 

. .K..v.. .4,
K a. .,
ft . .71
a

,

 

3.» _

. Hum»:
. lind-

£11.11.“

. .

1*“ .t 7:: a a
1 3.13.1 .
1- .—
‘ .11
.. ‘
‘4.

... .35
1.3. ...\\

34.3..
x 5.5.7?

 

olAn.
itsc
1. pl a

13.21.

:8. II}. x...
'v.‘..‘ll‘.

‘ a 3“: Gian.

a. it
tuna“? .

 

 

 

This is to certify that the
dissertation entitled

EVOLUTIONARY AND ECOLOGICAL FORCES
SHAPING PATTERNS OF COOPERATION
AMONG SPOTTED HYENAS

presented by

JENNIFER ELAINE SMITH

has been accepted towards fulﬁllment
of the requirements for the

Doctor of degree in Zoology and Ecology,
Philosophy Evolutionary Biology
and Behavior

 

 

   
 

/;RAajor Professor’ 5 Séﬁature
{MILK 2;;/o

Date

 

MSU is an Afﬁrmative Action/Equal Opportunity Employer

 

 

LIBRARY
Michigan State
University

 

 

 

—.-ono-l-o-o—u—-—-—o--s-n-u-n-u—

 

‘---.---.-’-.-.-1-‘---_A‘A

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Aoc&Pres/ClRC/DaleDm.indd

 

 

 

EVOLUTIONARY AND ECOLOGICAL FORCES SHAPING PATTERNS
OF COOPERATION AMONG SPOTTED HYENAS

By

Jennifer Elaine Smith

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

2010

ABSTRACT

EVOLUTIONARY AND ECOLOGICAL FORCES SHAPING PATTERNS
OF COOPERATION AMONG SPOTTED HYENAS

By

Jennifer Elaine Smith

The evolution of cooperation is a central problem in biology, but the
current body of theory often fails to explain complex patterns found in nature.
Biological market theory extends traditional paradigms by predicting that the
relative values of social partners depend upon the current state of the
marketplace. My dissertation tested this prediction using data from a long-term
study (1988-present) of spotted hyenas (Crocuta crocuta) in Kenya. These
carnivores live in ﬁssion-fusion societies, called clans, which contain up to 80
members. Dominance hierarchies structure clans, and all adult females are
dominant to all immigrant males. Although all clan members defend a common
territory, individuals make active decisions to leave (ﬁssion) and to join (fusion)
subgroups of individuals within the clan. This ﬂexible social structure, together
with rank-related variation in partner value, offers an excellent system in which to
test biological market theory. First, I tested the theoretical prediction that social
and ecological circumstances inﬂuence ﬁssion-fusion dynamics in the spotted
hyena. Whereas cooperative defense of shared resources during inter-clan

competition and conﬂicts with lions promoted social cohesion, feeding
. competition was a strongly disruptive force that enhanced the tendency for

hyenas to separate temporarily from group members. Second, market theory

predicts that, when social partners vary in their relative value, individuals should
favor partners of the highest value. As predicted, high-ranking adult females were
most gregarious, and choosy subordinate females gained the most social and
feeding tolerance, but not coalitionary support, from dominants. Third, I evaluated
the evolutionary forces favoring intragroup coalitions among adult female spotted
hyenas. I found evidence that adult females beneﬁted indirectly via kin selection
and directly by reinforcing the status quo, but not from reciprocal altruism. Fourth,
I found that greeting ceremonies occurring outside of agonistic contexts
represent honest signals of long-term social bond strength and that the act of
greeting immediately promotes coalition formation among adult females. Overall,
these results elucidate the market forces shaping the dynamics of cooperation
and competition among spotted hyenas, and broaden our understanding of the

rules governing social decision-making among gregarious animals.

Copyright by
JENNIFER ELAINE SMITH

2010

ACKNOWLEDGEMENTS

My doctoral research is the product of a long journey, full of twists and
turns, much of which started before I ever set foot on the Michigan State
University (MSU) campus. Growing—up in the small town of Cushing, Maine, my
interest in biology stemmed from my early introduction to coastal marine life by
my high school biology teacher, Phil Marcoux. Phil has since passed away, but
his fascinating introduction to the thrill of making scientiﬁc discoveries has
provided me with the strong foundation from which I still draw from today. In my
early years, my parents and my sister, Christian, encouraged my passion for
animals, always inspiring me to do my best while accepting me for who I am. To
this day, my family continues to offer me a strong support network from which to
grow and to challenge myself. For this, I cannot thank them enough.

As an undergraduate at Colby College in Maine, I left North America for
the ﬁrst time in my life. Under the amazing instruction of Russ Cole, I went to the
British West Indies to study the behavior of land hermit crabs. l was shocked to
learn that these animals were nocturnal and spent most of their time residing in
the local chicken-coop, but was even more amazed to learn that people actually
made a career out of studying the behavioral ecology of wild animals in remote
locations. Thanks to Russ, l have been hooked on this type of research
experience ever since. Upon completing his course, I immediately secured
scholarship support to spend my entire junior year abroad in Australia and in

Kenya. After pursuing a variety of internships, I then went on to pursue my

Master’s at the University of Illinois at Urbana-Champaign (UIUC) with George
Batzli. George took me under his wing and graciously introduced me to the art of
scientiﬁc design and writing. I continue to implement George’s sound guidance
when I challenge myself to improve my writing. It was also at UIUC where I ﬁrst
met, Chris Strelioff. Chris has offered me his steady support and encouragement
throughout my doctoral degree; albeit from across the ocean and across state
lines during different stages of our relationship. I appreciate all of the sacriﬁces
he has made for me. For his patience, insights, and kindness, I will forever be
thankful to have him at my side.

After having spent a semester in Kenya as an undergraduate, I had
always hoped to return, but I had never dreamed that I would one day have the
unique experience of living in the bush for a year to pursue doctoral research on
spotted hyenas, one of the most bizarre and fascinating species in existence. For
this amazing opportunity and for all of her support at the many stages of my
doctoral research, I thank Kay Holekamp. Kay is truly an outstanding advisor,
passionate scientist and phenomenal female role model. She brings unparalleled
aptitude and scholarly enthusiasm for learning to her research program and
generously shares her insights with her students. For these reasons, and many
more for which I lack the 'space to adequately enumerate, Kay has been truly an
outstanding mentor. Although her life is surely a juggling act, Kay handles all of
these responsibilities and competing demands marvelously, and always makes
time to help me achieve my professional goals. I thank her for our countless

meetings, her careful answers to my billions and the billions of emails, and her

vi

thoughtful guidance throughout every stage of this degree process. Kay’s ﬁrm,
yet encouraging, mentoring style continuously challenges me to strive for
excellence in all of my pursuits. Kay is unquestionably one of the most
intellectually curious and generous people that I have ever met. I thank her for
giving me the opportunity to be her student, and to learn from her incredible
example. As I develop into an independent scientist, I look fonrvard to becoming
her colleague and will always strive to similarly impact the lives of my own
students in the future in the way she has impacted mine.

My other committee members, Laura Smale, Tom Getty and Andrew
McAdam, have also generously offered me their support and insights at every
critical stage of my doctoral program. Without their encouragement and wisdom,
this dissertation would surely be a different product. Laura Smale, along with
Kay, was responsible for devising the general observational protocols employed
here and for collecting much of the long-term data examined here. Due to her
ﬁrst-hand knowledge of hyena biology, Laura has been an excellent resource,
providing me with interesting insights with respect to patterns of coalition
formation and sex differences in these animals. Tom Getty has also been an
outstanding mentor. As his student, and then later as his teaching assistant, in
his Behavioral Ecology course, Tom opened-up a new world for me and
continuously challenged me to effectively communicate difﬁcult concepts to my
students and colleagues. Tom’s rich understanding of the theoretical
underpinnings of behavioral ecology has provided me with a fruitful foundation. I

can only hope to emulate these qualities in my own mentoring of student

vii

research in the future. Andrew’s support and insights have also been
extraordinary. During the tenure of this dissertation, the distances between our
ofﬁces have ranged from being separated by a single wall to being separated by
an international border. Despite that fact that he currently resides at the
University of Guelph in Canada, Andrew’s enthusiasm for science remains clear
and his insights continue to be instrumental in shaping my view of evolutionary
ecology. Finally, although Richard Hill is not a member of my dissertation
committee, I thank him for training me to educate the next generation of
undergraduates. It has been an honor to assist him in his courses, and I thank
him for taking the time to make personal connections with his students at this
large university. His lessons on Thomas Kuhn’s scientiﬁc revolutions and Charles
Darwin’s pluralism have deeply shaped my research and teaching philosophies.
The Holekamp lab is a large, constantly changing body of motivated
individuals who have collectively offered me many insights along the way. During
my six years at MSU, I have had the opportunity to share a lab with Sarah
Benson-Amram, Pat Bills, Andy Booms, Katy Califf, Leslie Curren, Ben Dantzer,
Andy Flies, Suzanne La Croix, Joe Kolowski, Eva-Maria Muecke, Wiline Pangle,
Kate Shaw, Greg Stricker, Eli Swanson, Jaime Tanner, Kevin Theis, Ryan
Taylor, Russ Van Horn, Page Van Meter, Aaron Wagner, Soﬁ Wahaj, and
Heather Watts. l have learned a lot from these folks and thank them for their
guidance along the way. In particular, the research of Anne Engh and Soﬁ Wahaj

inspired much of my doctoral work. Although Anne and I never overlapped here

viii

on campus, I thank both of these women for their tremendous intelleCtual
contributions and for bringing so many fascinating questions to the forefront.

In Kenya, Jaime Tanner was an excellent resource. She introduced me to
the Mara Hyena Project, trained me on all of relevant hyena protocols, and was
amazingly helpful in identifying all of the hyenas. I also thank Joe Kolowski and
Soﬁ Wahaj for inviting me to join them at Mara lntrepids Lodge to enjoy some of
life’s creature comforts amidst the African savannah. I thank John Keshe and
Moses Nairori for making me feel at home while living in the African bush. In
particular, I thank Moses for chasing baboons away from my tent and always
greeting me with a smile regardless of how ﬂooded our driveway became during
the rainy season. I thank John for sharing my love of cooking and for
encouraging me to pursue my research on cooperation, even though I thought for
sure at the time that we already knew everything there was to know about
coalition formation among spotted hyenas. Boy was I wrong!

In addition to the Holekamp lab members mentioned above, many
additional people participated in the collection and analysis of the data presented
in this dissertation. Pat Bills has been outstanding in helping me to extract and
organize data from the ﬁeld notes. lmportantly, two of my dissertation chapters
were in collaboration with Joe Kolowski and Russ Van Horn. For Chapter 2, Joe
tirelessly spent many continuous hours during the day and night collecting the
24-hour follow data. For Chapter 4, Russ diligently conducted many genetic
analyses used to assign the paternal relationships among adult females.

lmportantly, Dr. Kim T. Scribner was critical in assisting Russ to identify these

microsatellite markers. Also for Chapter 4, Pat Bills, Page Van Meter, and
Michelle Fisher were instrumental in helping me to update and maintain the
aggression table. Laura Sams and Wiline Pangle were extremely helpful in timing
the duration of many of the greetings presented in Chapter 5. Due to their
outstanding contributions, the following undergraduates have co-authored the
research presented here: Alison Cole, Steph Dawes, Jill Estrada, Katie Graham,
Adrienne Hopper, Sandra Memenis, Katie Powning, and Stacey Piotrowski.
Heather Richards and Karl Mitsos continue to be immensely helpful in two
ongoing projects that are not included in this dissertation.

