DEVELOPMENTAL FLEXIBILITY IN SPOTTED HYENAS (CROCUTA CROCUTA):
THE ROLE OF MATERNAL AND ANTHROPOGENIC EFFECTS
By
Julia Rachel Greenberg

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Integrative Biology -- Doctor of Philosophy
Ecology, Evolutionary Biology, and Behavior -- Dual Major
2017

ABSTRACT
DEVELOPMENTAL FLEXIBILITY IN SPOTTED HYENAS (CROCUTA CROCUTA):
THE ROLE OF MATERNAL AND ANTHROPOGENIC EFFECTS
By
Julia Rachel Greenberg
Whereas human activities are leading to declines in many wildlife populations, some species
appear to be faring relatively well in the face of environmental change by demonstrating a great
deal of phenotypic plasticity. I investigated the mechanisms underlying such flexibility in spotted
hyenas (Crocuta crocuta), a species that persists in areas rapidly changing due to human activity.
Specifically, I tested an hypothesis that the developmental trajectory of spotted hyenas is shaped
by anthropogenic disturbance, and that hyena mothers transmit information about
anthropogenic disturbance in their environment to their offspring via behavioral and/or
physiological maternal effects. To test this hypothesis, I used longitudinal analyses of hyena
behavior and physiology as well as cross-sectional analyses comparing hyenas in “highdisturbance” and “low-disturbance” areas of the Masai Mara National Reserve in Kenya.
First, I conducted field experiments to assess differences in personality among juvenile
spotted hyenas living in high- and low-disturbance areas. I found that juveniles in the highdisturbance area were significantly less neophobic and more exploratory, but also less bold, than
those in the low-disturbance area. Because most of the subjects tested were residing at the
communal den, where cubs are not directly exposed to anthropogenic activity, these findings
suggested that maternal effects may be shaping these differences. Next, I determined socioecological correlates of juvenile fecal glucocorticoid (fGC) concentrations and tested the

hypothesis that disturbance-related differences in fGC concentrations mediate differences in
juvenile personality. I did not find strong support for this hypothesis because, although juvenile
females showed higher fGC concentrations in high-disturbance than low-disturbance areas,
there were no significant disturbance-related differences in fGC concentrations among juvenile
males, lactating females, or pregnant females. Finally, I explored variation in maternal behavior
among spotted hyenas. I found that mothers showed significant individual variation in the
amount of time they spent grooming cubs, but little variation in the amount of time they spent
nursing and in close proximity to cubs, or in how they maintained proximity with cubs. I also
found support for the hypothesis that disturbance-related differences in maternal behavior
mediated the differences I documented in juvenile personality. Using spatial data from mothers
fitted with GPS collars, I found that mothers in the high-disturbance area showed significantly
lower rates of den attendance than did mothers in low-disturbance areas. While at the den,
mothers in high-disturbance spent more time nursing their cubs, but also less time close to their
cubs and higher rates of leaving their cubs, than did mothers in low-disturbance.
Overall, our results provide insights into how plasticity may be affording spotted hyenas
the ability to persist in disturbed areas. None of the maternal behaviors measured were
significantly associated with offspring survival, although maternal effects on offspring boldness
may positively affect cub survivorship in high-disturbance. Therefore, juvenile resilience to
variation in maternal behavior and an early period buffered from disturbance at the communal
den may allow mothers to alter their behavior in response to disturbance without negatively
affecting offspring survivorship.

Copyright by
JULIA RACHEL GREENBERG
2017

For my father, Gary Greenberg, who told me from a young age, that I would one day earn a
PhD just like he did, and fulfill my dream of studying animals in beautiful, far away places.

v

ACKNOWLEDGEMENTS

I have been fortunate in my academic career to learn from a number of “big thinkers.”
First and foremost, I want to thank Dr. Kay Holekamp, who, as an expansive scientist and
thinker, has encouraged me to “think big” in my approach to my dissertation and to science. I
am particularly grateful that Kay gave me the freedom to pursue my curiosities, take risks, and
combine my varied research interests in creative ways. I have learned so much from studying
with a scholar whose expertise cannot be easily classified into one sub-field or methodological
approach. I am grateful to have studied with a scientist who values keen observational skills,
encouraged me to ground my ideas in direct observations, and to spend time simply watching
my study subjects before making too many specific plans. I will always cherish the time that I
spent observing hyenas with Kay in the field and learning things that cannot be taught in any
textbook. Finally, I am beyond grateful to have spent seven years working with a woman who
sees me as a person and not simply a scientist. In an academic climate in which there is
increasing pressure to specialize, to de-value the art of deep observation in favor of quick
results, and to adopt a cut-throat mentality towards peers, I am flooded with the recognition of
how fortunate I have been. I have also been humbled by the opportunity to utilize a long-term
dataset that you and your students built over the course of three decades. Thank you.
Similarly, the opportunity to work among lab-mates whose interests and expertise are
so incredibly diverse has helped to expand my thinking into scientific realms I could not have
foreseen. I cannot express how much I have learned from my many lab-mates and fellow
researchers in Kenya throughout the years. I know how lucky I am to have worked in an
environment where people want to help each other, share ideas and data, and collaborate with
vi

one another even if it does not have some immediate reward. I specifically want to thank several
researchers and friends who have helped me in specific aspects of my dissertation work. Tracy
Montgomery and Sarah Jones, I honestly do not know how I would have survived lab work
without you. Jenna Parker, Lily Johnson-Ulrich, Wes Binder, Benson Pion, and Wilson Kilong,
you not only made my field experiments and observations possible, but I became a better field
researcher and observer from working alongside you. Andrew Denhardt (aka “Stats Man”), I
started meeting with you at a time when I was at my wits end with stats and you gave me the
confidence and skills to overcome analytical hurdles. Zach Laubach, if it weren’t for your
interest and skill in working with messy behavioral data, I might currently be summing up the
data for two chapters of this dissertation on pencil and paper. Thank you for being such a great
collaborator and I’m looking forward to continuing to work with you in the future. Joseph
Kamaamia, Jackson Kamaamia, Philomen Naigurian, Moses Naigurian, and George Kilinyet,
thank you for opening up your hearts and worlds to me in Kenya. The things I learned from
living and learning with you cannot be expressed in the text of a dissertation. I also am grateful
to have worked with numerous talented undergraduates, several of which not only made my
research possible, but also helped to inform my thinking, specifically Shannon Carvey, Kaycee
Morra, Samantha Patterson, and Karee Lesko.
There is not enough room for me to describe how each of the following people have
helped me along the way, but I want to thank them for not only being great colleagues, but also
for being supportive friends who have made my years in graduate school and Kenya an insane
amount of fun: Leslie Curren, Katy Califf, Andy Flies, Andy Booms, David Green, Marc
Wiseman, Siri Okamoto, Emily Dittmar, Eli Swanson, Eli Strauss, Kenna Lehmann, Julie
Turner, Nora Lewin, Hadley Couraud, Erin Pearson, Aaron Sandel, Rachna Reddy, Bethany
vii

Hansen, Elizabeth Johnson, Rachael Eaton, Bene Bachelot, Rohan Maddamsetti, Kari
Dammerman, Austin Dreyer, Pat Bills, and Dee White.
I want to thank my committee members for their guidance. Tom Getty, the person who
always asks that tough question at seminar, I have benefited from your discerning eye. Joe
Lonstein, I’m grateful you made sure I read foundational papers for my comprehensive exams,
and I’ve greatly benefited from having my research critiqued by someone who studies similar
topics in a laboratory setting. Laura Smale, your class and our discussions over the years have
made me think deeply about developmental processes and I am grateful to have had the
opportunity to utilize some of the data you collected in the early days of the project.
I also want to thank several other faculty members who helped in my training and
development during graduate school. Dr. Jeff French and Dr. Jacinta Beehner, thank you for
opening up your endocrinology labs to me and helping me to understand and resolve numerous
thorny issues involving hormone assays and long-term hormonal data sets. Thanks also to your
lab managers, Drew Birnie and Teera Parr for assisting in my training. Dr. John Mitani, your
Friday afternoon lab meetings served as a kind of surrogate academic home for me while I was
conducting research in Ann Arbor. I want to thank you for being so welcoming, for your
enthusiasm and interest in talking about my research, and for the numerous times that you
shared your own “big thoughts” about science and the world with me.
My work would not have been possible without funding to myself and to my advisor
from several sources. My research was supported by an NSF Graduate Research Fellowship, the
Hensley Fellowship from the MSU College of Natural Sciences, and grants from MSU’s
Program in Ecology, Evolutionary Biology, and Behavior. This work was also supported by
NSF Grants OISE 1556407 and DEB 1353110. I also want to thank the Kenyan National
viii

Council for Science and Technology, the Narok County Council, the Kenya Wildlife Service,
and the Mara Conservancy for permission to conduct field research with hyenas.
Finally, I want to thank several people in my non-academic life. Michelle (mom), you
have always believed in me. In graduate school, the fact that you have seen such value in me
pursuing a PhD, and this line of research in particular, has helped me to believe in myself and
my abilities. Your “pep talks” and writing advice have been invaluable. Jared, my biggest fan,
thank you for your enthusiasm, and thanks to you and Jordie for making such a memorable visit
to Kenya. Aura, you came into my life at a difficult time in my PhD career, and you helped me
move through it. Thank you for forcing me to leave my computer and go for beautiful walks
and swims with you, and for always humbling me when I begin to think I know anything about
animal behavior. John, meeting you mid-way through grad school and your presence in my life
ever since has kept me grounded in why I did this to begin with. Moving through the world
with you helps me to remember that no scientific analysis can replace what is learned by being in
nature, and that I should never let an academic matter distract me from my fundamental
curiosity and connection to the natural world. Your help formatting this dissertation wasn’t bad
either! Finally, thank you to my father. Although you couldn’t be here during this time in my
life, I have imagined you with me in all phases of this process-- visiting me in Kenya and playing
“research assistant” like you always said you would, meeting my colleagues and friends, or
having long chats on the phone about the details of research that most people don’t want to
hear. My PhD has been the fulfillment of a childhood dream that you helped to foster.

ix

TABLE OF CONTENTS

LIST OF TABLES...................................................................................................................................xii
LIST OF FIGURES.................................................................................................................................xiv
CHAPTER 1 GENERAL INTRODUCTION.....................................................................................1
OVERVIEW OF CHAPTERS ................................................................................................................4
CHAPTER 2 HUMAN DISTURBANCE AFFECTS PERSONALITY
DEVELOPMENT IN A WILD CARNIVORE ..................................................................................8
ABSTRACT..................................................................................................................................................8
INTRODUCTION ....................................................................................................................................9
METHODS................................................................................................................................................12
Subjects & Study Populations ...................................................................................................................12
Experimental Stimuli ................................................................................................................................14
Ethical Note ...............................................................................................................................................16
Data Collection ..........................................................................................................................................16
Data extraction ...........................................................................................................................................17
Statistical Analysis ......................................................................................................................................18
Independent Variables ..............................................................................................................................18
Models of variation in neophobia, exploration and boldness.................................................................20
Consistency across time and context ........................................................................................................22
Relationship between personality traits and survival ..............................................................................23
RESULTS ..................................................................................................................................................24
Neophobia ..................................................................................................................................................24
Exploration .................................................................................................................................................25
Boldness ......................................................................................................................................................27
Consistency across time and context ........................................................................................................29
Relationship between personality traits and survival ..............................................................................31
DISCUSSION ..........................................................................................................................................32
ACKNOWLEDGEMENTS ..................................................................................................................38
CHAPTER 3 SEX- AND AGE-SPECIFIC CORRELATES OF FECAL
GLUCOCORTICOID CONCENTRATIONS IN SPOTTED HYENAS....................................39
INTRODUCTION ................................................................................................................................. 39
METHODS ...............................................................................................................................................44
Study Site and Subjects...............................................................................................................................44
Fecal sample collection, extraction, and immunoassay of hormone concentrations...........................46
Statistical Analysis ......................................................................................................................................47
Longitudinal analysis of juvenile fGC concentrations 1993-2014 ........................................................48
x

Cross-sectional comparison of fGC concentrations 2009-2014...........................................................48
Relationship between early life fGC concentrations and fitness...........................................................50
RESULTS..................................................................................................................................................52
Longitudinal analysis of socio-ecological factors affeting juvenile fGC concentrations...................52
Cross-sectional comparison of fGC concentrations in high- and low-disturbance areas...................52
Juveniles.......................................................................................................................................................52
Lactating and Pregnant Females...............................................................................................................55
Relationship between early life fGC concentrations and fitness............................................................57
DISCUSSION....................................................................................................................................58
CHAPTER 4 VARIATION IN MATERNAL BEHAVIOR AND
MOTHER-OFFSPRING INTERACTIONS......................................................................................64
INTRODUCTION..........................................................................................................................64
METHODS ...............................................................................................................................................69
Subjects & Study Populations...................................................................................................................69
Data Collection .........................................................................................................................................71
Statistical Analysis ....................................................................................................................................72
Individual variability and repeatability of maternal behavior.................................................................75
Relationship between maternal behavior and offspring fitness.............................................................76
RESULTS.....................................................................................................................76
Models of time spent nursing, grooming, and in close proximity to offspring ....................................76
Models of proximity maintenance...........................................................................................................82
Individual variability and repeatability of maternal behavior.................................................................82
Relationship between maternal behavior and offspring fitness.............................................................85
DISCUSSION ...........................................................................................................................................85
CHAPTER 5 ANTHROPOGENIC EFFECTS ON MATERNAL
BEHAVIOR IN SPOTTED HYENAS ..............................................................................................91
INTRODUCTION...............................................................................................................................91
METHODS..................................................................................................................................97
Subjects & Study Populations....................................................................................................................97
Data Collection...........................................................................................................................................99
Maternal attendance at dens ......................................................................................................................99
Maternal behavior at dens .......................................................................................................................102
Statistical analysis of maternal den attendance and behavior ...............................................................103
Maternal den attendance and offspring survivorship...........................................................................105
RESULTS .................................................................................................................................................106
Maternal attendance at dens ...................................................................................................................106
Maternal behavior at dens ......................................................................................................................108
Maternal den attendance and offspring survivorship ..........................................................................113
DISCUSSION .........................................................................................................................................115
REFERENCES .......................................................................................................................................120
xi

LIST OF TABLES
Table 2.1. Results of a Cox proportional hazards model for variation in latency to contact the
novel object (N=58). Asterisks indicate P ≤ 0.051.............................................................................25
Table 2.2. Results of a logistic regression model for variation in exploration of the novel object
(N=58). Asterisks indicate P ≤ 0.051....................................................................................................27
Table 2.3. Results of a logistic regression model for variation in boldness (N=60). Asterisks
indicate P ≤ 0.051......................................................................................................................................29
Table 3.1. Factors explaining variation in fGC concentrations in juvenile spotted hyenas from
the Talek West clan 1993-2014 (N=382 samples from 171 individuals). An asterisk indicates
P<0.05.........................................................................................................................................................53
Table 3.2. Results of preliminary analysis of variation in fGC concentrations among juveniles,
lactating, and pregnant females in the low disturbance area. An asterisk indicates
P<0.05.........................................................................................................................................................54
Table 4.1. Results of a zero-inflated negative binomial model explaining variation among
mothers in duration of nursing. An asterisk indicates P<0.05..........................................................79
Table 4.2. Results of a zero-inflated negative binomial model explaining variation in time
mothers spent in close proximity to their cubs. An asterisk indicates P<0.05...............................79
Table 4.3. Results of a negative binomial model explaining variation in time mothers spent
grooming offspring. An asterisk indicates P<0.05..............................................................................81
Table 4.4. Results of a negative binomial model explaining variation in how often mothers
approached their cub. An asterisk indicates P<0.05...........................................................................83
Table 4.5. Results of a negative binomial model explaining variation in how often mothers
moved more than 1 m from their cub. An asterisk indicates P<0.05..............................................84
Table 5.1. Results of zero-inflated negative binomial model explaining variation in
maternal den attendance. ** indicates P <0.05. * indicates P <0.1................................................107
Table 5.2. Tukey pairwise comparisons of rates of nursing, grooming, and close proximity
among mothers living in the three clans in the undisturbed area....................................................110

xii

Table 5.3. Results of models explaining variation in how much FAS time mothers spent: a)
nursing cubs (zero-inflated negative binomial model), b) in close proximity to cubs
(negative binomial model), and c) grooming cubs (negative binomial model). ** indicates
P <0.05. * indicates P <0.1...................................................................................................................111
Table 5.4. Results of a zero-inflated negative binomial model explaining variation in how often
mothers approached their cubs to within 1m. ** indicates P <0.05. * indicates P <0.1.............114
Table 5.5. Results of a poisson model explaining variation in how often mothers left their cubs.
** indicates P <0.05. * indicates P <0.1...............................................................................................114

xiii

LIST OF FIGURES
Figure 2.1. The four objects used for the novel object test to measure neophobia and exploration.
Each object was approximately 0.3m tall when deployed. The cooler, funnel, and stool were
plastic, and the bucket was metal. Novel objects were chosen that juveniles would be highly
unlikely to encounter otherwise..............................................................................................................15
Figure 2.2. The metal box used for the baited box test to measure boldness (70cm long x 40cm
wide x 44cm high). The box had no bottom surface except for a metal tray (20cm X 40cm) at
the back end of the box where we could sprinkle a layer of powdered milk (15ml). The box was
open on one side, allowing a subject to walk in and fully enter the box in order to access the
powdered milk at the back.......................................................................................................................16
Figure 2.3. A Cox proportional hazards model indicated that: a) juveniles in high disturbance
were significantly less neophobic than subjects in low disturbance (P=0.05). b) social rank was
inversely related to latency to contact the object (P=0.002). Social rank was analyzed statistically
as a continuous variable but is shown here by dividing the clan’s hierarchy into three dominance
classes (the high rank category includes ranks 1.00 to 0.33, the middle rank category includes
ranks 0.33 to -0.33, and the low rank category includes ranks -0.33 to -1.00).................................26
Figure 2.4. A logistic regression model for variation in exploration score indicated that: a)
juveniles in high disturbance were significantly more exploratory than those in low disturbance
(P=0.003) and b) high-ranking juveniles were significantly more exploratory than low-ranking
juveniles (P=0.020; N=58). ....................................................................................................................28
Figure 2.5. A logistic regression model for variation in boldness score indicated that juveniles in
high disturbance were less bold than juveniles in low disturbance (P=0.019;
N=60).........................................................................................................................................................30
Figure 3.1. A longitudinal analysis of juvenile fGC concentrations in the Talek West clan
1993-2014 indicated that fGC concentrations significantly decrease with age among juveniles of
both sexes, and that juvenile females have significantly higher fGC concentrations than do
juvenile males. Only samples collected in the morning are shown (N=188 samples)...................53
Figure 3.2. Residual juvenile fGC concentrations in high- and low-disturbance areas after
controlling for time of day and age at sample collection. N=55 females (82 samples) and 67
males (112 samples). An asterisk indicates p < .05..............................................................................56
Figure 4.1. Photograph of a spotted hyena mother nursing two cubs. As is typical, the
dominant member of the litter is laying up against its mother’s stomach while nursing,
whereas the subordinate litter-mate is strattled between her legs. Photo courtesy of Lily
Johnson-Ulrich. .........................................................................................................................................70
xiv

Figure 4.2. The total number of Focal Animal Surveys (FAS) collected on mothers with cubs
10-15 months. Data are binned by the cub’s age in months. Numbers above bars indicate the
number of cubs observed in that age
category. .....................................................................................................................................................77
Figure 4.3. Developmental changes in three maternal behaviors among spotted hyena mothers
with cubs 0-15 months old. Out of the total minutes a mother and cub were both present in an
FAS, figures show the proportion of time a mother spent a) nursing the cub; b) grooming the
cub and; c) in close proximity to the cub. Labels indicate the number of FAS averaged for each
month of development.............................................................................................................................78
Figure 4.4. Developmental changes in the percentage of approaches to within 1m that were
initiated by the mother. Labels indicate the number of FAS averaged for each month of
development. ............................................................................................................................................83
Figure 4.5. Developmental changes in the percentage of departures to further than 1m apart that
were initiated by the mother. Labels indicate the number of FAS averaged for each month of
development...............................................................................................................................................84
Figure 4.6. Individual variation in the proportion of a mother’s time spent nursing, grooming,
and in close proximity to cubs. Y-axis is the best linear unbiased predictor (BLUP) for each of
the 34 mothers calculated from our models for proportion of time spent in each of the three
maternal behaviors.....................................................................................................................................86
Figure 5.1. Map of the territories of the four study clans monitored in the Masai Mara National
Reserve, Kenya. Study clans were Talek West (TW), Serena North (SN), Serena South (SS), and
Happy Zebra (HZ). The three clan in the shaded area are located in the Mara Conservancy.
Star indicates the town of Talek..............................................................................................................98
Figure 5.2. Map of one female’s GPS locations while she was at a natal den, showing a high
concentration of points in close proximity to the natal den. Rings indicate 10m and 25m from
the natal den. Approximately 50% of her total GPS points for the one-month period were
within 10m of the natal den...................................................................................................................100
Figure 5.3. Plot showing mean maternal den attendance as cubs mature in low- and highdisturbance areas. Data points represent monthly means of mothers’ daily den attendance
(the proportion of mothers’ total points that were within 100m of the communal den). ..........107
Figure 5.4. Plot showing mean proportion of time mothers spent a) nursing and b) in close
proximity to cubs as cubs mature in low- and high-disturbance areas. Data points represent
monthly means of mothers’ behavior...................................................................................................109

xv

CHAPTER 1
GENERAL INTRODUCTION
Across the globe, there are few ecosystems that remain unaffected by human population growth,
changes in land-use, or climate change. Species that appear best able to adjust to such changes
are those expressing a great deal of phenotypic plasticity (Reed et al. 2010, Sih 2013). Numerous
mechanisms of plasticity have been documented in mammalian systems, including those
involving neural circuitry, endocrine function, behavior, immunological responses, and gene
expression. Plasticity is often shaped by developmental processes (Lindström 1999, Bateson and
Gluckman 2011), and mothers can play a significant role in shaping developmental plasticity
through maternal effects, in which a mother’s phenotype directly influences her offspring’s
phenotype (Bernardo 1996). In recent years, there has been great interest in the potential of
maternal effects to be adaptive for offspring, whereby mothers transmit information to
offspring about the environment that in turn allows offspring to maximize their survival or
reproduction (Mousseau and Fox 1998, West-Eberhard 2005). The extent to which maternal
effects serve an adaptive function for offspring is controversial (Marshall and Uller 2007).
However, their potential to shape offspring phenotypes in situations of rapid environmental
change is intriguing, because it could serve as a powerful mechanism of plasticity that facilitates
species persistence in disturbed environments (Jablonka et al. 1995, Bernardo 1996).
In my dissertation, I aim to develop spotted hyenas as a model species for studying
mammalian developmental plasticity in the context of environmental change. I test an
hypothesis suggesting that the developmental trajectory of spotted hyenas is shaped by
anthropogenic disturbance, and that spotted hyena mothers transmit information about
1

anthropogenic disturbance in their environment to their offspring via behavioral and/or
physiological maternal effects. Spotted hyenas make an excellent species in which to examine
developmental flexibility because they show an unusually protracted period of physical and
behavioral development, and because mothers invest very heavily in a small number of offspring
(Watts et al. 2009). Mothers give birth to 1 or 2 cubs and nurse cubs until approximately 14
months. Even after being weaned, offspring survival is dependent on maternal support at
carcasses, where young hyenas must compete against adults for food despite the fact that the
feeding apparatus of young spotted hyenas is not fully developed for eating meat and bone until
36 months of age (Tanner et al. 2009). Individuals do not reach full proficiency at hunting until
roughly five years of age (Holekamp et al. 1997b).
Adult spotted hyenas show remarkable flexibility in many aspects of their biology and
ecology. The most abundant large carnivores in sub-Saharan Africa, spotted hyenas exist, and
often thrive, in a diverse range of habitats. Hyenas’ ecological flexibility allows them to feed on
multiple food sources (at least 43 types of prey in the Masai Mara alone; Cooper et al. 1999);
hunt up to 95% of their own food at some sites but also exist as opportunistic scavengers and
trash-eaters at others; and fill a bone-eating niche that no other animal can (Honer et al. 2002,
Holekamp and Dloniak 2010). They also show pronounced behavioral flexibility, exhibiting the
ability, for example, to follow migratory prey in seasonal patterns on “commuting” trips for up
to 6 days or to shift their daily biorhythms in response to local human activity (Hofer and East
1993a, Kolowski et al. 2007). Spotted hyenas’ impressive physiological flexibility affords them the
ability to conceive and reproduce year round and survive exposure to many pathogens that
compromise the immune systems of other carnivores, such as canine distemper and rabies
2

(Holekamp et al. 1996, Flies et al. 2012). Finally, they show a great deal of social flexibility in that
their fission-fusion social system allows them to adjust their group size and composition in order
to take advantage of highly ephemeral carcasses and kills (Frank 1986a, Smith et al. 2008).
Each of these aspects of spotted hyenas’ flexibility has likely played a role in allowing
them to adjust to increasing anthropogenic disturbance over the past twenty-nine years at our
study site in the Masai Mara National Reserve, Kenya (MMNR). In the territory of the Talek
West clan, which lives along the border of the MMNR, the human population has grown
exponentially since the late 1990s, resulting in increased wildlife-human conflict, and numbers of
livestock being grazed within the reserve (Kolowski and Holekamp 2006, Holekamp and
Dloniak 2010, Green et al. 2017). As human disturbance has increased, adult spotted hyenas
have become more nocturnal, expanded their home ranges, and increased the distances they
travel to hunt (Boydston et al. 2003b). Compared with hyenas living in pristine areas of the
MMNR, hyenas in the disturbed Talek region show a greater preference for protective
vegetation (Kolowski and Holekamp 2009), are more vigilant (Pangle and Holekamp 2010), shift
their activity patterns to avoid interaction with humans (Kolowski et al. 2007), travel greater
distances in the mid-day heat (Ikime 2015), and travel faster overall (Green 2015).
The marked flexibility that has been documented among adult hyenas raises questions
regarding the development of such plasticity. When during a hyena’s lifetime do we begin to see
disturbance-related differences? How is plasticity shaped throughout the course of a hyena’s
development? What cues provide information to a developing hyena about anthropogenic
disturbance in its environment? These questions are even more intriguing given that young
spotted hyenas reside at a communal den, in an environment that is distinct from that
3

experienced by their mothers and other adults. Much of the research in my dissertation focuses
on juveniles when they are residing at communal dens at approximately 1-9 months old. Each
communal den is composed of a series of burrows and tunnels that protect hyenas from dangers
present in the rest of the clan’s territory (Boydston et al. 2006). However, the communal den is
also a social focal point of the clan, and cubs emerge daily to interact with their mothers and
other hyenas, who are directly experiencing those dangers (Holekamp and Smale 1998a). In
species that are long-lived and slow-growing, like spotted hyenas, the ability to respond to strong
maternal signals of anthropogenic disturbance, starting very early in life, while still inhabiting a
relatively safe environment, may be critical to their success in disturbed environments.

OVERVIEW OF CHAPTERS
Each of the remaining chapters of my dissertation helps to test the hypothesis that the
developmental trajectory of spotted hyenas is shaped by anthropogenic disturbance, and that
spotted hyena mothers transmit information about anthropogenic disturbance in their
environment to their offspring via behavioral and/or physiological maternal effects.
Throughout, I employ both longitudinal analyses of the hyena population that has been
monitored since the MSU Hyena Project began in 1988, as well as cross-sectional analyses
comparing hyenas in “high-disturbance” and "low-disturbance” areas of the MMNR. In each
chapter, I also examine the influence of other socio-ecological factors on cub development, and
whether anthropogenic disturbance interacts with these other factors to affect spotted hyena
behavior or physiology. Finally, in each chapter, I test whether the juvenile or maternal trait of
interest is significantly associated with offspring survivorship and/or longevity. When sample
4

size allows, I also test whether the relationship between these traits and offspring fitness differs
in areas of high- and low- disturbance.
In Chapter 2, I examine variation in personality among juvenile spotted hyenas, and test
the hypothesis that juvenile spotted hyenas, like birds and small mammals, show consistent
differences in personality between high- and low-disturbance areas. I use field experiments to
assess individual variation and consistency in neophobia, exploration, and boldness. I find that
juveniles in the high-disturbance area are significantly less neophobic and more exploratory, but
also less bold, than those in low-disturbance areas. Because most of the subjects tested were
residing at the communal den, where cubs are not directly exposed to anthropogenic activity,
these findings raise the question of what mechanisms mediate these behavioral differences. It
suggests that disturbance-related differences in maternal behavior and/or physiology may be
shaping these differences in hyenas, starting very early in life.
In Chapter 3, I inquire which anthropogenic and socio-ecological factors are associated
with variation in juvenile fecal glucocorticoid (fGC) concentrations. I also test the hypothesis
that disturbance-related differences in fGC, either among juveniles or their mothers, mediate the
behavioral differences documented in Chapter 2. I do not find strong support for this
hypothesis because, although juvenile females show higher fGC concentrations in highdisturbance than low-disturbance areas, there were no significant disturbance-related differences
in fGC concentrations among juvenile males, lactating females, or pregnant females. This
suggests that juvenile females may show a particular sensitivity to anthropogenic disturbance and
that there may be important sex differences in how early stressors affect hyenas. However, it
also suggests that stress physiology, either among juveniles themselves or their mothers, are
5

likely not a major mediator of behavioral differences between juveniles inhabiting high- and lowdisturbance areas. This suggests that disturbance-related differences in maternal behavior may
be the more important factor.
In Chapter 4, I use archived focal animal surveys (FAS) to document naturally-occurring
variation in maternal behavior among individuals living in the Talek West clan before
anthropogenic disturbance was prevalent. This provides the first study documenting the degree
to which spotted hyena mothers vary in basic measures of maternal care, such as nursing,
grooming, and maintaining close proximity with offspring. Interestingly, I find that mothers
show little consistent individual variation in the amount of time they spend nursing and in close
proximity to cubs, or in how they maintain proximity with cubs. However, they show
significant individual variation with respect to how much time they spend grooming cubs.
Unlike most aspects of hyena behavior that have been studied to date, a female’s social rank is
not significantly associated with the amount of time she spends nursing, grooming, or in close
proximity to offspring. My findings also indicate that a cub’s litter size significantly shape its
experience of maternal care.
Having investigated naturally-occurring variation in maternal behavior, in Chapter 5, I go
on to ask how maternal behavior is influenced by anthropogenic disturbance and test the
hypothesis that differences in maternal behavior between the high- and low-disturbance areas
mediate differences in juvenile personality. First, I use spatial data from mothers fitted with
GPS collars to compare rates of maternal attendance at communal dens in the high- and lowdisturbance areas. Second, to inquire whether mothers in high- and low-disturbance areas, when
present at communal dens, behave differently towards their cubs, I conducted FAS on mother6

offspring pairs. Spotted hyena mothers in the high disturbance area showed significantly lower
rates of den attendance than did mothers in low disturbance areas. Perhaps to compensate,
mothers in the high- disturbance area spent a higher proportion of their time nursing their cubs
than did mothers in low-disturbance areas. Mothers in the high-disturbance area also showed
differences regarding the ways in which they interacted with their cubs, spending less time close
to their cubs and more often leaving their cubs rather than staying close to them. Our findings
suggest that, although spotted hyenas show a remarkable ability to persist in disturbed areas,
such conditions present a challenge to mothers. Our findings also support the hypothesis that
disturbance-related differences in maternal behavior are associated with differences in juvenile
personality. The association between maternal behavior and offspring personality shows some
consistency with results from studies of Old World Monkeys. In these primates, mothers who
show a low investment maternal strategy or are more “rejecting” tend to have offspring that are
less neophobic and more exploratory than do mothers who show high investment in offspring
or are more “protective” (reviewed in Fairbanks, 1996).
All of the research presented in this dissertation was the result of collaboration. My
analyses utilize data collected by the many research assistants and graduate students who have
contributed to the daily monitoring of spotted hyenas in Kenya since 1988. Additionally, each
of my chapters was prepared in manuscript form, with significant input from co-authors.
Therefore, I use the first person plural, rather than the first person singular, throughout the
remainder of this dissertation.

