strains. Biological Psychiatry, 59, 415-423.
Schulke, O., Bhagavatula, J., Vigilant, L., & Ostner, J. 2010. Social Bonds Enhance Reproductive
Success in Male Macaques. Current biology : CB, 20, 2207-2210.
Sikes, R. S., & Gannon, W. L. 2011. Guidelines of the American Society of Mammalogists for the
use of wild mammals in research. Journal of Mammalogy, 92, 235-253.
Silk, J. B., Alberts, S. C., & Altmann, J. 2003. Social bonds of female baboons enhance infant
survival. Science, 302, 1231-1234.
Silk, J. B., Alberts, S. C., & Altmann, J. 2004. Patterns of coalition formation by adult female
baboons in Amboseli, Kenya. Animal Behaviour, 67, 573-582.
Silk, J., Altmann, J., & Alberts, S. 2006. Social relationships among adult female baboons (Papio
cynocephalus) I. Variation in the strength of social bonds. Behavioral Ecology and
Sociobiology, 61, 183-195.
Silk, J. B., Beehner, J. C., Bergman, T. J., Crockford, C., Engh, A. L., Moscovice, L. R., Wittig, R.
M., Seyfarth, R. M., & Cheney, D. L. 2010. Strong and consistent social bonds enhance the
longevity of female baboons. Current Biology, 20, 1359-1361.
Smale, L., Frank, L. G., & Holekamp, K. E. 1993. Ontogeny of dominance in free-living spotted
hyaenas: juvenile rank relations with adult females and immigrant males. Animal
Behaviour, 46, 467-477.
Smith, B. R., & Blumstein, D. T. 2008. Fitness consequences of personality: a meta-analysis.
Behavioral Ecology, 19, 448-455.
Smith, J. E., Van Horn, R. C., Powning, K. S., Cole, A. R., Graham, K. E., Memenis, S. K., &
Holekamp, K. E. 2010. Evolutionary forces favoring intragroup coalitions among spotted
hyenas and other animals . Behavioral Ecology , 21 , 284-303.

136

Smith, J. E., Kolowski, J. M., Graham, K. E., Dawes, S. E., & Holekamp, K. E. 2008. Social and
ecological determinants of fission–fusion dynamics in the spotted hyaena. Animal
Behaviour, 76, 619-636.
Smith, J., Memenis, S., & Holekamp, K. 2007. Rank-related partner choice in the fission–fusion
society of the spotted hyena (Crocuta crocuta). Behavioral Ecology and Sociobiology, 61,
753-765.
Smith, J. E., Powning, K. S., Dawes, S. E., Estrada, J. R., Hopper, A. L., Piotrowski, S. L., &
Holekamp, K. E. 2011. Greetings promote cooperation and reinforce social bonds among
spotted hyaenas. Animal Behaviour, 81, 401-415.
Svartberg, K., & Forkman, B. 2002. Personality traits in the domestic dog (Canis familiaris).
Applied Animal Behaviour Science, 79, 133-155.
Taylor, R. W., Boon, A. K., Dantzer, B., Reale, D., Humphries, M. M., Boutin, S., Gorrell, J. C.,
Coltman, D. W., & McAdam, A. G. 2012. Low heritabilities, but genetic and maternal
correlations between red squirrel behaviours. Journal of Evolutionary Biology, 25, 614-624.
R Development Core Team, 2011. R: A Language and Environment for Statistical Computing.
Van Horn, R. C., Engh, A. L., Scribner, K. T., Funk, S. M., & Holekamp, K. E. 2004. Behavioural
structuring of relatedness in the spotted hyena (Crocuta crocuta) suggests direct fitness
benefits of clan-level cooperation. Molecular Ecology, 13, 449-458.
Van Horn, R. C., McElhinny, T. L., & Holekamp, K. E. 2003. Age estimation and dispersal in the
spotted hyena (Crocuta crocuta). Journal of Mammalogy, 84, 1019-1030.
Van Meter, P. 2009. Hormones, stress and aggression in the spotted hyena (Crocuta crocuta).
Michigan State University.
Wahaj, S. A., Guse, K. R., & Holekamp, K. E. 2001. Reconciliation in the spotted hyena (Crocuta
crocuta). Ethology, 107, 1057-1074.
Watts, H. E., & Holekamp, K. E. 2009. Ecological determinants of survival and reproduction in
the spotted hyena. Journal of Mammalogy, 90, 461-471.
Watts, H. E., Scribner, K. T., Garcia, H. A., & Holekamp, K. E. 2011. Genetic diversity and
structure in two spotted hyena populations reflects social organization and male dispersal.
Journal of Zoology, 285, 281-291.
Wilson, D. S., Clark, A. B., Coleman, K., & Dearstyne, T. 1994. Shyness and boldness in humans
and other animals. Trends in Ecology and Evolution, 9, 442-446.

137

Wilson, A. J., Réale, D., Clements, M. N., Morrissey, M. M., Postma, E., Walling, C. A., Kruuk, L.
E. B., & Nussey, D. H. 2009. An ecologist’s guide to the animal model. Journal of Animal
Ecology, 79, 13-26.
Wilsson, E., & Sundgren, P.E. 1997. The use of a behaviour test for selection of dogs for service
and breeding. II. Heritability for tested parameters and effect of selection based on service
dog characteristics. Applied Animal Behaviour Science, 54, 235-241.

138