I am also very grateful to Han de Vries, Vincent Carey, Julian Faraway,
Charlotte Hemelrijk, Andrew McAdam, Brian McGill, Wei Pan, Chris Strelioff. and
Jun Yan for helping me to implement statistical models. Along with my committee
members, Peter Henzi, Wiline Pangle, Robert Seyfarth, Gabrielle Ramos- .
Fernandez, and multiple anonymous reviewers provided comments that greatly
improved the overall quality of the published portions of this dissertation.

Many people have offered me friendship during my time at MSU. Although
I will refrain from listing them all, I would like to thank Katie Wharton Boes, Carrie
DeJaco, Shana Dumont, Carrie Kissman, Wiline Pangle, Amelia Kantrovitz, and
Anita Stone for their continued support despite my strange obsessions with
scholarly pursuits and my constant inability to answer the phone. I also thank all
of my ofﬁce mates for their encouragement over the years and for respecting my
silence “when the headphones go on” (my version of the thinking cap). In

particular, Wiline warmly welcomed me when I ﬁrst arrived to MSU and continues

to offer me advice and friendship when l have needed it most. Without the
collective humor, support and words of wisdom from these friends and others,
surely none of this research would have been possible.

Finally, I thank the Ofﬁce of the President of Kenya, the Kenya Wildlife
Service, the Kenyan Ministry for Science Education and Technology, the Narok
County Council, and the Senior Warden of the Masai Mara National Reserve for
giving me permission to live and conduct research in Kenya. Because my
research relied upon the long-term data set, this work was made possible by
National Science Foundation grants awarded to Kay Holekamp (BNSB706939,
BN39021461, IBN9309805, IBN9606939, IBN9630667, IBN9906445,
lBNO113170, IBN0343381, lOBOG18022, and lOS-0819437). My research was
also supported by grants from the PEG. International Sisterhood and the
American Society of Mammalogists. I also thank MSU for their generous support
during all phases of my dissertation work. The College of Natural Science
awarded me Recruiting, Graduate Summer Research, Dissertation Completion,
and Barnett Rosenberg Fellowships. I received the University Enrichment,
Research Enhancement, and Travel Fellowships from the Graduate School. The
Department of Zoology provided me with a John R. Shaver Award and Research
Grants. I also received fellowships from the Quantitative Biology Initiative and the
Program in Ecology, Evolutionary Biology, and Behavior. Without this ﬁnancial

support, none of this research would have been possible.

xi

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. xii

LIST OF FIGURES .............................................................. ' ................ xiv

CHAPTER 1:

GENERAL INTRODUCTION .................................................................... 1

CHAPTER 2:

SOCIAL AND ECOLOGICAL DETERMINANTS OF FlSSlON-FUSION

DYNAMICS IN THE SPOTTED HYENA .................................................... 13
Introduction ................................................................................ 14
Methods .................................................................................... 19
Results ..................................................................................... 32
Discussion ................................................................................. 50

CHAPTER 3:

RANK-RELATED PARTNER CHOICE IN THE FlSSlON-FUSION SOCIETY

OF THE SPOTTED HYENA .................................................................... 62
Introduction ................................................................................ 63
Methods .................................................................................... 66
Results ..................................................................................... 75
Discussion ................................................................................. 83

CHAPTER 4:

EVOLUTIONARY FORCES FAVORING INTRAGROUP COALITIONS AMONG

SPOTTED HYENAS AND OTHER ANIMALS ............................................. 94
Introduction ................................................................................ 95
Methods .................................................................................. 102
Results .................................................................................... 1 1 1
Discussion ............................................................................... 145

CHAPTER 5:

GESTURAL COMMUNICATION REINFORCES ALLIANCES AMONG

GREETING PARTNERS IN THE SPOTTED HYENA ................................. 154
Introduction .............................................................................. 1 55
Methods .................................................................................. 161
Results .................................................................................... 169
Discussion ............................................................................... 192

LITERATURE CITED ........................ . .................................................. 202

xii

LIST OF TABLES

Table 2.1. Each session was assigned to one of the following behavioural
contexts ................................................................................................ 25

Table 2.2. Mean E SE subgroup sizes and estimates of ﬂuidity for males (N = 5)
and females (N = 11) based on entire 24-h composite follows ........................ 34

Table 3.1. Spearman rank correlations between association indices or rank
distances and hourly rates of A) non-food and B) food-related dyadic aggression
directed by dominants towards subordinates, C) feeding tolerance of
subordinates by dominates, and D) hourly rates at which dominants provided
coalitionary support to subordinates within dyads of unrelated adult females....81

Table 4.1. Patterns of coalition formation during intragroup conﬂicts ............. 112

Table 4.2. Patterns of coalition formation among spotted hyenas belonging to
different age-sex classes within a single, large clan in the Masai Mara National
Reserve, Kenya ................................................................................. 123

Table 4.3. Independent variables predicting patterns of coalitionary support
donated by adult females on behalf of adult female beneﬁciaries during
interventions in dyadic ﬁghts ................................................................. 130

Table 4.4. Independent variables predicting patterns of coalitionary attacks
directed by adult females towards adult female victims during interventions in
dyadic ﬁghts ....................................................................................... 131

Table 5.1. Independent variables predicting the duration of greetings in which
focal hyenas participated ..................................................................... 171

Table 5.2. Independent variables predicting the duration of greetings based on
the relationships between initiators and recipients of greetings ..................... 172

Table 5.3. Independent variables predicting whether or not arriving adult females
greeted particular females present in joined groups at each opportunity to do so.
....................................................................................................... 187

Table 5.4. Independent variables predicting the symmetry of non-conciliatory
greetings among adult females .............................................................. 193

Table 5.5. Directional consistency (DC) of behavioral interactions among
gregarious mammals ........................................................................... 196

xiii

   

LIST OF FIGURES

Figure 2.1. Variation in subgroup size experienced by one adult female spotted
hyena during two long-term follows, A and B, conducted at different times. Follow
A consisted of 5 segments lasting an average of 4.8 hours each. Follow B
consisted of 3 segments, lasting an average of 8.0 hours each. Breaks between
continuous follow segments are indicated by asterisks (follow A) and plus signs
(follow B). Plotted values include the focal animal, and represent subgroup size
recorded every 10 min; a subgroup size of one indicates the focal animal was
alone. Shaded and open bars indicate hours of darkness and daylight,
respectively ......................................................................................... 33

Figure 2.2. Frequency distribution of subgroup sizes in which we found members
of the Talek clan throughout our longitudinal study (N = 34,848 observation
sessions). The inset box shows a magniﬁed view of the frequency of subgroup
sizes ranging from 19 to 39 .................................................................... 36

Figure 2.3. Mean :1: SE percentage of observation sessions in which (A) natal
animals were found alone during each life history stage, and (B) adult females (N
= 45) and immigrant males (N = 40) were found alone as a function of intrasexual
social rank. By convention, the highest possible rank is one. Sample sizes in (A)
shown above each bar represent numbers of individuals. Different letters above
bars indicate statistically signiﬁcant differences after correcting for multiple
testing ............................................................................................... 38

Figure 2.4. Mean :l: SE subgroup size (left vertical axis and histogram bars) and
proportion of observations in which Crocuta were found in subgroups containing
more than one individual (right vertical axis and open circles) as a function of the
context in which groups formed. Sample sizes, shown below each bar, represent
numbers of observation sessions assigned to each context. Different letters
indicate statistically signiﬁcant differences between contexts after correcting for
multiple testing. The shaded bar represents the baseline value of subgroup sizes
occurring in ‘other’ sessions, against which other groups were compared ........ 40

Figure 2.5. Mean 2!: SE percentage of observations in which A) adult females (N
= 38) were found alone or only with their dependent offspring as a function of the
female’s reproductive state, and B) natal animals were found with their mothers
during each life history stage. Sample sizes, shown above each bar in (B),
represent numbers of individuals observed in each life history stage. Different
letters indicate statistically signiﬁcant differences between categories after
correcting for multiple testing .................................................................. 42

xiv

Figure 2.6. (A) Monthly mean :I: SE numbers of prey animals counted each
month during biweekly ungulate censuses (left vertical axis and histogram bars)
and percentage of observation sessions in which Crocuta were found in
subgroups containing more than one individual (right vertical axis and open
circles). (B) Mean :l: SE subgroup size as a function of prey mass [logs of values
reported by (Kingdon 1997) and (Oindo 2002)] available at sessions with scraps
(N = 1315) or fresh kills. Sample sizes for kills were: Thomson’s gazelle (N =
382), impala (N = 53), wildebeest (N = 706), topi (N = 108), zebra (N = 193),
giraffe (N = 29), and elephant (N = 13) ...................................................... 46

Figure 2.7. Feeding subgroup size during the ﬁrst 15 min after solo hunters or
pairs of hunters killed ungulates of similar size (N = 9 matched pairs of 'hunts)..48

Figure 2.8. (A) Per capita daily energy gain as a function of the number of adult
females present at fresh ungulate kills (N = 41). (B) Mean :1: SE percentage of
adult females observed feeding per scan as a function of the number of adult
females present within each subgroup at kills (N = 426 sessions) ................... 49

Figure 2.9. Mean 1: SE percentage of scans in which focal females were
observed feeding at kills while holding (A) an absolute social rank position within
the clan (N = 24 absolute ranks), or (B) a relative social rank position within the
current subgroup (N = 14 relative ranks) ................................................... 51

Figure 3.1. Mean 1 SE intrasexual association indices plotted as a function of A)
intrasexual social rank and B) intrasexual rank distance for unrelated adult

Crocuta. Squares represent adult females and circles represent adult males that
held a particular rank while observed with an unrelated adult of the same sex. By
convention, the highest rank possible is 1 .................................................. 76

Figure 3.2. Mean :l: SE association indices of Crocuta of each sex with animals
of the same sex occupying rank positions either immediately above, or
immediately below, them in the hierarchy of the clan. The asterisk over the
bracket indicates P < 0.05, and NS. indicates P > 0.87 ................................ 77

Figure 3.3. Proportion of unrelated subordinate and dominant adult females
available to join that actively joined subgroups containing focal females, divided
by the number of hours during which we monitored each focal female. The
asterisk over the bracket indicates P < 0.05 ............................................... 79

XV

 

Figure 3.4. Hourly rates of dyadic aggression directed by dominant females
towards subordinate females when food was A) absent and B) present (N = 367
dyads for both). C) Feeding tolerance, measured as the percent of scans in
which a dominant female tolerated a subordinate female feeding at a kill when
both were concurrently present, plotted against association indices within 37
dyads ................................................................................................ 82

Figure 3.5. A) Proportion of subordinate and dominant females present during
ongoing ﬁghts that intervened to provide coalitionary support to focal female
aggressors per hour. The asterisk over the bracket indicates P < 0.05. B) The
relationship between association indices and rates at which dominant females
provided coalitionary support to subordinate females within 375 dyads ........... 84

Figure 4.1. Mean :1: SE interventions per opportunity in which the same adult
females (N = 24) intervened in ﬁghts between two other adult females based on
the context of the original ﬁght. Different letters above bars indicate statistically
signiﬁcant differences after correcting for multiple testing ........................... 128

Figure 4.2. Proportion of ﬁghts in which adult females supported beneﬁciaries
subordinate or dominant to themselves, out of all opportunities to intervene in the
presence or absence of food. Bar color indicates kinship of each dyad as non-kin
(white), distant kin (gray), and close kin (black). Sample sizes over each bar
indicate the number of opportunities potential donors had to support potential
beneﬁciaries. Error bars represent 1 1 standard deviation for binomial trials... 132