7

CHAPTER 2
HUMAN DISTURBANCE AFFECTS PERSONALITY DEVELOPMENT IN A WILD
CARNIVORE
Julia R. Greenberg & Kay E. Holekamp. 2017. Human disturbance affects personality
development in a wild carnivore. Animal Behaviour 132:303-312
ABSTRACT
Human activity can dramatically affect personality traits in birds and small mammals. However,
we know very little about how anthropogenic disturbance shapes personality in mammalian
carnivores, and whether the personality traits that may be affected have fitness consequences in
human-dominated landscapes. We adapted standard experiments commonly used to assess
personality in captive animals to compare three personality traits in 72 wild juvenile spotted
hyenas (Crocuta crocuta) living in either areas heavily disturbed by human activity or areas with low
levels of disturbance. We examined neophobia, defined as the tendency to avoid unfamiliar
things, exploration, defined as the number of different ways an individual interacts with an
object, and boldness, defined as an individual’s tendency to take risks. To assess neophobia and
exploration, we measured individuals’ responses to a novel object, and to assess boldness, we
measured hyenas’ propensity to enter a wire-mesh box to obtain food. Juvenile spotted hyenas
living in low disturbance areas were significantly more neophobic and less exploratory than
individuals living in high disturbance areas. This is consistent with results obtained with birds
and small mammals; however, unlike these other taxa, hyenas living in low disturbance areas
were bolder than individuals living in high disturbance areas. The expression of some of these
personality traits was also affected by the subject’s social rank and the presence of a litter-mate,
8

but not by subject age or sex. Of the three traits, only boldness predicted survival to adulthood;
less bold individuals were significantly more likely to survive than bolder individuals, in both
high disturbance and low disturbance habitats. We propose that behavioral or physiological
maternal effects may be shaping offspring temperament differences related to disturbance.

Keywords: personality, temperament, anthropogenic disturbance, human-wildlife conflict, social
rank, neophobia, spotted hyena, Crocuta crocuta

INTRODUCTION
Human activity has been found to affect the behavior of animals in a variety of ways. Some of
the most pronounced effects have been documented on animals’ personality or temperament
traits, which are individual differences in behavior that are stable across contexts and over time
(Réale et al. 2007, Miranda et al. 2013, Sol et al. 2013). Changes in personality traits due to
anthropogenic disturbance can result from individual behavioral plasticity or micro-evolutionary
changes (Miranda et al. 2013). Across numerous bird and small mammal species, individuals
tend to show more neophilic, exploratory, aggressive, and bold personalities in urban than rural
areas (Miranda et al. 2013, Sol et al. 2013). However, we know virtually nothing about whether
or how human disturbance shapes personality in most other animals, including large mammalian
carnivores.
There is a critical need to understand how human activity influences carnivore behavior
and populations (Baker et al. 2008, Darrow and Shivik 2009). Worldwide, carnivores are
increasingly living in close proximity to humans (Treves and Karanth 2003, Treves et al. 2006,
9

Bateman and Fleming 2012), resulting in more frequent predation on livestock, injury to
humans, and retaliatory killing of carnivores (Baker et al. 2008, Ripple et al. 2014).
Understanding whether variation in personality affects carnivore survival in disturbed landscapes
can help us to predict the effects of human activity on carnivore populations. Consistent interindividual differences in behavior have been documented in a few wild adult carnivores (e.g.,
coyotes, Canis latrans: Harris and Knowlton 2001, Heffernan et al. 2007), but most studies have
been conducted on captive individuals (maned wolves, Chrysocyon brachyurus: Silva and Azevedo
2013; gray wolves, Canis lupus: Fox 1972, MacDonald 1983, Moretti, Hentrup, Kotrschal, &
Range, 2015; coyotes, Mettler and Shivik 2007; European mink, Mustela lutreola: Haage et al.
2013; American mink, Neovison vison: Noer et al. 2015; kowaris, Dasyuroides byrnei: Russell and
Pearce 1971, Wynne and McLean 1999 or domestic animals (dogs, Canis familiaris: Draper 1995,
Jones and Gosling 2005; cats, Felis catus: Durr and Smith 1997. Understanding variation in
personality among wild carnivores may also help researchers and managers to discern whether
certain individuals consistently display behavior patterns that put them at risk of conflict with
humans, and to target such “problem animals” for intervention (Caro 1999, McDougall et al.
2006, Shivik 2006).
Few studies have assessed the fitness consequences of personality traits in areas
disturbed by human activity (Archard and Braithwaite 2010) , although a number of recent
studies have found that personality traits may be heritable and affect fitness in undisturbed
habitats (Dingemanse et al. 2002, Dall et al. 2004, Smith and Blumstein 2008, Nicolaus et al.
2012, Taylor et al. 2012, Kortet et al. 2014, Petelle et al. 2015, Yoshida et al. 2016). It has been
hypothesized that, in disturbed areas, birds and small mammals with more neophilic,
10

exploratory, and bold personalities may have a fitness advantage, as these traits may allow them
to exploit novel habitats, live at high population densities, and take advantage of new resources
(Miranda et al. 2013). However, the fitness consequences of these personality traits in areas with
human activity may be dramatically different for carnivores than for birds or small mammals, as
such tendencies would likely bring carnivores into direct conflict with humans.
Understanding whether personality traits in juvenile carnivores predict survivorship may
be particularly important because young animals may be especially likely to engage in risky
behavior and end up in conflict with humans (Anderson 1981, Saberwal et al. 1994). Lack of
hunting experience, tendency to prey on animals that are easy to kill, and poor body condition
relative to adults, are all factors that might make juvenile carnivores prone to human-wildlife
conflicts, such as depredation of livestock (Stirling and Latour 1978, Payne and Jameson 1984,
Matlack and Evans 1992, Seidensticker and McDougal 1993, Caro 1994, Holekamp et al. 1997b,
Litvaitis et al. 2017). Young dispersing males may be particularly prone to engage in such
activities (Linnell et al. 1999).
Here we used an experimental approach to explore variation in three personality traits in
wild juvenile spotted hyenas (Crocuta crocuta), and inquire whether these traits differ between
hyenas reared in habitat heavily disturbed by anthropogenic activity and those reared in areas
with very low anthropogenic activity. Specifically, we examined neophobia, defined as the
tendency to avoid or fear unfamiliar things (Barnett 1958), exploration, defined as the number of
different ways an individual interacts with an object (Sroges and Glickman 1966), and boldness,
defined as an individual’s tendency to take risks (Réale et al. 2007). We predicted that, if juvenile
hyenas living in areas with high human disturbance behaved like “urbanized” birds and small
11

mammals (Miranda et al. 2013, Sol et al. 2013), they would be less neophobic, more exploratory,
and bolder than juveniles in low disturbance areas. In addition to human disturbance, we
explored the effects of sex and social rank on juvenile personality traits because these variables
begin to shape hyena behavior early in life (Smale et al. 1995, Dloniak et al. 2006, Holekamp et
al. 2013). We also explored the effects of age on expression of these personality traits because,
even though all of our subjects were juveniles, neophobia, exploration, and boldness change in
some species as individuals approach reproductive maturity (Kendal et al. 2005, Biondi et al.
2010). Finally, we tested for consistency in behavior across time and context, and inquired
whether any of these personality traits predicted survival to reproductive maturity.

METHODS
Subjects & Study Populations
Study subjects were 72 juvenile spotted hyenas inhabiting two protected areas in the MaraSerengeti ecosystem in southwestern Kenya. Fifty-eight of these juveniles were subjects in our
tests of neophobia and exploration and 60 of them were subjects in our test of boldness; 45
participated in both tests. Roughly half of the subjects (59% for tests of neophobia and
exploration; 50% for test of boldness) lived in three clans whose territories were located in The
Mara Conservancy, a pristine area managed by a private non-profit organization that strictly
prohibits cattle grazing and human presence except in tour vehicles. We will therefore refer to
this area as “low disturbance.” The remaining subjects were from a clan that has been
continuously monitored since 1988, and lives just inside the border of the Masai Mara National
Reserve (henceforth, “the Reserve”). We refer to this area as “high disturbance” because, since
12

the late 1990s, there has been exponential human population growth along the border of the
Reserve (Watts and Holekamp 2009), humans are active with their livestock both day and night
inside the Reserve in this area, and direct conflict between hyenas and livestock is common both
inside and outside the Reserve (Kolowski and Holekamp 2006). Since 2005, humans have been
responsible for the majority of hyena deaths (through spearing, snaring, and poisoning) for
which mortality sources can be determined (Holekamp and Dloniak 2010). Through historical
analyses and comparative studies with populations living in more pristine areas, numerous
behavioral effects of increased human activity have been documented in this population,
including increased nocturnality, increased daily travel, lower rates of den attendance by
mothers, active avoidance of livestock and herders, and a preference for areas with dense
vegetative cover (Boydston et al. 2003b, Kolowski et al. 2007, Kolowski and Holekamp 2009).
Our subjects ranged in age from 50-463 days old (mean = 171 days, median = 153 days).
Spotted hyenas do not reach reproductive maturity until after ~720 days of age, nor do they
reach full morphological maturity before 1000 days of age. All subjects could be individually
identified based on unique spot patterns and ear damage. Age was initially estimated when cubs
were first observed, based on their appearance and size (Holekamp et al. 1996), and sex was
determined based on the shape of the glans of the erect phallus (Frank et al. 1990). However,
two subjects died before we could determine their sex and were excluded from our models.
Juveniles were assigned social ranks based on our observations of their mother’s position
in the clan’s dominance hierarchy. Observations of each adult female’s aggressive and
submissive behaviors during dyadic agonistic interactions were used to construct the hierarchy,
as described previously (Martin and Bateson 1988, Smale et al. 1993). Juveniles “inherit”
13

dominance ranks immediately below those of their mothers in a process of social learning that is
not complete until at least 18 months of age (Holekamp and Smale 1993, Smale et al. 1993).
Hyenas of both sexes retain their maternal ranks as long as they reside in the natal clan, which
females do throughout their lives, but most males disperse to new clans at 2-5 years of age
(Frank 1986a, Holekamp and Smale 1998a).
We conducted our tests on groups of individuals at communal dens, where juvenile
hyenas live together and are rarely found alone. Spotted hyenas give birth to litters containing 1
or 2 cubs at isolated natal dens. However, when cubs are 3-4 weeks of age, mothers move them
to a communal den, where they live for the next 7-12 months. The communal den is a complex
of underground tunnels and chambers that cannot be entered by adults. However, juveniles
emerge from the warmth and safety of the den daily to nurse from their mothers and socialize
near the den with clan-mates (Holekamp and Smale 1998a, Holekamp and Dloniak 2010).
Mothers select protective den sites (Boydston et al. 2006) such that hyenas in neither high nor
low disturbance areas, experience direct exposure to humans or anthropogenic activity at their
dens aside from occasional visitation by tour vehicles, to which subjects in both high and low
disturbance areas are well habituated.

Experimental Stimuli
We used a novel object test to measure neophobia and exploration and a baited box test to measure
boldness. In the novel object test, we used one of four objects (Figure 2.1). Each object was
approximately 0.3m tall when deployed. Novel objects were chosen that juveniles would be
highly unlikely to encounter otherwise, even if they had begun to spend time away from the
14

	
	
	
	
Figure 2.1. The four objects used for the novel object test to measure neophobia and exploration.
Each object was approximately 0.3m tall when deployed. The cooler, funnel, and stool were
plastic, and the bucket was metal. Novel objects were chosen that juveniles would be highly
unlikely to encounter otherwise.

communal den and even if they lived in areas with human activity. Because test sessions often
involved multiple individuals, we varied the object being used to ensure that it was novel to the
individuals being tested.
In the baited box test, we used a metal mesh box (70cm long x 40cm wide x 44cm high)
with one open side large enough for cubs to enter (Figure 2.2). The box had no bottom surface
except for a metal tray (20cm X 40cm) at the back end of the box where we could sprinkle a
layer of powdered milk (15ml), which young hyenas prefer over other food rewards. Cubs could
only reach the powdered milk to feed if they fully entered the box. Prior to baited box test
sessions, we familiarized individuals with powdered milk by liberally distributing it around each
active communal den on three separate occasions when no hyenas were present above ground.
Although we did not systematically observe and record whether each subject consumed
powdered milk prior to testing, we assumed that any juveniles actively using the den would be
exposed to the scent and taste of milk on at least one of these three occasions.

15

	

Figure 2.2. The metal box used for the baited box test to measure boldness (70cm long x 40cm
wide x 44cm high). The box had no bottom surface except for a metal tray (20cm X 40cm) at
the back end of the box where we could sprinkle a layer of powdered milk (15ml). The box was
open on one side, allowing a subject to walk in and fully enter the box in order to access the
powdered milk at the back.
Ethical Note
All subjects were free-living, and were thus free to choose whether or not to interact with the
stimuli we presented. The study was conducted in compliance with Kenyan law and guidelines
for work with mammals provided by the American Society of Mammalogists (Sikes et al. 2011) .
The work was conducted under IACUC approval #05/14-087-00 from Michigan State
University and Research Clearance #NACOSTI/P/14/ 2154/1323 from the Kenyan
Commission on Science, Technology and Innovation.

Data Collection
Experiments were conducted opportunistically during our two daily observation periods, in the

16

morning (0600-1000) and evening (1700-2000). Stimuli were deployed from our research
vehicles, to which all subjects were well habituated. In both the novel object test and the baited
box test, we placed the stimulus approximately 20m away from the den entrance. The stimulus
was deployed either when no individuals were present above ground, or by positioning the
research vehicle to block the view of any individuals present above ground. In the baited box
test, the open side of the box was oriented towards the den entrance. All trials were videotaped
from approximately 20m away from the stimulus, using a Sony HDD Handycam Super Steady
Shot HDR-SR11 mounted on a tripod affixed to our research vehicle. Trials were terminated
when no new individuals entered within a 10m radius of the stimulus for 5 minutes, or due to
other logistical constraints (e.g., darkness or to prevent an individual from destroying an object).

Data extraction
Behaviors were coded from the video footage using JWatcher 1.0 (Blumstein & Daniel, 2007).
In each test session, any individual approaching within 10m of the stimulus was considered a
subject, and we coded each subject’s behavior for the entire time it was within 10m of the
stimulus. This cut-off distance allowed us to obtain high-quality footage of subjects’ interactions
with the stimuli.
In the novel object test, we operationally defined neophobia as the number of minutes it
took a subject to contact the object after it had come within 5m of the object (i.e. latency), as has
been done in previous work assessing spotted hyenas’ approaches to novel stimuli (BensonAmram and Holekamp 2012). If a subject never contacted the object, its latency was calculated
as the total time the subject was within 5m of the object during the trial.
17

In the novel object test, we defined exploration as a subject’s tendency to interact with
the novel object in multiple different ways. We assigned each subject an “exploration score,”
ranging from 0-4, based on how many exploratory behaviors it exhibited. Subjects received a
score of “0” if they came within 10m of the object but never contacted it. Subjects received a
score of “1” if they contacted the object with their snout. Subjects received a score of “2” if
they bit or licked the object in addition to contacting it with their snout. Subjects received a
score of “3” if they displayed these two behaviors and, in addition, either pawed the object or
picked it up. The majority of the subjects who received a score of “3” (seven of nine subjects)
picked up the object but did not use their paws; nevertheless, we gave both behavioral
combinations a score of “3.” Subjects with a score of “4” displayed all four of these behaviors.
In the baited box test, we defined boldness as a subject’s willingness to enter the box to
feed on milk. We assigned subjects a “boldness score,” ranging from 0-3. Subjects received a
score of “0” if they came within 10m of the box but never contacted or entered it. Subjects
received a score of “1” if they contacted the box but did not go inside. Subjects received a score
of “2” if they entered the box partially, or entered the box fully but never attempted to feed.
Subjects received a score of “3” if they entered the box completely and fed on the milk.

Statistical Analysis
Independent Variables
We first ran a Cox proportional hazard model to inquire whether there were differences among
the three clans in the low disturbance area with respect to neophobia, and Kruskal-Wallis Rank
Sum Tests to determine whether there were any significant differences in either boldness or
18

exploration among the three clans in the low disturbance area. We then inquired whether
neophobia, exploration and boldness were predicted by whether the subject lived in a high or
low disturbance area, as well as by its age, sex and maternal rank. Age in days was included as a
continuous variable. We assigned each subject the current social rank of its mother within the
clan’s dominance hierarchy on the day of testing. Ranks were standardized from -1 to 1, with the
most dominant individual having a standardized rank of 1. None of our results changed if we
calculated a subject’s social rank relative only to the other individuals present in its specific test
session. We also included time of day at testing (morning or evening) as a predictor variable.
We conducted the novel object and baited box tests on groups of individuals because
juvenile hyenas are almost never found alone at the communal den, and therefore, we examined
effects of several variables related to the social context of the testing situation. First, in our
models for neophobia, exploration and boldness, we included as a variable the total number of
subjects present in order to determine whether the presence of other hyenas facilitated or
interfered with an individual’s interaction with the stimulus. In a small percentage of test
sessions (4 of 18 novel object sessions and 2 of 17 box test sessions), in addition to test subjects,
an adult was present and within 10m of the stimulus at some point during the session. Although
we did not analyze the behavior of these individuals as subjects, we included them when
assessing the social context of the test session by increasing the number of subjects present to
account for them. Second, we included as a binary variable whether or not the subject had a
litter-mate who was also a subject during the test because, in other species, litter-mates may be
less fearful or inhibited when together than when apart (Stöwe et al. 2006b, Bergmüller and
Taborsky 2010, Hudson et al. 2011).
19

In our models for neophobia and boldness, we additionally included whether or not the
subject had seen another individual contact the novel object or baited box before doing so itself
or, if the subject did not contact the stimulus, whether it observed another subject contact the
stimulus earlier during that test session. We did not include this as a variable in our model for
exploration because we assumed that seeing another individual contact the novel object would
be likelier to influence whether or not the subject contacted the object than to affect the
diversity of behaviors emitted during interactions with the object, which was our dependent
variable. Furthermore, because our measures of neophobia and exploration were assessed from
the same novel object test, we chose to examine the effect of seeing another individual contact
the novel object in our analysis of neophobia, where the main behavior of interest was a
subject’s willingness to approach and contact the object. However, our results did not change if
we included this variable in our model for exploration.

Models of variation in neophobia, exploration and boldness
In statistical analysis of neophobia, we modeled variation in latency to contact the novel object
using a Cox proportional hazards model in the “survival” package in R. Our full model included
the following predictor variables: age (in days), disturbance (high or low), social rank, sex, time
of day (morning or evening), number of subjects present in the session, presence of a litter-mate,
and whether the subject observed another individual contact the object. A Cox proportional
hazards model, originally developed for survival analyses, provides a method to analyze right
censored “time to event” data. This approach allowed us to analyze the responses of those
subjects that never touched the object together with those that did, instead of performing two
20

separate analyses or assigning “time out scores” to subjects that never contacted the object
(Budaev 1997, Jahn-Eimermacher et al. 2011). Right-censored latencies for the subjects that
never contacted the novel object were specified in the model. This model assumes no underlying
distribution of the latency measure, but does assume that effects of predictor variables on the
latency to respond are constant and additive. These assumptions were satisfied in our dataset.
In statistical analysis of exploration and boldness, we treated subjects’ scores as ordered
categorical variables and used the “ordinal” package in R to perform cumulative link models.
These models, also known as proportional odds models or ordered regression models, assume
intrinsic ordering in the levels of the response measure (in our case, from least bold or
exploratory, to most bold or exploratory). Our full model for exploration included the following
predictor variables: age (in days), disturbance (high or low), social rank, sex, time of day
(morning or evening), number of subjects in the session, and presence of a litter-mate. Our
model for boldness included these same predictor variables plus the binary variable of whether
or not the subject observed another individual contact the box. In both models, we also
included each subject’s total trial time as a covariate to control for variation among subjects in
trial duration. We calculated the odds ratio for each predictor variable from the output of the
models. Preliminary analyses indicated no effect of the object type (Figure 1) on neophobia or
exploration, so we did not include it as a factor in our analyses. All analyses were conducted in
R (R version 3.3.1; Development Core Team, 2015). Parameter estimates were considered
significant when P≤ 0.051.

21

Consistency across time and context
Due to the opportunistic nature of conducting these experiments, a subset of individuals
participated twice in either the novel object test (N=14) or the baited box test (N=14). To
examine individual consistency over time in neophobia, we looked exclusively at individuals who
had participated in two novel object tests, each with a different object (N=14). We did not
analyze data for subjects who participated in two novel object tests with the same object, as the
stimulus was no longer novel to the subject in its second test. Because we had data points that
were right-censored when an individual never contacted the object, we used an extension of the
Cox proportional hazards model from our analysis of initial trials, described above. We
incorporated a random effect of subject ID into a Cox proportional hazards model using a
shared gamma frailty model (Wienke 2011). We also included exposure number (first or second)
as a fixed effect, and any variables that had significant parameter estimates in our analysis of first
trial responses. We then used Akaike’s Information Criterion corrected for small sample sizes
(AICc) to compare the fit of this frailty model to the fit of a Cox proportional hazards model
that did not include subject identity. A smaller AICc value indicated a better fit. We also
calculated a parameter estimate for the effect of subject identity using a likelihood ratio test.
To examine individual consistency over time in exploration and boldness, we again
analyzed responses for only the 14 subjects that participated twice in a test. We used a repeated
measures ANOVA, with the subject’s exploration or boldness score as the response measure.
We included as a predictor variable whether the trial was a subject’s first or second exposure, as
well as the predictor variables that had significant parameter estimates in our analysis of first trial
responses. To estimate repeatability, defined as the proportion of total variance accounted for by
22

the subject’s identity (Lessells and Boag 1987), we calculated intra-class correlation coefficients
(ICC), and associated confidence intervals according to Wolak, Fairbairn, & Paulsen (2011)
using the “ICC” package in R.
We examined consistency in neophobia across contexts for subjects who were tested in
both a novel object test and a box test (N=45). To do this, we treated the box as a fifth novel
stimulus and inquired whether there was a significant correlation between a subject’s latency to
approach a novel object and its latency to approach the box baited with food. We calculated
each subject’s latency to contact or enter the box, whichever came first, after coming within 5m
of it. We then fit a shared gamma frailty model, similar to the one described above, modeling
subject as a random effect, and including as fixed effects the type of test (novel object or box),
exposure number (first or second), and variables that had significant parameter estimates in our
analysis of neophobia in first trial responses. Using AICc, we compared this model to a model
that did not incorporate subject identity, and calculated a parameter estimate for subject identity
using a likelihood ratio test.

Relationship between personality traits and survival
Using our demographic records, we determined whether each subject survived to reproductive
maturity (2 years of age); all subjects could potentially have reached this age by the time this
analysis was conducted. We then used three different generalized linear models (GLM) to
determine whether performance in each of the temperament tests predicted the log odds of a
subject surviving to reproductive maturity. We used the “glm” package in R and specified a
binomial distribution. In our model for neophobia, latencies to contact the novel object were
23

included as a predictor variable and were right-censored for subjects that never contacted the
object. In our analyses testing whether exploration or boldness predicted survival to maturity,
exploration scores or boldness scores were included as predictor variables in the GLM model.
In each of the three models, we also included maternal rank, sex, and level of disturbance (high
or low) as predictor variables. We also included the age at testing as a covariate in each model,
but it did not significantly correlate with survival, so we report the results of our models without
controlling for age at testing. To further test whether the relationship between each of the three
personality traits and survival was mediated by the degree of human disturbance, or the subject’s
sex, or social rank, we used AICc to compare the fit of our full models to models including
interaction terms between each of these factors and the score on the personality test (latency to
contact the object for neophobia, exploration score or boldness score).

RESULTS
Preliminary analysis indicated that subjects did not differ among the three clans in the lowdisturbance area in their neophobia (Wald Test=2.68, df=3, P=0.44), exploration (H(2)=4.75, P
= 0.09), or boldness (H(2)=3.77, P = 0.15). In subsequent analysis, we therefore treated all
individuals from clans in low-disturbance areas together as we compared them to juveniles in the
clan living in the high disturbance area.

Neophobia
Fifty-eight juveniles participated in the novel object test across 18 test sessions, and 69% (40 of
58) of these individuals contacted the object. The Cox proportional hazards model was
24

significant (Wald Test=26.2, df=8, N=58, P<0.001), indicating that subjects in the high
disturbance area were significantly less neophobic than subjects in low disturbance areas
(P=0.051; Table 1; Figure 2.3a). The model also indicated that high-ranking juveniles were less
neophobic than low-ranking juveniles (P=0.002; Figure 2.3b), and that juveniles with a littermate present were less neophobic than other juveniles (P=0.032). Subjects were also less
neophobic in evening than in morning sessions (P=0.011), and in sessions where more hyenas
were present (P=0.043). Surprisingly, subjects were significantly more neophobic if they saw
another individual contact the object earlier in the session (P=0.036).

Exploration
Exploration scores in the novel object test varied from 0-4 (N=58). Eighteen subjects received
a score of 0 (31%), 13 subjects received a score of 1 (22%), 7 subjects received a score of 2
(12%), 9 subjects received a score of 3 (16%), and 11 subjects received a score of 4 (19%)

Age
Disturbance
Social Rank
Sex
Time of day
# Subjects in session
Litter-mate present
Observe another
subject make contact
	

Hazard ratio
(exp(beta))
1.00
1.96
2.78
1.21
2.96
1.16
2.36
0.39

Coefficient

Std. Error

z

0.0002
0.67
1.02
0.19
1.08
0.15
0.86
-0.95

0.002
0.35
0.33
0.36
0.43
0.08
0.40
0.45

0.12
1.95
3.07
0.54
2.53
2.02
2.15
-2.10

P
0.936
0.051*
0.002*
0.597
0.011*
0.043*
0.032*
0.036*

Table 2.1. Results of a Cox proportional hazards model for variation in latency to contact the
novel object (N=58). Asterisks indicate P ≤ 0.051.

25

Figure 2.3. A Cox proportional hazards model indicated that: a) juveniles in high disturbance
were significantly less neophobic than subjects in low disturbance (P=0.05). b) social rank was
inversely related to latency to contact the object (P=0.002). Social rank was analyzed statistically
as a continuous variable but is shown here by dividing the clan’s hierarchy into three dominance
classes (the high rank category includes ranks 1.00 to 0.33, the middle rank category includes
ranks 0.33 to -0.33, and the low rank category includes ranks -0.33 to -1.00).
26

(Figure 2.4). The logistic regression model indicated significant effects of disturbance and rank
on exploration behavior (Table 2.2). Specifically, individuals in the high disturbance area were
significantly more exploratory than individuals in the low disturbance area (P=0.003; Figure
2.4a), and high-ranking individuals were more exploratory than low-ranking individuals
(P=0.020; Figure 2.4b). There was a significant positive association between exploration and
trial duration (P=0.0005), and subjects were significantly more exploratory in the evenings than
mornings (P=0.006).

Boldness
Sixty juveniles participated in the boldness box test, and boldness scores varied from 0-3.
Nineteen subjects received a score of 0 (31%), 10 subjects received a score of 1 (17%), 10
subjects received a score of 2 (17%), and 21 subjects received a score of 3 (35%). The logistic
regression model indicated that subjects in low disturbance areas were significantly more bold

Trial duration
Age
Disturbance
Social Rank
Sex
Time of day
# Subjects in session
Litter-mate present
	

Odds Ratio
1.22
1.00
6.37
2.93
1.87
5.62
1.21
1.70

Coefficient Std. Error
0.20
0.06
0.001
0.003
1.85
0.63
1.08
0.46
0.62
0.58
1.72
0.63
0.19
0.13
0.53
0.60

z
3.49
0.36
2.95
2.32
1.08
2.73
1.52
0.89

P
0.0005*
0.718
0.003*
0.020*
0.282
0.006*
0.129
0.375

Table 2.2. Results of a logistic regression model for variation in exploration of the novel object
(N=58). Asterisks indicate P ≤ 0.051.
27

Figure 2.4. A logistic regression model for variation in exploration score indicated that: a)
juveniles in high disturbance were significantly more exploratory than those in low disturbance
(P=0.003) and b) high-ranking juveniles were significantly more exploratory than low-ranking
juveniles (P=0.020; N=58).

28

than individuals in high disturbance areas (P=0.019; Table 2.3; Figure 2.5). There was a
significant positive association between boldness and trial duration (P=0.0006), and a trend for
females to be bolder than males (P=0.066), but we observed no effects of rank on boldness.

Consistency across time and context
For neophobia, our frailty model containing the random effect of subject ID was a significantly
better fit than the model without subject ID as a factor, indicating significant repeatability in
subjects’ neophobia across trials (2=10.99, P=0.006). Subjects’ latency to contact a novel object
did not change significantly across exposures (Hazard ratio = 0.94, P = 0.92). For exploration,
the repeated measures ANOVA similarly indicated that subjects’ exploratory behavior did not
change significantly across exposures, as indicated by a non-significant within-subjects parameter

Trial duration
Age
Disturbance
Social Rank
Sex
Time of day
# Subjects in session
Litter-mate present
Observe another
subject enter or
contact box
	

Odds Ratio
1.15
0.99
0.22
0.74
0.28
1.05
0.97
1.35
2.45

Coefficient
0.15
-0.01
-1.50
-0.35
-1.31
-0.02
-0.01
0.36
0.79

Std. Error
0.04
0.005
0.64
0.46
0.71
0.88
0.12
0.57
0.87

z
3.44
-1.40
-2.34
-0.77
-1.84
-0.03
-0.12
0.63
0.91

P
0.0006*
0.161
0.019*
0.440
0.066
0.980
0.910
0.529
0.365

Table 2.3. Results of a logistic regression model for variation in boldness (N=60). Asterisks
indicate P ≤ 0.051.

29

Figure 2.5. A logistic regression model for variation in boldness score indicated that juveniles in
high disturbance were less bold than juveniles in low disturbance (P=0.019; N=60).
estimate for exposure number (F1,11 = 0.258, P =0.62). However, the repeatability estimate for
exploration was also non-significant (r = 0.20, CI= -0.34, 0.64). For boldness, the repeated
measures ANOVA indicated that subjects’ boldness did not change significantly across repeated
trials, as indicated by a non-significant within-subjects parameter estimate for exposure number
(F1,12 = 0.222, P =0.424). The repeatability estimate for boldness was significant (r = 0.78, CI=
0.45, 0.92). Thus, juvenile spotted hyenas appear to show consistency over time in their
neophobia and boldness, but not in their degree of exploration.
In our assessment of consistency in neophobia in two different contexts, namely a
subject’s latency to contact the novel object and latency to contact or enter the baited box, our

30

frailty model with subject ID as a random effect was a significantly better fit than our model
without subject ID as a factor (2=28.1, P=0.002). This indicates significant individual
consistency in subjects’ neophobia when encountering novel stimuli that offer a food reward
and those that do not. Subjects had significantly shorter latencies to approach the stimulus on
the second test they received than on their first test (Hazard ratio = 2.33, P= 0.001) regardless
of whether it was the novel object test or baited box test (Hazard ratio = 1.20, P= 0.513).

Relationship between personality traits and survival
Of the 58 subjects in the novel object test, 39 (67%) survived to reproductive maturity. Latency
to contact the object (B=1.20, P=0.22) did not predict survival to maturity, but juveniles in low
disturbance were more likely to survive than juveniles in high disturbance (B=0.18, P=0.018),
and juveniles of high rank were more likely to survive than juveniles of low rank (B=4.59,
P=0.02). Sex did not predict survival (B=0.61, P=0.49). Exploration score also failed to predict
survival to reproductive maturity (B=0.91, P=0.70), but again subjects in low disturbance were
significantly more likely to survive than subjects in high disturbance (B=0.18, P=0.02), and
survival was significantly positively correlated with social rank (B=4.34, P=0.015). Sex did not
predict survival to reproductive maturity (B=0.65, P=0.54). Models that included interactions
between the temperament traits and the other predictor variables did not fit the data significantly
better than our model without interactions.
Of the 60 subjects tested in the baited box test, 40 (67%) survived to reproductive
maturity. Boldness affected survival similarly in both disturbed and undisturbed areas. Subjects
who were less bold were significantly more likely to survive (B=0.49, P=0.05), and a model that
31

included boldness was a significantly better fit than one that did not (likelihood ratio test,
=-4.51, P=0.034). In this sample, individuals in low disturbance were again significantly more

2

likely to survive than individuals in high disturbance (B=0.075, P=0.004). Higher ranking
individuals were also more likely to survive than lower ranking individuals (B=3.91, P=0.03), but
sex did not predict survival to reproductive maturity (B=0.47, P=0.36). Models that included
interactions between boldness and the other predictor variables failed to fit the data significantly
better than our model without interactions.

DISCUSSION
Our findings are consistent with the growing literature demonstrating the profound influence
that human disturbance can have on personality traits in wild populations, and permit
comparisons with other taxa that have been more thoroughly studied. We found significant
effects of human disturbance on all three of the juvenile personality traits we measured in
spotted hyenas. Patterns in neophobia and exploration were similar to those documented in
comparisons of urban and rural individuals among birds and small mammals (Smith and
Blumstein 2008, Miranda et al. 2013, Sol et al. 2013); juvenile hyenas in low disturbance areas
were significantly more neophobic and less exploratory than juveniles in high disturbance areas.
However, in contrast to these other taxa, juvenile spotted hyenas in low disturbance areas were
significantly more bold than those in high disturbance areas.
Our results also demonstrate that personality traits, although often challenging to
measure in lone individuals in the field, can be effectively assessed in a group context.
Conducting personality tests in group contexts rather than in isolation enhances ecological
32

validity for individuals who are always in a social group (Drea 2006), and better indicates how
personality traits are actually expressed in free-living animals than testing subjects alone
(Webster and Ward 2010). We were able to account for the effects of social context by
measuring situational variables and including them in our statistical models. Neophobia and
boldness, but not exploration, showed significant repeatability within the juvenile period. We
also found evidence for consistency in neophobia across contexts. Among juveniles that
participated in both the novel object and baited box tests, latencies to contact the stimuli were
highly consistent.
In addition to human disturbance, maternal rank emerged as an important factor shaping
juvenile personality traits in spotted hyenas. The relationship between social rank and
personality varies among species, and our finding that higher-ranking juveniles were less
neophobic and more exploratory than low ranking juveniles is consistent with some previous
studies (Mettler and Shivik 2007, David et al. 2011), but not others (Gómez-Laplaza 2002, Fox
et al. 2009, An et al. 2011). That there was still a significant effect of rank when we calculated
subjects’ ranks relative only to those of the other subjects present in a specific test session
suggests that the influence of rank was not dependent on the immediate social context of the
test; low-ranking individuals were more neophobic and less exploratory even when they
happened to be among the highest ranking subjects present in a particular test session. Maternal
rank, therefore, appears to shape personality traits from a young age, as it does offspring growth,
aggression, and play(Hofer and East 2003, Dloniak et al. 2006, Holekamp et al. 2013, Lewin et
al. 2017).
It is perhaps surprising that we found no relationship between maternal rank and
33

boldness, as high-ranking individuals could potentially have dominated access to the powdered
milk in the test box by attacking lower ranking subjects. Yoshida et al. (2016) found that rank
and boldness, as measured by individuals’ closest approach to lions (Panthera leo) during lionhyena interactions, were significantly positively related in females but not males. Our results
might be due to the fact that juveniles are still in the process of learning which conspecifics they
can appropriately dominate or displace from food. Additionally, although a low-ranking juvenile
subject may have had to wait for a higher-ranking subject to enter the box and feed first, when
the higher-ranking animal left the box, the scent of milk likely still attracted lower- ranking
subjects, which could still scrounge for remaining milk and thus receive the highest boldness
score. It seems unlikely that juveniles in low disturbance areas are bolder than those in high
disturbance areas because they are more hungry, as has been found in other studies (Biro and
Booth 2009, Chapman et al. 2010). If hunger were primarily driving the expression of boldness
in our test, we would expect lower-ranking individuals to be bolder than higher-ranking
individuals, who are generally better fed. We would also expect individuals to be less bold in
trials conducted in the morning than in the evening, when juveniles typically have not nursed
since the morning, so they should be more highly motivated to reach the milk in our test
apparatus. However, neither time of day nor rank significantly predicted boldness.
Neophobia was the only trait significantly affected by the presence and behavior of
conspecifics. Subjects were less neophobic when more subjects were present in the test session,
which is consistent with the behavior of hyenas solving a puzzle box problem (Benson-Amram
and Holekamp 2012) , the important role of social facilitation in hyena social behavior
(Glickman et al. 1997), and results of novel object tests administered to other species (Cadieu et
34

al. 1995, Moscovice and Snowdon 2006, Webster et al. 2007, Moretti et al. 2015). The presence
of a hyena litter-mate, in particular, significantly facilitated approach to the novel object,
suggesting that litter-mates may have especially strong effects on juvenile hyenas’ confidence and
propensity to explore new things.
Interestingly, individuals were more neophobic if they saw another individual contact the
novel object, which is consistent with findings that a conspecific’s presence or behavior can
inhibit approach to novel objects (Ryer and Olla 1991, Brown and Laland 2002, Stöwe et al.
2006a). In our test, this inhibitory effect may have been due to the tendency of juvenile hyenas
to startle or become skittish when others do. Juveniles are strongly influenced by the behavior
of their peers and it is common to see many cubs run into the communal den after a single cub
startles. In our test, subjects often backed away or startled after contacting the novel object for
the first time, so it seems possible that viewing this may have either inhibited other subjects
from approaching at all or slowed their approach.
Notably, our study is one of only a few to directly relate personality traits to fitness in a
wild mammal (Archard and Braithwaite 2010). We found that boldness, but not neophobia or
exploration, predicted survival to adulthood, with less bold individuals showing enhanced
survivorship. These findings are consistent with a meta-analysis by Smith and Blumstein (2008),
demonstrating that boldness consistently shows a significant negative correlation with survival
across species. Furthermore, we found a negative relationship between boldness and juvenile
survivorship regardless of disturbance, suggesting that the benefits of showing restraint in the
face of risk accrue regardless of the level of human activity.
Differences in juvenile survivorship and predation pressure in the high and low
35

disturbance areas may be shaping the differences we observed in juvenile boldness. When
looking at hyenas of all ages, Yoshida et al. (2016) found evidence for stabilizing selection with
regard to boldness during lion-hyena interactions, with reduced longevity for individuals who
were either highly prone to take risks or who completely avoided them. However, we found
that juveniles in high disturbance were significantly less bold than juveniles in low disturbance.
Although surprising, this is likely due to decreased predation by lions in the high disturbance
area, where there has been a significant decline in lion densities and where juvenile survivorship
was significantly greater than in the low disturbance area from 2009-2013 (Green 2015). This
suggests that a reduction in predation pressure may be relaxing the selection pressure against shy
individuals in the high disturbance area.
Several questions should be answered by future research to make our results of the most
use to applied carnivore conservation and the emerging field of conservation behavior (Greggor
et al. 2016). First, we must determine whether a juvenile’s performance in our personality
experiments predicts its behavior and success in disturbed landscapes, particularly its likelihood
of leaving protected areas, attacking livestock, or being killed by humans. Second, to determine
whether juvenile personality traits remain stable into adulthood, we could measure these
personality traits in individuals several times across the lifespan. We cannot assume stability in
these traits beyond the juvenile period, as juvenile personality traits may represent adaptations
specific to immature life-stages (Bell and Stamps 2004, Sinn et al. 2008, Stamps and Groothuis
2010, Petelle et al. 2013).
Third, given our findings regarding boldness, disturbance, and juvenile survivorship, it
would be interesting to explore whether hyena personality traits have fitness trade-offs over the
36

lifetime of an individual, and whether such trade-offs differ in areas of high and low disturbance.
In their meta-analysis, Smith and Blumstein (2008) found that, although bold individuals showed
significantly lower survival, they also had higher reproductive success than less bold individuals.
To test for such a trade-off in spotted hyenas, we could relate boldness in the juveniles tested
here to their future reproductive success. Furthermore, this would allow us to explore whether
the fitness costs and benefits of boldness differ in areas of high and low disturbance, and
whether disturbance may be acting as a selective pressure on the life-history strategies of
individuals inhabiting these areas.
Overall, our study not only reveals significant behavioral differences between animals in
disturbed and pristine areas, but also raises important questions regarding how animals achieve
behavioral adjustments in response to anthropogenic disturbance. Juvenile spotted hyenas in
both high and low disturbance areas are remarkably well-buffered from direct human
disturbance at their communal dens. Even juveniles living in areas that are generally disturbed by
people, are not directly bothered by people while they live at the den, and in fact, they
experience virtually no direct exposure at their dens to any anthropogenic activity, including
livestock grazing. Therefore, it seems unlikely that human disturbance is directly shaping the
behavior of juveniles residing at the communal den. Nevertheless, the juveniles tested here
showed marked disturbance-related behavioral variation very early in life. We are currently
investigating the hypothesis that, such behavioral differences are shaped indirectly, via maternal
effects, as the behavior and stress physiology of adult female spotted hyenas can be greatly
affected by human disturbance (Boydston et al. 2003b, Kolowski and Holekamp 2009, Van
Meter et al. 2009). In many species, variation in maternal care or hormone exposure in utero
37

shapes offspring personality, providing a potentially adaptive signal of environmental quality to
offspring (Maestripieri and Mateo 2009, Dantzer et al. 2013). In species that are long-lived and
slow-growing, like spotted hyenas, it may be that responding early in life to strong maternal
signals of disturbance is critical to persistence in changing environments.

ACKNOWLEDGEMENTS
We thank the Kenyan National Council for Science and Technology, the Narok County Council,
the Kenya Wildlife Service, and the Mara Conservancy for permission to conduct field research
with hyenas. We thank Wes Binder, David Green, Lily Johnson-Ulrich, Wilson Kilong, Charles
Kolodziejski, Kevin McCormick, Jenna Parker, and Benson Pion for assistance in the field. We
also thank Andrew Dennhardt of MSU’s Center for Statistical Training and Consulting for
guidance with statistical analysis and Lily Johnson-Ulrich for consultation about video coding.
Finally, we thank Tom Getty, Joe Lonstein, Laura Smale, and two anonymous reviewers for
helpful comments on earlier drafts of the manuscript. This work was supported by NSF Grants
OISE 1556407 and DEB 1353110 to KEH, and by a NSF Graduate Research Fellowship to
JRG. JRG was also supported by fellowships from the College of Natural Sciences and the
program in Ecology, Evolutionary Biology, and Behavior at Michigan State University. This
work was made possible by support from the BEACON Center for the Study of Evolution in
Action, funded by NSF grant OIA 0939454.

38

CHAPTER 3
SEX- AND AGE-SPECIFIC CORRELATES OF FECAL GLUCOCORTICOID
CONCENTRATIONS IN SPOTTED HYENAS
Julia R. Greenberg, Stephanie M. Dloniak, Page E. Van Meter, Jeffrey A. French, Jacinta C.
Beehner, & Kay E. Holekamp
INTRODUCTION
Stress can have dramatic effects on an animal’s morphology, physiological functioning, and
behavior, and stress experienced early in life often has particularly profound effects on the
animal’s adult phenotype and fitness (Lindström 1999, Tung et al. 2016). Glucocortioids (GCs;
e.g. cortisol, corticosterone) are metabolic hormones that serve multiple physiological functions.
Circulating GCs support metabolic processes, mediate glucose availability, and maintain body
mass (Hau et al. 2016). GCs are often referred to as “stress hormones” due to their role in
responses mediated by the hypothalamic-pituitary-adrenal (HPA) axis to environmental or
physiological perturbations. GCs released in response to a stressor stimulate gluconeogenesis,
inhibit energy storage, and promote cardiovascular activity (Sapolsky et al. 2000). However,
continual exposure to an acute stressor can result in chronic activation of the HPA axis,
dysregution of the negative feedback mechanisms that minimize long-term exposure to GCs
(McEwen and Wingfield 2003), and negative effects on the animal’s immune function (Hoffman
et al. 2011), cardiovascular system (Sapolsky and Share 1994), and reproduction (Sheriff et al.
2009).
Exposure to early stress can shape a variety of phenotypic traits that may last into
adulthood. For example, individual humans who are exposed to chronic early stressors, such as
39

low-socioeconomic status or abuse, are likelier than unexposed individuals to suffer from
chronic disease as adults (Barker et al. 2002, O’Rand and Hamil-Luker 2005, Kittleson et al.
2006, Gluckman et al. 2008). Early in life, glucocorticoids also shape the development of
neuroendocrine pathways associated with emotion regulation and personality. In laboratory
rodents and primates, individuals who experience an early stressful environment are often less
neophobic, more exploratory, and less risk-averse (i.e. “bold”) than those who do not (Caldji et
al. 1998, Carere et al. 2003, Parker and Buckmaster 2004, Martins et al. 2007, Parker et al. 2007,
Guenther et al. 2014). Finally, due to their role in resource allocation, early exposure to elevated
glucocorticoids can affect offspring growth (Dantzer et al. 2013), and life-history trade-offs
(McGlothlin and Ketterson 2008, Crespi et al. 2012, Sheriff and Love 2013). For example,
individuals who experience early stressful environments sometimes show accelerated lifehistories, reaching menarche earlier than peers who do not experience early stress (Maestripieri
2005, Cameron et al. 2008a).
Maternal stress, either during pregnancy or lactation, can also shape phenotypic traits in
offspring. For example, in non-human primates, prenatal stress is associated with offspring
impairments in learning, attention, and neuromotor ability (Schneider et al. 1999), and in
humans, stress experienced prenatally has been identified as a risk factor for attention deficit and
anxiety disorders (Bergh et al. 2004, Huizink et al. 2004). During the post-natal period, ingestion
of GCs in a mother’s milk can also shape offspring phenotype. For example, in rhesus
macaques, mothers who produce high concentrations of GCs in their milk have sons who are
more cautious and nervous than those who produce milk containing lower GC concentrations
(Hinde et al. 2015). Maternal stress during lactation can also shape offspring traits indirectly by
40

affecting maternal behavior. For example, in several free-ranging primate species, GC
concentrations in lactating mothers are associated with their protectiveness of offspring (Bahr et
al. 1998, Nguyen et al. 2008, Maestripieri et al. 2009). These maternal behaviors, in turn, can
influence the pace at which offspring achieve independence, their exploratory tendencies, their
sociability, and even their own parenting behavior when they become adults (Fairbanks 1996,
Bardi and Huffman 2002, Maestripieri 2005).
Despite the well-documented role played by GCs in influencing behavior and physiology,
few studies of free-living animal populations have tested whether early stress is associated with
fitness. Although a variety of early-life factors have been found to correlate negatively with
fitness, including high population density, low resource availability, or sub-optimal maternal
condition (Descamps et al. 2008, Rodel et al. 2009, Hamel et al. 2010, Tung et al. 2016), few
studies have specifically associated particular environmental conditions with fGC concentrations
in juveniles themselves to confirm that such conditions are experienced as physiological
stressors. Furthermore, although early stress is often assumed to be associated with fitness costs,
it could actually be adaptive if early stress prepares offspring for their environment later in life
(Monaghan 2008, Beery and Francis 2011, Champagne 2011). For instance, individuals that
experience high early stress are often more reactive to environmental circumstances and exhibit
greater behavioral flexibility, traits that may be beneficial in an unpredictable environment
(Sheriff and Love 2013, Crino and Breuner 2015).
Human activity is an increasingly common stressor in the lives of wild animals, as
indicated by individuals’ concentrations of glucocorticoid stress hormone (reviewed in (Busch
and Hayward 2009, Dantzer et al. 2014). However, few studies of anthropogenic stress and its
41

consequences have focused on juveniles to inquire whether anthropogenic stressors experienced
early in life are associated with phenotypic traits later in life, or whether early stress has longterm fitness consequences. Here our primary goal was to determine whether anthropogenic
disturbance affects GC concentrations in free-living juvenile spotted hyenas (Crocuta crocuta) or
their mothers. We measured GC concentrations in feces (fGC), which provide an integrated
measure of GCs released in the body over approximately 24 hours (Dloniak 2004). We tested a
hypothesis suggesting that fGC concentrations, either in juveniles themselves or in their mothers
during pregnancy or lactation, differ between animals exposed to anthropogenic disturbance and
those living in undisturbed areas.
Previous research showed that human disturbance is associated with elevated fGC
concentrations in free-living adult spotted hyenas and that pastoralist activity in particular, rather
than tourism, is associated with elevated fGC concentrations. Van Meter et al. (Van Meter et al.
2009) compared fGC concentrations of adult spotted hyenas living in four territories that varied
with respect to lion densities, rainfall, tourist activity, and pastoralism. Adults that lived on the
border of the Masai Mara National Reserve in Kenya were exposed to the most pastoralist
activity and had the highest fGC concentrations. Furthermore, a twelve-year longitudinal
analysis of one clan exposed to human disturbance indicated that fGC concentrations among
adult males had significantly increased over time as the neighboring human population expanded
and livestock grazing increased (Van Meter et al. 2009). This finding is consistent with a recent
meta-analysis showing that males are particularly sensitive to the stressor of human disturbance
(Dantzer et al. 2014).
In the current study, we first used a longitudinal analysis to test whether a variety of
42

environmental and social factors, all examined previously in adult hyenas, correlate with fGC
concentrations among juveniles. We inquired whether fGC concentrations in juveniles increased
between 1993 and 2014, as human disturbance in our study area increased. We also inquired
whether juvenile fGC concentrations were associated with prey density or clan size, as resource
competition and population density have been found to correlate positively with fGC
concentrations in other species (Harper and Austad 2000, Foley et al. 2001, Ethan Pride 2005,
Beehner and McCann 2008, Dantzer et al. 2013). We also assessed the effects of a juvenile’s sex
and dominance relationship to its litter-mate, because these factors were found to correlate with
fGC concentrations in a Tanzanian population of spotted hyenas (Benhaiem et al. 2013). We
tested whether a juvenile’s social rank was associated with fGC concentrations. Across species,
the relationship between rank and fGC concentrations is variable and appears to be mediated by
environmental conditions, the stability of the social hierarchy, the breeding system, how
individuals maintain their rank, and whether subordinate individuals have opportunities for
social support (Creel 2001, Abbott et al. 2003, Goymann and Wingfield 2004).
After completing our longitudinal analysis to identify salient predictors of fGC
concentrations, we compared fGC concentrations between hyenas living in disturbed and
undisturbed areas of a single wildlife reserve from 2009-2014. Specifically, we used a crosssectional analysis to test the hypothesis that stress hormone concentrations, either in juveniles
themselves or in their mothers during pregnancy or lactation, differ between animals exposed to
anthropogenic disturbance and those living in pristine areas of the Mara-Serengeti ecosystem in
eastern Africa. This hypothesis predicts that juveniles in high-disturbance areas should have
significantly higher fGC concentrations than juveniles in low disturbance areas. If maternal
43

stress is affecting offspring phenotypes, then pregnant or lactating females in high-disturbance
areas should have significantly higher fGC concentrations than those in low disturbance areas.
Finally, we tested whether early life fGC concentrations, regardless of the causes of its variation,
predict survivorship or longevity in spotted hyenas.

METHODS
Study Site and Subjects
Study subjects were spotted hyenas inhabiting two protected areas in southwestern Kenya; one
area was undisturbed, whereas the other was highly disturbed by anthropogenic activity. Data
from the “low disturbance” area were collected from hyenas belonging to three clans inhabiting
the pristine Mara Conservancy since monitoring began there in 2008. The Mara Conservancy is
managed by a private non-profit organization that strictly prohibits cattle grazing or any human
activity except wildlife viewing from tour vehicles. Data collected from hyenas inhabiting a
highly-disturbed area were those taken from the Talek West clan, which lives along the northern
border of the Masai Mara National Reserve (henceforth, the Reserve), near the burgeoning town
of Talek. This clan has been monitored closely and continuously since 1988. Since the late
1990s, there has been exponential human population growth along the border of the Reserve
(Watts and Holekamp 2009), humans are active with their livestock both day and night inside
the Reserve in this area, and direct conflict between hyenas and livestock is common both inside
and outside the Reserve (Kolowski and Holekamp 2006). Since 2005, humans have been
responsible for the majority of hyena deaths (through spearing, snaring, and poisoning) for
which mortality sources could be determined (Holekamp and Dloniak 2010). Numerous
44

behavioral effects of increased human activity have been documented in this population,
including increased nocturnality, increased daily travel, lower rates of den attendance by
mothers, active avoidance of livestock, and a preference for areas with dense vegetative cover
(Boydston et al. 2003a, Kolowski et al. 2007, Kolowski and Holekamp 2009).
All subjects were individually identified based on their unique spot patterns and ear
damage. Age was initially estimated when cubs were first observed above ground, based on their
appearance and size (Holekamp et al. 1996), and sex was determined based on the shape of the
glans of the erect phallus (Frank et al. 1990). Spotted hyenas usually give birth to litters of 1 or 2
cubs. Shortly after the birth of twin litters, litter-mates fight for priority of access to the
mother’s teats for nursing and, within the first month of life, establish a consistent dominance
relationship that typically remains intact as long as both litter-mates remain in the natal clan.
Here each cub born in a twin litter could therefore be assigned an “intra-litter rank” based on
observations of aggressive and appeasement behavior between litter-mates (Smale et al. 1995,
Holekamp and Smale 1998a). Individuals were considered juveniles until they reached two years
of age. A female’s first parturition could be determined reliably because her pseudopenis tears
when she gives birth to her first litter, and leaves an unambiguous stripe of pink scar tissue on
the posterior surface of the phallus (Frank and Glickman 1994, Holekamp et al. 1996).
The reproductive state of adult females was continuously monitored to determine
periods of pregnancy and lactation, as described by Holekamp et al. (1996). The gestation
period for spotted hyenas is 110 days (Kruuk 1972). Conception dates were determined by
back-calculating from a cub’s date of birth, and based on observations of matings, or on sighting
a fresh tear in a phallus, which indicated parturition. To delineate lactation intervals for adult
45

females, we calculated weaning dates for juveniles based on observations of nursing conflicts
and observations of cubs subsequently seen with their mother when no nursing occurred
(Holekamp et al. 1996).
Spotted hyenas live in fission-fusion societies of 5-130 individuals (Holekamp et al. 2015)
that are structured according to a strict matrilineal hierarchy (Frank 1986b). The social rank of
adult females here was calculated on a yearly basis based on outcomes of dyadic agonistic
interactions (Martin and Bateson 1988, Smale et al. 1993). Juveniles “inherit” dominance ranks
immediately below those of their mothers in a process of social learning that is not complete
until at least 18 months of age (Holekamp and Smale 1993, Smale et al. 1993). We therefore
assigned each juvenile the social rank of its mother. Hyenas of both sexes retain their maternal
ranks as long as they reside in the natal clan, which females do throughout their lives, but most
males disperse to new clans at 2-5 years of age (Boydston et al 2005; Smale et al. 1997). Each of
the clans analyzed in the cross-sectional analysis experienced at least one period of instability in
the social hierarchy during the period in which they were compared (2009-2014), when the alpha
female died or was in the process of being overtaken by lower ranking females.

Fecal sample collection, extraction, and immunoassay of hormone concentrations
We measured excreted hormone concentrations by assaying fecal samples collected from
individuals in four different clans. Fecal samples were collected whenever a known hyena was
observed defecating during our routine morning (0600-1000h) and evening (1600-2000h)
observation sessions. Samples were stored in liquid nitrogen, and carried to Michigan State
University on dry ice for storage in -80°freezers. An ethanol extraction procedure was used that
46

has been described previously (Dloniak 2004, Dloniak et al. 2004, 2006, Van Meter et al. 2008,
2009). Samples were assayed in duplicate using a corticosterone radioimmunoassay kit
(ImmuChem Double Antibody Corticosterone 125I RIA kit, MP Biomedicals), and were diluted
to a 1:20 concentration with RIA buffer supplied with the commercial kit. The detection range
for the assay was 10-1200 ng/g. Any samples that had fGC concentrations below 10 ng/g were
assigned the lower detection limit as their value. An analytical validation was performed for use
of this assay with hyena fecal samples and the assay demonstrated accuracy, precision, and
linearity (Dloniak, unpublished; Greenberg et al. in prep). For our longitudinal analysis, the
intra-assay coefficient of variation was 4.66 ± 2.32 and the inter-assay coefficient of variation
was 10.27% (N=39 assays). For our cross-sectional analysis, the intra-assay coefficient of
variation was 4.28 ± 0.98 and the inter-assay coefficient of variation was 10.73% (N=22 assays).

Statistical Analysis
We modeled variation in fGC concentrations using generalized linear mixed models (GLMM)
with the “lmer” package in R, and used the package “multcomp” to perform Tukey’s post-hoc
tests for pairwise comparisons. fGC concentrations were log transformed to better approximate
a normal distribution. We used Variance Inflation Factors (VIF) to assess multi-colinearity of
predictor variables in all of our models. No variable had a VIF above 3, indicating that there
was no significant multi-colinearity among variables in our models. To assess potential outliers
or data points with high leverage, we calculated Cook’s Distances, used visual inspection of
graphs, and compared results with and without inclusion of these data points.

47

Longitudinal analysis of juvenile fGC concentrations 1993-2014
To determine which ecological and social variables predicted fGC concentrations in juveniles,
we analyzed samples collected in the Talek West clan from 1993-2014. Because we had 1-8
samples per individual hyena, we included subject ID as a random effect. To control for the
time of day at which the sample was collected, we included a binary variable, morning or
evening, to indicate during which observation session the sample was collected. Age (in
months), sex, and intra-litter rank were included as fixed effects. Clan size was included as a
fixed effect and calculated as the number of breeding females present in the clan during the year
in which the sample was collected. To determine whether fGC concentrations changed over the
course of the study, we included study year as a fixed effect. Following Van Meter et al. (2009),
prey density was calculated as the average density of prey counted within 100m of three 4-km
transect lines, run biweekly, during the month before the sample was collected.
To explore whether a juvenile’s rank mediated the effect of socio-ecological variables on
fGC concentrations, we tested our model containing no interaction terms against models that
contained two-way interactions between rank and each of our other variables. To compare the
fit of models, we used Akaike’s Information Criterion corrected for small sample sizes (AICc),
with the lowest AICc score indicating the best model fit. We used likelihood ratio tests to
obtain a parameter estimate for the best-fitting model.