Figure 4.3. Cooperation networks among females present concurrently as adults
from 1997 to 1998 that were A) non-kin, B) distant kin, and C) close kin. Each
circle (node) represents an adult female. Node size is proportional to the social
rank held by that female (e.g. alpha female is largest). Connected dyads within
each network represent dyads belonging to a particular kinship group based on
maternal and paternal kinship. Whereas dyads connected by solid black arcs
formed coalitions, dyads connected by dashed gray lines failed to support one
another in ﬁghts despite the fact that both dyad members had opportunities to do
so. Solid, black arrows (arcs) originate at donors of support, and arrowheads
point towards beneﬁciaries of support. Double-headed arcs represent reciprocal
support within dyads, whereas single-headed arcs represent unilateral support.
Densities reﬂect the number of realized arcs (number of solid black lines) divided
by the total number of possible connections (number of dashed gray lines and
solid black lines for each kin group). Line thickness (weighted edges) indicates
the proportion of times adult females intervened out of all opportunities to do so.
...................................................................................................... 133

xvi

Figure 4.4. Variation among kinship categories regarding the proportion of dyads
in which both members supported one another (reciprocal support: black), only
one member supported the other member (unilateral support: gray), or neither
member provided support to the other (no support: white). Sample sizes over
bars indicate the number of dyads in which both members had opportunities to
provide reciprocal support (N = 106 dyads) .............................................. 140

Figure 4.5. Proportion of ﬁghts in which adult females attacked victims
subordinate or dominant to themselves, given the opportunity to intervene in the
presence or absence of food. Bar colors indicate whether females intervened in
ongoing dyadic ﬁghts of low (black) or high (white) intensity. Dyads belonging to
all kinship categories are pooled because kinship failed to protect victims from
attacks. Sample sizes over bars indicate the number of opportunities females
had to attack targets already under attack by another adult female. Error bars
represent 1 1 standard deviation for binomial trials .................................... 141

Figure 4.6. The effect of absolute social rank on the proportion of ﬁghts in which
females intervened on behalf of other adult females out of all opportunities in
which females holding a particular social rank might have supported potential
beneﬁciaries. By convention, the highest possible rank Is one (e.g., the alpha
female) ............................................................................................. 143

Figure 4.7. Agonistic network: A directed, weighted attack network structured by
the same adult females as in Figure 4.3. Solid, black arrows (arcs) originate at
intervening aggressors, and arrowheads terminate at victims of coalitionary

- attack. Line thickness (weighted edges) indicates the proportion of times adult
females targeted the other member of the dyad out of all opportunities to do so.
All dyads are shown on a single network because kinship failed to protect
females from coalitionary attacks ........................................................... 144

Figure 5.1. Mean i SE hourly rates at which hyenas in each life history stage
participated in greeting ceremonies at kills, dens, and “other” contexts (both away
from kills and away from dens). Because we detected no sex differences in the
rates at which den cubs or den-independent natal animals greeted, we pooled
the values within each life history stage. Sample sizes shown above each bar
represent numbers of individuals in each category. Letters above bars indicate
statistically signiﬁcant differences after correcting for multiple testing at P < 0.05.
....................................................................................................... 175

xvii

Figure 5.2. Percentage of greetings with adults initiated by juveniles. Den cubs
were those juveniles still residing at the den whereas subadults were those
juveniles that were independent of the den but not yet adults. Sample sizes
shown above each bar represent numbers of individuals involved in greetings
with partners in each adult category. Letters above bars indicate statistically
signiﬁcant differences after correcting for multiple testing ............................ 176

Figure 5.3. A dominance matrix based on the outcomes of 717 dyadic ﬁghts,
each of which had a clear winner and loser. Each row of the matrix represents a

' different adult female (N = 19 hyenas), all of which were present together as
adults both at food and away from food. At the intersection of each row (the
winner) and column (the loser), a cell shows the number of ﬁghts won by the
aggressor against the loser. We listed individuals based on their rank order, with
the alpha female (bsh) represented in the top row and the left most column of the
matrix. Similarly shaded adjacent cells within the ﬁrst row and the left most
column represent adult females belonging to the same matriline. For example,
the alpha matriline contains bsh, mrph, sein, gil, bb, kip, who, and mali. A total of
ﬁve matrilines are represented within the matrix. Dyadic relationships were either
unidirectional (71.9 %, N = 123), bidirectional (4.1 %, N = 7), or unknown (24.0
%, N = 41). Black squares indicate that one adult female member of the dyad
won the majority of ﬁghts within that dyad. White squares indicate dyads for
which an equal number or no agonistic interactions were observed ............... 179

Figure 5.4. A greeting interaction matrix based on 354 greetings, each of which
had a clear initiator and recipient. Because the individuals included here (N = 19
hyenas) were the same adult females shown in Figure 5.5, and because the act
of initiating greetings failed to produce a signiﬁcantly non-random ordering of
relationships, we ranked females based on the dominance rank order established
in Figure 5.5. As before, adult females belonging to the same matriline are shown
in adjacent cells in the top row and left-most column are shaded in the same
color. At the intersection of each row (the recipient of greetings) and column (the
solicitor of greetings), a cell shows the number of greeting interactions received
by a particular individual from the initiating hyena. Dyadic relationships were
either unidirectional (39.7 %, N = 68), bidirectional (21.1 %, N = 36), or unknown
(39.2 %, N = 67). Black squares indicate that one member of the dyad
preferentially received greetings more often than it initiated greetings with the
other adult female within that dyad. White squares indicate dyads for which an
equal number of greetings or no initiations were observed. ......................... 181

Figure 5.5. Percentage of all non-conciliatory greetings (N = 11,759) that

occurred among spotted hyenas within the 5 minute-intervals directly after
subgroup reunions (“fusion events") throughout our longitudinal study ........... 185

xviii

Figure 5.6. Proportion of fusion events in which adult females greeted other adult
females within 10 minutes of subgroup fusion, out of all opportunities to do so, at
kills (food context) and away from kills (non-food context). Coalition partners
formed at least one aggressive coalition directed towards another hyena or
towards a lion during the minutes after a fusion event, non-coalition partners
failed to form a coalition in the minutes after a fusion event. Although bar color
here indicates whether members of each dyad were non-kin (white) and kin
(black), we entered the coefﬁcient of relatedness for each dyad as a continuous
variable in our statistical model (Table 5.2). Sample sizes over each bar indicate
the number of opportunities available to arriving females to greet potential adult
female partners. Error bars represent :l: 1 standard deviation for binomial trials.
....................................................................................................... 1 88

, Figure 5.7. Proportion of fusion events in which adult females greeted other adult
females within 10 minutes of subgroup fusion, out of all opportunities to do so, as
a function of how often pairs of adult females associated with each other. For the
purposes of visual representation only, we present patterns of association binned
into discrete categories, but entered values as a continuous variable into our
statistical model (Table 5.2). Sample sizes over each bar indicate the number of
opportunities arriving adult females had to greet potential adult female partners at
fusion events. Error bars represent 1: 1 standard deviation for binomial trials
..................................................................................................... 1 90

xix

Chapter 1

GENERAL INTRODUCTION

Understanding the evolution of sociality in general, and cooperation in particular,
is a central problem in biology, with a rich history tracing back to Charles Danlvin
(reviewed by CIutton-Brock, 2009; Dugatkin, 2002; Noe, 2006; Nowak, 2006;
Pennisi, 2005; Queller, 1985; Sachs et al., 2004; West et al., 2007a; West et al.,
2007b). The traditional approach to studying the evolution of social behavior is to
recognize that, although gregarious animals often gain ﬁtness beneﬁts from living
in groups, sociality can be costly when group members compete for access to
the resources necessary for survival and reproduction (Alexander, 1974; Ridley
et al., 2005; Silk, 2007a; Sterck et al., 1997b; Vehrencamp, 1983). Thus, most
empirical workers employ a cost-beneﬁt approach, testing optimization theory, to
evaluate the extent to which individuals make decisions that minimize the costs
of group life and maximize its beneﬁts.

Biological market theory, like traditional optimization theory, predicts that
the beneﬁts of cooperative partnerships should outweigh the costs (Noe and
Hammerstein, 1994; Noé and Hammerstein, 1995; Noe et al., 1991). However, it
further predicts that the relative values of partnerships depend upon the current
state of the biological marketplace. Within this framework, the cooperative
decisions that individuals make are expected to vary dynamically in response to
local variation in social and ecological conditions. Speciﬁcally, market theory
accounts for partner switching by extending two-player games to include multiple

players. Within this paradigm, natural selection is expected to favor those

individuals best able to track changes in their environments, and to make
adaptive decisions based on their immediate circumstances.

Market forces should be of particular importance in those societies in
which the immediate number and quality of available social partners is constantly
in ﬂux (Noe and Hammerstein, 1994). Such circumstances are prevalent in
societies structured by ﬁssion-fusion dynamics (Aureli et al., 2008); members of
these societies are continuously faced with decisions regarding whether to
separate from (ﬁssion) or join (fusion) conspeciﬁcs who are members of their
permanent social group, but who are usually present in small subgroups
(Kummer, 1971; Wittemyer et al., 2005). As a result, although all group members
recognize each other and belong to a single overarching social unit, all group
members are rarely, if ever, concurrently together. Instead, individuals are
distributed across the landscape within temporary subgroups that may change
size and composition on an hour-to-hour basis.

This ﬂexible lifestyle theoretically permits individuals to separate
temporarily from one another when the costs of grouping are high and to
aggregate when the costs of grouping are low or the beneﬁts of sociality are high
(Aureli et al., 2008; Chapman et al., 1995; Schino, 2000; Wrangham et al., 1993).
Although this societal structure permits individuals to reduce conﬂicts of interest
by temporarily separating from group members, it simultaneously imposes a
unique set of challenges on group members who are regularly subject to
prolonged separation (Aureli et al., 2008). As a result, individuals living in these

continuously changing social environments must cope with uncertain social

relationships at reunions (Barrett et al., 2003).

Why study biological market theory in spotted h yenas?

My dissertation tests several predictions derived from market theory, primarily
using data from a long-term study initiated in 1988 on a single, large social group
of spotted hyenas (Crocuta crocuta) that defend a common territory in the Masai
Mara Reserve in Kenya, East Africa. Spotted hyenas live in ﬁssion-fusion
societies structured by dominance hierarchies in which group members vary
dramatically in their value as social partners, and they are subject to extreme
seasonal variation in local resource availability. Therefore, these animals offer an
excellent system in which to test predictions of biological market theory. The
longitudinal data available here provide a unique opportunity to investigate the
dynamical nature of social relationships among a natural population of known
individuals.

Why focus on adult female spotted hyenas?

My dissertation research focuses mainly on explaining how evolutionary and
ecological forces favor patterns of cooperation among adult female spotted
hyenas. I focused on adult females in particular because these animals are
philopatric (Smale et al., 1997), highly gregarious (Holekamp et al., 1997a; Smith
et al., 2007), long—lived (up to 19 years in the wild, Drea and Frank, 2003), and
demonstrate many of the same cognitive abilities as monkeys (Holekamp et al.,
2007), permitting them to remember earlier interactions. Further, adult females
are the most powerful animals in these female-dominated societies (Frank,

1986), and they use an extraordinarily bizarre, fully erectile penile clitoris to

update relationship status at subgroup reunions (East et al., 1993; Kruuk, 1972).
Social lives of spotted h yenas resemble those of monkeys
Interestingly, the hierarchical structure and size of spotted hyena societies, called
clans, are more similar to those characteristics of troops of cercopithecine
monkeys than to those characterizing groups of other social carnivores (reviewed
by Drea and Frank, 2003; Holekamp et al., 2007). Although most social
carnivores live in kin groups, hyena clans are multigenerational groups that
usually contain up to 80 members, many of whom are unrelated to one another.
Spotted hyena clans contain multiple matrilines of adult females and their
offspring, as well as multiple adult immigrant males born elsewhere (Kruuk,
1972). Like most troops of monkeys (reviewed by Gouzoules and Gouzoules,
1987; Silk, 1987), each hyena clan is structured by a linear dominance hierarchy
(Frank, 1986; Kruuk, 1972). As are many monkey societies (e.g. Bercovitch,
1988; Seyfarth and Cheney, 1984; Silk et al., 2004; Widdig et al., 2000), coalition
formation plays an important role in the acquisition and maintenance of rank
positions within this hierarchy (Engh et al., 2000; Holekamp and Smale, 1991;
Smale et al., 1995). Further, the social rank of each hyena in the clan determines
its priority of access to food and other resources (Frank, 1986; Tilson and
Hamilton, 1984).