Cross-sectional comparison of fGC concentrations 2009-2014
Our cross-sectional analysis inquired whether juveniles, pregnant females, or lactating females in
the highly disturbed Talek region showed significantly higher fGC concentrations than
48

individuals in the low disturbance area. We first conducted preliminary analysis of fGC
concentrations to determine which socio-ecological variables to include in our cross-sectional
analysis. Because previous analyses have explored the socio-ecological correlates of fGC
concentrations in the high-disturbance area (Van Meter et al. 2009), we tested whether the same
or different variables as in those earlier analyses were associated with fGC concentrations among
hyenas in the low disturbance area. In preliminary analysis of juvenile fGC concentrations in the
low disturbance area, we tested the same variables that we examined in our longitudinal analysis
of the Talek region; these were time of day, age, sex, clan size, prey, maternal rank, and litter
size.
In our preliminary analyses of pregnant and lactating females, we separately modeled
fGC concentrations among females in each of these two reproductive states, because in previous
studies, reproductive state was a significant predictor of fGC concentrations (Goymann et al.
2001, Van Meter et al. 2009). In our preliminary analysis of fGC concentrations among lactating
females in the low disturbance area, we replicated Van Meter et al.’s (2009) analysis by testing
whether the following variables significantly predicted fGC concentrations: social rank, prey
density, rainfall (mm), tourism (high or low), and social instability (occurring or not occurring).
Social rank and prey density were calculated as described above for analysis of juvenile samples.
Daily rainfall (mm) was measured using a plastic rain gauge, and we included as a variable in our
models the daily mean rainfall during the 30 days prior to the collection date of each fecal
sample. Following Van Meter et al (2009), we treated tourism as a binary variable; samples
collected June-October and December were considered “high tourism season,” and samples
collected in the remaining six months of each year were considered “low tourism season.”
49

Social instability was a binary fixed factor indicating whether the subject’s clan was experiencing
social upheaval at the time of sample collection.
In our preliminary analysis of fGC concentrations among pregnant females in the low
disturbance area, we could not include all these variables in our model due to our small sample
size. We therefore only tested the effects of variables that we hypothesized were most likely to
correlate with fGC concentrations, namely, time of day, day of gestation, social rank, prey
density, and whether or not social instability was occurring. Finally, Van Meter et al. (2009)
tested whether fGC concentrations were elevated due to recent presence of lions, although this
was found to have no effect on ffGC concentrtions. We did not test for such an effect among
lactating or pregnant females in our low disturbance area because none of our samples were
collected when lions had recently interacted with our subjects.
In our cross-sectional comparison of fGC concentrations between high-and lowdisturbance areas, we separately modeled fGC concentration for juveniles, pregnant females, and
lactating females. We included disturbance as a binary fixed factor (high versus low), time of day
as a binary fixed factor (morning or evening), and subject ID and year as random effects, as well
as any variables determined to be significant in our preliminary analyses. To examine whether
the association between fGC concentrations and disturbance was mediated by other socioecological factors, we also compared each of our models to models that contained a two-way
interaction with disturbance and each of the other predictor variables.

Relationship between early life fGC concentrations and fitness
We focused our analysis on the relationship between early life fGC concentration and fitness of
50

hyenas in the Talek clan, as we have monitored and collected fecal samples from this clan for 21
years (1993-2014). Our preliminary analysis indicated that fGC concentrations were significantly
correlated with age for individuals younger than one year, but that age was not significant if we
restricted the analysis to individuals 6-12 months of age. The time of day in which the sample
was collected was also not a significant predictor of fGC concentrtions for individuals 6-12
months. Therefore, we used samples collected from individuals in this age range to relate fGC
concentrations to our fitness measures. When we had more than one fecal sample for an
individual during this 6 month time period, we averaged the sample values.
We used our demographic records to determine whether individuals survived to
reproductive maturity and to calculate their longevity. For longevity, we restricted our analysis to
females in the Talek clan, as males can often not be monitored after they disperse from their
natal clan. We included all females who died within the course of the study (1988-2014), and for
which we had at least one fecal sample during her first 6-12 months life. Using the glmer
package in R, we used a GLMM with a binomial distribution to test whether an individual’s fGC
concentrations during this period of early life was associated with its likelihood of surviving to 2
years. Using the lmer package in R, we used a linear mixed-effects model to test whether an
individual’s fGC concentrations during this period of early life were associated with its longevity
(in days). In both models, we log-transformed the fGC value and in our longevity model, we
also log-transformed longevity. We also included each model, as a fixed effect, the social rank
of the female’s mother in the year the female was born, because social rank has been found to
predict numerous measures of fitness in spotted hyenas (Holekamp et al. 1996, Swanson et al.
2011). Additionally, we included the female’s birth year as a random effect to account for
51

particularly high or low mortality in certain years due to time-specific environmental conditions.
RESULTS
Longitudinal analysis of socio-ecological factors affecting juvenile fGC concentrations
Our longitudinal analysis of juvenile fGC concentrations included 382 samples collected from
171 juveniles in the Talek clan between 1993 and 2014. Our model indicated significant effects
of time of day, age, sex, and prey density on fGC concentrations (Table 3.1). Specifically,
juvenile fGC concentrations were higher in the morning than in the evening (P<0.001), higher in
females than in males (P=0.032), decreased with age as individuals approached the end of their
second year of life (P<0.001; Figure 3.1), and were positively correlated with prey density
(P=0.034). None of our models containing two-way interactions between rank and our other
predictor variables fit the data significantly better than the models containing no interaction
terms.
Because sex was a significant predictor of fGC concentrations, we next analyzed fGC
concentrations in juvenile males and females in separate models including the same predictor
variables as in the prior model. Females, but not males, showed a significant positive association
between fGC concentrations and prey density (females β=0.14, SE=0.07, P= 0.051; males
β=0.08, SE=0.07, P= 0.26).
Cross-sectional comparison of fGC concentrations in high- and low-disturbance areas
Juveniles
Preliminary analysis of juveniles in the low disturbance area (Table 3.2) showed that fGC
concentrations significantly declined as individuals matured (P<0.001), that juveniles had
52

(Intercept)
Study year
Time of day
Age
Sex
Clan Size
Prey Density
Maternal Rank
Litter Size

Estimate
4.09
0.02
-0.36
-0.37
-0.25
-0.13
0.11
-0.06
-0.07

Std. Error
0.22
0.01
0.10
0.05
0.12
0.08
0.05
0.10
0.13

P
<0.001
0.163
<0.001*
<0.001*
0.032*
0.096
0.034*
0.552
0.583

Table 3.1. Factors explaining variation in fGC concentrations in juvenile spotted hyenas from
the Talek West clan 1993-2014 (N=382 samples from 171 individuals). An asterisk indicates
P<0.05

Figure 3.1. A longitudinal analysis of juvenile fGC concentrations in the Talek West clan
1993-2014 indicated that fGC concentrations significantly decrease with age among juveniles of
both sexes, and that juvenile females have significantly higher fGC concentrations than do
juvenile males. Only samples collected in the morning are shown (N=188 samples).

53

Estimate

Std. Error

P

Juveniles
(Intercept)
Time of day
Age
Sex
Maternal Rank
Prey Density
Litter Size

4.20
-1.01
-0.48
0.48
-0.03
-0.13
-0.07

0.28
0.21
0.12
0.23
0.20
0.08
0.23

<0.001
<0.001*
<0.001*
0.043*
0.90
0.13
0.78

Lactating Females
(Intercept)
Time of day
Rank
Prey Density
Social Instability
Rainfall
Tourism Season

4.8
-0.41
0.27
-0.09
0.53
-0.01
-0.26

0.24
0.22
0.19
0.11
0.41
0.06
0.24

<0.001
0.07
0.16
0.41
0.21
0.85
0.27

	

Pregnant Females
(Intercept)
4.35
0.54
<0.001
Time of day
-0.87
0.52
0.12
Rank
0.17
0.41
0.68
Prey Density
-0.28
0.32
0.41
Social Instability
1.22
0.6
0.07
Day of gestation
0.01
0.01
0.14
Table 3.2. Results of preliminary analysis of variation in fGC concentrations among juveniles,
lactating, and pregnant females in the low disturbance area. An asterisk indicates P<0.05

significantly higher fGC concentrations in the mornings than in the evenings (P<0.001), and that
males had significantly higher fGC concentrations than did females (P=0.043). These variables
were also significant correlates of juvenile fGC concentrations in our longitudinal analysis of the
Talek West clan (although with the direction of the sex effect in the opposite direction), so we
included them here in our cross-sectional analysis for juveniles. Although prey density was not a
significant predictor of juvenile fGC concentrations in the low disturbance area, we included it
as a fixed effect in our cross-sectional analysis because it had a significant parameter estimate in
54

our longitudinal analysis of juvenile females the high-disturbance area.
Our cross-sectional analysis of juvenile fGC concentrations included 194 samples
collected from 123 juveniles (49 juveniles from the low-disturbance area and 74 juveniles from
the high-disturbance area). Our model indicated that fGC concentrations were significantly
higher for samples collected in the morning than in the evening (β=-0.35, SE=0.14, P = 0.011),
and were significantly negatively correlated with age (β=-0.34, SE=0.07, P < 0.001). No other
variables were significant predictors of juvenile fGC concentrations (disturbance, β= 0.18,
SE=0.16, P = 0.256; sex, β= -0.12, SE=0.15, P = 0.442; prey density, β= 0.00, SE=0.00, P =
0.767). However, the model containing an interaction between disturbance and sex was our best
fitting model and a significantly better fit than the model without interactions (2(1)= 9.65, P =
0.002). We therefore analyzed males and females separately. Juvenile females in the highdisturbance area showed significantly higher fGC concentrations than did those in the low
disturbance area (β=0.68, SE=0.24, P=0.007), but fGC concentrations did not vary with
disturbance among juvenile males (β=-0.24, SE=0.19, P=0.223; Figure 3.2). Among juveniles of
both sexes, age (females β= -0.22, SE=0.11, P=0.040; males β= -0.43, SE=0.09, P<0.001) and
time of day (females β= -0.54, SE=0.21, P=0.013; males β= -0.28, SE=0.17, P=0.11) were again
significant predictors of fGC concentrations. There was no significant association between prey
density and fGC concentrations among either males (β=0.00, SE=0.00, P=0.374) or females
(β=0.00, SE=0.00, P=0.230).

Lactating and Pregnant Females
In our preliminary analysis of fGC concentrations among pregnant (N=11 samples from 10
55

Figure 3.2. Residual juvenile fGC concentrations in high- and low-disturbance areas after
controlling for time of day and age at sample collection. N=55 females (82 samples) and 67
males (112 samples). An asterisk indicates p < .05.
individuals) and lactating (N=70 samples from 29 individuals) females in the low disturbance
area, social rank, prey density, and social instability were not significantly correlated with fGC
concentrations (Table 3.2). Among lactating females, rainfall and tourism were also not
significant predictors of fGC concentrations. Therefore, in our cross sectional analyses for
pregnant and lactating females, we did not include any of these variables as fixed effects.
Although day of gestation was not significantly correlated with fGC concentration among
pregnant females in low disturbance areas (P=0.14), we included day of gestation as a
continuous fixed effect in our cross-sectional model because previous hyena studies have found
that fGC concentrations significantly increase over the course of gestation (Goymann et al.
2001, Van Meter et al. 2009).
56

Our cross-sectional analysis of fGC concentrations among pregnant females included 42
samples collected from 31 females. Disturbance level did not significantly predict fGC
concentrations (β=-0.11, SE=0.27, P=0.68). fGC concentrations significantly increased over
the course of gestation (β=0.44, SE=0.12, P<0.001), but the time of sample collection (morning
or evening) was not associated with fGC concentration (β=-0.30, SE=0.26, P=0.27). Neither
our model containing an interaction between disturbance and social rank nor our model
containing an interaction between disturbance and prey density fit the data significantly better
than our model containing neither interaction.
Our cross-sectional analysis of fGC concentrations among lactating females included 191
samples collected from 65 females. Samples collected in the morning had significantly higher
fGC concentrations than samples collected in the evening (β=-0.46, SE=0.14, P=0.002), but
disturbance level did not significantly predict fGC concentrations (β=-0.21, SE=0.18, P=0.24).
Neither our model containing an interaction between disturbance and social rank nor our model
containing an interaction between disturbance and prey density fit the data significantly better
than our model containing no interaction term.

Relationship between early life fGC concentrations and fitness
Seventy-nine individuals in the Talek West clan met the criteria for inclusion in our analysis of
whether early life fGC concentrations predicted the likelihood of surviving to reproductive
maturity. There was not a significant association between early life fGC concentration and
survivorship (β=-0.67, SE=0.41, P=0.104). There was a postive trend between rank and
survivorship (β=1.00, SE=0.60, P=0.074).
57

Thirty-one Talek West females met the criteria for inclusion in our analysis of the
relationship between juvenile fGC concentrations and longevity. Longevity in this sample
ranged from 1.07-14.81 years. There was a trend for females with higher fGC concentrations in
early life to have lower longevity (β=-0.24, SE=0.14, P=0.089) and the association was weaker
when we removed one data point from an individual that had a relatively high fGC
concentration and a short lifespan (β=-0.19, SE=0.15, P=0.20). The mother’s rank in the year
the juvenile was born was also a significant predictor of longevity (β=0.72, SE=0.20, P= 0.001).
DISCUSSION
Our study indicates important sex- and age-specific effects of ecological and anthropogenic
variables on stress hormone concentrations in a wild carnivore. Juvenile females had
significantly higher fGC concentrations than juvenile males, with fGC concentrations
significantly decreasing in both sexes as they approached reproductive maturity. Furthermore, a
juvenile’s sex mediated the effects of environmental variables on stress hormone concentrations.
The fGC concentrations of female juveniles, but not of male juveniles or adult females, were
affected by human disturbance and prey density. Overall, these results suggest that females may
be more sensitive to environmental stressors than males and are consistent with studies of
laboratory rats showing that females show greater HPA reactivity than males from birth
(Patchev et al. 1999, Romeo 2010). This finding has important implications for studying effects
of early stress in wild populations because it suggests that wild spotted hyenas, like laboratory
rodents, may also show sex differences in the long-term behavioral and physiological effects of
early stress (McCormick et al. 1995, Reznikov et al. 2001, Papaioannou et al. 2002, Park et al.

58

2003).
Our cross-sectional analysis indicated sex and age-specific effects of human disturbance
on fGC concentrations. Juvenile females had higher fGC concentrations in the highdisturbance area than in the low disturbance area. However, fGC concentrations among
juvenile males, pregnant females, and lactating females did not differ significantly between highand low-disturbance areas. This is surprising given that, across species, adult males show a
greater increase in stress hormone concentrations due to human disturbance than do adult
females (Dantzer et al. 2014). However, these results are again consistent with findings from
laboratory rats that the response of the HPA axis to environmental events shows age-related
variation (Jacobson-Pick and Richter-Levin 2010, Romeo 2010). Our findings demonstrate that
individuals of different ages and sexes within a population may show different sensitivities to
anthropogenic disturbance and suggest the importance of looking for such effects when
studying the effects of disturbance on stress physiology.
Our longitudinal analysis provided further evidence that the stress physiology of juvenile
females may be particularly sensitive to environmental variables because juvenile females, but
not males, showed a significant positive association between prey density and fGC
concentrations. A previous study at our site found no relationship between prey density and
fGC concentrations among adult hyenas of both sexes (Van Meter et al. 2009), nor did we find a
significant relationship between prey density and fGC concentrations in our preliminary analysis
of adult females in the low-disturbance area (Figure 2). In the current study, it is perhaps
surprising that juvenile females showed a significant positive, not negative, association between
prey density and fGC concentrations. Spotted hyenas are found in larger groups when prey
59

densities are high than when prey is scarce. Perhaps it is larger group sizes, and not high prey
per se, that is stressful for juvenile females. We did not see a significant association between prey
density and fGC concentrations among juvenile females in our cross-sectional analysis, where
maximum prey densities tended to be around 400-500 animals/meter less than in our
longitudinal analysis. This suggests that samples collected during times of very high prey density
were primarily driving the effect in our longitudinal analysis and that this situation may only
serve as a stressor for juvenile females when prey density is very high.
Our results regarding fGC concentrations in adult females differ from previous findings,
which were that adults in the highly disturbed Talek area had significantly higher fGC
concentrations than adults in low disturbance regions (Van Meter et al. 2009). However, the
authors of this previous study sampled from different low disturbance areas than we did, and
fGC concentrations from individuals in low disturbance clans in the current study appear to be
higher than those from individuals in the low disturbance clans studied by Van Meter et al.
(2009). We suspect this might be due to an unusually high frequency of social instability in the
low disturbance clans studied here compared to those studied by Van Meter et al (2009). We
used a binary variable to indicate social instability due to sudden events, such as the death of an
alpha female. However, we are currently exploring several continuous measures of group
stability that better reflect more subtle social dynamics, and that might reveal associations with
fGC concentrations.
Another possible explanation for why we found no significant effect of disturbance on
fGC concentrations among pregnant and lactating females in the current study is that hyenas in
the high-disturbance Talek area may be undergoing long-term adjustments in their HPA activity
60

in response to the chronic stress of pastoralism that has been increasing since the late 1990s.
Chronically stressed animals may show attenuation in HPA activity due to habituation,
desensitization, or HPA axis exhaustion. If hyenas living in the high-disturbance Talek area
have been chronically stressed and attenuation has occurred, we would not necessarily expect
fGC concentrations to differ between individuals sampled in high- and low-disturbance areas.
Changes in HPA activity due to chronic stress are particularly difficult to predict and detect in
wild populations (Cyr and Michael Romero 2007, Cyr and Romero 2009, Dickens and Romero
2013). However, we are currently exploring this hypothesis by investigating whether changes in
fGC concentrations indicate attenuation among hyenas who have been repeatedly sampled and
who have lived in the Talek area as pastoralism has increased.
The data from our study suggest that differences in circulating GC concentrations among
mothers or juveniles are likely not mediating the behavioral differences we previously
documented between juvenile hyenas living in high- and low-disturbance areas. Juveniles in the
high- and low-disturbance areas studied here display consistent differences in personality traits,
with juveniles in high-disturbance areas being less neophobic, more exploratory, and less bold
than juveniles in low disturbance areas (Greenberg and Holekamp 2017). It remains possible
that stress physiology plays a role in mediating these behavioral differences, but that a different
measure of physiological stress is functionally related to personality traits. For example,
biomarkers of stress in the autonomic system have been found to relate to personality traits and
offspring stress physiology in humans (Ponirakis et al. 1998, Nater and Rohleder 2009).
Additionally, numerous studies have found a significant relationship between personality and
HPA axis reactivity, or how quickly and to what degree GCs are released in response to a
61

stressor (Oswald et al. 2006, Schoech et al. 2011, Baugh et al. 2012). In future research, we hope
to compare HPA axis reactivity of spotted hyenas in high- and low-disturbance areas by
performing ACTH challenges during routine immobilizations, and collecting serial blood
samples to be assayed for GC concentrations.
Examination of other aspects of HPA functionality, and not simply baseline fGC
concentrations, should also be conducted to further explore the relationship between stress
physiology and fitness in this species. Although the associations between early life fGC
concentrations and the fitness measures we investigated were in the predicted directions, the
associations were not statistically significant. Our study, along with several other recent studies
and meta-analyses, therefore challenge the assumptions that baseline stress hormone
concentrations are negatively associated with fitness (Bonier et al. 2009a, 2009b), and that
elevations of stress hormone concentrations due to anthropogenic disturbance in particular are
associated with fitness costs (Dantzer et al. 2014). This recent work, along with the current
study, therefore raise important questions regarding the role of non-invasive monitoring of
stress hormone concentrations for wildlife management and conservation. However, as other
researchers have argued, measures like survivorship and longevity may be related to other
aspects of HPA sensitivity or feedback, rather than the crude measure of glucocorticoids
excreted in feces (Romero and Wikelski 2010, Breuner et al. 2012).
In conclusion, the data from our current study allow us to rule out an hypothesis
suggesting that maternal or offspring stress physiology shape personality differences observed
between high- and low-disturbance areas. We are currently testing an alternative hypothesis that
maternal behavior influences these juvenile personality differences. Previous research in the
62

high-disturbance Talek area has shown that, as human activity has increased, mothers spend less
time at communal dens (Kolowski et al. 2007), and preferentially select dens in thicketed areas,
which can be costly when this means they must travel farther to hunt and avoid livestock than
females in undisturbed areas (Boydston et al. 2003a, Kolowski and Holekamp 2009). Research
in other species, both in captivity and the wild, has indicated that changes in environmental
quality or predictability can shape maternal investment strategies and maternal care styles,
resulting in subsequent changes in juvenile traits, such as personality (Rosenblum and Paully
1984, Andrews and Rosenblum 1991, Fairbanks 1996). We are testing this alternative
hypothesis by comparing maternal den attendance and maternal care in areas of high- and lowdisturbance.

63

CHAPTER 4
VARIATION IN MATERNAL BEHAVIOR AND MOTHER-OFFSPRING
INTERACTIONS
Julia R. Greenberg, Zach M. Laubach, Shannon Carvey, Kaycee Morra, Laura Smale, & Kay E.
Holekamp
INTRODUCTION
Observers of mammalian behavior have long noted that mothers often vary in the way they care
for offspring (Hinde 1976, Altmann 1980, Clutton-Brock 1991), and that such variation seems
to be consistent across generations (Fairbanks 1989, Berman 1990, Maestripieri and Wallen
2003). In recent years, scientists have demonstrated that laboratory rodents show “maternal
styles,” revealing natural variation in the way they groom and nurse pups that is consistent over
development and across multiple litters (Liu et al. 1997). This variation in maternal behavior
affects offspring personality, stress reactivity, reproductive strategies, and sociality (Moore 1984,
1992, Coplan et al. 1996, Caldji et al. 1998, Francis et al. 1999, Cameron et al. 2008b, Stevens et
al. 2009, Andrews and Rosenblum 2010). Maternal styles in this laboratory model system are
transmitted across generations via epigenetic mechanisms, and a female’s early experience of
maternal care shapes her own reproductive strategies and patterns of maternal behavior towards
her own pups (Francis, Diorio, & Liu, 1999).
To understand the ecological and evolutionary factors shaping and maintaining such
variation, we need a better understanding of maternal styles in natural systems. Maternal styles
have been documented in some free-living animals (yellow baboon Nguyen et al. 2008, red
squirrel Dantzer et al. 2011, grey seal Twiss et al. 2012, gray-cheeked mangabeys Arlet et al.
64

2016). However, it remains unclear how widespread they are in nature, where variation in
ecological and social conditions is far greater than in the laboratory (Stewart and Mcadam 2014).
Additionally, determining whether maternal styles are directly related to offspring fitness would
shed light on whether the epigenetic mechanisms documented in the laboratory are also likely to
occur in nature.
The causes and consequences of variation in maternal behavior have been most
thoroughly studied in Old World Monkeys (reviewed in Fairbanks 1996). As infants mature,
mothers generally spend less time in contact with offspring, and offspring become more
responsible for mediating changes in contact and proximity with their mothers, using their
mothers as a “secure base” from which to explore their environment (Hinde and Atkinson 1970,
Nash 1978). However, beyond this ontogenetic trend, mothers often show marked variation in
how much time they spend in contact with offspring, to what degree they initiate contact with
and departures from, offspring, and how often they groom offspring. Highly protective mothers
spend a relatively large proportion of their time in contact with offspring, frequently initiate
contact with their offspring, and restrain offsprings’ attempts to break contact. Highly rejecting
mothers spend a relatively small proportion of their time in contact with offspring, often reject
offspring attempts to make contact, and show high rates of breaking contact with offspring
(Fairbanks 1996). In some species, a third dimension of maternal behavior emerges, sometimes
referred to as warmth, which is positivity correlated with maternal grooming and cradling of
offspring (pigtail macaques, Maestripieri 1998).
Although a large proportion of the variation in maternal behavior among cercopithecines
is due to individual variation, maternal styles may also be shaped by the degree to which a
65

mother’s physical or social environment is challenging. For example, mothers that are resourcelimited, live in unpredictable environments, or experience extreme environmental events tend to
be more rejecting of offspring than those living in more favorable environments (Hauser and
Fairbanks 1988, Andrews and Rosenblum 1991, Fairbanks and McGuire 1995, Lycett et al.
1998). Such effects may be mediated by effects of the environment on maternal condition. On
the other hand, if the mother experiences a challenging social environment that poses risks to
offspring, such as aggression, harassment, or kidnapping, mothers tend to be more protective
(rhesus macaque Maestripieri 1995, 2001). This is consistent with the findings of some primate
studies that low-ranking mothers, who experience more social challenges, are more protective
than high-ranking mothers (savannah baboon Altmann 1980, rhesus macaque Berman 1990,
yellow baboon Wasser and Wasser 1995). However, in other primate studies, the opposite
relationship between rank and maternal behavior has been found (Schino et al. 1995, BentleyCondit 2003). It is also unclear how the degree of environmental or social risk to young relates
to the common finding that primiparous mothers tend to be more protective than multiparous
mothers (Tanaka 1989, Maestripieri 1998). Instead, differences in maternal behavior due to
parity have largely been interpreted as the result of maternal experience.
Across species, mothers may also behave differently towards offspring based on
offspring traits. For example, maternal investment is predicted to be greater in the sex that
shows more variation in reproductive success (Trivers 1972, 1974, Smith 1980, Clutton-Brock et
al. 1981). This may explain why male offspring sometimes nurse at higher rates than females
(Lee and Moss 1986), initiate contact with their mothers more often than female offspring
(Nguyen et al. 2012), or receive more grooming from mothers than do female offspring (Moore
66

1984, 1992). Additionally, an individual’s litter size may mediate mother-offspring relationships
not only because an individual competes with siblings for maternal resources (Hudson and
Trillmich 2008, Hudson et al. 2011), but also because mothers may behave differently towards
some members of a litter than toward others (Mendl 1988). For example, even among same-sex
siblings, mothers groom some rat pups significantly more frequently than their siblings (Cavigelli
and McClintock 2003, Ragan et al. 2011, 2016).
In this study, we explored natural variation in maternal behavior and its fitness
consequences for offspring in a free-living population of spotted hyenas. In terms of their basic
developmental biology and life-history, spotted hyenas show both similarities and differences to
the primates and rodents that have been the focus of most earlier studies of maternal behavior.
Understanding patterns of maternal variation in this wild carnivore can therefore help us to
better understand the ecological and evolutionary factors shaping variation in maternal care
across species. Spotted hyenas show a degree of maternal investment and a protracted period of
offspring growth and development that is more typical of primates than other carnivores (Watts
et al. 2009). Mothers give birth to small litters (1-2 cubs) at isolated natal dens and cubs do not
wean until an average of 14 months of age at our site (Holekamp et al. 1996, Holekamp and
Dloniak 2010). A mother’s presence and defense remain critical for offspring survival even after
weaning, as the feeding apparatus of young spotted hyenas is not fully developed for eating meat
and bone until 36 months of age (Tanner et al. 2009), and they do not reach full proficiency at
hunting until roughly five years of age (Holekamp et al. 1997b).
Whereas patterns of maternal investment among spotted hyenas are similar to many
primates, young spotted hyenas spend much less time with their mothers than do primates, and
67

instead reside at a communal den that is more similar to the nests utilized by rodents. When
cubs are approximately 3-4 weeks of age, mothers move cubs from the natal den to communal
dens where they live for the next 7-12 months. The communal den is composed of a series of
burrows and tunnels that protect young hyenas from dangers present in the rest of the clan’s
territory (Boydston et al. 2006). Hyena cubs emerge from the den to nurse from their mothers
and socialize with other hyenas. Whereas primate mothers carry their offspring for many
months, young hyenas only have the opportunity to interact with their mothers when mothers
visit the communal den; this can occur at intervals ranging from a few hours, as at our study site,
to 3-4 days at other sites (Hofer & East, 1993b). Furthermore, whereas juvenile primates
typically treat their mothers as a “secure base” from which to explore their environment and flee
to her during times of danger, the communal den may also serve this function for young hyenas.
Spotted hyenas live in matrilineal societies nearly identical in size and structure to those
of many Old World Monkey species; however unlike those primate species, female spotted
hyenas are dominant to males and therefore primate and hyena mothers may vary in their
treatment of male and female offspring. Juvenile hyenas “inherit” dominance ranks immediately
below those of their mothers in a process of social learning that is not complete until at least 18
months of age (Holekamp & Smale 1993; Smale et al. 1993). Hyenas of both sexes retain their
maternal ranks as long as they reside in the natal clan, which females do throughout their lives,
but most males disperse to new clans at 2-5 years of age (Boydston et al 2005; Smale et al. 1997;
Höner et al. 2007). High-ranking spotted hyena mothers intervene on behalf of their offspring
in aggressive interactions more often and more effectively than do low-ranking mothers (Engh
et al. 2000)
68

To examine variation in maternal behavior, we first used focal animal surveys (FAS) on
mother-cub dyads to determine the proportion of time that mothers nursed their cub(s), were in
close proximity to their cub(s), groomed their cub (s), and to what degree the mother initiated
being close to cub(s). To determine sources of this variation, we examined the effects of a
mother’s parity and social rank, cub age, cub sex, litter size, and external factors such as time of
day and prey density. Next, we explored whether spotted hyena mothers show consistent
individual variation in their maternal behavior, and to what extent variation in maternal behavior
is due to these consistent individual differences. Finally, we inquired whether measures of
maternal behavior predicted offspring survivorship to reproductive maturity.

METHODS
Subjects & Study Populations
Study subjects lived in a single large clan of spotted hyenas inhabiting the Talek region of the
Masai Mara National Reserve (MMNR) in southwestern Kenya during two different time
periods, 1988-1991 and 1993-1996. Individuals could be identified by their unique spot patterns
and ear damage. Subjects were juveniles who had not yet been weaned, and their mothers. Age
was initially estimated when cubs were first observed, based on their appearance and size
(Holekamp et al. 1996), and sex was determined based on the shape of the glans of the erect
phallus (Frank et al. 1990). The date on which a cub was last seen was assigned as its death date
because female hyenas spend their entire lives in their natal clan and male hyenas do not
typically disperse before two years of age (Smale et al. 1997, Holekamp and Smale 1998a). In
litters of two cubs, we determined the cub’s intra-litter rank based on nursing position and
69

aggressive and submissive interactions between the siblings. Spotted hyena mothers have only
two teats, and once the dominance relationship between litter-mates has been established,
typically by 4 weeks of age, a cub’s position while nursing consistently varies with its intra-litter
rank. As shown in Figure 4.1, the dominant cub typically nurses while lying against its mother’s
stomach, whereas the subordinate sibling typically nurses while strattled between its mother’s
hind legs.
Spotted hyenas live in clans of up to 130 individuals (Holekamp et al. 2015) that are
structured according to a strict matrilineal hierarchy (Frank 1986). Here the social ranks of adult
females were calculated on a yearly basis, based on outcomes of dyadic agonistic interactions,
and were used to construct a dominance hierarchy each year (Martin & Bateson 1988; Smale et

	

Figure 4.1. Photograph of a spotted hyena mother nursing two cubs. As is typical, the dominant
member of the litter is laying up against its mother’s stomach while nursing, whereas the
subordinate litter-mate is strattled between her legs. Photo courtesy of Lily Johnson-Ulrich.

70

al. 1993). Ranks were standardized from -1 to 1, with the most dominant individual having a
rank of 1.
The reproductive state and parity of adult females was continuously monitored as
described by Holekamp et al. (1996). A female’s first parturition could be determined reliably
because a female’s pseudopenis tears when she gives birth to her first litter, leaving an
unambiguous stripe of pink scar tissue on the posterior side of the phallus (Frank and Glickman
1994, Holekamp et al. 1996).