As is true in many cercopithecine primates (e.g. Bergman et al., 2003;
Cheney and Seyfarth, 1982; Dasser, 1988), spotted hyenas recognize their
group mates as unique individuals (Holekamp et al., 1999a), and they also

recognize third-party kin and rank relationships among their clan mates (Engh et

al., 2005). Moreover, like monkeys (reviewed by Silk, 2002; Widdig, 2007),
spotted hyenas discriminate among potential social partners on the basis of
maternal and paternal kinship, and associate most often with their close kin
(Holekamp et al., 1997a; Van Horn et al., 2004b; Wahaj et al., 2004). Finally, as
do many species of monkeys, spotted hyenas engage in intimate and risky
behavioral interactions, called greetings, in which individuals inspect the genitalia
of their social partners (East et al., 1993; Kruuk, 1 972).

These similarities present a unique opportunity to compare the
evolutionary forces shaping the societies of primates and carnivores, taxonomic
groups that last shared a common ancestor 90-100 MYA (Springer et al., 2003;
Springer et al., 2005). The convergent evolution in the social behavior of these
animals will allow me to test the generalizability of current socioecological
models. Although most of these models were originally proposed to explain
social behavior in primates, they should theoretically also explain the social
behavior of non-primates. Therefore, by testing their predictions in this model
social carnivore, the broad goal of my dissertation research is to evaluate the
widespread utility of current theoretical explanations for the evolution of sociality
in general, and cooperation in particular.

Important diﬁ'erences between spotted h yenas and monkeys

Despite all of the similarities outlined above, several key differences exist
between the social lives of spotted hyenas and those of cercopithecine primates.
First, unlike monkeys, which often live in highly cohesive groups, spotted hyena

clans are ﬁssion fusion societies in which individuals make active decisions to

leave (ﬁssion) and to join (fusion) subgroups of clan mates forming within the
larger social unit. Individual members travel, rest, and forage in subgroups that
change membership roughly every hour (Kruuk, 1972; Mills, 1990). Second,
although monkeys feed mainly on fruits and vegetable matter (e.g. Chapman and
Pavelka, 2005; Wrangham et al., 1998), spotted hyenas prey upon ungulates that
they mainly hunt themselves (Cooper, 1990; Holekamp et al., 1997b; Honer et
al., 2002; Kruuk, 1972; Mills, 1990). Because ungulate carcasses are both
energetically rich and highly ephemeral, feeding competition is extremely intense
among hyenas (Frank, 1986; Kruuk, 1972). Third, in contrast to the status of
relationships in most mammals, all adult female hyenas are socially dominant to
all adult immigrant males (Kruuk 1972; Frank 1986). Moreover, the reproductive
anatomy of hyenas is unique in that adult females possess a fully erectile clitoris
(Frank, 1983; Matthews, 1939; Neaves et al., 1980). Females urinate, copulate,
and give birth through this “pseudo-penis” and the erect phallus plays a
prominent role during reunion displays, called greeting ceremonies (East et al.,
1993; Kruuk, 1972).

Overview of the chapters

Here I outline the rationale and major ﬁndings of the four data chapters in this
dissertation. In Chapter 2, I tested hypotheses to identify the social and
ecological determinants shaping ﬁssion-fusion dynamics in spotted hyenas. In
particular, I tested the theoretical prediction that individuals living in ﬁssion-fusion
societies, in which group members frequently change subgroups, should modify

grouping patterns in response to varying social and environmental conditions. I

provide the ﬁrst detailed description of ﬁssion-fusion dynamics available for this
species. Then, because social and ecological circumstances can inﬂuence the
cohesiveness of animal societies, l evaluated the extent to which speciﬁc
circumstances promote the formation of hyena subgroups of various sizes. I
found that cooperative defense of shared resources during interclan competition
and protection from lions were cohesive forces that promoted formation of large
subgroups. Finally, I tested multiple hypotheses, each of which suggests factors
that might limit subgroup size. I found evidence for all three hypotheses, which
are not mutually exclusive. First, as predicted by the infant safety hypothesis
(Otali and Gilchrist, 2006), mothers with small cubs avoided conspeciﬁcs, thereby
reducing risk of infanticide. Second, as predicted by the dispersive conﬂict
resolution hypothesis (Schino, 2000), victims of aggression either reconciled
ﬁghts or separated from former opponents to reduce the immediate costs of
escalated aggression in the absence of food. Most importantly, however, as
predicted by the ecological constraints hypothesis (Chapman et al., 1995),
hyenas adjusted their grouping patterns over both short and long time scales in
response to feeding competition. Spotted hyenas were most gregarious during
periods of abundant prey, joined clanmates at ephemeral kills in numbers that
correlated with the energetic value of the prey and gained the most energy when
foraging alone because kills made during cooperative hunts attracted numerous
competitors. Overall, these ﬁndings indicate that whereas intergroup competition
promotes group augmentation, intense intragroup competition over limited

resources constrains grouping in this species. These ﬁndings were published in

Animal Behavior (Smith et al., 2008).

In Chapter 3, I investigated how social rank inﬂuences individual
differences with respect to the ﬁssion-fusion dynamics documented in Chapter 2.
In particular, I studied rank-related partner choice by testing hypotheses
suggesting the beneﬁts of mutual partner choice among unrelated adult females.
This study revealed that adult females actively join subgroups containing
preferred social partners, and that market forces shape the decision-making
process regarding whether or not to join a particular subgroup. Because patterns
of association among spotted hyenas reﬂect social preferences, I calculated
association indices (Als, Calms and Schwager, 1987) to assess the effects of
social rank on intrasexual partner choice among unrelated adults. Among adult
females, the highest-ranking individuals were generally most gregarious; females
associated most often with dominant and adjacent-ranking females. Females
joined subgroups based on the presence of particular conspeciﬁcs such that
subordinates joined focal females at higher rates than did dominants. Dominants
beneﬁt from associations with subordinates by enjoying priority of access to
resources obtained and defended by multiple group members, but the beneﬁts of
these associations to subordinates were previously unknown. To investigate this,
I tested three hypotheses suggesting how subordinates might beneﬁt from rank-
related partner choice among unrelated females. Subordinates who initiated
group formation beneﬁted by gaining social and feeding tolerance from
dominants. However, the extent to which females associated failed to predict the

hourly rates at which the dominant member of the dyad intervened during

ongoing ﬁghts to provide coalitionary support to the subordinate member of each
dyad. Overall, my data resemble those documenting patterns of association
among cercopithecine primates. I consider our results in light of optimal
reproductive skew theory (Clutton-Brock, 1998; Hamilton,'2000; Reeve et al.,
1998), Seyfarth’s rank attractiveness model (1977), and biological market theory
(Noé, 2001; N06 and Hammerstein, 1995). These data are more consistent with
the predictions of Seyfarth’s model and of biological market theory than with
those of skew theory. This research was published in Behavioral Ecology and
Sociobiology (Smith et al., 2007).

Because market forces failed to explain patterns of coalitionary support
among unrelated adult females in Chapter 3, I then initiated Chapter 4 to
elucidate the evolutionary forces that are acting to favor coalitionary interventions
among adult female spotted hyenas. First, I performed a comprehensive
literature review of 49 vertebrate species to test Harcourt's (1992) hypothesis that
intragroup coalitions formed by primates differ systematically from those formed
by non-primates. However, I found that patterns of intragroup coalition formation
are in fact remarkably similar between primates and non-primates. Then, I tested
hypotheses suggesting kin selection (Hamilton, 1964), reciprocal altruism
(Trivers 1971), and direct beneﬁts (also called by-product mutualisms, Brown,
1983; Connor, 1995; West-Eberhard, 1975) as adaptive explanations for
coalitionary interventions among adult female spotted hyenas. As predicted by
kin selection theory, female hyenas supported close kin most often, and the

density (connectedness) of cooperation networks increased with genetic

relatedness. Nevertheless, kinship failed to protect females from coalitionary
attacks. I found no evidence of enduring alliances based on reciprocal support
among unrelated adult females. Instead, donors generally minimized costs to
themselves, intervening most often during low-intensity ﬁghts and when feeding
opportunities were unavailable. Females also gained direct beneﬁts by directing
coalitionary attacks toward subordinates. Finally, females monitored the number
of dominant bystanders present in the "audience" at ﬁghts, and modiﬁed their
level of cooperation based on this knowledge. Overall, hyenas made ﬂexible
decisions regarding whether or not to intervene in ﬁghts, modifying their tendency
to cooperate based on multiple types of information about their immediate social
and ecological environments. Taken together, these ﬁndings indicate that the
combined evolutionary forces of kin selection and direct beneﬁts derived from
reinforcing the status quo drive coalitionary interventions among adult female
spotted hyenas. The resUlts of this chapter were recently publishedrin Behavioral
Ecology (Smith et al., 2010).

Finally, in Chapter 5, I investigated the function of non-conciliatory
greetings among adult female spotted hyenas during subgroups reunions. These
interactions among spotted hyenas are of particular interest because, although
their flexible lifestyle permits females to reduce conﬂicts of interest, ﬁssion-fusion
sociality simultaneously imposes a unique set of challenges on group members
that are regularly subject to prolonged separation (Aureli et al., 2008; Barrett et
al., 2003). Theory predicts that animals whom separate frequently from one

another should evolve ritualized displays to quickly update uncertain

10

relationships at reunions (Endler, 1993; Maynard Smith and Price, 1973). I tested
predictions derived from the submission (de Waal, 1986; de Waal and Luttrell,
1985; Preuschoft and van Schaik, 2000), tension reduction (Aureli and Schaffner,
2007; Dias et al., 2008; Kutsukake et al., 2006; Schaffner and Aureli, 2005), and
social bonding hypotheses (Smuts, 2002; Smuts and Watanabe, 1990; Zahavi,
1977b). Because the directional consistency of hyena greetings was low, and
because females greeted most per opportunity when food was absent, our data
fail to support the submission or tension reduction hypotheses. Whereas rank
effects were relatively weak, the best statistical model was consistent with
predictions of the social bonding hypothesis; females greeted coalition partners
and close associates, including kin, most often per opportunity at fusion events.
Further, the act of greeting coordinated cooperation by promoting coalition
formation among allies. Overall, the results of Chapter 5 indicate that risky
greeting gestures permit hyenas to effectively communicate their immediate
commitment to alliance afﬁliations within a continuously shifting social milieu.
Collaborative nature of this research project

The results of my dissertation research elucidate the evolutionary and ecological
forces shaping the dynamics of cooperation. More broadly, these ﬁndings
deepen our understanding cf the rules governing social decision-making in free-
living vertebrates, and they inform theoretical approaches to the evolution of
sociality by testing the generalizability of socioecological models across
taxonomic groups. lmportantly, these results and their potential inﬂuences on the

ﬁelds of ecology, evolutionary biology and behavior would not have been

11

possible without the efforts of multiple collaborators.

Drs. Kay E. Holekamp, Laura Smale, numerous ﬁeld assistants and
graduate students working on the Mara Hyena Project helped me to collect the
long-term data analyzed in this dissertation. Chapters 2 and 4 were generated in
collaboration with two former graduate students in the Holekamp lab, Drs.
Joseph M. Kolowski and Russell C. Van Horn, respectively. An amazing team of
undergraduates also assisted in enriching the conceptual frameworks tested
throughout my dissertation and helped me to extract each of the relevant data
sets from archived ﬁeld notes. Many of these individuals served as on each of
the co-authors the published chapters or will co-author of Chapter 5.

Because of the outstanding contributions of the aforementioned
collaborators, I will use the term “we” in each of the following data chapters to

indicate that my dissertation research was, in fact, a collaborative effort.

12

Chapter 2
Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE, 2008. Social and

ecological determinants of ﬁssion-fusion dynamics in the spotted hyaena.
Animal Behaviour 76:619-636.

13

Chapter 2
SOCIAL AND ECOLOGICAL DETERMINANTS OF FlSSION-FUSION
DYNAMICS IN THE SPOTTED HYENA

INTRODUCTION
Most mammalian carnivores are solitary, spending their lives alone except when
breeding (Gittleman, 1989). Among the roughly 20% of carnivore species that are
at least somewhat gregarious, a few species live in groups that are highly
cohesive, such as wild dogs (Lycaon pictus) and various species of mongooses.
However, like elephants (Loxodonta spp.), cetaceans (e.g. bottlenose dolphins,
Tursiops truncatus), and certain primates, most gregarious carnivores live in
groups commonly referred to as ﬁssion-fusion (FF) societies. FF societies are
stable social units in which individual group members are often found alone or in
small subgroups, and in which subgroup size and composition change frequently
over time. In the FF societies of hamadryas (Papio hamadryas), gelada baboons
(Theropithecus gelada), and elephants, stable subgroups that contain multiple
individuals join (fusion) and break away from (ﬁssion) other stable subgroups
belonging to the larger social unit (Kummer, 1971; Wittemyer et al., 2005). By
contrast, the FF societies of gregarious carnivores are typically individual-based
(Rodseth et al., 1991) such that individual group-members are commonly found
alone, and individually make decisions to join or leave subgroups (Gittleman,
1989).

Gregarious carnivores living in FF societies typically know each other as

individuals and defend a common territory, but all group members rarely occur

14

together concurrently (reviewed by Holekamp et al. 2000). Although group
members seldom exhibit signs of distress when separating from group-mates,
they typically engage in reunion displays upon subgroup fusion (Holekamp et al.
2000). Terrestrial carnivores that live in societies with these characteristics
Include lions (Panthera leo), coatis (Nasua spp.), European badgers (Me/es
meles), dingoes (Canis lupus dingo), coyotes (C. latrans), dholes (Cuon alpinus),
kinkajous (Potos avus), brown hyenas (Parahyaena brunnea) and spotted
hyenas (Crocuta crocuta).

Spotted hyenas are long-lived carnivores that reside in permanent social
groups, called clans, in which individual members travel, rest, and forage in
subgroups (Kruuk, 1972; Mills, 1990) that change membership multiple times per
day (Kolowski et al., 2007). Virtually all males permanently disperse from their
natal clans after puberty, whereas females are philopatric (East and Hofer, 2001;
Mills, 1990; Smale et al., 1997). Clans contain one to several matrilines of adult
females and their offspring, as well as multiple adult immigrant males. Individuals
choose to join subgroups containing particular clan members, and they vary in
the extent to which they associate with conspeciﬁcs (Smith et al., 2007; Szykman
et al., 2001). Hyenas associate most often with kin (Holekamp et al., 1997a;
Wahaj et al., 2004). Among non-kin, hyenas prefer to join subgroups containing
potential mates (Szykman et al., 2001) and same-sexed social companions who
are higher-ranking than, but close in rank to, themselves (Smith et al., 2007).
Although up to 80 individuals may belong to a single clan concurrently (Henschel

and Skinner, 1991; Kruuk, 1972; Mills, 1990), all clan members are rarely, if ever,

15

found together in one place (Holekamp et al., 2000). Here we provide the ﬁrst
detailed desoription of FF dynamics in the spotted hyena. Because social and
ecological circumstances can promote or constrain the cohesiveness of animal
societies (Chapman, 1990; Chapman et al., 1995; Wrangham et al., 1993), we
ﬁrst evaluate the extent to which speciﬁc circumstances promote the formation of
hyena subgroups of various sizes, and the tendency for individual hyenas to be
found alone or with conspeciﬁcs. We then test three hypotheses suggesting
factors limiting subgroup size in this species.

First, the infant safety hypothesis (Otali and Gilchrist, 2006) predicts that
reproduction is a disruptive force in FF societies in which offspring are vulnerable
to infanticide, the direct killing of infants by older conspeciﬁcs. Because adult
Crocuta are known to commit infanticide (East and Hofer, 2002; Kruuk, 1972;
White, 2005), we test two predictions derived from this hypothesis. Mammalian
offspring are especially vulnerable to infanticide immediately after parturition
(Agrell et al., 1998), so we expected adult females to spend the most time away
from other conspeciﬁcs during early lactation. We also expected young hyenas to
be found most often with their mothers during early life history stages in which
young are most vulnerable to infanticide.

Second, the dispersive conﬂict resolution hypothesis (Schino, 2000)
proposes that costs associated with physical combat, such as energy
expenditure and risk of injury or death, limit gregariousness among animals living
in FF societies. Crocuta frequently direct aggression towards clan-mates to

establish and maintain rank relationships even in the absence of food (Kruuk,

16

 

1972; Smale et al., 1993). Because Crocuta are well armed with massive teeth
and jaws, victims of aggression risk injury resulting from continued or escalated
ﬁghting during within-group conﬂicts (Kruuk, 1972). Thus, individuals might -
reduce the short-term costs of conﬂict by relying on dispersive mechanisms to
avoid or resolve ﬁghts. If this is the case, then hyenas should leave subgroups
more often after receiving aggression than when conspeciﬁcs direct no
aggression towards them. Further, hyenas sometimes resolve conﬂicts by
engaging in conciliatory behaviors such as greetings and/or non-aggressive
approaches (Hofer and East, 2000; Wahaj et al., 2001). If reconciliation promotes
social cohesion in this species, then targets of aggression should remain in
subgroups more often when they reconcile with former opponents than when no
reconciliation occurs.

Finally, the ecological constraints hypothesis (Chapman et al., 1995)
posits that resource competition, as affected by both short-tenn and seasonal
ﬂuctuations in resource availability, limits subgroup size among animals that
otherwise beneﬁt from grouping. This hypothesis explains FF dynamics in a
number of non-human primates, but should also theoretically be able to explain
grouping patterns in a broad range of gregarious taxa, including spotted hyenas.
Crocuta beneﬁt from grouping because multiple, often unrelated (Van Horn et al.,
2004a), clan members cooperate to obtain and defend resources from
kleptoparasitism by neighboring hyena clans and lions, and also to protect
clanmates from direct killing by lions (Boydston et al., 2001; Henschel and

Skinner, 1991; Kruuk, 1972). Nevertheless, hyenas compete intensely with

17

group-mates for limited food, comprised mainly of ungulate prey they have killed
themselves (Engh et al., 2000; Frank, 1986; Kruuk, 1972; Tilson and Hamilton,
1984). Ungulate carcasses represent energy-rich food patches that are both
ephemeral and usurpable, and an individual’s priority of access to food is
determined by its social rank (Engh et al., 2000; Frank, 1986; Tilson and
Hamilton, 1984). If the ecological constraints hypothesis is correct, then given
their reduced priority of access to resources, low-ranking hyenas should spend
more time alone than high-ranking ones. We also expected heterogeneity in the
foraging environment to inﬂuence grouping patterns (Ramos-Fernandez et al.,
2006). If feeding competition in particular constrains subgroup size, then hyenas
should congregate at food patches in numbers proportional to the amount of
energy contained within patches, and they should spend relatively more time with
conspeciﬁcs than alone when prey are superabundant. Further, low-ranking
hyenas should be particularly vulnerable to costs associated with feeding in large
subgroups. To test this prediction, we replicate earlier work (Frank, 1986) by
inquiring whether social rank determines feeding success in adult female
Crocuta, and extend it by examining how a female’s relative rank within her
current subgroup inﬂuences her ability to feed. Finally, although energy gain
increases with group size in some carnivores living in cohesive societies (e.g.
wild dogs, Creel & Creel 1995; Creel 1997), the ecological constraints hypothesis
predicts that per capita energy intake and the proportion of time individuals
spend feeding should decline with increasing subgroup size among spotted

hyenas.

18

METHODS

Study Populations

We monitored two large Crocuta clans inhabiting the Masai Mara National
Reserve, Kenya. From July 1988 through December 2004, we monitored hyenas
in the Talek clan. From August 2002 to March 2004, we also studied the Mara
River clan, located 8 km west of Talek, in an area with habitat types and prey
abundance that did not differ signiﬁcantly from those in the Talek area (Kolowski
et al., 2007). We identiﬁed individuals in both clans by their unique spots. From
1988 to 1999, the Talek clan defended a stable group territory covering an area
of 62 km2 (Boydston et al., 2001). Starting in 2000, the original Talek clan
permanently split to form two new clans, Talek East and Talek West, defending
adjacent territories of 19 km2 and 28 kmz, respectively (Kolowski et al., 2007).
Members of the Mara River clan defended a territory of 31 kmz. Subjects in the
current study were members of the original Talek, the Talek West, and the Mara
River clans.

Resident ungulates grazing year round in the study areas include
Thomson's gazelle (Gaze/la thomsonii, average body mass: 25 kg), impala
(Aepyceros melampus, 53 kg), and topi (Damaliscus korn’gum, 119 kg). Large
migratory herds of wildebeest (Connochaetes taun'nus, 132 kg) and zebra
(Equus burchelli, 235 kg) join resident ungulates annually between June and
September; the superabundance of prey during these months relaxes feeding
competition among hyenas (Holekamp et al. 1993, 1996). Crocuta in our study

areas hunt all of these species and occasionally also scavenge carcasses of

19

adult giraffe (Giraffe cameIOpardalis, 935 kg) and elephants (3550 kg). Mean
masses reported here are from Kingdon (1997) and Oindo (2002).

Here we estimated (to i- 7 days) the ages of cubs upon ﬁrst observing them
above ground (Holekamp et al., 1996). We sexed hyenas based on the
morphology of the erect phallus (Frank et al., 1990). Adult females bear young in
isolated natal dens and transfer them to a communal den when cubs are 2 to 5
weeks old (East et al., 1989; Kruuk, 1972). There is no allonursing or communal
care of young in this species (Mills, 1985). We considered cubs to be
independent of dens when we found them more than 200 m from the current
communal den on at least 4 consecutive occasions; this occurred when cubs
were around 9 months of age (Boydston et al., 2005). On average, den-
independent cubs nurse from their mothers until they are 14 months old
(Holekamp et al., 1996). Here we assigned weaning dates (to i 10 days) based
on observed weaning conﬂicts and the cessation of nursing (Holekamp et al.,
1996). We considered natal animals older than 24 months to be reproductively
mature adults (Glickman et al., 1992).