Data Collection
From 1988-1991, we conducted focal animal surveys (FAS) on every juvenile in the clan
approximately every two weeks. From 1993-1996, we collected FAS data on adult females
opportunistically, making every attempt to balance the dataset with regards to the mother’s
social rank, parity, and litter size. FAS were conducted during our routine daily morning
(0600-1000h) and evening (1600-2000h) observation periods. We followed the focal subject,
either the cub or the adult female, for 30 minutes or until the subject had been out of sight for 5
minutes. During FAS, we recorded all affiliative interactions between individuals, such as,
sniffing, grooming, playing, and ritualitized “greetings” in which two individual lift a leg to allow
the other to sniff their genital region). We also recorded all aggressive interactions, ranging from
displacements to chasing and biting, counterattacks, and coalitionary attacks. Lastly, we
recorded all instances in which the focal subject approached or was approached by another
individual to within 1m.
We extracted data on maternal behavior from FAS on cubs in which the cub’s mother
71

was present, and from FAS on adult females in which one or more of her dependent cubs was
present. If a female had a litter of two cubs and both were present during a FAS conducted on
her, we extracted two different sets of FAS data, one for each cub. From each FAS, we extracted
the number of minutes in which the mother engaged in three maternal behaviors: nursing,
grooming the cub, or being in close proximity to the cub (within 1m or in contact). We then
calculated the total duration of each behavior in each FAS. For the time that a mother spent in
close proximity to the cub, we excluded any minutes when the mother was nursing her cub, as
nursing necessitates contact with the cub. Additionally, whenever the mother-cub pair moved in
or out of close proximity, we recorded which individual initiated the change as an “event” (e.g.
cub approach mother, mother approach cub, cub leave mother, or mother leave cub). In some
cases, both the mother and cub simultaneously approached one another (0.4% of all
approaches) or left one another (0.3% of all departures), or we could not determine which
individual approached (5% of all approaches) or departed (2% of all departures). However,
because these cases represented a small percentage of all proximity changes, and because we
were primarily interested in whether the mother or cub was primarily responsible for proximity
changes, we excluded these cases from analysis.

Statistical Analysis
We used generalized linear mixed models to model variation in each of the three maternal
behaviors of interest (nursing, grooming, and time spent in close proximity to the cub), and our
event measures that indicated proximity maintenance. All analyses were conducted using the
glmmADMB package in R (R version 3.3.1; Development Core Team, 2015), which allows for
72

the incorporation of both random effects and zero-inflated predictor variables. We used the
multicomp package to perform post-hoc tests with a Tukey correction for pairwise comparisons.
For our models involving durations of maternal behaviors, we used the number of minutes in
which the mother engaged in each behavior during a particular FAS as our response measure.
To model variation in how often mothers approached their offspring, we limited our analysis to
sessions in which either the mother or the cub approached the other at least once during the
FAS. We used the number of times a mother approached her cub in the FAS as our response
measure and included the total number of approaches by either mother or cub in the FAS as a
predictor variable. To determine the extent to which mothers left their cubs, we limited our
analysis to sessions in which the mother and cub left one another at least once, used the number
of times a mother left her cub within a FAS as our response measure, and included the total
count of times the mother or cub left one another as a predictor variable. To account for the
fact that our FAS varied in the amount of time when both a mother and her cub were present,
we included as a logged offset the number of minutes in which the mother and cub were both
present during the FAS in each model. We logged the offset value as this is the standard way to
incorporate offsets in glmmADMB and helps to scale the offset to be proportional to the
predictor variable. We excluded from all analyses any FAS in which the mother and cub were
present together for less than 5 min. In each model, we also included the mother’s ID as a
random effect.
In all models, we included several ecological, social, and demographic variables as
predictor variables. As binary fixed effects, we included the time of day (morning or evening),
and the mother’s parity (primiparous or multiparous), and litter size (i.e. whether the focal cub
73

did or did not have a living litter-mate alive on the day of the FAS). As continuous fixed effects,
we included the cub’s age, the mother’s social rank (standardized from -1 to 1, as described
above), and prey density. Prey density was calculated as the average number of prey counted
within 100m of three 4-km prey transects sampled within two-weeks of the FAS. For the first
three months of the study in 1988, we did not have data from prey transects and we therefore
used the values averaged from the other years of the study during the corresponding two-week
intervals. Cub’s age and prey density were scaled and centered in our analyses. If models
indicated a significant effect of whether the cub had a litter-mate or not, we then restricted our
dataset to include only cubs that had a living litter-mate, and tested whether a cub’s intra-litter
rank (dominant or subordinate) was a significant predictor of the maternal behavior of interest.
In all models except our model for nursing, we also included a random effect of study
year. We removed this random effect in our models for nursing in order for our models to
converge. Our diagnostics suggested that this term was problematic, as random effects with
relatively few levels can sometimes cause non-convergence especially when they have low
variance, as was the case for this term. There are several suggestions regarding what to do in
this situation, including removing the term from the model and fitting the model with the term
as a fixed effect instead of a random effect (Barr, D.J., Levy, R., Scheeperes, C., Tily 2014). We
did both of these and our results did not differ when using these alternative approaches.
Because we were mainly interested in controlling for year-to-year variation in our models of
maternal behavior, we report the results of our nursing model that does not include study year as
a factor.
For each maternal behavior, we assessed the fit of four types of count models; a negative
74

binomial model, a poisson model, a zero-inflated negative binomial model, and a zero-inflated
poisson model. We then compared the fit of the four models using Akaike’s Information
Criterion corrected for small sample sizes (AICc). A smaller AICc value indicated a better fit,
and we report the results from the best-fitting model for each maternal behavior.
glmmADMB is a relatively new R package, and model diagnostic methods are still being
developed. We therefore focused our evaluation of model fit on the residual deviance. We also
evaluated overdispersion using the theta value provided in the model outputs. We used Variance
Inflation Factors (VIF) to assess multi-colinearity of predictor variables in all of our models. No
variable had a VIF above 3, indicating that there was no significant multi-colinearity among
variables in our models.

Individual variability and repeatability of maternal behavior
We assessed the contribution of individual variation among mothers to our models of nursing,
close proximity, and grooming durations, as well as our models of proximity maintenance in two
ways. First, for each of these models we determined whether the random effect of maternal ID
significantly improved the fit of our models. We used a likelihood ratio test to compare each
model to a model that did not include the mother’s ID as a random effect, and calculated a
parameter estimate for the mother’s identity. Second, we calculated the proportion of variance
explained by the random effect of maternal ID by dividing the proportion of variance explained
by maternal ID by the total residual variance of the model. This offers a measure of
repeatability in maternal behavior (Lessells and Boag 1987, Dantzer et al. 2011).

75

Relationship between maternal behavior and offspring fitness
To determine whether an individual’s experience of maternal behavior in early life predicted it
likelihood of surviving to adulthood, we first used our demographic records to determine
whether each cub survived to reproductive maturity (2 years of age). We calculated the
proportion of time that a cub nursed, was groomed by its mother, or was in close proximity to
its mother from 3-6 months of age. To control for differences in behavior based on time of
day, we only included evening sessions. We chose this time interval in order to maximize the
number of individuals we could include and we only included individuals that had multiple
evening observations in each of the three months. In three separate general linear models, we
then modeled the likelihood of surviving (Y/N) to reproductive maturity as a function of the
proportion of FAS time a cub spent nursing, being groomed by its mother, or in close proximity
to its mother. We used the “glm” package in R and specified a binomial response measure.

RESULTS
Our data on mothers with cubs 0-15 months included a total of 1075 FAS collected on 34
mothers and 122 cubs (Figure 4.2). An average of 8.83 FAS (range 1-32) were conducted on
each mother-offspring pair. Figure 4.3 shows the average percent of the time that a mother and
cub were both present in a FAS during which the mother was nursing her cub (a), in close
proximity to her cub (b), and grooming her cub (c).

Models of time spent nursing, grooming, and in close proximity to offspring
A zero-inflated negative binomial model best described variation in nursing among mothers
76

	

Figure 4.2. The total number of Focal Animal Surveys (FAS) collected on mothers with cubs
10-15 months. Data are binned by the cub’s age in months. Numbers above bars indicate the
number of cubs observed in that age category.

(Table 4.1). Mothers spent a significantly greater proportion of their time nursing as cubs
matured, and in the evening than in the morning. There was a trend for primiparous mothers to
nurse their cubs more than did multiparous females. Mothers of twin litters spent a greater
proportion of their time nursing than did mothers of singletons. A model including only cubs
who had a living litter-mate indicated no significant effect of intra-litter rank on the proportion
of time a mother spent nursing (B = -0.04, SE = 0.05, P =0.50).
A zero inflated negative binomial model best explained variation in how much time
mothers spent in close proximity to their cubs (Table 4.2). Mothers spent a significantly lower
proportion of their time close to cubs as the cubs matured, and a significantly greater proportion
of their time in close proximity in the mornings than in the evenings. Mothers spent a
77

Figure 4.3. Developmental changes in three maternal behaviors among spotted hyena mothers
with cubs 0-15 months old. Out of the total minutes a mother and cub were both present in an
FAS, figures show the proportion of time a mother spent a) nursing the cub; b) grooming the
cub and; c) in close proximity to the cub. Labels indicate the number of FAS averaged for each
month of development.
78

(Intercept)
Cub Age
Cub Sex
Time of Day
Maternal Rank
Parity
Litter Size
Prey Density

Odds Ratio
0.54
1.16
0.94
1.20
1.00
0.88
1.18
1.01

Estimate
-0.62
0.14
-0.06
0.18
0.00
-0.12
0.16
0.01

Std. Error
0.06
0.03
0.05
0.05
0.05
0.06
0.06
0.02

P
<0.001
<0.001*
0.221
<0.001*
0.964
0.057
0.005*
0.598

	
Table 4.1. Results of a zero-inflated negative binomial model explaining variation among
mothers in duration of nursing. An asterisk indicates P<0.05.

(Intercept)
Cub Age
Time of Day
Cub Sex
Maternal Rank
Parity
Litter Size
Prey Density

Odds Ratio
0.42
0.88
0.72
1.19
1.04
0.91
0.73
0.95

Estimate
-0.88
-0.13
-0.33
0.17
0.04
-0.10
-0.31
-0.05

Std. Error
0.13
0.04
0.07
0.07
0.08
0.12
0.08
0.04

P
<0.001
0.001*
<0.001*
0.020*
0.573
0.405
<0.001*
0.169

	
Table 4.2. Results of a zero-inflated negative binomial model explaining variation in time
mothers spent in close proximity to their cubs. An asterisk indicates P<0.05.

79

significantly greater proportion of their time in close proximity to male than female cubs, and a
significantly greater proportion of their time in close proximity to the focal cub if it was a
singleton than if it had a litter-mate. A model including only cubs who had a litter-mate
indicated that a cub’s intra-litter rank was not significantly associated with the proportion of
time a mother spent in close proximity to the cub (B = 0.05, SE = 0.08, P =0.54).
Because our analysis indicated a significant effect of cub sex on how much time a mother
spent in close proximity to the cub, we stratified the dataset according to cub sex and ran the
model separately for male and female cubs. We included the same offset, random effects, and
independent variables, but excluded cub sex as a variable. Mothers of males cubs spent a greater
proportion of their time in close proximity to the focal cub if it was a singleton than if it had a
litter-mate (B = -0.48, SE = 0.10, P < 0.001). A model including only male cubs that had a
litter-mate indicated that a cub’s intra-litter rank was not associated with the proportion of time
a mother spent in close proximity to the cub (B = 0.11, SE = 0.14, P =0.42). Among mothers of
female cubs, there was no significant effect of litter size on the proportion of time a mother
spent in close proximity to the cub (B = 0.04, SE = 0.16, P =0.80).
A negative binomial model best described variation in grooming among mothers (Table
4.3). Mothers spent a significantly lower proportion of their time grooming cubs as the cubs
matured and spent a significantly greater proportion of their time grooming cubs in the evening
than in the morning. Multiparous mothers spent a significantly greater proportion of their time
grooming cubs than did primiparous mothers. Mothers spent a significantly greater proportion
of their time grooming the focal cub when the cub was a singleton than when it had a littermate. A model including only cubs who had a litter-mate indicated that a cub’s intra-litter rank
80

(Intercept)
Cub Age
Time of Day
Cub Sex
Maternal Rank
Parity
Litter Size
Prey Density

Odds Ratio
0.02
0.41
1.79
1.23
0.68
0.48
0.39
0.88

Estimate
-3.96
-0.90
0.58
0.21
-0.39
-0.74
-0.93
0.12

Std. Error
0.31
0.12
0.21
0.22
0.29
0.36
0.25
0.11

P
<0.001
<0.001*
0.005*
0.341
0.179
0.041*
<0.001*
0.254

	
Table 4.3. Results of a negative binomial model explaining variation in time mothers spent
grooming offspring. An asterisk indicates P<0.05.
was not significantly associated with the proportion of time a mother spent grooming it (B =
-0.36, SE = 0.25, P =0.16).
Because our analysis indicated a significant effect of maternal parity on how often a
mother groomed her offspring, we stratified the dataset according to parity and ran the model
separately for primiparous and multiparous mothers. We included the same offset, random
effects, and independent variables, but excluded parity as a predictor variable. Results suggest
that variation among multiparous mothers was driving the effect of litter size on maternal
grooming because there was a significant effect of litter size on the proportion of time
multiparous mothers spent grooming their cubs, but this effect was not observed among
primiparous mothers. Multiparous mothers spent a significantly greater proportion of their time
grooming the focal cub when the cub was a singleton than when it had a litter-mate (B = -0.94,
SE = 0.27, P = 0.001), but there was no significant effect of a cub’s intra-litter rank on grooming
(B = -0.29, SE = 0.26, P = 0.268). Grooming by primiparous mothers did not vary with litter
size (B = -0.82, SE = 0.66, P = 0.214).

81

Models of proximity maintenance
Our model of variation in how often a mother approached her cub included 541 surveys
collected from 34 mothers and 116 cubs (Figure 4.4). Each mother-cub pair had an average of
5.92 FAS (range: 1-22). Our best fitting model was a negative binomial model (Table 4.4). The
total number of approaches within a session was a significant predictor of the number of times
mothers approached their cub and mothers approached singletons significantly more often than
cubs that had a litter-mate. A model including only cubs who had a litter-mate indicated that a
cub’s intra-litter rank was not significantly associated with the number of times a mother
approached her cub (B = -0.24, SE = 0.21, P =0.25).
Our model for variation in how often a mother left her cub included 712 surveys
collected from 34 mothers and 115 cubs (Figure 4.5). Each mother-cub pair had an average of
6.21 FAS sessions (range: 1-23). Our best fitting model was a negative binomial model (Table
4.5). Mothers left their cubs significantly more as cubs matured and the total number of
departure events within a session was also a significant predictor of the number of times
mothers left their cub.

Individual variability and repeatability of maternal behavior
In regard to nursing, the model that included maternal ID was not a significantly better fit than
the model without maternal ID as a random effect. In fact, our model without maternal ID was
a significantly better fit than our model with maternal ID (likelihood ratio test, 2=4.88, P
=0.027), and variation among mothers explained only a small proportion (0.17%) of the residual
variance in the model. In regard to time spent in close proximity, the model that included
82

Figure 4.4. Developmental changes in the percentage of approaches to within 1m that were
initiated by the mother. Labels indicate the number of FAS averaged for each month of
development.

Odds Ratio
0.01

Estimate
-4.32

Std. Error
0.26

P
<0.001

(Intercept)
Total # of
approaches
1.18
0.16
0.03
<0.001*
Cub Age
1.08
0.08
0.09
0.386
Time of Day
0.92
-0.08
0.17
0.620
Sex
0.99
-0.01
0.18
0.960
Maternal Rank
1.07
0.07
0.17
0.684
Parity
1.28
0.25
0.27
0.349
Litter Size
0.38
-0.96
0.18
<0.001*
Prey Density
1.04
0.04
0.09
0.671
Table 4.4. 	Results of a negative binomial model explaining variation in how often mothers

approached their cub. An asterisk indicates P<0.05.

83

Figure 4.5. Developmental changes in the percentage of departures to further than 1m apart that
were initiated by the mother. Labels indicate the number of FAS averaged for each month of
development.

(Intercept)
Total # of leaves
Cub Age
Time of Day
Cub Sex
Maternal Rank
Parity
Litter Size
Prey Density

Odds Ratio
0.02
1.09
1.40
1.00
0.94
0.84
1.04
0.93
0.96

Estimate
-3.86
0.09
0.34
0.00
-0.06
-0.17
0.04
-0.07
-0.05

Std. Error
0.19
0.01
0.05
0.10
0.10
0.10
0.17
0.11
0.06

P
<0.001
<0.001*
<0.001*
0.973
0.556
0.095
0.797
0.505
0.441

	

Table 4.5. Results of a negative binomial model explaining variation in how often mothers
moved more than 1 m from their cub. An asterisk indicates P<0.05.

84

maternal ID was not a significantly better fit than the model without maternal ID as a random
effect (likelihood ratio test, 2=2.28, P =0.131), and variation among mothers explained only 2%
of the residual variance of the model. However, in regard to grooming, the model including
maternal ID was a significantly better fit than the model without maternal ID as a random effect
(likelihood ratio test, 2=9.86, P =0.002), and variation among mothers explained 48% of the
residual variance of the model (Figure 4.6). Neither model for proximity maintenance was a
better fit when including maternal ID as a random effect (approaches by mother: 2=1.61,
P=0.20; departures by mother: 2=1.62, P =0.20). Maternal ID explained 5% of the residual
variance of the model for maternal approaches and 2% for maternal departures from the cub.

Relationship between maternal behavior and offspring fitness
Only nine individuals met our criteria for inclusion in the analysis of whether maternal behaviors
predict offspring survivorship. Neither the proportion of time a mother spent nursing her cubs,
time she was in close proximity to her cubs, nor the proportion of time she spent grooming her
cubs significantly predicted a cub’s likelihood of surviving to reproductive maturity (nursing, B
= 7.87, SE = 7.41, P = 0.29; close proximity, B = 9.79, SE = 9.24, P = 0.29; grooming B = 4.43,
SE = 34.91, P = 0.90).

DISCUSSION
Our study documents natural variation in maternal behavior among free-living carnivores, and
adds to a growing body of literature elucidating the factors that shape intra-specific variation in
maternal behavior. Compared to the Old World Monkey and rodent species that have been
85

Figure 4.6. Individual variation in the proportion of a mother’s time spent nursing, grooming,
and in close proximity to cubs. Y-axis is the best linear unbiased predictor (BLUP) for each of
the 34 mothers calculated from our models for proportion of time spent in each of the three
maternal behaviors.

86

most thoroughly studied, spotted hyenas showed both important similarities and differences in
their patterns of maternal behavior. Spotted hyena mothers showed significant and consistent
inter-individual variation in how much time they spent grooming cubs, suggesting that, like
laboratory rodents, the degree to which a mother grooms her cubs is a component of maternal
styles in spotted hyenas.
Unlike Old World Monkeys, spotted hyenas did not show significant and consistent
inter-individual variation in how much time they spent close to cubs and how they mediated
proximity with their cubs. It is possible that this difference stems from the fact that
cercopithecines, unlike spotted hyenas, carry their young for extended periods of time and that
young can physically cling to mothers. Additionally, while juveniles of these primate species
often flee to their mothers in times of danger, or use their mother as a secure base from which
to explore their environments, spotted hyena cubs often utilize the communal den as a secure
base, and this could mean that changes in proximity between mother and offspring do not
reflect maternal protectiveness or offspring attachment as meaningfully in hyenas as monkeys.
Future research should investigate other aspects of maternal behavior that may contribute to
individual maternal styles. For example, the degree to which a mother is aggressive (Dwyer and
Lawrence 2000), or playful (Blehar et al. 1977) towards offspring can vary consistently and
significantly among mothers of other species, and it would be interesting to look at this in
spotted hyenas.
Interestingly, a female’s social rank was not a significant predictor of any of the maternal
behaviors we measured. This is in contrast to most other aspects of spotted hyena behavior that
have been studied at our field site, including patterns of aggression (Dloniak et al. 2006), space
87

use (Boydston et al. 2003a), personality (Yoshida et al. 2016), and social bond formation
(Holekamp et al. 1997a, Smith et al. 2007). Spotted hyenas live in a more flexible fission-fusion
society than cercopithecines, and previous studies have found that low-ranking mothers are able
to avoid threatening social situations by visiting the den when few high-ranking females are
present (White 2007). Furthermore, even within cercopithecines, some studies have found that
rank and maternal care are not significantly associated (Schino et al. 1995, Bentley-Condit 2003),
or that rank and maternal behaviors are significantly related among male but not female
offspring (Schino et al. 1999). Our finding suggests that rank-related maternal effects on cub
growth and development are not mediated by how much time mothers spend nursing,
grooming, or in close proximity to their offspring, suggesting that these rank effects are due
solely to differences in maternal condition (Hofer and East 2003), gestational hormones
(Dloniak et al. 2006), and aggressive defense of offspring (Engh et al. 2000).
Our study also indicated that a spotted hyena’s experience of maternal care is profoundly
influenced by litter size. Mothers spent a significantly greater proportion of their time grooming,
or in close proximity to a cub if it was a singleton than if it had a litter-mate. However, mothers
of twin litters did not bias their behavior towards either the dominant or subordinate cub for
either behavior. This suggests that mothers of twin litters divide their maternal care between
litter-mates. Regarding nursing, mothers spent a significantly greater proportion of their time
nursing when they had twin litters as opposed to singletons. Again, there was no difference in
the proportion of time mothers spent nursing dominant versus subordinate litter-mates, which is
surprising given previous findings that subordinate litter-mates sustain more nursing
interruptions due to sibling aggression (Benhaiem et al. 2012), and are often considerably smaller
88

than their dominant litter-mates (Holekamp and Dloniak 2009). This suggests that, at least in
our dataset, such nursing interruptions were brief and subordinates were able to offset their
effects. It also suggests that, although dominant litter-mates are often larger than subordinate
litter-mates, this pattern is not due to differences in durations of nursing, but perhaps
differences due to nursing position (for example, differences in milk let-down from different
teats or differences in the degree of body contact with the mother). In addition to a cub’s litter
size, its sex was also a significant predictor how much time it spent in close proximity to its
mother. Mothers appeared to bias this aspect of their maternal care towards male singletons.
Studies in other species have found similar evidence for sex-biased maternal care in some
behavioral measures, but not others (Schino et al. 1999, Dettmer et al. 2015, King and Goldizen
2016)
Contrary to findings among Old World Monkeys, we found that primiparous mothers
showed behaviors indicative of less protectiveness than multiparous mothers. Primiparous
spotted hyena mothers spent a significantly lower proportion of their time grooming cubs than
did multiparous mothers. And unlike multiparous mothers, who groomed singletons more than
members of twin litters, primiparous mothers groomed cubs equally regardless of litter size.
These differences could be due either to primiparous mothers having less experience rearing
offspring or being in poorer condition than multiparous mothers. Consistent with studies on
Old World Monkeys, there was a trend for primiparous mothers to nurse their cubs more than
multiparous mothers, suggesting that primiparous mothers may attempt to compensate for
deficits in milk quality or nursing ability.
Our study raises questions regarding how variation in maternal behavior influences
89

offspring and how it is maintained within this population. Future research could determine
whether offspring of “high-grooming” and “low-grooming” spotted hyena mothers show
consistent differences in personality or stress reactivity, as they do in laboratory rodents. Given
that we found notable differences in how mothers behave towards singletons versus members
of twin litters, it would also be interesting to inquire whether individuals show personality
differences into adulthood based on their litter status. If maternal behaviors such as grooming
and spending time close to offspring influence offspring traits as they do in cercopithecines, we
would predict that singletons would be more neophobic, fearful, and less exploratory than
individuals who were in twin litters (Fairbanks 1993, Schino et al. 2001). Finally, our study raises
the question of whether the natural variation in grooming among spotted hyena mothers is
transmitted across generations through epigenetic mechanisms, as it is among laboratory
rodents. Future research should examine cross-generational consistency in these maternal
behaviors.

90

CHAPTER 5
ANTHROPOGENIC EFFECTS ON MATERNAL BEHAVIOR IN SPOTTED HYENAS
Julia R. Greenberg, Zach M. Laubach, & Kay E. Holekamp
INTRODUCTION
Maternal effects, whereby a mother influences her offspring phenotype independently of
offspring genotype, can profoundly affect offspring fitness, physiology, and behavior (Mousseau
and Fox 1998). Mothers alter their patterns of maternal care in response to environmental
challenges or altered resource availability (Ono et al. 1987, Hauser and Fairbanks 1988, Lee et al.
1991, Burkanov et al. 2011). Such changes can, in turn, affect offspring growth, physiology,
behavior, or survivorship (Rosenblum and Paully 1984, Coplan et al. 2001, Stevens and Boness
2003, Arroyo and Razin 2005, Gibbs et al. 2010, Champagne 2011, French et al. 2011, Johnson
et al. 2012, Almasi et al. 2015). Maternal effects can therefore be an important mechanism
through which organisms respond flexibly to changing conditions in their environment
(Jablonka et al. 1995, Räsänen and Kruuk 2007).
Anthropogenic disturbance affects the behavior of a range of wildlife species (reviewed
in Blumstein, Fernández-Juricic, Zollner, & Garity, 2005, Miranda, Schielzeth, Sonntag, &
Partecke, 2013, Sol, Lapiedra, & González-Lagos, 2013), and therefore has the potential to affect
offspring through behavioral maternal effects. Among mammalian carnivores, which are
threatened worldwide (Ripple et al. 2014), human activity has both lethal and sub-lethal effects
on behavior. Compared to carnivores in undisturbed areas, carnivores in disturbed areas are
often more nocturnal (Ciucci et al. 1997), actively avoid humans (Riley et al. 2003, Gosselink et

91

al. 2007), show shifts in habitat preference and diet (reviewed in (Bateman and Fleming 2012),
and changes in home range size (Gehrt et al. 2009). Anthropogenic disturbance may also shape
variation in carnivore social tolerance, social structure (Boyd et al. 1995; Macdonald, Courtenay,
Forbes, & Mathews, 1999), and personality (Greenberg and Holekamp 2017). Despite the
potential for disturbance-induced maternal effects on the behavior of offspring, most studies of
disturbance-related maternal effects in carnivores have focused on offspring survivorship. In
several species, studies have shown that disturbance may influence maternal behavior and
offspring development without negatively affecting offspring survival (Amstrup 1994, Engelhard
et al. 2002, Stevens and Boness 2003, James and Myfanwy 2013).
Here we inquired whether anthropogenic activity shapes maternal behavior among freeliving spotted hyenas, and tested an hypothesis suggesting that disturbance-related maternal
effects influence offspring behavior. Maternal effects have the potential to profoundly affect
offspring fitness and behavior in this species because spotted hyena mothers show an unusually
protracted period of maternal investment in comparison to most other mammalian carnivores
(Watts et al. 2009). Spotted hyenas give birth to litters containing 1 or 2 cubs at isolated natal
dens, and move cubs to a communal den when they are 3-4 weeks of age. There they reside for
the next 7-14 months (Holekamp & Dloniak 2010), emerging from the den to nurse from their
mothers and socialize with clan-mates (Holekamp & Smale, 1998). Mothers must significantly
alter their space use to visit the communal den (Hofer and East 1993a, Boydston et al. 2003a),
from which they effectively become central place foragers while their cubs reside there.
Maternal presence at dens is critical for cubs’ physical and social development. In the Serengeti,
maternal attendance rates at dens are significantly positively correlated with cub growth rates,
92

which in turn predict cub survivorship to reproductive maturity (Hofer and East 1993a).
Additionally, it is at the communal den that cubs learn their dominance ranks. Spotted hyenas
live in fission-fusion social groups, called clans, that are structured according to a matrilineal
dominance hierarchy (Frank 1986a). Maternal presence at dens is critical for the process of rank
acquisition, as mothers reinforce cub’s appropriate aggressive behavior and intervene on behalf
of their offspring (Holekamp and Smale 1993, Engh et al. 2000).
Spotted hyenas also show remarkable flexibility in the face of anthropogenic disturbance
(Holekamp and Dloniak 2010). Where we conducted our study, in Kenya’s Masai Mara National
Reserve (MMNR), the human population along the borders of the reserve has grown
exponentially since the late 1990s, resulting in increased numbers of livestock grazing both day
and night inside the protected area (Green et al. in press). Direct conflict between hyenas and
livestock is common (Kolowski and Holekamp, 2006), resulting in high rates of hyenas being
speared, snared, and poisoned (Holekamp and Dloniak, 2010). Since 2005, humans have been
responsible for the majority of hyena deaths for which mortality sources could be determined
(Holekamp and Dloniak 2010). In addition to these lethal effects, increased human activity
along the borders of the MMNR has also had indirect effects on the behavior of adult spotted
hyenas. As human disturbance has increased, individuals have become more nocturnal,
expanded their home ranges, and increased the distances they travel to hunt (Boydston et al.
2003b). Compared with hyenas living in pristine areas of the MMNR, hyenas in the disturbed
area show a greater preference for protective vegetation (Kolowski and Holekamp 2009), are
more vigilant (Pangle and Holekamp 2010), shift their activity patterns to avoid interaction with
humans (Kolowski et al. 2007), travel greater distances in the mid-day heat (Ikime 2015), and
93

travel faster overall (Green 2015).
In a recent study, Greenberg and Holekamp (2017) found that behavioral differences
between juvenile hyenas living in high- and low-disturbance areas emerge early in life, when cubs
reside at communal dens, despite experiencing virtually no direct exposure to human
disturbance at dens. Using field experiments, they found that juveniles in high- and lowdisturbance areas showed consistent differences in personality traits, specifically, neophobia,
exploration, and boldness. When exposed to a novel object, juveniles in the high-disturbance
area were significantly less neophobic and more exploratory than juveniles in the lowdisturbance area. When exposed to a box they could enter to feed on a reward, juveniles in the
high-disturbance area were less bold than juveniles in the low-disturbance area. Greenberg and
Holekamp proposed that such behavioral differences might be shaped indirectly, by differences
in maternal behavior between high- and low-disturbance areas. For example, among Old World
Monkeys, which show many similarities to spotted hyenas in their maternal investment patterns,
life-histories, and social behavior, the offspring of mothers who are the most “rejecting” or
“laissez-faire” in their mothering style tend to be least neophobic and most exploratory
(Andrews and Rosenblum 1991, Fairbanks 1996). Moreover, female primates are less attentive
and protective of their offspring in more challenging or unpredictable environments (Andrews
and Rosenblum 1991). The degree to which mothers are “protective” versus “rejecting” has
been assessed in numerous primate species by measuring how much time a mother spends close
to offspring (in contact or within 1m) and how often a mother approaches and leaves her
offspring from this close proximity range. However, in other taxanomic groups, such as rodents,
it is mothers who are more protective, as measured by rates of grooming and the use of particular
94

nursing positions, that tend to have offspring who are less afraid to approach and explore new
aspects of their environment (Caldji et al. 1998).
Despite the many effects of human disturbance previously documented on the behavior
of spotted hyenas, Green et al. (2017) recently found that survivorship of juvenile hyenas was
actually greater in a highly disturbed area than in low-disturbance areas with the Masai Mara
National Reserve, Kenya, and as a result, their study clan in the disturbed area grew to be the
largest on record at 130 individuals. They demonstrated that this surprising demographic pattern
was likely due, at least in part, to a decline in lion densities in the high-disturbance area,
presumably also due to anthropogenic activity, and an increase in the hyenas’ depredation of
livestock. However, it is also possible that differential maternal behavior in high- and lowdisturbance areas contributes to the success of juveniles in the disturbed habitat.
Previous studies using VHF telemetry and direct observation indicated that
anthropogenic disturbance negatively affects maternal den attendance among spotted hyenas
living in a highly disturbed area of the MMNR. As human disturbance has increased, adult
females have been found significantly farther from the communal den (Boydston et al. 2003b),
and they arrive at communal dens later in the evening and leave them earlier in the morning than
in undisturbed areas (Kolowski et al. 2007). However, these studies were limited by the coarse
scale of the spatial data and the difficulty of obtaining data during the night.
Here we inquired how maternal behavior is shaped by human disturbance and tested the
hypothesis that differences in maternal behavior mediate the disturbance-related patterns in
juvenile personality at this site. If this hypothesis is correct, then we expected to see
unambiguous differences in maternal den attendance and behavior between high- and low95

disturbance areas. First, we compared maternal attendance rates at communal dens in high- and
low- disturbance areas of the MMNR using spatial data from spotted hyena mothers fitted with
GPS collars transmitting hourly locational data. We also examined the effects of offspring age,
litter size, prey density, and social rank on maternal den attendance rates. We predicted that
maternal den attendance would significantly decrease as cubs matured, and that mothers with
litters containing two cubs would show higher den attendance rates than mothers with
singletons. We also predicted that den attendance would be positively correlated with local prey
density, as has been found in other populations (Hofer and East 1993a), because spotted hyenas
travel significantly greater distances from the den when prey are scarce than when they are
abundant (Boydston, Kapheim, & Szykman, 2003; Hofer & East, 1993; Kolowski & Holekamp,
2009). Previous research on space use at our study site has shown that, among females with dendwelling cubs, there is no rank-related variation in how far females are found from the
communal den (Boydston et al. 2003a). However, we suspected that low-ranking mothers might
have lower attendance rates than high-ranking mothers in areas affected by human disturbance
because low-ranking mothers bear a higher cost of den visitation (White 2007); therefore, we
also assessed the effects of maternal social rank on den attendance.
To compare how mothers in high- and low-disturbance areas interacted with offspring at
communal dens, we conducted Focal Animal Surveys (FAS) on mother-offspring pairs.
Specifically, we determined the proportion of a mother’s FAS time when she was present at a
den with her cubs that she spent nursing her cubs, in close proximity to her cubs, or grooming
her cubs. We also determined to what extent mothers were responsible for maintaining close
proximity with their cubs. In a previous study (Greenberg et al. Ch.4), we analyzed naturally
96

occurring variation among spotted hyena mothers during a time period in which there was no
human disturbance. We found that mothers significantly altered their participation in these
activities based on cub age, litter size and time of day, but that social rank and prey density did
not appear to shape these maternal behaviors.
Finally, because we found that human disturbance does affect some aspects of maternal
behavior, we next inquired whether these maternal behaviors predict offspring survivorship and
whether it might also help explain the differences that have been documented in juvenile
survivorship in the high- and low- disturbance area. Greenberg et al. (chapter 4) previously
found that variation in how much time a mother spent nursing, grooming, or in close proximity
to offspring did not significantly predict either juvenile survivorship to reproductive maturity or
longevity. We therefore focused our analysis on whether maternal den attendance significantly
predicted offspring survivorship to reproductive maturity.