We ranked adults in a linear dominance hierarchy, based on outcomes of
dyadic agonistic interactions (Holekamp and Smale, 1993; Smale et al., 1993).
All adult female spotted hyenas breed, but high-ranking females enjoy greater
reproductive success than do low-ranking females (Frank et al., 1995; Hofer and
East, 2003; Holekamp et al., 1996). All adult females are dominant to all
immigrant males. Here we ranked immigrant males and adult females in separate

hierarchies, with one being the highest possible rank in each. We assigned

20

relative ranks to adult females at each kill based on their positions within the
dominance hierarchy relative to those of the other females present at that kill.
Behavioral Data Collection
We used two general methods to collect behavioral data: long-term focal animal
‘follows,’ and short-term observation ‘sessions.’ From 2002 to 2004, we
conducted focal follows on 19 adults (11 females, 8 males) ﬁtted with radiocollars
(T elonics lnc., Mesa, Arizona). Focal animals were members of either the Talek
West clan (N = 9) or the Mara River clan (N = 10). Focal animals spanned a wide
range of social ranks. Follows were focal animal samples (Altmann, 1974) with
continuous recording of behavior, lasting from 2-15 h. Using methods described
by Kolowski et al. (2007), we conducted follows at all times of day and night with
the aid of night-vision binoculars and infrared spotlights. In addition to continuous
monitoring of behavior, every 10 minutes we recorded the total number of hyenas
present in the subgroup of the focal hyena. We recorded the identity of every
hyena in the subgroup whenever possible. Subgroups were comprised of one or
more hyenas separated from other hyenas by at least 200 m. The 19 hyenas
followed were in view of observers, on average, for 98 i 0.56% of follow minutes
(N = 100 follow segments). We terminated a follow when the focal animal
remained out of sight for more than 30 minutes.

We were unable to follow hyenas for complete 24-hour periods due to
constraints .imposed by terrain and vegetation. Instead, we documented the 24—
hour pattern of social activity for each individual hyena by observing it during

shorter follow segments that together generated a composite 24-hour cycle. We

21

attempted to complete this cycle as quickly as possible after its onset, with the
average time necessary for completion being 31 days. All analyses below
requiring equal sampling throughout the 24-hour period utilize only data from A
composite 24-hour cycles. However, other analyses utilize all recorded follow
segments, 21% of which did not contribute to a composite 24-hour cycle (e.g.
due to hyena death or collar failure before cycle completion). In all analyses
based on follows, the sampling unit was the individual hyena. Averaged
estimates represent any individual observed duringmore than one composite
follow. Because females with den-dwelling cubs spend much of their time at the
communal den (Holekamp et al., 1996), where subgroup size is often large, we
only followed females without den-dwelling cubs to allow for appropriate
comparisons between the sexes.

Fluidity is a measure of how often subgroup composition changes over
time (Kummer, 1971 ). To describe the size and ﬂuidity of subgroups here, we
averaged subgroup size, subgroup duration, and minimum number of changes in
subgroup size during each composite follow across all focal animals. In addition,
we calculated the mean number of different clan members encountered per hour,
and the total numbers of clan members encountered by focal animals during
follow segments conducted between 1800 and 0900 hours. We used individual
follow segments for these calculations because accurate estimation of these two
variables required use of continuous monitoring. We focused exclusively on the
1800-0900 period here because Mara hyenas spend most of their daylight hours

lying in cool, shaded spots, and move very little (Kolowski et al., 2007).

22

Our second method of data collection was based on short-tenn
observation sessions involving members of the Talek and Talek West clans. We
collected these data daily around dawn and dusk, between 0530 and 0900 hours
and between 1700 and 2000 hours, respectively, throughout our 16-year study.
We initiated a session each time we encountered one or more hyenas separated
from others by at least 200 m. Sessions ranged in duration from 5 minutes to
several hours, and ended when we left an individual or subgroup. Every 15-20
minutes throughout each session, we conducted a scan in whichwe recorded the
identity and activity of every hyena present. We also recorded the geographic
location, relative to known landmarks, at which subgroups were found, subgroup
size (total number of hyenas observed in the seSsion), the primary activity in
which hyenas present were engaged, and whether or not food, alien hyenas, or
lions were also present. In 2003 and 2004, we recorded subgroup locations using
GPS units. From GPS data, we calculated distances between successive
observation sessions occurring within the same morning or evening sampling
period to estimate distances among subgroups in our study area.

Tendency to be Alone

Based on session data, we evaluated how the tendency to be alone varied
across the lifespan by calculating the percent of sessions in which natal males
and females were found alone during each of the following life history stages: 1)
natal den, 2) communal den, 3) den-independent but still nursing, 4) weaned but
pre-reproductive, and 5) reproductively mature adults. To calculate the percent of

observations alone, we divided the number of sessions in which an individual

23

was found alone by the total number of sessions in which we observed that
individual during a particular life history stage, and then multiplied by 100. We
used this same method to calculate the percent of sessions in which immigrant
males were found alone. We also assessed the relationship between an adult’s
intrasexual social rank and its tendency to be alone. For reasons detailed
elsewhere (Holekamp et al., 1997a; Smith et al., 2007), we calculated an overall
mean proportion of observations alone for each intrasexual rank position by
summing proportions across all individuals holding that rank during the study,
and divided this value by the total number of individuals holding that rank
position. We limited analyses based on rank to years (July 1988 to July 1989,
1991 to 1999, and 2002 to 2004) in which rank relationships were known to be -
stable.

Social and Ecological Inﬂuences on Subgroup Size

We assigned each session to one of nine behavioural contexts (Table 2.1) in
order to evaluate the extent to which speciﬁc circumstances inﬂuence subgroup
size and the tendency for individuals to be found alone or with conspeciﬁcs. To
compare conﬂicts with lions or alien hyenas to situations in which lions or aliens
were present but no conﬂict occurred, we also assigned sessions in which lions
and aliens were present, but no between-group ﬁghting occurred, to their own
‘non-conﬂict’ contexts. Conﬂicts occurred if we observed at least one agonistic
interaction between our study animals and lions or alien hyenas, respectively.
Conﬂict sessions occurring at kills were only assigned to the conﬂict context.

The ‘other’ category (Table 2.1) provided a ‘baseline’

24

Table 2.1 . Each session was assigned to one of the following behavioural

contexts.

Behavioural context

Social and ecological circumstance

 

Hunting

Natal den

Kill

Courtship/mating

Communal den

Border patrol

Clan war

Conﬂict with lion(s)

Other

one or more resident hyenas chased a selected prey
animal for at least 50 m, regardless of the outcome of

the hunting attempt1'2*

one or more resident hyenas observed at an isolated
den used by only one mother for shelter of a single litter
until her cubs reach 2 to 5 weeks of age; no food

present1 '3'4*

one or more resident hyenas observed feeding on at
least one fresh ungulate carcass1'2'3; no hunting
observed*

immigrant male(s) engaged in mating tactics such as
shadowing, defending, harassing, or mounting a

sexually mature female1'5'6; no food present*

one or more resident hyenas observed at a den or den

complex used concurrently by several litters ranging up

to 12 months of age1'3'7'8; no food present*

resident hyenas engaged in high rates of scent-marking
and socially facilitated defecation along territory

boundaries1'9; no food present*

agonistic interactions observed between resident and
alien hyenas at territory boundaries1'9; no lions present

agonistic interactions observed between resident

hyenas and at least one lion1'10'11; regardless of
location or other activity

one or more resident hyena(s) traveling or resting
when no food present; none of the contexts above

applied*

 

Asterisks (*) represent contexts assigned only when both Iions‘and alien hyenas
were absent. Superscript numbers refer to published works explaining contexts in

more detail than presented here. Kruuk 19721, Holekamp et al. 1997 , Mills
1990’, East et al. 19894, Hofer & East 20035, Szykman 20076, White 20077,
Boydston et al. 20058, Boydston et al. 20019, Cooper 1991“, Honer et al.

2005".

25

measure of social activity for Crocuta because neither resources (e.g. food or
mates) nor threats (e.g. natural enemies or competitors) that might attract or
repel conspeciﬁcs were present during these sessions. We considered subgroup
sizes that differed from ‘baseline’ to represent the formation of smaller or larger
subgroups. To ensure consistency in data collection across years, we limited our
analyses to sessions in which hyenas were located without the aid of radio
telemetry because radiocollars were not used in Talek until 1991.

Tests of Hypotheses Suggesting Forces Limiting Subgroup Size

Infant Safety Hypothesis

We compared the tendency for adult females to be alone, or only with their
dependent offspring, across reproductive states; offspring were considered to be
dependent until they were weaned. We divided the lactation interval into two
parts: the ﬁrst two weeks of lactation (‘early lactation’) and the remainder of the
lactation interval (‘Iate lactation’). We then assigned each female in each session
to one of the following reproductive states: the ﬁrst, second, or third trimesters of
pregnancy, early lactation, or late lactation. We divided the total number of
sessions during which a female in a particular reproductive state was found
alone, or with only her dependent offspring, by the total number of sessions in
which we observed that adult female in that reproductive state, and multiplied by
100. Next, we evaluated the relative impacts of reproductive state and prey
abundance on the tendency for adult females to be alone or only with their
dependent offspring. Focusing here on the subset of females observed at three

or more observation sessions across all ﬁve reproductive states, we compared

26

female behaviour between months when migratory prey were present and
absent. Finally, focusing on offspring, we evaluated how the tendency to be with
one’s mother varied across the lifespan by calculating the percent of sessions
each hyena spent with its mothers during each life history stage.

Disgersive Conﬂict Resolution Hypothesis

From 1988 to 2001, we conducted 30-minute focal animal ‘surveys’ on natal
animals of both sexes that were no longer living at dens to evaluate the effect of
within-group conﬂict on subgroup cohesion. During surveys in which the focal
animal was the target of at least one dyadic aggression (e.g. lunge, snap, bite,
chase, displace, push, stand over, and intention movement to bite), we recorded
whether focal animals engaged in conciliatory behaviors in response to
aggression within 15 minutes after each ﬁght. We ended all surveys when the
focal animal moved at least 200 m away from its subgroup or 15 minutes passed
after the ﬁght started, whichever occurred ﬁrst. We then compared the tendency
for a focal animal to leave its subgroup between surveys in which it received
aggression (but did not reconcile) and surveys in which that same animal
received no aggression. We also compared the tendency for a victim of
aggression to remain within its current subgroup between surveys in which it
initiated a conciliatory interaction with a former opponent and surveys in which it
failed to reconcile its ﬁghts with that same opponent. We required that sessions
containing matched surveys occur within 45 days of one another, that they
contain similar numbers (1 4) of individuals, and that either both occur at kills or

both occur away from kills. Due to small sample size, surveys assigned to other

27

behavioural contexts were excluded (Table 2.1).

Ecolpgical Constraints Hypothesis

We estimated local abundance of prey and assessed the extent to which this
inﬂuenced the tendency for hyenas to be with conspeciﬁcs (versus alone) and to
hunt with particular numbers of conspeciﬁcs. To do this, we performed biweekly
counts between 08:00 and 10:00 hours of all prey animals found within 100 m of
two 4-km transect lines in different parts of the Talek area, and averaged
biweekly counts to determine mean monthly prey counts. We also inquired
whether the mass of the ungulate carcass available in each kill session predicted
the numbers of hyenas present. Here we focused only on sessions in which the
prey species was known or where only ‘scraps’ (e.g. scattered bones, horns,
and/or small pieces of skin) were present that contained little nutritional or
energetic value. We used mean mass values for each prey species (Kingdon,
1997; Oindo, 2002) to estimate food amounts available at kill sessions. Following
Henschel & Tilson (1988), we assumed scraps weighed 2 kg.

Because hunting success is signiﬁcantly higher when hyenas‘hunt in
groups than when they hunt alone (Holekamp et al., 1997b), we also asked
whether hunting and feeding subgroup sizes were related. We did this to
evaluate the relative costs and beneﬁts associated with feeding or hunting with
conspeciﬁcs. The addition of a second hunter increases hunting success by 19%,
but the addition of subsequent hunters does not signiﬁcantly increase hunting
success further (Holekamp et al., 1997b). Therefore, we compared numbers of

new arrivals and total competitors present 5, 10, and 15 minutes after solo

28

hunters or pairs of hunters successfully captured prey. To control for effects of
prey size, we limited this analysis to matched sessions in which solo hunters and
pairs of hunters acquired ungulates of similar size. Lone hyenas regularly kill
antelope as large as wildebeest or topi in this population (Holekamp et al.,
1997b)

We inquired about the effects of intraspeciﬁc competition on feeding
success using two types of data from adult females. We ﬁrst assessed how per
capita energy gain varied among females as a function of subgroup size. We
also asked how the proportion of scans in which females fed varied as function of
subgroup size and social rank. Using methods developed by Creel & Creel
(2002), we determined the energetic value of each prey animal consumed by
each hyena subgroup of known size based on the species, sex, and age class of
that prey animal. Here we used only sessions in which the amount of food (e.g.
proportions of ﬂesh, viscera, and skin) consumed by known individuals was
visually quantiﬁable. We considered amounts to be quantiﬁable only when our
ﬁeld notes indicated exact changes in the amount of food (e.g. forelimbs, hind
limbs, pelvis, lumbar spine, ribcage, neck, and head) consumed over time. These
amounts corresponded directly to published data on the masses of East African
prey (Blumenschine and Caro, 1986; Sachs, 1967). We subtracted the amount of
edible biomass present at the end of the session or when no meat remained on
the carcass, whichever occurred ﬁrst, from that present when the food was ﬁrst
acquired, or when the session began if we arrived on the scene after the prey

animal had been captured.

29

We standardized subgroups containing hyenas of various ages by
calculating rates of energy gain by juveniles relative to those of adult females, as
has been done for other 'camivores (Baird and Dill, 1996; Mills and Biggs, 1993;
Packer et al., 1990). Frank (1986) found that cubs less than ﬁve months of age
and immigrant males rarely feed at kills with adult females. Therefore, we ignored
food intake by animals in the former two categories in our calculations. We
assumed that individuals 6 to 24 months of age consumed only half as much
food mass per unit time as did adult females (Frank, 1986). When changes in

feeding subgroup sizes occurred, we calculated a weighted average of the

number of adults present as: {Z (# of minutes each subgroup size lasted)*[(# of

adult females in each subgroup) + (0.5 * # of subadults present in each

subgroup)]}/(# of minutes each session lasted). We calculated per capita energy
intake per minute by dividing the number of kilojoules the entire subgroup
consumed by the number of adult females present and the number of minutes in
which feeding was observed. Kolowski et al. (2007) found that adult females in
the Masai Mara spend, on average, 9.2 minutes feeding at fresh kills during each
24-hour period. Therefore, we multiplied energy intake per minute for each
session by 9.2 minutes to convert to daily rates, allowing for comparison with
values reported elsewhere.

Within each scan made at each kill session, we quantiﬁed the proportion
of the adult females present that were actually feeding from the carcass, and
asked whether this proportion varied as a function of the numbers of adult

females present. In each kill session between 1988 and 1999 at which neither

30

alien hyenas nor lions were present, we averaged the proportion of adult females
feeding over all scans within each session and multiplied this value by 100.
Because food is unequally divided among adult females (Frank, 1986), we also
assessed how the proportion of time adult females fed at kills varied with their
absolute social ranks within the clan, and with their relative ranks within their
current subgroups. This was based on the number of scans in which each adult
female fed at kills, divided by the number of scans in which that female was
present at kills while she held a particular rank position, and then multiplied by
100. Here again, we used values averaged over all individuals that held each
particular rank position.

Statistical Analyses

We employed non-parametric statistics throughout most of this study due to low
sample sizes, the inability to transform non-normally distributed data, or both. We
used Mann-Whitney U tests to compare means between two independent
samples, and Kruskal-Wallis ANOVAs to compare means among multiple
groups. We used Wilcoxon-signed rank tests and Friedman’s ANOVA for
repeated measures when comparing the means of two or more than two
dependent groups, respectively. We calculated correlation coefﬁcients,
Spearrnan’s R, to examine correlational relationships. We compared effects of
multiple independent variables using Analysis of Covariance (ANCOVA) on log-
transforrned, normally distributed data. We report partial eta-squared values as
measures of effect size. We performed all statistical tests using STATISTICA 6.1

(StatSoft, lnc., Tulsa, OK, USA.) We used only two-tailed tests, and considered

31

differences to be statistically signiﬁcant at alpha < 0.05. We corrected for multiple
testing using the sequential Bonferroni adjustment (Rice, 1989). We report P-
values in their adjusted form and critical values following (Mundry and Fischer,

1998). Wherever appropriate, we report means i standard error (SE).

RESULTS
Fluidity, Duration, and Size of Subgroups
During long-term focal follows, we monitored the behavior of 19 hyenas for a total
of 624 hours, and completed composite 24-hour follow cycles for 16 hyenas (N =
5 males, N = 11 females). Of those hyenas, we followed six hyenas for two
complete cycles. Variation in subgroup size experienced by one typical adult
female spotted hyena during her two composite long-term follows is shown in
Figure 2.1. Hyenas from both Talek West and Mara River clans encountered
similar numbers of different conspec'rﬁcs during each hour followed (1.7 :l: 0.3 and
1.7 :l: 0.2 individuals/hour, respectively, Mann-Whitney U test: U = 38.0, P =
0.825). Because we detected no signiﬁcant differences for this, or any other
variables, based on clan membership, we pooled follow data from both clans
(Mann-Whitney U tests, U 3 0.20 and P > 0.20 for all variables). We detected no
sex differences in any measure calculated from follow data (Mann-Whitney U
tests, U 3 18.0 and P _>_ 0.28 in all cases, Table 2.2).

Although clan sizes ranged from 47 to 55 for Talek West and from 28 to
41 for Mara River during the period in which we conducted long-term follows,

subgroup size averaged only 3.6 i 0.4 individuals over all 24-hour composite

32

—— Follow A
t.; .2 . - . Follow B

Subgroup size

 

 

 

 

 

o
t
ooooooooooo is: so

 

 

0 . . . . . . . . . . . L
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200

 

 

Time (h)

Figure 2.1. Variation in subgroup size experienced by one adult female spotted
hyena during two long-term follows, A and B, conducted at different times. Follow
A consisted of 5 segments lasting an average of 4.8 hours each. Follow B
consisted of 3 segments, lasting an average of 8.0 hours each. Breaks between
continuous follow segments are indicated by asterisks (follow A) and plus signs
(follow B). Plotted values include the focal animal, and represent subgroup size
recorded every 10 min; a subgroup size of one indicates the focal animal was
alone. Shaded and open bars indicate hours of darkness and daylight,
respectively.

33

Table 2.2. Mean 3: SE subgroup sizes and estimates of ﬂuidity for males (N = 5)

and females (N = 11) based on entire 24-hour composite follows

 

Females Males U-statistic P-value
Subgroup Size 3.9 i 0.6 3.1 i 0.3 24.0 0.692
# Subgroup Size As 26.7 i 3.3 23.5 i 3.3 20.5 0.427
Subgroup Duration (min) 54.3 i 7.4 60.7 i 10.1 18.0 0.282
% Time Spent Alone 27.2 :t 7.2 35.1 i 10.7 24.0 0.692

 

34

follows. The largest subgroup size recorded during any of the follows was 23
hyenas (44% of the clan) during a conﬂict with lions. Subgroup size was highly
ﬂuid such that focal hyenas, on average, experienced a minimum of 25.7 :I: 2.5
changes in subgroup size during the course of a 24-hour period. Subgroup sizes
experienced by each focal individual varied dramatically both within and between
days (e.g. Figure 2.1). The maximum number of observed changes in subgroup
size during a 24-hour period was 48. During follows, subgroup compositions
lasted for an average of 56.3 :I: 5.9 minutes, and adults spent approximately one
third (29.6 i 5.8 %) of their time alone. On average, adults (N = 19) encountered
a minimum of 8.2 i: 0.7 other clan members (19.7 i 1.6 % of the entire clan)
within a single follow segment during the active period; these segments lasted an
average of 5.4 i 0.4 hours. Thus, each hyena encountered roughly 1.5 new
conspeciﬁcs per hour during the active period.

Observation session data generated patterns that were generally
consistent with data from long-term follows. On average, the distance between
subgroups occurring at two different observation sessions observed sequentially
within the same morning or evening sampling period was 1.11 i 0.03 km (N =
1291 distances; ranging from 201 m to 9.8 km). Mean subgroup size was 3.70 :t
0.02 hyenas (N = 34,848 sessions). Modal subgroup size, however, was only one
hyena, and almost half of our observation sessions (45.3%) involved lone hyenas
(Figure 2.2). Excluding transient males (immigrants remaining in the clan for less
than six months), our clan in the Talek area contained 39 to 74 members, and

the mean size of the clan during the entire study was 57 :l: 3 hyenas, based on

35

 

 

 

 

 

45 m
E 0.4
40 E 0.3
E
g 0.2
35 .o
E 0.1
E 30 2 0 .
'9‘ e 19 21 23 25 27 29 31 33 35 37 39
g 25 Subgroup size
3
,2 20
0
°\ 15
10 5I
!l!III-.---- "‘IllllllllllllllllJ_
01

7 9 111315171921232527293133353739
Subgroupsize

Figure 2.2. Frequency distribution of subgroup sizes in which we found members
of the Talek clan throughout our longitudinal study (N = 34,848 observation
sessions). The inset box shows a magniﬁed view of the frequency of subgroup
sizes ranging from 19 to 39.

36

monthly population estimates. Subgroups ranged in size from 1 to 39 individuals
(Figure 2.2) and were always less than current clan sizes. For example, when we
observed 39 hyenas together, the current clan size was 67 individuals. We never
observed the entire clan together concurrently during a single session. The
frequency with which we encountered subgroups of speciﬁc sizes decreased as
subgroup size increased (Spearman rank correlation: R3 = -0.997, P < 0.00001,
N = 39 sizes, Figure 2.2).

Variation in the Tendency to be Alone

The tendency for hyenas to be alone increased signiﬁcantly with each successive
life history stage (Kruskal-Wallis test: H4595 = 421.3, P < 0.00001; Figure 2.3A).

Cubs were alone signiﬁcantly more often at communal than at natal dens (Mann-
Whitney U test: Z = -2.62, P < 0.009). Cubs independent of the communal den
were alone more often than were those still residing at the communal den, but
less often than weaned, pre-reproductive animals (Z = -11.1 and -8.29,
respectively, P < 0.00001 for both). We detected no sex differences in the
proportion of sessions spent alone within any life history stage before adulthood
(Mann-Whitney U tests: 21 = 0.00, 22 = -0.60, 23 = -1.59, 24 = -0.73, P 3 0.56 in
all cases, Figure 2.3A). However, reproductively mature natal males were
signiﬁcantly more likely to be alone than were either adult females (Z = -3.31, P =
0.006) or weaned, pre-reproductive natal males (N = 85, Z = -2.99, P = 0.014).
By contrast, adult females were no more likely to be alone than were weaned,
pre-reproductive females (2 = -1.13, P = 0.52). On average, immigrant males (N

= 67) were found alone during a signiﬁcantly greater proportion of their

37

 

 

 

 

 

 

 

 

 

 

25 -
(a) E] Natal females E
I Natal males 74
w 20 -
8 D D E.
E 85
a .3 +
.9 15 '
‘5
E
3 C
.o
3 10 ' C 64
O 72
g
5 _ = =
F 10:1 17 93 104
A B B
0 l I 1 [+1 1- l I , I I I l I I I
K. K.
of? 8‘59 875‘ 60 ~o° .453 e
t} '9 0 x c6 '0‘ '9 0‘
'5‘ 0 0 S" C. 09 ,5"
e 6‘0 68 Q0 ,9 8° 02> to
1 ¢ 9 0‘ $ ﬂ, Q‘
()0 > ‘0 ngQ Q3’
0"“ ‘5
Life history stage
35
(b) D Adult females
30 L I Immigrant males
E T
.9
2 25- i
I:
o
I:
E 20 -
31
s j I
a 15 ' I I
as t
10 -
5 1 I l I
0 5 10 15 20 25

Social rank

Figure 2.3. Mean 1 SE percentage of observation sessions in which A) natal
animals were found alone during each life history stage, and B) adult females (N
= 45) and immigrant males (N = 40) were found alone as a function of intrasexual
social rank. By convention, the highest possible rank is one. Sample sizes in A)
shown above each bar represent numbers of individuals. Different letters above
bars indicate signiﬁcant differences after correcting for multiple testing.