METHODS
Subjects & Study Populations
Study subjects were spotted hyenas inhabiting two areas within the Mara-Serengeti ecosystem in
southwestern Kenya. As described above, one area was undisturbed, whereas the other was
highly disturbed by anthropogenic activity (Figure 5.1). Hyenas living in the high-disturbance
area were members of the Talek West clan, which lives along the Northern border of the Masai
Mara National Reserve, near the burgeoning town of Talek. This clan has been continuously
monitored on a daily basis since 1988. Hyenas living in the low-disturbance area were members
of three clans living in the pristine Mara Conservancy. The Mara Conservancy is managed by a
97

Figure 5.1. Map of the territories of the four study clans monitored in the Masai Mara National
Reserve, Kenya. Study clans were Talek West (TW), Serena North (SN), Serena South (SS), and
Happy Zebra (HZ). The three clan in the shaded area are located in the Mara Conservancy.
Star indicates the town of Talek.
private non-profit organization that strictly prohibits cattle grazing or any human activity except
wildlife viewing from tour vehicles or hot-air balloons. These clans have been monitored daily
since 2008.
Spotted hyenas live in fission-fusion societies, each containing up to 130 individuals, and
each structured according to a strict matrilineal hierarchy (Frank 1986). Females are dominant to
males, which disperse from their natal clans at 2-5 years of age, and assume the lowest possible
ranks in the new clans into which they immigrate (Holekamp and Smale 1998b, van Horn et al.
2003). The social rank of adult females was calculated here on a yearly basis based on outcomes
of dyadic agonistic interactions used to construct an annual hierarchy for each clan (Martin &
Bateson 1988; Smale et al. 1993). To account for variation in clan size, ranks were standardized
from -1 to 1, with the most dominant individual having a standardized rank of 1.

98

Age was initially estimated when cubs were first observed, based on their appearance and
size (Holekamp et al. 1996), and sex was determined based on the shape of the glans of the erect
phallus (Frank et al. 1990). The reproductive state of adult females was continuously monitored
to determine periods of pregnancy and lactation, as described by Holekamp et al (1996). The
gestation period for spotted hyenas is 110 days (Kruuk, 1972). Conception dates were
determined by back-calculating from a cub’s date of birth, and based on observations of matings
and fresh tears in females’ phalluses, indicating recent parturition.

Data Collection
Maternal attendance at dens
GPS radio collars (Vectronic Aerospace, Berlin, Germany) were deployed in 2012 on 18 sexually
mature adult females. Of these, 10 females were members of two clans in the low-disturbance
area (5 from the Serena North clan and 5 from the Serena South clan), and 8 were members of
the Talek West clan, located in the high-disturbance area. In the low-disturbance area, 7 of the
collared females were high ranking and 3 were low ranking. In the high-disturbance area, 4 of
the collared females were high ranking and 4 were low ranking. To fit females with collars, we
anesthetized hyenas using Telazol (6.5 mg/kg) administered in a plastic dart fired from a CO2
powered rifle (Telinject Inc., Saugus, California). Collars were programmed to record GPS
locations hourly from 1600 h to 1000 h, and also once at 1300 h, for a total of 20 location fixes
per 24 hour period.
For each female, we defined periods in which she had a litter residing at the communal
den. We defined the start date for each period as the day on which the mother moved her litter
99

to the communal den from the natal den. It was not always possible to confirm the mother’s
presence at a natal den in the field because mothers often used dens in areas that were difficult
for our field team to access, and were often wary of humans during this period such that they
sometimes left the den when they saw our research vehicle approach. There was also often a lag
time in when we received GPS fixes due to spotty cell tower coverage. We identified the natal
den used by each mother by inspecting her GPS locations and looking for high concentrations
of points approximately one month prior to the cubs’ date of birth (see Figure 5.2). Then, using
a mother’s GPS locations along with our field notes, we determined the date on which each
mother brought her cub(s) to the active communal den.

Figure 5.2. Map of one female’s GPS locations while she was at a natal den, showing a high
concentration of points in close proximity to the natal den. Rings indicate 10m and 25m from
the natal den. Approximately 50% of her total GPS points for the one-month period were
within 10m of the natal den.

100

We defined the end date of the period during which a mother had den-dwelling cubs as
the date on which her cubs were eight months of age, which is the average age at which cubs
become independent of the den in the MMNR (Holekamp and Dloniak 2010). Not all 18
collared females had location fixes for the entire period during which their cubs lived at the
communal den because collars were sometimes deployed when females already had den-dwelling
litters, and some collars malfunctioned before the end of the cub’s den-dwelling period. If the
mother or cub died before the date on which cubs reached 8 months of age, we defined the end
date as the day on which the mother or cub was last seen alive.
We determined which communal den was actively being used by each clan at any given
time using archived behavioral records. Spotted hyenas may re-use dens within their territories,
and on average, change communal dens monthly (Boydston et al. 2006). During our routine
observation sessions, we typically visited the communal den being actively used by each clan
twice per day. Dens here were defined as being actively used when cubs were observed entering
and exiting the den entrance. If more than one den hole was being used within 200m, we
characterized the entire area as a “den complex.” In some cases, a clan was actively using two
dens that were more than 200m apart, or were clearly in transition as mothers moved their cubs
from one den to another. In these cases, we considered both dens to be actively used. If we
were unable to identify the active communal den via direct observation in the field, we
determined which den was being used by inspecting the GPS points of females with dendwelling cubs during that period. This was typically the case only for dens in areas that were
impossible for us to access due to rocks or dense vegetation.
We used QGIS to create a 100m buffer around each communal den. We then calculated
101

the proportion of each female’s total fixes that occurred within 100m of the active den during
the time period defined by the above start and end dates. In the case of den complexes, we
defined the mother as present at the den if her location fix was within 100m of any of the den
holes included in the complex.
Maternal behavior at dens
During our routine morning (0600-1000h) and evening (1600-2000h) observation periods, we
conducted focal animal surveys (FAS) on mother-cub pairs. We initiated a FAS when a mother
and cub were both present. We followed the pair for 30 minutes or until either the mother or
cub had been out of site for five minutes. During a FAS, we recorded all minutes during which
a mother was nursing, grooming, or in close proximity (within 1m or in contact) to the cub.
Additionally, whenever the mother-cub pair moved in or out of close proximity, we recorded
which individual initiated the change as an “event” (e.g. cub approach mother, mother approach
cub, cub leave mother, or mother leave cub). FAS were conducted opportunistically, but we
made every attempt to balance the dataset with regards to the mother’s social rank, mother’s
parity, and litter size. Although we had no den attendance (GPS) data from primiparous
mothers, we conducted FAS on primiparous mothers in addition to parous mothers. We also
conducted FAS on mother-cubs pairs in three study clans in the low-disturbance area, rather
than two. That is, in addition to collecting FAS data from the Serena North (SN) and Serena
South (SS) clans, we also collected them from mother-infant pairs in the Happy Zebra (HZ)
clan.

102

Statistical analysis of maternal den attendance and behavior
We used the glmmADMB package in R (R version 3.3.1; Development Core Team, 2015) to
model both variation in maternal attendance at dens and maternal behaviors during FAS. The
glmmADMB package allows for the incorporation of both random effects and zero-inflated
predictor variables. For maternal den attendance, we modeled the number of points during each
24 hour period when a mother was at the den while she had den-dependent cubs as our
response measure, and her total number of points for that 24 hr period (a maximum of 20 fixes)
as a logged-offset. We compared the fit of four types of count models; a negative binomial
model, a poisson model, a zero-inflated negative binomial model, and a zero-inflated poisson
model. The model with the lowest Akaike’s Information Criterion corrected for small sample
sizes (AICc) was chosen as our best fitting model. In each of these candidate models, we
included the mother’s ID as a random effect. As fixed effects, we included the cub’s age as a
continuous measure, the disturbance level (high or low), the mother’s social rank as a
continuous variable, the number of cubs the mother had alive at the den on that day (1 or 2),
and prey density. Prey density was calculated as the average density of prey (including livestock)
counted within 100m of multiple prey transects run bi-monthly within the territorial boundaries
of our study clans, during each two-week interval in which maternal behavior data were
collected. Prey density and cub age were centered and scaled for analyses.
We next asked whether disturbance mediated the relationship between other variables
and den attendance. To do this, we tested for a significant two-way interaction between
disturbance and each of our other variables of interest (litter size, maternal rank, and prey
density) by performing a likelihood ratio test between our model without interactions and each
103

of our models that included an interaction.
To model variation in maternal behavior, we calculated the total number of minutes
during which mother and cub were both present in each FAS as well as the number of FAS
minutes in which the mother was nursing her cub, in close proximity her cub, and grooming her
cub. For the number of minutes a mother spent in close proximity to her cub, we excluded any
minutes during which she was nursing the cub because she would necessarily be in contact with
the cub while nursing. We then constructed a separate model for each of these three maternal
behaviors, with the time spent engaged in the behavior as the response measure and the total
time mother and cub were present together as a logged offset. In each model, we included the
mother’s ID as a random effect, and disturbance as a fixed effect (high-disturbance or lowdisturbance). We also included the following fixed effects: time of day (morning or evening),
litter size (singleton or twins), cub sex, maternal rank, maternal parity (primiparous or
multiparous), and prey density. We again compared the fit of count models with four different
distributions (negative binomial, poisson, zero-inflated negative binomial, and zero-inflated
poisson), and report the model with the lowest AICc score. We excluded from analysis any FAS
in which the mother and cub were present together for less than 5 min, and only included FAS
in which the cub was less than 9 months old, the average age at which cubs begin to spend the
majority of their time away from the den in our study area.
We used our data on whether the mother or cub initiated changes in proximity to model
the extent to which mothers were responsible for maintaining proximity with their cubs. To
model how often mothers approached their offspring, we limited our analysis to sessions in
which either the mother or the cub had approached the other at least once during the FAS. We
104

used the number of times a mother approached her cub in the FAS as our response measure and
included the total number of approaches by either mother or cub in the FAS as a predictor
variable. To determine the extent to which mothers left their cubs, we limited our analysis to
sessions in which the mother and cub left one another at least once, used the number of times a
mother left her cub within a FAS as our response measure, and included the total count of times
the mother or cub left one another as a predictor variable. For each of our maternal behaviors,
we again asked whether disturbance mediated the relationship between the behavior and our
other variables. To do this, we tested for a significant two-way interaction between disturbance
and each of our other variables of interest (litter size, maternal rank, cub sex, maternal parity,
and prey density) by performing a likelihood ratio test between our model without interactions
and each of our models that included an interaction. We included mother ID as a random effect
and cub age and time of day as fixed effects in each model.
glmmADMB is a relatively new R package and model diagnostic methods are still being
developed. We therefore focused our evaluation of model fit on the residual deviance. We also
evaluated overdispersion using the theta values provided in model outputs. We used Variance
Inflation Factors (VIF) to assess multi-colinearity of predictor variables in all of our models. No
variable had a VIF above 3, indicating that there was no significant multi-colinearity among
variables in our models.

Maternal den attendance and offspring survivorship
We used our demographic records to determine whether the offspring of mothers fitted with
GPS collars survived to reproductive maturity, which occurs at 24 months of age in spotted
105

hyenas. All youngsters could have reached this age by the time our analysis was conducted. For
each juvenile, we calculated the proportion of its mothers’ location fixes that were at the den
when the cub was 30-120 days old, or roughly the first three months in which a cub resided at
the communal den. We excluded from analysis any juveniles whose mothers’ collars did not
transmit at least 10 location fixes during each of the three months.

RESULTS
Maternal attendance at dens
Preliminary analysis showed that, within the low disturbance area, hyenas in the two clans did
not show significantly different den attendance rates (zero inflated negative binomial model,
B=0.070, SE=0.187, P = 0.707). Therefore, we did not include clan as a factor in subsequent
analyses of spatial data.
Our analysis included a total of 40,317 points, with an average of 14.39 points per female
per day. The dataset included an average of 4.94 months of data from each mother per litter
(range 0.57-8.26 months). A zero-inflated negative binomial distribution best fit our den
attendance data (Table 5.1). Den attendance was significantly higher among mothers in lowthan high-disturbance areas (P = 0.002; Figure 5.3). Den attendance significantly decreased as
cubs matured (P < 0.001; Figure 5.3), high-ranking mothers had significantly lower den
attendance than low-ranking mothers (P = 0.024), and we also found a significant positive
association between maternal attendance and prey density (P < 0.001). There was a trend for
mothers to show higher den attendance when they had twins than singletons (P = 0.057).
We observed a marginally significant interaction between disturbance and litter size (2(1)
106

(Intercept)
Cub Age
Disturbance
Social Rank
Prey Density
Litter Size

Odds Ratio
0.351
0.765
0.625
0.804
1.077
1.096

Estimate
-1.048
-0.267
-0.470
-0.218
0.074
0.092

Std. Error
0.103
0.019
0.151
0.096
0.013
0.048

P
<0.001
<0.001**
0.002**
0.024**
<0.001**
0.057 *

	

Table 5.1. Results of zero-inflated negative binomial model explaining variation in maternal den
attendance. ** indicates P <0.05. * indicates P <0.1.

Figure 5.3. Plot showing mean maternal den attendance as cubs mature in low- and highdisturbance areas. Data points represent monthly means of mothers’ daily den attendance (the
proportion of mothers’ total points that were within 100m of the communal den).

107

= 2.96, P = 0.085). Looking at mothers of singletons versus twins separately indicated that the
magnitude of the disturbance effect was greater among mothers of singletons (odds ratio =
0.722, B = -0.326, SE = 0.139, P = 0.019) than of twin litters (odds ratio = 0.536, B = -0.623,
SE = 0.212, P = 0.003). A likelihood ratio test indicated a significant interaction between
disturbance and prey density (2(1) = 34.72, P < 0.001). However, when we removed data points
from one out-lying two-week interval with high prey densities in one clan’s territory, this effect
was no longer significant (2(1) = 2.68, = 0.102). The interaction between disturbance and rank
was not significant (2(1) = 1.3, P = 0.254).

Maternal behavior at dens
Our analysis included a total of 333 FAS collected on 43 mothers (24 from low disturbance
areas; 19 from the high disturbance area) and 62 cubs (33 from low disturbance areas; 29 from
the high disturbance area). An average of 5.32 FAS (range 1-16) were conducted on each
mother-offspring pair. Preliminary analysis showed that, within the undisturbed area, hyenas in
the three study clans did not show significantly different rates of nursing, grooming, or being in
close proximity to offspring (Table 5.2). Therefore, we did not include clan as a factor in
subsequent analyses.
A zero-inflated negative binomial model best explained variation in nursing among
mothers (Table 5.3a). Mothers in the high-disturbance area spent significantly more of their
time during FAS nursing their cubs than did mothers in the low-disturbance areaa (P < 0.001;
Figure 5.4a). Mothers also nursed their cub significantly more as the cub matured (P < 0.001),
and significantly less as prey density increased (P < 0.029). There were non-significant trends
108

Figure 5.4. Plot showing mean proportion of time mothers spent a) nursing and b) in close
proximity to cubs as cubs mature in low- and high-disturbance areas. Data points represent
monthly means of mothers’ behavior.

109

Estimate

Std. Error

P

Nursing
North - Happy Zebra
South - Happy Zebra
North - South

-0.02
-0.08
-0.06

0.15
0.20
0.25

0.98
0.91
0.97

Close Proximity
North - Happy Zebra
South - Happy Zebra
North - South

-0.12
0.15
0.27

0.29
0.37
0.48

0.91
0.91
0.84

Grooming
North - Happy Zebra
South - Happy Zebra
North - South

-0.30
-0.06
0.25

0.57
0.71
0.94

0.85
1.00
0.96

	

Table 5.2. Tukey pairwise comparisons of rates of nursing, grooming, and close proximity
among mothers living in the three clans in the undisturbed area.

for high-ranking mothers to spend more time nursing cubs than did low-ranking mothers (P <
0.060), and mothers of female cubs to spend more time nursing than did mothers of male cubs
(P < 0.071). Likelihood ratio tests comparing models for nursing with and without interaction
terms indicated a significant interaction between disturbance and maternal rank (2(1) = 5.2, P =
0.023). Looking at high-ranking mothers and low-ranking mothers separately indicated that,
although mothers of all ranks spent more FAS time nursing in high- than low-disturbance areas,
the magnitude of the disturbance effect was larger among low-ranking mothers (odds ratio =
1.39, B = 0.331, SE = 0.164, P = 0.04) than high-ranking mothers (odds ratio = 1.36, B = 0.306,
SE = 0.077, P < 0.001). All other interactions were non-significant (disturbance X cub sex (2(1)
= 0.52, P = 0.471; disturbance X litter size (2(1) = 0.4, P = 0.527; disturbance X litter size ((2(1)
110

Odds Ratio

Estimate

Std. Error

P
	

A. Nursing
(Intercept)
Cub age
Time of day
Disturbance
Litter Size
Cub sex
Maternal rank
Maternal parity
Prey density

0.679
1.136
1.053
1.278
0.993
0.909
0.926
1.102
1.000

-0.39
0.13
0.05
0.25
-0.01
-0.10
-0.08
0.10
0.00

0.07
0.03
0.05
0.05
0.05
0.05
0.04
0.06
0.00

<0.001
<0.001**
0.340
<0.001**
0.895
0.071*
0.060*
0.116
0.029**

B. Close Proximity
(Intercept)
Cub age
Time of day
Disturbance
Litter Size
Cub sex
Maternal rank
Maternal parity
Prey density

0.255
0.687
0.775
0.554
0.855
1.282
1.017
1.169
1.000

-1.37
-0.38
-0.26
-0.59
-0.16
0.25
0.02
0.16
0.00

0.21
0.07
0.13
0.15
0.16
0.17
0.12
0.19
0.00

<0.001
<0.001**
0.055 *
<0.001**
0.331
0.133
0.886
0.418
0.991

C. Grooming
(Intercept)
Cub age
Time of day
Disturbance
Litter Size
Cub sex
Maternal rank
Maternal parity
Prey density

0.087
0.600
1.226
1.243
0.379
0.596
0.653
0.699
1.000

-2.44
-0.51
0.20
0.22
-0.97
-0.52
-0.43
-0.36
0.00

0.40
0.12
0.25
0.34
0.34
0.34
0.25
0.43
0.00

<0.001
<0.001**
0.419
0.520
0.004**
0.131
0.094*
0.404
0.152

Table 5.3. Results of models explaining variation in how much FAS time mothers spent: a)
nursing cubs (zero-inflated negative binomial model), b) in close proximity to cubs (negative
binomial model), and c) grooming cubs (negative binomial model). ** indicates P <0.05. *
indicates P <0.1.

111

= 0.4, P = 0.527; disturbance X maternal parity (2(1) = 0.12, P = 0.729; disturbance X prey
density (2(1) = 0.06, P = 0.807).
A negative binomial model best explained variation in how often mothers were in close
proximity to their cub (s; Table 5.3b). Mothers in the high-disturbance area spent significantly
less FAS time in close proximity than did mothers in the low-disturbance area (P < 0.001; Figure
5.4b). Mothers also spent significantly less time in close proximity to cubs as they matured (P <
0.001), and mothers tended to spend more time in close proximity to cubs in the mornings than
in the evenings (P < 0.055).
Likelihood ratio tests comparing models for FAS time spent in close proximity to cubs
with and without interaction terms indicated a significant interaction between disturbance and
maternal parity (2(1) = 4.77, P = 0.029). Looking at primiparous and multiparous mothers
separately indicated that the magnitude of the disturbance effect was greater among multiparous
mothers (odds ratio = 0.66, B = -0.41, SE = 0.15, P = 0.006) than primiparous mothers (odds
ratio = 0.29, B = -1.25, SE = 0.43, P = 0.004). There was also a marginally significant
interaction between disturbance and maternal rank (2(1) = 3.13, P = 0.077). Looking at highranking and low-ranking mothers separately indicated that the disturbance effect was significant
among high-ranking mothers (odds ratio = 0.79, B = -0.65, SE = 0.20, P = 0.001), but not lowranking mothers (odds ratio = 0.52, B = -0.24, SE = 0.29, P = 0.406). All other interactions
were non-significant (disturbance X prey density (2(1) = 0.50, P = 0.481; disturbance X cub sex 2
(1) = 0.14, P = 0.710; disturbance X litter size (2(1) = -0.22, P = 0.99).
A negative binomial model best explained variation in grooming among mothers (Table
5.3c). There was no significant difference between high- and low-disturbance areas with respect
112

to how much FAS time mothers spent grooming their cubs (P = 0.520). Mothers spent
significantly less time grooming cubs as they matured (P < 0.001), and significantly more time
grooming singletons than cubs in twin litters (P = 0.004). There was a non-significant trend for
low-ranking mothers to spend more time grooming cubs than high-ranking mothers (P = 0.094).
Likelihood ratio tests indicated that no interactions with disturbance were significant
(disturbance X prey density 2(1) = 1.52, P = 0.218; disturbance X maternal parity 2(1) = 0.69, P
= 0.407; disturbance X maternal rank 2(1) = 0.62, P = 0.432; disturbance X litter size 2(1) = 0.49,
P = 0.499; disturbance X cub sex 2(1) = 0.10, P = 0.747).
A zero-inflated negative binomial model based on 160 FAS best explained variation in
how often a mother approached her cub. Only the total number of approaches in the survey
significantly predicted how often mothers approached cubs (Table 5.4). A poisson model that
included 156 surveys best explained variation in how often a mother left her cub (Table 5.5).
Mothers in the high-disturbance area left their cubs significantly more often than did mothers in
the low-disturbance area (P < 0.001), and mothers of females showed higher rates of leaving
their cubs than did mothers of males (P = 0.053). Due to the small size of these datasets, we did
not test for significant interactions between disturbance and our other predictor variables.

Maternal den attendance and offspring survivorship
Twenty-one individuals met our criteria for inclusion in the analysis. Of these 21 individuals, 4
did not survive to reproductive maturity (3 in the high-disturbance area and 1 in the lowdisturbance area). Maternal attendance at dens did not significantly predict offspring
survivorship (B = 9.43, SE = 7.46, P = 0.206).
113

(Intercept)
Total number of
approaches
Cub age
Time of day
Disturbance
Litter size
Cub sex
Maternal rank
Maternal parity
Prey density

Odds Ratio
0.018

Estimate
-4.05

Std. Error
0.47

1.132
1.252
0.995
1.296
1.183
0.985
1.099
1.032
1.165

0.12
0.23
-0.01
0.26
0.17
-0.02
0.09
0.03
0.15

0.04
0.15
0.23
0.31
0.31
0.32
0.23
0.47
0.13

P
<0.001
<0.001**
0.146
0.984
0.404
0.585
0.961
0.683
0.947
0.231

	

Table 5.4. Results of a zero-inflated negative binomial model explaining variation in how often
mothers approached their cubs to within 1m. ** indicates P <0.05. * indicates P <0.1.

(Intercept)
Total number of
departures
Cub age
Time of day
Disturbance
Litter size
Cub sex
Maternal rank
Maternal parity
Prey density

Odds Ratio
0.024

Estimate
-3.74

Std. Error
0.30

1.086
1.559
1.155
2.978
0.879
0.673
0.783
1.003
0.893

0.08
0.44
0.14
1.09
-0.13
-0.40
-0.25
0.003
-0.11

0.09
0.10
0.17
0.22
0.23
0.21
0.17
0.28
0.10

P
<0.001
0.003**
<0.001**
0.398
<0.001**
0.568
0.053**
0.143
0.991
0.272

	

Table 5.5. Results of a poisson model explaining variation in how often mothers left their cubs.
** indicates P <0.05. * indicates P <0.1.

114

DISCUSSION
Our study demonstrates that, although spotted hyenas show a remarkable ability to persist in
areas with human-wildlife conflict and disturbance, such conditions present a challenge to
mothers, who show different patterns of den attendance and maternal care in high and low
disturbance areas. When den attendance rates were averaged over the time period during which
cubs resided at communal dens, we found that mothers in high-disturbance areas had den
attendance rates 12% lower than mothers in low-disturbance areas. Previous studies have
demonstrated that hyenas living in disturbed areas have larger home ranges and travel farther to
hunt than hyenas in undisturbed areas (Boydston et al. 2003b). Although this behavioral change
may allow hyenas to effectively hunt while avoiding confrontation with humans, our results
likely demonstrate a cost to this strategy. In the high-disturbance area, mothers spent a greater
proportion of their time at the den nursing their cubs than did mothers in the low-disturbance
areas. Mothers in the high-disturbance area also showed differences in the ways in which they
interacted with their cubs, spending less time close to their cubs and more often leaving their
cubs than mothers in the low-disturbance area. Although our maternal den attendance data were
collected throughout the 24-hr cycle, we could not conduct FAS in total darkness. Therefore,
we cannot rule out the possibility that anthropogenic disturbance affected the circadian pattern
of when mothers engage in the behaviors we studied rather than the overall amount of time that
mothers spent engaging in these activities.
Taken together, our findings are consistent with the hypothesis that maternal behavior is
mediating the disturbance-related differences in juvenile personality documented by (Greenberg
and Holekamp 2017). Here, we found that, in the area where spotted hyena cubs are less
115

neophobic and more exploratory, mothers also show low rates of attendance at dens, and, when
present, spend less time close to offspring and leave their cubs at higher rates. In other species,
such a pattern of maternal care has sometimes been interpreted as a “low investment” maternal
strategy (Hauser and Fairbanks 1988, Champagne 2011, Fairbanks and Hinde 2012). In the lowdisturbance area on the other hand, mothers are able to invest more, showing higher rates of
den attendance and time spent near cubs, and as predicted, cubs in this area are more neophobic
and less exploratory than cubs in the high-disturbance area.
A recent study documenting a trade-off in the quality and quantity of spotted hyena milk
may shed light on our results regarding effects of disturbance and rank on maternal behavior.
We found that mothers in high-disturbance areas, who otherwise show low investment in
offspring, spent a greater proportion of their time nursing than mothers in the low-disturbance
area. This may allow mothers in high-disturbance areas to compensate for their lower den
attendance rates and is consistent with the finding that spotted hyena mothers in the Serengeti
who visit the den less often also engage in longer nursing bouts (Hofer et al. 2016). However, a
mother’s time spent nursing does not take into account variation in milk quality and can be a
surprisingly poor indicator of actual milk intake by offspring (Cameron 1998). Hofer et al.
(2016) analyzed the nutritional content of spotted hyena milk and weighed cubs to measure milk
intake. They found a trade-off between milk quality and quantity; mothers transferred high
quantities of low-quality milk or small quantities of high-quality milk. Therefore, it would be
interesting to determine whether the consumption of more livestock by hyenas in the highdisturbance (Green et al. 2017) affects the nutritional quality of hyena milk. If consumption of
livestock causes the milk of mothers in the high-disturbance area to be of lower quality than the
116

milk of mothers in the low-disturbance area, this might explain our finding that mothers in the
high-disturbance area spent a greater proportion of their time nursing than did mothers in the
low-disturbance area.
Examining variation in milk quality as it relates to a mother’s social rank could also help
to shed light on our surprising finding that high-ranking mothers showed lower den attendance
rates than low-ranking mothers. We predict that high-ranking females, who have priority access
to food, would have more nutritious milk than low-ranking females, who are typically in poorer
condition. If this is the case, low-ranking mothers may provide their offspring with a relatively
large volume of low-quality milk during frequent visits to the den whereas high-ranking mothers
may provide offspring with a lower volume of high-quality milk at less frequent visits to the den.
Our findings also demonstrate that anthropogenic disturbance may interact with effects
of other socio-ecological variables on maternal behavior. High-ranking and more experienced
mothers appear better able to adopt flexible maternal strategies in response to disturbance than
mothers who are low-ranking or inexperienced, and typically also in poorer condition. Whereas
high-ranking mothers spent significantly less time close to cubs in high- than low-disturbance
areas, this measure did not differ for low-ranking mothers between low- and high-disturbance
areas. One interpretation of this finding is that high-ranking mothers can afford to be less
protective of their offspring in high-disturbance areas, whereas low-ranking mothers, who
experience a more threatening social environment, cannot. Similarly, whereas all mothers in
high-disturbance areas spent less time close to cubs than did mothers in low-disturbance, the
magnitude of this disturbance effect was greater among multiparous mothers than among
primiparous mothers. Again, it might be that maintaining close proximity with offspring is
117

particularly important for inexperienced mothers, who cannot alter their behavior as flexibly in
response to disturbance. Although we do not know whether these maternal behaviors take time
away from other important activities, our findings suggest that disturbance may
disproportionately affect certain individuals who are less able to flexibly alter their maternal
behavior.
Whereas our results support the hypothesis that maternal behavior is mediating
differences in juvenile personality, they do not support the hypothesis that these differences are
driving Green et al.’s (2017) finding that juvenile survivorship is higher in the high-disturbance
area than in the low-disturbance area. First, den attendance and the maternal behaviors we
monitored (Chapter 1) were not significantly correlated with offspring survival. Second,
mothers in high-disturbance areas showed lower, not higher, rates of den attendance and time
spent close to cubs than mothers in low-disturbance areas. Therefore an increase in maternal
investment cannot explain higher survivorship in the high- than low-disturbance areas. This
result provides support for a recent hypothesis suggesting that spotted hyenas are able to persist
in areas of high anthropogenic disturbance as long as competition with lions is reduced and prey
densities are high (M’soka et al. 2016). That spotted hyena mothers are able to maintain high
cub survivorship in disturbed areas even with significantly reduced investment in cubs suggests
that these ecological factors alone may predict which areas are habitable for spotted hyenas in
landscapes increasingly affected by human activity.
Our study raises important questions regarding the role of maternal effects in shaping
population responses to anthropogenic disturbance and the consequences of these maternal
effects for mothers and offspring living in disturbed areas. Numerous researchers have
118

hypothesized that a low investment maternal strategy, such as the strategy that appears to be
adopted by spotted hyena mothers in high disturbance areas, could provide a signal to offspring
of a challenging or unpredictable adult environment, leading offspring to develop traits that will
be adaptive in these environments, such as a fast life-history and enhanced risk-taking behavior.
However, it remains controversial how often maternal effects function as adaptive signals of
environmental quality, allowing offspring to optimize their fitness, and some researchers have
argued that selection more often acts on maternal effects to maximize a female’s fitness across
her lifespan (Marshall and Uller 2007). Our results indicate that the maternal behaviors we
studied are not associated with offspring survivorship. Similarly, Greenberg and Holekamp
(2017) did not find evidence that the juvenile personality traits observed in the high-disturbance
area conferred any type of survivorship benefit. Future research should assess whether there are
longer-term impacts of these disturbance-related changes in maternal and offspring behavior on
offspring longevity, reproductive success, or life-history. However, it is also possible that any
potential effects of disturbance-related changes in maternal behavior on offspring are masked by
the effects of other changes in the area, such as the decline in lion densities, and increased
depredation of livestock by hyenas.

119

REFERENCES

120

REFERENCES

Abbott, D. H., E. B. Keverne, F. B. Bercovitch, C. A. Shively, S. P. Mendoza, W. Saltzman, C.
T. Snowdon, T. E. Ziegler, M. Banjevic, J. Garland T, and R. M. Sapolsky. 2003. Are
subordinates always stressed? a comparative analysis of rank differences in cortisol levels
among primates. Hormones and Behavior 43:67–82.
Almasi, B., P. Béziers, A. Roulin, and L. Jenni. 2015. Agricultural land use and human presence
around breeding sites increase stress-hormone levels and decrease body mass in barn owl
nestlings:89–101.
Altmann, J. 1980. Baboons Mothers and Infants. The University of Chicago Press, Chicago.
Amstrup, S. C. 1994. Human Disturbances of Denning Polar Bears in Alaska 46:246–250.
An, Y. S., B. Kriengwatana, A. E. Newman, E. A. MacDougall-Shackleton, and S. A.
MacDougall-Shackleton. 2011. Social rank, neophobia and observational learning in blackcapped chickadees. Behaviour 148:55–69.
Anderson, J. L. 1981. The re-establishment and management of a lion (Panthera leo) population
in Zululand, South Africa. Biological Conservation 19:107–117.
Andrews, M. W., and L. A. Rosenblum. 1991. Attachment in monkey infants raised in variableand low-demand environments. Child Development 62:686–693.
Andrews, M. W., and L. A. Rosenblum. 2010. The Development of Affiliative and Agonistic
Social Patterns in Differentially Reared Monkeys. Child Development:1–8.
Archard, G. A., and V. A. Braithwaite. 2010. The importance of wild populations in studies of
animal temperament. Journal of Zoology 281:149–160.
Arlet, M. E., L. . Veromann, R. Mand, and A. Lemasson. 2016. Call Rates of Mothers Change
With Maternal Experience and With Infant Characteristics in Free-Ranging Gray-Cheeked
Mangabeys. American journal of primatology 991:983–991.
Arroyo, B., and M. Razin. 2005. Effect of human activities on bearded vulture behaviour and
breeding success in the French Pyrenees 8:0–8.
Bahr, N. I., C. R. Pryce, M. Döbeli, and R. D. Martin. 1998. Evidence from urinary cortisol that
maternal behavior is related to stress in gorillas. Physiology & Behavior 64:429–437.
Baker, P. J., L. Boitani, S. Harris, G. Saunders, and P. C. L. White. 2008. Terrestrial carnivores
and human food production: impact and management. Mammal Review 38:123–166.

121

Bardi, M., and M. A. Huffman. 2002. Effects of Maternal Style on Infant Behavior in Japanese
Macaques ( Macaca fuscata ). Developmental Psychobiology 41:364–372.
Barker, D. J. P., J. G. Eriksson, T. Forsén, and C. Osmond. 2002. Fetal origins of adult disease:
strength of effects and biological basis. International Journal of Epidemiology 31:1235–
1239.
Barnett, S. A. 1958. Experiments on “neophobia” in wild and laboratory rats. British Journal of
Psychology 49:195–201.
Barr, D.J., Levy, R., Scheeperes, C., Tily, H. J. 2014. Random effects structure for confirmatory
hypothesis testing: Keep it maximal. Journal of Memory and Language 68:1–43.
Bateman, P. W., and P. A. Fleming. 2012. Big city life: carnivores in urban environments.
Journal of Zoology 287:1–23.
Bateson, P., and P. D. Gluckman. 2011. Plasticity, Robustness, Development and Evolution.
Cambridge University Press, New York, NY.
Baugh, A. T., S. V Schaper, M. Hau, J. F. Cockrem, and P. De Goede. 2012. Corticosterone
responses differ between lines of great tits (Parus major) selected for divergent
personalities. General and Comparative Endocrinology 175:488–494.
Beehner, J., and C. McCann. 2008. Seasonal and altitudinal effects on glucocorticoid metabolites
in a wild primate (Theropithecus gelada). Physiology & Behavior 95:508–514.
Beery, A. K., and D. D. Francis. 2011. Adaptive significance of natural variations in maternal
care in rats: A translational perspective. Neuroscience & Biobehavioral Reviews 35:1552–
1561.
Bell, A. M., and J. A. Stamps. 2004. Development of behavioural differences between individuals
and populations of sticklebacks, Gasterosteus aculeatus. Animal Behaviour 68:1339–1348.
Benhaiem, S., H. Hofer, M. Dehnhard, J. Helms, and M. L. East. 2013. Sibling competition and
hunger increase allostatic load in spotted hyaenas. Biology Letters 9:20130040.
Benhaiem, S., H. Hofer, S. Kramer-Schadt, E. Brunner, and M. L. East. 2012. Sibling rivalry:
training effects, emergence of dominance and incomplete control. Proceedings of the Royal
Society B: Biological Sciences 279:3727–3735.
Benson-Amram, S., and K. E. Holekamp. 2012. Innovative problem solving by wild spotted
hyenas. Proceedings of the Royal Society B: Biological Sciences 279:4087–4095.
Bentley-Condit, V. K. 2003. Sex Differences in Captive Olive Baboon Behavior During the First
Fourteen Days of Life. International Journal of Primatology 24:1093–1112.

122

Bergh, V. Den, B. R. H. Van Den Bergh, and A. Marcoen. 2004. High antenatal maternal anxiety
is related to ADHD symptoms , externalizing problems , and anxiety in 8- and 9-year-olds
High Antenatal Maternal Anxiety Is Related to ADHD Symptoms , Externalizing
Problems , and Anxiety in 8- and 9-Year-Olds. Child Development 75:1085–1097.
Bergmüller, R., and M. Taborsky. 2010. Animal personality due to social niche specialisation.
Trends in Ecology & Evolution 25:504–511.
Berman, C. M. 1990. Consistency in Maternal-Behavior Within Families of Free-Ranging
Rhesus-Monkeys - an Extension of the Concept of Maternal Style. American Journal of
Primatology 22:159–169.
Bernardo, J. 1996. Maternal Effects in Animal Ecology. American Zoologist 36:83–105.
Biondi, L. M., M. S. Bó, and A. I. Vassallo,. 2010. Inter-individual and age differences in
exploration , neophobia and problem-solving ability in a Neotropical raptor (Milvago
chimango). Animal Cognition 13:701–710.
Biro, P. A., and D. J. Booth. 2009. Extreme boldness precedes starvation mortality in six-lined
trumpeter (Pelates sexlineatus ). Hydrobiologia 635:395–398.
Blehar, M. C., A. F. Lieberman, and M. D. S. Ainsworth. 1977. Early Face-to-Face Interaction
and Its Relation to Later Infant-Mother Attachment. Child Development 48:182–194.
Blumstein, D. T., E. Fernández-Juricic, P. A. Zollner, and S. C. Garity. 2005. Inter-specific
variation in avian responses to human disturbance. Journal of Applied Ecology 42:943–953.
Bonier, F., P. R. Martin, and I. T. Moore. 2009a. Do baseline glucocorticoids predict fitness?
Trends in Ecology & Evolution 24:634–642.
Bonier, F., I. T. Moore, P. R. Martin, and R. J. Robertson. 2009b. The relationship between
fitness and baseline glucocorticoids in a passerine bird. General and Comparative
Endocrinology 163:208–213.
Boydston, E. E., K. M. Kapheim, and K. E. Holekamp. 2006. Patterns of den occupation by the
spotted hyaena (Crocuta crocuta). Journal of African Ecology 44:77–86.
Boydston, E., K. Kapheim, and M. Szykman. 2003a. Individual variation in space use by female
spotted hyenas. Journal of Mammalogy 84:1006–1018.
Boydston, E., K. Kapheim, H. Watts, M. Szykman, and K. E. Holekamp. 2003b. Altered
behaviour in spotted hyenas associated with increased human activity. Animal Conservation
6:207–219.
Breuner, C. W., B. Delehanty, and R. Boonstra. 2012. Evaluating stress in natural populations of

123

vertebrates: total CORT is not good enough. Functional Ecology 27:24–36.
Brown, C., and K. N. Laland. 2002. Social learning of a novel avoidance task in the guppy:
conformity and social release. Animal Behaviour 64:41–47.
Budaev, S. V. 1997. The statistical analysis of behavioural latency measures. International Society
for Comparative Psychology Newsletter 14:1–4.
Burkanov, V., E. Gurarie, A. Altukhov, E. Mamaev, P. Permyakov, A. Trukhin, J. Waite, and T.
Gelatt. 2011. Environmental and biological factors influencing maternal attendance
patterns of Steller sea lions ( Eumetopias jubatus ) in Russia. Journal of Mammalogy
92:352–366.
Busch, D. S., and L. S. Hayward. 2009. Stress in a conservation context: A discussion of
glucocorticoid actions and how levels change with conservation-relevant variables.
Biological Conservation 142:2844–2853.
Cadieu, J. C., N. Cadieu, and J. Lauga. 1995. Local enhancement and seed choice in the juvenile
canary , Serinus canarius. Animal Behaviour 50:793–800.
Caldji, C., B. Tannenbaum, S. Sharma, D. Francis, P. M. Plotsky, and M. J. Meaney. 1998.
Maternal care during infancy regulates the development of neural systems mediating the
expression of fearfulness in the rat. Proceedings of the National Academy of Sciences of
the United States of America 95:5335–5340.
Cameron, E. Z. 1998. Is suckling behaviour a useful predictor of milk intake? A review. Animal
Behaviour 56:521–532.
Cameron, N., A. Del Corpo, J. Diorio, K. McAllister, S. Sharma, and M. J. Meaney. 2008a.
Maternal Programming of Sexual Behavior and Hypothalamic-Pituitary-Gonadal Function
in the Female Rat. PLoS ONE 3:e2210.
Cameron, N. M., D. Shahrokh, A. Del Corpo, S. K. Dhir, M. Szyf, F. A. Champagne, and M. J.
Meaney. 2008b. Epigenetic Programming of Phenotypic Variations in Reproductive
Strategies in the Rat Through Maternal Care. Journal of Neuroendocrinology 20:795–801.
Carere, C., T. G. G. Groothuis, E. Möstl, S. Daan, and J. M. Koolhaas. 2003. Fecal
corticosteroids in a territorial bird selected for different personalities: daily rhythm and the
response to social stress. Hormones and Behavior 43:540–548.
Caro, T. 1994. Cheetahs of the Serengeti Plains: Group living in an asocial species. University of
Chicago Press, Chicago.
Caro, T. 1999. The behaviour-conservation interface. Trends in Ecology & Evolution.
14:366-369.

124

Cavigelli, S. A., and M. K. McClintock. 2003. Fear of novelty in infant rats predicts adult
corticosterone dynamics and an early death. Proceedings of the National Academy of
Sciences 100:16131–16136.
Champagne, F. A. 2011. Maternal imprints and the origins of variation. Hormones and Behavior
60:4–11.
Chapman, B. B., L. J. Morrell, J. Krause, C. Biology, and I. Fisheries. 2010. Unpredictability in
food supply during early life influences boldness in fish. Behavioral Ecology 21:501–506.
Ciucci, P., L. Boitani, F. Francisci, and G. Andreoli. 1997. Home range, activity and movements
of a wolf pack in central Italy. Journal of Zoology 243:803–819.
Clutton-Brock, T. H. 1991. The evolution of parental care. Princeton University Press,
Princeton, N.J.
Clutton-Brock, T. H., S. D. Albon, and F. E. Guinness. 1981. Parental investment in male and
female offspring in polygynous mammals. Nature 289:487–489.
Cooper, S. M., K. E. Holekamp, and L. Smale. 1999. A seasonal feast: long-term analysis of
feeding behaviour in the spotted hyaena (Crocuta crocuta). African Journal of Ecology
37:149–160.
Coplan, J. D., E. L. Smith, M. Altemus, B. A. Scharf, M. J. Owens, C. B. Nemeroff, J. M.
Gorman, and L. A. Rosenblum. 2001. Variable foraging demand rearing: sustained
elevations in cisternal cerebrospinal fluid corticotropin-releasing factor concentrations in
adult primates. Biological psychiatry 50:200–204.
Coplan, J. R., M. W. Andrews, L. A. Rosenblum, M. . Owens, S. Friedman, J. M. Gorman, and
C. B. Nemeroff. 1996. Persistent elevations of cerebrospinal fluid concentrations of
corticotropin-releasing factor in adult nonhuman primates exposed to early-life stressors:
Implications for the pathophysiology of mood and anxiety disorders. PNAS 93:1619–1623.
Creel, S. 2001. Social dominance and stress hormones. Trends in Ecology & Evolution 16:491–
497.
Crespi, E. J., T. D. Williams, T. S. Jessop, and B. Delehanty. 2012. Life history and the ecology
of stress: how do glucocorticoid hormones influence life-history variation in animals?
Functional Ecology 27:93–106.
Crino, O. L., and C. W. Breuner. 2015. Developmental stress: evidence for positive phenotypic
and fitness effects in birds. Journal of Ornithology:1–10.
Cyr, N. E., and L. Michael Romero. 2007. Chronic stress in free-living European starlings
reduces corticosterone concentrations and reproductive success. General and Comparative

125

Endocrinology 151:82–89.
Cyr, N. E., and L. M. Romero. 2009. Identifying hormonal habituation in field studies of stress.
General and Comparative Endocrinology 161:295–303.
Dall, S. R. X., A. I. Houston, and J. M. McNamara. 2004. The behavioural ecology of
personality: consistent individual differences from an adaptive perspective. Ecology Letters
7:734–739.
Dantzer, B., Q. E. Fletcher, R. Boonstra, and M. J. Sheriff. 2014. Measures of physiological
stress: a transparent or opaque window into the status, management and conservation of
species? Conservation Physiology 2:1–18.
Dantzer, B., A. G. Mcadam, R. Palme, M. M. Humphries, S. Boutin, and R. Boonstra. 2011.
Maternal androgens and behaviour in free-ranging North American red squirrels. Animal
Behaviour 81:469–479.
Dantzer, B., A. E. M. Newman, R. Boonstra, R. Palme, S. Boutin, M. M. Humphries, and A. G.
McAdam. 2013. Density Triggers Maternal Hormones That Increase Adaptive Offspring
Growth in a Wild Mammal. Science 340:1215–1217.
Darrow, P. A., and J. A. Shivik. 2009. Bold, shy, and persistent: Variable coyote response to light
and sound stimuli. Applied Animal Behaviour Science 116:82–87.
David, M., Y. Auclair, and F. Cézilly. 2011. Personality predicts social dominance in female
zebra fi nches , Taeniopygia guttata , in a feeding context. Animal Behaviour 81:219–224.
Descamps, S., S. Boutin, D. Berteaux, A. G. McAdam, and J.-M. Gaillard. 2008. Cohort effects
in red squirrels: the influence of density, food abundance and temperature on future
survival and reproductive success. Journal of Animal Ecology 77:305–314.
Dettmer, A. M., S. S. K. Kaburu, K. L. Byers, and A. M. Murphy. 2015. First-Time Rhesus
Monkey Mothers , and Mothers of Sons , Preferentially Engage in Face-to-Face
Interactions With Their Infants:1–9.
Dickens, M. J., and L. M. Romero. 2013. A consensus endocrine profile for chronically stressed
wild animals does not exist. General and Comparative Endocrinology 191:177–189.
Dingemanse, N., C. Both, P. Drent, and K. Van Oers. 2002. Repeatability and heritability of
exploratory behaviour in great tits from the wild. Animal Behaviour 64:929–938.
Dloniak, S. M. 2004. Socioendocrinology of Spotted Hyenas. Michigan State University.
Dloniak, S. M., J. A. French, and K. E. Holekamp. 2006. Rank-related maternal effects of
androgens on behaviour in wild spotted hyaenas. Nature 440:1190–1193.

126

Dloniak, S. M., J. A. French, N. J. Place, M. L. Weldele, S. E. Glickman, and K. E. Holekamp.
2004. Non-invasive monitoring of fecal androgens in spotted hyenas (Crocuta crocuta).
General and Comparative Endocrinology 135:51–61.
Draper, T. W. 1995. Canine analogs of human personality factors. The Journal of general
psychology 122:241–252.
Drea, C. M. 2006. Studying primate learning in group contexts: Tests of social foraging,
response to novelty, and cooperative problem solving. Methods 38:162–177.
Durr, R., and C. Smith. 1997. Individual differences and their relation to social structure in
domestic cats. Journal of Comparative Psychology 111:412–418.
Dwyer, C. M., and A. . Lawrence. 2000. Maternal Behaviour in Domestic Sheep ( Ovis aries ):
Constancy and Change with Maternal Experience. Behaviour 137:1391–1413.
Engelhard, G. H., A. N. J. Baarspul, M. Broekman, J. C. S. Creuwels, and P. J. H. Reijnders.
2002. Human disturbance , nursing behaviour , and lactational pup growth in a declining
southern elephant seal ( Mirounga leonina ) population 1886:1876–1886.
Engh, A. L., K. Esch, L. Smale, and K. E. Holekamp. 2000. Mechanisms of maternal rank
“inheritance” in the spotted hyaena, Crocuta crocuta. Animal Behaviour 60:323–332.
Ethan Pride, R. 2005. High faecal glucocorticoid levels predict mortality in ring-tailed lemurs
( Lemur catta ). Biology Letters 1:60–63.
Fairbanks, L. 1993. Maternal protectiveness and response to the unfamiliar in vervet monkeys.
American Journal of Primatology 30:119–129.
Fairbanks, L. A. 1989. Early experience and cross-generational continuity of mother-infant
contact in vervet monkeys. Developmental psychobiology 22:669–681.
Fairbanks, L. A. 1996. Individual Differences in Maternal Style:: Causes and Consequences for
Mothers and offspring. Advances in the Study of Behavior 25:579–611.
Fairbanks, L. A., and K. Hinde. 2012. Behavioral Response of Mothers and Infants to Variation
in Maternal Condition: Adaptation, Compensation, and Resilience. Pages 281–302Building
babies. Springer New York, New York, NY.
Fairbanks, L. A., and M. T. McGuire. 1995. Maternal condition and the quality of maternal care
in vervet monkeys. Behaviour 132:733–754.
Ficetola, G. F., R. Sacchi, S. Scali, A. Gentilli, F. De Bernardi, and P. Galeotti. 2007. Vertebrates
respond differently to human disturbance: implications for the use of a focal species
approach. Acta Oecologica 31:109–118.

127

Flies, A. S., C. K. Grant, L. S. Mansfield, E. J. Smith, M. L. Weldele, and K. E. Holekamp. 2012.
Development of a hyena immunology toolbox. Veterinary Immunology and
Immunopathology 145:110–119.
Foley, C. A. H., S. Papageorge, and S. K. Wasser. 2001. Noninvasive Stress and Reproductive
Measures of Social and Ecological Pressures in Free-Ranging African Elephants.
Conservation Biology 15:1134–1142.
Fox, M. W. 1972. Socio-ecological implications of individual differences in wolf litters: a
developmental and evolutionary perspective. Behaviour 41:298–313.
Fox, R. A., L. D. Ladage, T. C. Roth, and V. V Pravosudov. 2009. Behavioural profile predicts
dominance status in mountain chickadees , Poecile gambeli. Animal Behaviour 77:1441–
1448.
Francis, D., J. Diorio, and D. Liu. 1999. Nongenomic Transmission Across Generations of
Maternal Behavior and Stress Responses in the Rat. Science 286:1155–1158.
Frank, L. 1986a. Social organization of the spotted hyaena (Crocuta crocuta). I. Demography.
Animal Behaviour 34:1500–1509.
Frank, L. 1986b. Social organization of the spotted hyaena Crocuta crocuta. II. Dominance and
reproduction. Animal Behaviour 34:1510–1527.
Frank, L. G., and S. E. Glickman. 1994. Giving birth through a penile clitoris: parturition and
dystocia in the spotted hyena (Crocuta crocuta). J. Zool. Lond. 234:659–690.
Frank, L. G., S. E. Glickman, and I. Powch. 1990. Sexual dimorphism in the spotted hyaena
(Crocuta crocuta). Journal of Zoology 221:308–313.
French, S. S., M. Gonzales-Suarez, J. K. Young, S. Durham, and L. R. Gerber. 2011. Human
Disturbance Influences Reproductive Success and Growth Rate in California Sea Lions
( Zalophus californianus ). PLoS ONE 6:e17686.
Gehrt, S. D., C. Anchor, and L. a. White. 2009. Home Range and Landscape Use of Coyotes in
a Metropolitan Landscape: Conflict or Coexistence? Journal of Mammalogy 90:1045–1057.
Gibbs, M., C. J. Breuker, and H. Van Dyck. 2010. Flight during oviposition reduces maternal
egg provisioning and influences offspring development in Pararge aegeria (L.).
Physiological entomology 35:29–39.
Glickman, S., C. Zabel, S. Yoerg, M. Weldele, C. M. Drea, and L. G. Frank. 1997. Social
Facilitation, Affiliation, and Dominance in the Social Life of Spotted Hyenas. Annals of the
New York Academy of Sciences 807:175–184.

128

Gluckman, P. D., M. A. Hanson, C. Cooper, and K. L. Thornburg. 2008. Effect of in utero and
early-life conditions on adult health and disease. New England Journal of Medicine 359:61–
73.
Gómez-Laplaza, L. M. 2002. Social status and investigatory behaviour in the angelfish
(Pterophyllum scalare). Behaviour 139:1469–1490.
Gosselink, T. E., T. R. Van Deelen, R. E. Warner, and P. C. Mankin. 2007. Survival and CauseSpecific Mortality of Red Foxes in Agricultural and Urban Areas of Illinois Survival and
Cause-Specific Mortality of Red Foxes in Agricultural and Urban Areas of Illinois. Journal
of Wildlife Management 71:1862–1873.
Goymann, W., M. L. East, B. Wachter, O. P. Honer, E. Mostl, T. J. Van’t Holf, and H. Hofer.
2001. Social, state-dependent and environmental modulation of faecal corticosteroid levels
in free-ranging female spotted hyenas. Proceedings of the Royal Society B: Biological
Sciences 268:2453–2459.
Goymann, W., and J. C. Wingfield. 2004. Allostatic load, social status and stress hormones: the
costs of social status matter. Animal Behaviour 67:591–602.
Green, D. S. 2015. Anthropogenic disturbance, ecological change, and wildlife conservation at
the edge of the mara-serengeti ecosystem. Michigan State University.
Green, D. S., L. Johnson-Ulrich, H. E. Couraud, and K. E. Holekamp. 2017. Anthropogenic
disturbance induces opposing population trends in spotted hyenas and African lions.
Biodiversity and Conservation:1–19.
Greenberg, J. R., and K. E. Holekamp. 2017. Human disturbance affects personality
development in a wild carnivore. Animal Behaviour 132:303–312.
Greggor, A. L., O. Berger-tal, D. T. Blumstein, L. Angeloni, C. Bessa-gomes, B. F. Blackwell, C.
Cassady, S. Clair, K. Crooks, S. De Silva, E. Fernández-juricic, S. Z. Goldenberg, S. L.
Mesnick, M. Owen, C. J. Price, D. Saltz, C. J. Schell, A. V Suarez, R. R. Swaisgood, C. S.
Winchell, and W. J. Sutherland. 2016. Research Priorities from Animal Behaviour for
Maximising Conservation Progress. Trends in Ecology & Evolution 31:953–964.
Guenther, A., M. A. Finkemeier, and F. Trillmich. 2014. The ontogeny of personality in the wild
guinea pig. Animal Behaviour 90:131–139.
Haage, M., U. A. Bergvall, T. Maran, K. Kiik, and A. Angerbjörn. 2013. Situation and context
impacts the expression of personality: The influence of breeding season and test context.
Behavioural Processes 100:103–109.
Hamel, S., S. D. Côté, and M. Festa-Bianchet. 2010. Maternal characteristics and environment
affect the costs of reproduction in female mountain goats. Ecology 91:2034–2043.
129

Harper, J. M., and S. N. Austad. 2000. Fecal glucocorticoids: a noninvasive method of measuring
adrenal activity in wild and captive rodents. Physiological and Biochemical Zoology 73:12–
22.
Harris, C. E., and F. F. Knowlton. 2001. Differential responses of coyotes to novel stimuli in
familiar and unfamiliar settings. Canadian Journal of Zoology 79:2005–2013.
Hau, M., S. Casagrande, J. Q. Ouyang, and A. T. Baugh. 2016. Glucocorticoid-Mediated
Phenotypes in Vertebrates: Multilevel Variation and Evolution. Advances in the Study of
Behavior 48:1–75.
Hauser, M., and L. Fairbanks. 1988. Mother-offspring conflict in vervet monkeys: variation in
response to ecological conditions. Animal Behaviour 36:802–813.
Heffernan, D. J., W. F. Andelt, and J. A. Shivik. 2007. Coyote Investigative Behavior Following
Removal of Novel Stimuli. Journal of Wildlife Management 71:587–593.
Hinde, K., A. L. Skibiel, A. B. Foster, L. Del Rosso, S. P. Mendoza, and J. P. Capitanio. 2015.
Cortisol in mother’s milk across lactation reflects maternal life history and predicts infant
temperament. Behavioral Ecology 26:269–281.
Hinde, R. A. 1976. On Describing Relationships. Journal of Child Psychology and Psychiatry
17:1–19.
Hinde, R. A., and S. Atkinson. 1970. Assessing the roles of social partners in maintaining mutual
proximity, as exemplified by mother-infant relations in rhesus monkeys. Animal Behaviour
18:169-176.
Hofer, H., S. Benhaiem, W. Golla, and M. L. East. 2016. Trade-offs in lactation and milk intake
by competing siblings in a fluctuating environment. Behavioral Ecology 27:1567–1578.
Hofer, H., and M. L. East. 1993a. The commuting system of Serengeti spotted hyaenas: how a
predator copes with migratory prey. III. Attendance and maternal care. Animal
Behaviour:1–15.
Hofer, H., and M. L. East. 1993b. The commuting system fo Serengeti spotted hyaenas: how a
predator copes with migratory prey. I. Social organization. Animal Behaviour 46:547–557.
Hofer, H., and M. L. East. 2003. Behavioral processes and costs of co-existence in female
spotted hyenas: a life history perspective. Evolutionary Ecology 17:315–331.
Hoffman, C. L., J. P. Higham, M. Heistermann, C. L. Coe, B. J. Prendergast, and D.
Maestripieri. 2011. Immune function and HPA axis activity in free-ranging rhesus
macaques. Physiology & Behavior 104:507–514.

130

Holekamp, K. E., S. M. Cooper, C. I. Katona, N. A. Berry, L. G. Frank, and L. Smale. 1997a.
Patterns of Association among Female Spotted Hyenas (Crocuta crocuta). Journal of
Mammalogy 78:55–64.
Holekamp, K. E., B. Dantzer, G. Stricker, K. C. S. Yoshida, and S. Benson-Amram. 2015.
Brains, brawn and sociality: a hyaena’s tale. Animal Behaviour 103:237–248.
Holekamp, K. E., and S. M. Dloniak. 2009. Maternal Effects in Fissiped Carnivores. Pages 227–
255in D. Maestripieri and J. M. Mateo, editors.Maternal Effects in Mammals. The
University of Chicago Press, Chicago.
Holekamp, K. E., and S. M. Dloniak. 2010. Intraspecific Variation in the Behavioral Ecology of
a Tropical Carnivore, the Spotted Hyena. Advances in the Study of Behavior 42:189–229.
Holekamp, K. E., and L. Smale. 1993. Ontogeny of dominance in free-living spotted hyaenas:
juvenile rank relations with other immature individuals. Animal Behaviour 46:451–466.
Holekamp, K. E., and L. Smale. 1998a. Behavioral development in the Spotted Hyena.
BioScience 48:997–1005.
Holekamp, K. E., and L. Smale. 1998b. Dispersal Status Influences Hormones and Behavior in
the Male Spotted Hyena. Hormones and Behavior 33:205–216.
Holekamp, K. E., L. Smale, R. Berg, and S. M. Cooper. 1997b. Hunting rates and hunting
success in the spotted hyena (Crocuta crocuta). Journal of Zoology 242:1–15.
Holekamp, K. E., L. Smale, and M. Szykman. 1996. Rank and reproduction in the female
spotted hyaena. Reproduction 108:229–237.
Holekamp, K. E., E. M. Swanson, and P. E. Van Meter. 2013. Developmental constraints on
behavioural flexibility. Philosophical Transactions of the Royal Society B: Biological
Sciences 368:20120350.
Honer, O. P., B. Wachter, M. L. East, and H. Hofer. 2002. The response of spotted hyaenas to
long-term changes in prey populations: functional response and interspecific
kleptoparasitism. Journal of Animal Ecology 71:236-246.
van Horn, R. C., T. L. McElhinny, and K. E. Holekamp. 2003. Age estimation and dispersal in
the spotted hyena (Crocuta crocuta). Journal of Mammalogy 84:1019–1030.
Hudson, R., A. Bautista, V. Reyes-Meza, J. M. Montor, and H. G. Rödel. 2011. The effect of
siblings on early development: A potential contributor to personality differences in
mammals. Developmental Psychobiology 53:564–574.
Hudson, R., and F. Trillmich. 2008. Sibling competition and cooperation in mammals:

131

challenges, developments and prospects. Behavioral Ecology and Sociobiology 62:299–307.
Huizink, A. C., E. J. H. Mulder, and J. K. Buitelaar. 2004. Prenatal stress and risk for
psychopathology: specific effects or induction of general susceptibility? Psychological
bulletin 130:115.
Ikime, T. 2015. Diurnal movement patterns, habitat use and energy cost of locomotion by
spotted hyenas (Crocuta crocuta) in the Masai Mara National Reserve, Kenya. Michigan
State University.
Jablonka, E., B. Oborny, I. Molnar, E. Kisdi, J. Hofbauer, and T. Czaran. 1995. The adaptive
advantage of phenotypic memory in changing environments. Philosophical Transactions of
the Royal Society of London B: Biological Sciences 350:133–141.
Jacobson-Pick, S., and G. Richter-Levin. 2010. Differential impact of juvenile stress and
corticosterone in juvenility and in adulthood , in male and female rats. Behavioural Brain
Research 214:268–276.
Jahn-Eimermacher, A., I. Lasarzik, and J. Raber. 2011. Statistical analysis of latency outcomes in
behavioral experiments. Behavioural Brain Research 221:271–275.
James, H., and C. Myfanwy. 2013. Individual differences in maternal behaviour in the grey seal
(Halichoerus grypus) and the impact of disturbance at Donna Nook. Durham University.
Johnson, J. C., P. J. Trubl, and L. S. Miles. 2012. Black widows in an urban desert: city-living
compromises spider fecundity and egg investment despite urban prey abundance. The
American Midland Naturalist 168:333–340.
Jones, A. C., and S. D. Gosling. 2005. Temperament and personality in dogs (Canis familiaris): A
review and evaluation of past research. Applied Animal Behaviour Science 95:1–53.
Kendal, R. L., R. L. Coe, K. N. Laland, and P. Biology. 2005. Age Differences in Neophilia,
Exploration, and Innovation in Family Groups of Callitrichid Monkeys. American Journal
of Primatology 66:167–188.
King, W. J., and A. W. Goldizen. 2016. Few sex effects in the ontogeny of mother-offspring
relationships in eastern grey kangaroos. Animal Behaviour 113:59–67.
Kittleson, M. M., L. A. Meoni, N.-Y. Wang, A. Y. Chu, D. E. Ford, and M. J. Klag. 2006.
Association of childhood socioeconomic status with subsequent coronary heart disease in
physicians. Archives of internal medicine 166:2356–2361.
Kolowski, J., and K. Holekamp. 2009. Ecological and anthropogenic influences on space use by
spotted hyaenas. Journal of Zoology 277:23–36.

132

Kolowski, J., D. Katan, and K. Theis. 2007. Daily patterns of activity in the spotted hyena.
Journal of Mammology 88:1017–1028.
Kolowski, J. M., and K. E. Holekamp. 2006. Spatial, temporal, and physical characteristics of
livestock depredations by large carnivores along a Kenyan reserve border. Biological
Conservation 128:529–541.
Kortet, R., A. Vainikka, M. Janhunen, J. Piironen, and P. Hyvärinen. 2014. Behavioral variation
shows heritability in juvenile brown trout Salmo trutta. Behavioral Ecology and
Sociobiology 68:927–934.
Kruuk, H. 1972. The spotted hyaena. University of Chicago Press, Chicago.
Lee, P. C., P. Majluf, and I. J. Gordon. 1991. Growth, weaning and maternal investment from a
comparative perspective. Journal of Zoology 225:99–114.
Lee, P. C., and C. J. Moss. 1986. Early maternal investment in male and female African elephant
calves. Behavioral Ecology and Sociobiology 18:353–361.
Lessells, C., and P. T. Boag. 1987. Unrepeatable repeatabilities. The Auk 104:116–121.
Lewin, N., E. M. Swanson, B. L. Williams, and K. E. Holekamp. 2017. Juvenile concentrations
of IGF-1 predict life-history trade-offs in a wild mammal. Functional Ecology 31:894–902.
Lindström, J. 1999. Early development and fitness in birds and mammals. Trends in Ecology &
Evolution 14:343–348.
Linnell, J. D. C., J. Odden, M. E. Smith, R. Aanes, and J. E. Swenson. 1999. Large Carnivores
That Kill Livestock: Do “Problem Individuals” Really Exist? Wildlife Society Bulletin
27:698–705.
Litvaitis, J. A., A. G. Clark, and J. H. Hunt. 2017. Prey selection and fat deposits of bobcats
(Felis rufus) during autumn and winter in Maine. Journal of Mammalogy 67:389–392.
Liu, D., J. Diorio, B. Tannenbaum, C. Caldji, and D. Francis. 1997. Maternal care, hippocampal
glucocorticoid receptors, and hypothalamic-pituitary-adrenal responses to stress. Science
277:1659–1662.
Lycett, J. E., P. Henzi, and L. Barrett. 1998. Maternal investment in mountain baboons and the
hypothesis of reduced care. Behavioral Ecology and Sociobiology 42:49–56.
M’soka, J., S. Creel, M. S. Becker, and E. Droge. 2016. Spotted hyaena survival and density in a
lion depleted ecosystem : The effects of prey availability , humans and competition between
large carnivores in African savannahs. Biological Conservation 201:348–355.
Macdonald, D. W., O. Courtenay, S. Forbes, and F. Mathews. 1999. The red fox (Vulpes vulpes)
133

in Saudi Arabia: loose-knit groupings in the absence of territoriality. Journal of Zoology
249:383–391.
MacDonald, K. 1983. Stability of individual differences in a litter of wolf cubs (Canis lupus).
Journal of Comparative Psychology 97:99–106.
Maestripieri, D. 1995. Assessment of Danger to Themselves and Their Infants by Rhesus
Macaque ( Macaca mulatto ) Mothers 109:416–420.
Maestripieri, D. 1998. Social and demographic influences on mothering style in pigtail macaques.
Ethology 104:379-385.
Maestripieri, D. 2001. Intraspecific Variability in Parenting Styles of Rhesus Macaques (Macaca
mulatta): The Role of the Social Environment. Ethology 107:237–248.
Maestripieri, D. 2005. Early experience affects the intergenerational transmission of infant abuse
in rhesus monkeys. Proceedings of the National Academy of Sciences of the United States
of America 102:9726–9729.
Maestripieri, D., C. L. Hoffman, G. M. Anderson, C. S. Carter, and J. D. Higley. 2009. Mother–
infant interactions in free-ranging rhesus macaques: Relationships between physiological
and behavioral variables. Physiology & Behavior 96:613–619.
Maestripieri, D., and J. M. Mateo. 2009. Maternal effects in mammals. University of Chicago
Press, Chicago.
Maestripieri, D., and K. Wallen. 2003. Infant abuse runs in families of group-living pigtail
macaques:1–7.
Marshall, D. J., and T. Uller. 2007. When is a maternal effect adaptive? Oikos 116:1957–1963.
Martin, T., and P. Bateson. 1988. Measuring Behavior: an Introductory Guide. Cambridge
University Press, Cambridge.
Martins, T., M. Roberts, I. Giblin, R. Huxham, and M. Evans. 2007. Speed of exploration and
risk-taking behavior are linked to corticosterone titres in zebra finches. Hormones and
Behavior 52:445–453.
Matlack, C. R., and A. J. Evans. 1992. Diet and condition of bobcats, Lynx rufus, in Nova Scotia
during autumn and winter. Canadian Journal of Zoology 70:1114–1119.
McCormick, C. M., J. W. Smythe, S. Sharma, and M. J. Meaney. 1995. Sex-specific effects of
prenatal stress on hypothalamic-pituitary-adrenal responses to stress and brain
glucocorticoid receptor density in adult rats. Brain research. Developmental brain research
84:55–61.

134

McDougall, P. T., D. Réale, D. Sol, and S. M. Reader. 2006. Wildlife conservation and animal
temperament: causes and consequences of evolutionary change for captive, reintroduced,
and wild populations. Animal Conservation 9:39–48.
McEwen, B. S., and J. C. Wingfield. 2003. The concept of allostasis in biology and biomedicine.
Hormones and Behavior 43:2–15.
McGlothlin, J. W., and E. D. Ketterson. 2008. Hormone-mediated suites as adaptations and
evolutionary constraints. Philosophical Transactions of the Royal Society of London B:
Biological Sciences 363:1611–1620.
Mendl, M. 1988. The effects of litter size variation on mother-offspring relationships and
behavioural and physical development in several mammalian species (principally rodents).
Journal of Zoology 215:15–34.
Van Meter, P. E., J. A. French, K. Bidali, M. L. Weldele, J. L. Brown, and K. E. Holekamp.
2008. Non-invasive measurement of fecal estrogens in the spotted hyena (Crocuta crocuta).
General and Comparative Endocrinology 155:464–471.
Van Meter, P. E., J. A. French, S. M. Dloniak, H. E. Watts, J. M. Kolowski, and K. E.
Holekamp. 2009. Fecal glucocorticoids reflect socio-ecological and anthropogenic stressors
in the lives of wild spotted hyenas. Hormones and Behavior 55:329–337.
Mettler, A. E., and J. A. Shivik. 2007. Dominance and neophobia in coyote (Canis latrans)
breeding pairs. Applied Animal Behaviour Science 102:85–94.
Miranda, A. C., H. Schielzeth, T. Sonntag, and J. Partecke. 2013. Urbanization and its effects on
personality traits: a result of microevolution or phenotypic plasticity? Global Change
Biology 19:2634–2644.
Monaghan, P. 2008. Early growth conditions, phenotypic development and environmental
change. Philosophical Transactions of the Royal Society B: Biological Sciences 363:1635–
1645.
Moore, C. L. 1984. Maternal contributions to the development of masculine sexual behavior in
laboratory rats. Developmental Psychobiology 17:347–356.
Moore, C. L. 1992. The role of maternal stimulation in the development of sexual behavior and
its neural basis. Annals of the New York Academy of Sciences 662:160–77.
Moretti, L., M. Hentrup, K. Kotrschal, and F. Range. 2015. The influence of relationships on
neophobia and exploration in wolves and dogs. Animal Behaviour 107:159–173.
Moscovice, L. R., and C. T. Snowdon. 2006. The role of social context and individual experience
in novel task acquisition in cottontop tamarins, Saguinus oedipus. Animal Behaviour

135

71:933–943.
Mousseau, T., and C. Fox. 1998. The adaptive significance of maternal effects. Trends in
Ecology & Evolution 13:403–407.
Nash, L. T. 1978. The development of the mother-infant relationship in wild baboons (Papio
anubis). Animal Behaviour 26:746–759.
Nater, U. M., and N. Rohleder. 2009. Salivary alpha-amylase as a non-invasive biomarker for the
sympathetic nervous system : Current state of research. Psychoneuroendocrinology 34:486–
496.
Nguyen, N., L. Gesquiere, S. C. Alberts, and J. Altmann. 2012. Sex differences in the motherneonate relationship in wild baboons: social, experiential and hormonal correlates. Animal
Behaviour 83:891–903.
Nguyen, N., L. R. Gesquiere, E. O. Wango, S. C. Alberts, and J. Altmann. 2008. Late pregnancy
glucocorticoid levels predict responsiveness in wild baboon mothers (Papio cynocephalus).
Animal Behaviour 75:1747–1756.
Nicolaus, M., J. M. Tinbergen, K. M. Bouwman, S. P. M. Michler, R. Ubels, C. Both, B.
Kempenaers, and N. J. Dingemanse. 2012. Experimental evidence for adaptive
personalities in a wild passerine bird. Proceedings of the Royal Society B: Biological
Sciences 279:4885–4892.
Noer, C. L., E. K. Needham, A.-S. Wiese, T. J. S. Balsby, and T. Dabelsteen. 2015. Context
Matters: Multiple Novelty Tests Reveal Different Aspects of Shyness-Boldness in Farmed
American Mink (Neovison vison). PLoS ONE 10:e0130474.
Nussey, D., and A. Wilson. 2007. The evolutionary ecology of individual phenotypic plasticity in
wild populations. Journal of Evolutionary Biology 20:831–844.
O’Rand, A. M., and J. Hamil-Luker. 2005. Processes of cumulative adversity: Childhood
disadvantage and increased risk of heart attack across the life course. The Journals of
Gerontology Series B: Psychological Sciences and Social Sciences 60:S117–S124.
Ono, K. A., D. J. Boness, and O. T. Oftedal. 1987. The effect of a natural environmental
disturbance on maternal investment and pup behavior in the California sea lion. Behavioral
Ecology and Sociobiology 21:109–118.
Oswald, L. M., P. Zandi, G. Nestadt, J. B. Potash, A. E. Kalaydjian, and G. S. Wand. 2006.
Relationship between cortisol responses to stress and personality.
Neuropsychopharmacology 31:1583–1591.
Pangle, W. M., and K. E. Holekamp. 2010. Lethal and nonlethal anthropogenic effects on

136

spotted hyenas in the Masai Mara National Reserve. Journal of Mammalogy 91:154–164.
Papaioannou, A., K. Gerozissis, A. Prokopiou, S. Bolaris, and F. Stylianopoulou. 2002. Sex
differences in the effects of neonatal handling on the animal’s response to stress and the
vulnerability for depressive behaviour. Behavioural brain research 129:131–139.
Park, M. K., T. A. Hoang, J. D. Belluzzi, and F. M. Leslie. 2003. Gender specific effect of
neonatal handling on stress reactivity of adolescent rats. Journal of neuroendocrinology
15:289–295.
Parker, K. ., K. . Rainwater, C. . Buckmaster, A. F. Schatzberg, S. E. Lindley, and D. M. Lyons.
2007. Early life stress and novelty seeking behavior in adolescent monkeys.
Psychoneuroendocrinology 32:785–792.
Parker, K., and C. Buckmaster. 2004. Prospective investigation of stress inoculation in young
monkeys. Archives of General Psychiatry 61:933–941.
Patchev, V. K., S. Hayashi, C. Orikasa, and O. F. X. Almeida. 1999. Ontogeny of gender-specific
responsiveness to stress and glucocorticoids in the rat and its determination by the neonatal
gonadal steroid environment. Stress 3:41–54.
Payne, S. F., and R. J. Jameson. 1984. Early behavioral development of the sea otter, Enhydra
lutris. Journal of Mammalogy 65:527–531.
Petelle, M. B., J. G. A. Martin, and D. T. Blumstein. 2015. Heritability and genetic correlations
of personality traits in a wild population of yellow-bellied marmots ( Marmota flaviventris).
Journal of Evolutionary Biology 28:1840–1848.
Petelle, M. B., D. E. McCoy, V. Alejandro, J. G. A. Martin, and D. T. Blumstein. 2013.
Development of boldness and docility in yellow-bellied marmots. Animal Behaviour:1–8.
Ponirakis, A., E. J. Susman, and C. A. Stifter. 1998. Negative Emotionality and Cortisol during
Adolescent Pregnancy and Its Effects on Infant Health and Autonomic Nervous System
Reactivity. Developmental Psychobiology 33:163–174.
Ragan, C. M., K. M. Harding, and J. S. Lonstein. 2016. Hormones and Behavior Associations
among within-litter differences in early mothering received and later emotional behaviors ,
mothering , and cortical tryptophan hydroxylase-2 expression in female laboratory rats.
Hormones and Behavior 77:62–71.
Ragan, C. M., E. Loken, C. A. Stifter, and S. A. Cavigelli. 2011. Within-litter variance in early rat
pup-mother interactions and adult offspring responses to novelty. Developmental
Psychobiology 54:1–8.
Räsänen, K., and L. E. B. Kruuk. 2007. Maternal effects and evolution at ecological time-scales.

137

Functional Ecology 21:408–421.
Réale, D., S. M. Reader, D. Sol, P. T. McDougall, and N. J. Dingemanse. 2007. Integrating
animal temperament within ecology and evolution. Biological Reviews 82:291–318.
Reed, T. E., R. S. Waples, D. E. Schindler, J. J. Hard, and M. T. Kinnison. 2010. Phenotypic
plasticity and population viability: the importance of environmental predictability.
Proceedings of the Royal Society B: Biological Sciences 277:3391–3400.
Reznikov, A. G., N. D. Nosenko, L. V. Tarasenko, P. V. Sinitsyn, and L. I. Polyakova. 2001.
Early and long-term neuroendocrine effects of prenatal stress in male and female rats.
Neuroscience and behavioral physiology 31:1–5.
Riley, S. P. D., R. M. Sauvajot, T. K. Fuller, E. C. York, D. A. Kamradt, C. Bromley, and R.
Wayne. 2003. Effects of Urbanisation and Habitat Fragmentation on Bobcats and Coyotes
in Southern California. Conservation Biology 17:566–576.
Ripple, W. J., J. A. Estes, R. L. Beschta, C. C. Wilmers, E. G. Ritchie, M. Hebblewhite, J. Berger,
B. Elmhagen, M. Letnic, M. P. Nelson, O. J. Schmitz, D. W. Smith, A. D. Wallach, and A.
J. Wirsing. 2014. Status and Ecological Effects of the World’s Largest Carnivores. Science
343:1241484.
Rodel, H. G., D. von Holst, and C. Kraus. 2009. Family Legacies: Short- and Long-Term Fitness
Consequences of Early-Life Conditions in Female European Rabbits. Journal of Animal
Ecology 78:789–797.
Romeo, R. D. 2010. Frontiers in Neuroendocrinology Pubertal maturation and programming of
hypothalamic – pituitary – adrenal reactivity. Frontiers in Neuroendocrinology 31:232–240.
Romero, L. M., and M. Wikelski. 2010. Stress physiology as a predictor of survival in Galapagos
marine iguanas:3157–3162.
Rosenblum, L. A., and G. S. Paully. 1984. The effects of varying environmental demands on
maternal and infant behavior. Child Development 55:305–314.
Russell, E. ., and G. A. Pearce. 1971. Exploration of novel objects by marsupials. Behaviour
40:312–322.
Ryer, C. H., and B. L. Olla. 1991. Information transfer and the facilitation and inhibition of
feeding in a schooling fish. Environmental Biology of Fishes 30:317–323.
Saberwal, V. K., J. P. Gibbs, R. Chellam, and A. J. . Johnsingh. 1994. Lion-human Conflict in the
Gir Forest. Conservation Biology 8:501–507.
Sapolsky, R. M., L. M. Romero, and A. U. Munck. 2000. How do glucocorticoids influence

138

stress responses? Integrating permissive, suppressive, stimulatory, and preparative actions.
Endocrine Reviews 21:55.
Sapolsky, R., and L. . Share. 1994. Rank-related differences in cardiovascular function among
wild baboons: Role of sensitivity to glucocorticoids. American Journal of Primatology
32:261–275.
Schino, G., R. Cozzolino, and A. Troisi. 1999. Social Rank and Sex-Biased Maternal Investment
in Captive Japanese Macaques: Behavioural and Reproductive Data. Folia Primatologica
70:254–263.
Schino, G., F. R. D’Amato, and A. Troisi. 1995. Mother infant relationships in Japanese
macaques: sources of inter-individual variation. Animal Behaviour 49:151–158.
Schino, G., A. Troisi, and C. Psichiatria. 2001. Early Maternal Rejection and Later Social Anxiety
in Juvenile and Adult Japanese Macaques. Developmental Psychobiology 38:186–190.
Schneider, M. L., E. C. Roughton, and A. J. Koehler. 1999. Growth and Development
Following Prenatal Stress Exposure in Primates: An Examination of Ontogenetic
Vulnerability. Child Development 70:263–274.
Schoech, S. J., M. A. Rensel, and R. S. Heiss. 2011. Short- and long-term effect of
developmental corticosterone exposure on avian physiology, behavioral phenotype,
cognition, and fitness: A review. Current Zoology 4:514–530.
Seidensticker, J., and C. McDougal. 1993. Tiger predatory behaviour, ecology and conservation.
Symp. Zool. Soc. Lond. 65:105–125.
Sheriff, M. J., C. J. Krebs, and R. Boonstra. 2009. The sensitive hare: sublethal effects of
predator stress on reproduction in snowshoe hares. Journal of Animal Ecology 78:1249–
1258.
Sheriff, M. J., and O. P. Love. 2013. Determining the adaptive potential of maternal stress.
Ecology Letters 16:271–280.
Shivik, J. . 2006. Tools for the Edge: What’s New for Conserving Carnivores. Bioscience
56:253–259.
Sih, A. 2013. Understanding variation in behavioural responses to human-induced rapid
environmental change: a conceptual overview. Animal Behaviour 85:1077–1088.
Sikes, R. S., and W. L. Gannon. 2011. Guidelines of the American Society of Mammalogists for
the use of wild mammals in research. Journal of Mammalogy 92:235–253.
Silva, V. S., and C. Azevedo. 2013. Evaluating personality traits of captive maned wolves,

139

Chrysocyon brachyurus (Illiger, 1815) (Mammalia: Canidae), for conservation purposes.
Lundiana 11:35–41.
Sinn, D. L., S. D. Gosling, and N. A. Moltschaniwskyj. 2008. Development of shy/bold
behaviour in squid: context-specific phenotypes associated with developmental plasticity.
Animal Behaviour 75:433–442.
Smale, L., L. G. Frank, and K. E. Holekamp. 1993. Ontogeny of dominance in free-living
spotted hyaenas: juvenile rank relations with adult females and immigrant males. Animal
Behaviour 46:467–477.
Smale, L., K. E. Holekamp, M. Weldele, L. G. Frank, and S. E. Glickman. 1995. Competition
and cooperation between litter-mates in the spotted hyaena, Crocuta crocuta. Animal
Behaviour 50:671–682.
Smale, L., S. Nunes, and K. E. Holekamp. 1997. Sexually dimorphic dispersal in mammals:
patterns, causes, and consequences. Advances in the Study of Behavior 26:181–250.
Smith, B. R., and D. T. Blumstein. 2008. Fitness consequences of personality: a meta-analysis.
Behavioral Ecology 19:448–455.
Smith, J. E., J. M. Kolowski, K. E. Graham, S. E. Dawes, and K. E. Holekamp. 2008. Social and
ecological determinants of fission–fusion dynamics in the spotted hyaena. Animal
Behaviour 76:619–636.
Smith, J. E., S. K. Memenis, and K. E. Holekamp. 2007. Rank-related partner choice in the
fission – fusion society of the spotted hyena (Crocuta crocuta):753–765.
Smith, J. M. 1980. A new theory of sexual investment. Behavioral ecology and sociobiology
7:247–251.
Sol, D., O. Lapiedra, and C. González-Lagos. 2013. Behavioural adjustments for a life in the city.
Animal Behaviour 85:1101–1112.
Sroges, R. W., and S. E. Glickman. 1966. Curiosity in Zoo Animals. Behaviour 26:151–187.
Stamps, J., and T. G. G. Groothuis. 2010. The development of animal personality: relevance,
concepts and perspectives. Biological Reviews 85:301–325.
Stevens, H. E., J. F. Leckman, J. D. Coplan, and S. J. Suomi. 2009. Risk and Resilience. Journal
of the American Academy of Child & Adolescent Psychiatry 48:114–127.
Stevens, M. A., and D. J. Boness. 2003. Influences of habitat features and human disturbance on
use of breeding sites by a declining population of southern fur seals (Arctocephalus
australis):145–152.

140

Stewart, F. E. C., and A. G. Mcadam. 2014. Seasonal plasticity of maternal behaviour in
Peromyscus maniculatus 151:1641–1662.
Stirling, I., and P. B. Latour. 1978. Comparative hunting abilities of polar bear cubs of different
ages. Canadian Journal of Zoology 56:1768–1772.
Stöwe, M., T. Bugnyar, and B. Heinrich. 2006a. Effects of group size on approach to novel
objects in ravens (Corvus corax). Ethology 112:1079–1088.
Stöwe, M., T. Bugnyar, M. Loretto, and C. Schloegl. 2006b. Novel object exploration in ravens
(Corvus corax): effects of social relationships. Behavioural Processes 73:68–75.
Swanson, E. M., I. Dworkin, and K. E. Holekamp. 2011. Lifetime selection on a hypoallometric
size trait in the spotted hyena. Proceedings of the Royal Society of London B: Biological
Sciences 278:3277–3285.
Tanaka, I. 1989. Variability in the Development of Mother-infant Relationships Among Freeranging Japanese Macaques. Primates 30:477–491.
Tanner, J. B., M. L. Zelditch, B. L. Lundrigan, and K. E. Holekamp. 2009. Ontogenetic change
in skull morphology and mechanical advantage in the spotted hyena (Crocuta crocuta).
Journal of Morphology 271:353–365.
Taylor, R. W., A. K. Boon, B. Dantzer, D. Réale, M. M. Humphries, S. Boutin, J. C. Gorrell, D.
W. Coltman, and A. G. McAdam. 2012. Low heritabilities, but genetic and maternal
correlations between red squirrel behaviours. Journal of Evolutionary Biology 25:614–624.
Treves, A., and K. U. Karanth. 2003. Human-Carnivore Conflict and Perspectives on Carnivore
Management Worldwide. Conservation Biology 17:1491–1499.
Treves, A., R. B. Wallace, L. Naughton-Treves, and A. Morales. 2006. Co-Managing Human–
Wildlife Conflicts: A Review. Human Dimensions of Wildlife 11:383–396.
Trivers, R. L. 1972. Parental investment and sexual selection. Pages 136–179 in Sexual Selection
& the Descent of Man, Aldine de Gruyter, New York.
Trivers, R. L. 1974. Parent-Offspring Conflict. Integrative and Comparative Biology 14:249–264.
Tung, J., E. A. Archie, J. Altmann, and S. C. Alberts. 2016. Cumulative early life adversity
predicts longevity in wild baboons. Nature Communications 7:11181.
Twiss, S. D., C. Cairns, R. M. Culloch, S. A. Richards, and P. P. Pomeroy. 2012. Variation in
Female Grey Seal (Halichoerus grypus) Reproductive Performance Correlates to ProactiveReactive Behavioural Types PLoS ONE 7:28–30.
Wasser, L. M., and S. K. Wasser. 1995. Environmental Variation and Developmental Rate
141

Among Free Ranging Yellow Baboons ( Papio cynocephalus ). American Journal of
Primatology 35:15–30.
Watts, H. E., J. B. Tanner, B. L. Lundrigan, and K. E. Holekamp. 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.
Watts, H., and K. Holekamp. 2009. Ecological determinants of survival and reproduction in the
spotted hyena. Journal of Mammalogy 90:461–471.
Webster, M. M., and A. J. W. Ward. 2010. Personality and social context. Biological Reviews
86:759–773.
Webster, M. M., A. J. W. Ward, and P. J. B. Hart. 2007. Boldness is influenced by social context
in threespine sticklebacks (Gasterosteus aculeatus). Behaviour 144:351–371.
West-Eberhard, M. J. 2005. Phenotypic accommodation: adaptive innovation due to
developmental plasticity. Journal of Experimental Zoology Part B: Molecular and
Developmental Evolution 304B:610–618.
White, P. A. 2007. Costs and strategies of communal den use vary by rank for spotted hyaenas,
Crocuta crocuta. Animal Behaviour 73:149–156.
Wienke, A. 2011. Frailty models in survival analysis. CRC Press, Boca Raton, FL.
Wolak, M. E., D. J. Fairbairn, and Y. R. Paulsen. 2011. Guidelines for estimating repeatability.
Methods in Ecology and Evolution 3:129–137.
Wynne, C. D. L., and I. G. McLean. 1999. The comparative psychology of marsupials.
Australian Journal of Psychology 51:111–116.
Yoshida, K. C. S., P. E. Van Meter, and K. E. Holekamp. 2016. Variation among free-living
spotted hyenas in three personality traits. Behaviour 153:1665–1722.

142