38

sessions (21.2 :I: 1.5 %) than were adult females (N = 84, 15.8 i 1.3 %, Mann-
Whitney U test: 2 = 3.21, P = 0.001). Within each sex, low-ranking adults were
also found alone signiﬁcantly more often than were high-ranking individuals

(Spearman rank correlation: R3 = 0.77 and 0.85, N = 24 and 18 rank positions,

for adult females and immigrant males, respectively, P < 0.00001 for both; Figure
2.3B).

Social and Ecological Inﬂuences on Subgroup Size

The total numbers of hyenas present during sessions varied signiﬁcantly with the

contexts in which hyenas were observed (Kruskal-Wallis test: H3348“ =

11030.49, P < 0.0001, Figure 2.4). During baseline sessions, mean subgroup
size was only 2.2 i 0.2 hyenas. Hunting subgroups were signiﬁcantly smaller
than baseline (Mann-Whitney U test: Z = 5.00 and P = 0.0002), indicating that
hyenas typically leave subgroups to hunt alone or with a single companion. In
fact, solo hunters and pairs of hunters conducted 87.3 % of 393 hunts observed
here. Subgroups observed at natal dens were also generally small, but did not
differ signiﬁcantly from baseline (2 = 0.12, P = 0.91). Courtship interactions,
communal dens, kills, border patrols, conﬂicts with lions, and clan wars attracted
signiﬁcantly larger numbers of individuals than did baseline sessions (Z = -34.08,
-87.59, -47.74, -8.92, -9.92, -6.08, respectively, P < 0.0002 in all cases, Figure
2.4). Hyenas were observed with conspeciﬁcs in 81.5% of sessions where lions
were present but no interspeciﬁc agonistic interactions were observed; mean
subgroup size here (8.6 i 0.4, N = 390 sessions) was signiﬁcantly smaller than in

sessions in which agonistic interactions occurred between lions and hyenas (Z =

39

 

22
20

H
0‘1

Subgroup size
H
o
% Of observations with conspecifics '9'

 

 

 

 

 

 

 

 

   

        

42 13

 

1762 30

. to x‘: 9; é
o 1a Q 1.90 ‘99 9 $1»
‘6
so 0*
o°°

Figure 2.4. Mean :t SE subgroup size (left vertical axis and histogram bars) and
proportion of observations in which Crocuta were found in subgroups containing
more than one individual (right vertical axis and open circles) as a function of the
context in which groups formed. Sample sizes, shown below each bar, represent
numbers of observation sessions assigned to each context. Different letters
indicate statistically significant differences between contexts after correcting for
multiple testing. The shaded bar represents the baseline value of subgroup sizes
occurring in ‘other’ sessions, against which other groups were compared.

40

4.06, P < 0.0001). Mean subgroup size in sessions in which alien hyenas were
present, but no clan wars occurred, was 5.0 i 0.5 hyenas (N = 101 sessions), a
value signiﬁcantly lower than that observed during clan wars (2 = -4.87, P <
0.0001). Overall, both intra- and interspeciﬁc between-group conﬂicts promoted
the formation of large subgroups.

Testing the Infant Safety Hypothesis

In general, the tendency for adult females to be alone or only with their

dependent offspring varied signiﬁcantly among reproductive states (Friedman’s
ANOVA: F433 = 66.4, P < 0.00001, Figure 2.5A). Females were seen alone or

with only their dependent offspring signiﬁcantly more often during late pregnancy
and early lactation than during any of the other phases of the reproductive cycle
(Wilcoxon signed-ranks tests: 2 3 3.24 and P 5 0.016 in all cases). As predicted
by the infant safety hypothesis, females were seen alone with their dependent
offspring signiﬁcantly more often during early lactation than during other
reproductive states (2 = 4.87, P = 0.0001). However, the increasing tendency for
females to be found alone as pregnancy progressed (Figure 2.5A) was not
predicted by the infant safety hypothesis.

Our model containing the subset of females (N = 16) observed across all

reproductive states during both months of prey scarcity and abundance, with

social rank as a covariate(F1,149 = 3.685, Partial eta-squared = 0.024, P =
0.057), explained a signiﬁcant amount of variation (r2 = 0.316) in the tendency for

females to be alone or only with their dependent offspring (ANCOVA, F1o,149 =

6.879, P < 0.00001). As before (Figure 2.5A), the tendency for females to be

41

 

% Of observations alone/only with dependents

 

 

 

Pregnancy 1 Pregnancy 2 Pregnancy 3 Early lactation Late lactation
Reproductive state

 

(b) 27
100i-

% Of observations with mother

 

 

 

Life history stage

Figure 2.5. Mean :I: SE percentage of observations in which A) adult females (N
= 38) were found alone or only with their dependent offspring as a function of the
female’s reproductive state, and B) natal animals were found with their mothers
during each life history stage. Sample sizes, shown above each bar in B),
represent numbers of individuals observed In each life history stage. Different
letters indicate statistically signiﬁcant differences between categories after
correcting for multiple testing.

42

alone, or only with dependent offspring, varied signiﬁcantly among reproductive
states (F4,149= 15.089, Partial eta-squared = 0.288, P < 0.00001). Prey
abundance, however, did not signiﬁcantly predict this aspect of female behaviour

(F1449 = 0.028, Partial eta-squared = 0.0001, P = 0.867) nor did it interact with

the effect of reproductive state (F4349 = 1.181, Partial eta-squared = 0.031, P =

0.321). These results indicate that social rank and reproductive state are better
predictors of a female’s tendency to be alone, or only with dependent offspring,
than is local prey abundance.

Mothers were most likely to be present with their offspring during the life
history stages in which their offspring were most vulnerable to lnfanticidal

conspeciﬁcs. Mothers and infants spent progressively smaller proportions of their
time together as offspring matured (Kruskal-Wallis test: H4533 = 237.3, P <

0.0001; Figure 2.58). Mothers and cubs were observed together more often at
natal den than communal dens (Mann-Whitney U test: 2 = -8.35, P < 0.0001)
because newborn cubs rarely appeared above ground when their mothers were
absent. We observed mothers and cubs together more often when cubs resided
at communal dens than when cubs were den-independent but still nursing, or
weaned but pre-pubertal (Z = -4.00 and 5.52, respectively, P < 0.001 for both).
Mothers and offspring were also seen together more often when offspring were
weaned but pre-pubertal than when offspring were reproductively mature (Z =

3.67, P = 0.001). We found no sex differences in this measure within any life
history stage (Z1 = -0.43, 22 = 0.15, 23 = 1.36, 24 = 1.58, Z5 = -0.71, P_>_ 0.575 for

all).

43

 

Testing the Dispersive Conﬂict Resolution Hypothesis

Overall, we conducted a total of 211 focal animal surveys, 162 of which were in
the absence of food and 49 of which were at kill scenes, in which clan members
directed aggression towards 68 different focal animals in the absence of lions,
aliens, and courtship/mating. Within a single survey, a particular animal never
responded to aggression by both reconciling with a former opponent and
departing from its current subgroup. In response to aggression, focal hyenas left
their current subgroup after 19.8% of ﬁghts away from kills and 2.2 % of ﬁghts at
kills (N = 33 ﬁghts involving 21 focal animals). Focal animals were never
observed reconciling at kills, but did reconcile with former opponents after 11.7 %
of ﬁghts away from kills (N = 19 ﬁghts involving 17 focal animals).

We completed 145 pairs of matched surveys, occurring within 13 i 1 days
of one another, in which the same animal was present with conspeciﬁcs and
received aggression in one survey but not the other. None of these ﬁghts were
reconciled. Matched surveys differed in subgroup sizes by only 0.9 _t 0.1 hyenas.
In the absence of food, as predicted by the dispersive conﬂict resolution
hypothesis, the probability of immediate departure from the scenes was
signiﬁcantly higher when focal animals received aggression (23.8 i 5.2 %) than
when they did not (2.2 i: 1.0 %; Wilcoxon signed-ranks test: 2 = 3.68, P = 0.0002,
N = 47 focal animals). However, victims of aggression were signiﬁcantly less
likely to depart from their current subgroups at kills than from subgroups in which
no food was present (Mann-Whitney U test: 2 = -2.20, P = 0.028, N = 20 and 47

focal animals, respectively). Victims of aggression virtually always remained in

 

 

feeding subgroups; focal animals were no more likely to depart from feeding
subgroups after receiving aggression (5.0 i 5.0 %) than when they did not
receive aggression (5.0 i 5.0 %, N = 20 focal animals). Our sample did not
permit us to run a Wilcoxon signed-ranks test on these data because only two
focal animals differed in their responses between matched surveys at kills.

We completed 13 matched surveys during which the same victim of
aggression responded by initiating a conciliatory interaction with its former
opponent during one survey but failed to do so during the other. These matched
surveys were collected away from food within 20 i 5 days of one another, and
pairs of surveys differed in subgroup sizes by only 1.4 :t 0.3 hyenas. As predicted
by this hypothesis, reconciliation promoted subgroup cohesion. The tendency for
focal animals to remain in subgroups following ﬁghts was signiﬁcantly greater
when victims of aggression reconciled with former opponents~(100 i 0 %; no
departures) than when they did not (61.5 i 14.0 %; Wilcoxon signed-ranks test: T
= 0, P = 0.043, N = 13 focal animals).

Testing the Ecological Constraints Hypothesis
Consistent with the ecological constraints hypothesis, Crocuta adjusted
grouping patterns to match seasonal variation in local prey abundance and the
energy available at food sources. Crocuta were signiﬁcantly more likely to be
found with conspeciﬁcs during months when migratory prey were present (N =
60 months) than when migratory prey were absent (N = 122, Mann-Whitney U
test: Z = -3.36, P < 0.0001, Figure 2.6A). Variation in hunting subgroup size was

unable to explain this seasonal change in the tendency to be with conspeciﬁcs

45

 

 

 

 

 

 

 

 

 

 

 

1600 65
e S?
1400- _60 E
"U
Q)
I 1200- _55 can.
3 1000- 8
5 —50 5
g 800- E
3 ~45 E
‘3 600- .2
e .40 e:
a 400- E
200- -35 '8
O
30 5°
1 2 3 4 5 6 7 8 9 10 11 12
Month
20
b
18 - ( )
Elephant"
16 -
“é Giraffe 1
g1. 14 - Zebra :
go .21
.0 To 1
3 10 - p II Wildebeest
8 _ ImpalaI
GazelleI
6 " l:
4 scrapis I l I L J
0 0.5 1 1.5 2 2.5 3 3.5 4
Log prey mass (kg)

Figure 2.6. A) Monthly mean :1: SE numbers of prey animals counted each month
during biweekly ungulate censuses (left vertical axis and histogram bars) and
percentage of observation sessions in which Crocuta were found in subgroups
containing more than one individual (right vertical axis and open circles). B)
Mean i SE subgroup size as a function of prey mass [logs of values reported by
(Kingdon, 1997) and (Oindo, 2002)] available at sessions with scraps (N = 1315)
or fresh kills. Sample sizes for kills were: Thomson’s gazelle (N = 382), impala (N
= 53), wildebeest (N = 706), topi (N = 108), zebra (N = 193), giraffe (N = 29), and
elephant (N = 13). because mean numbers of individuals found hunting together
did not differ.

46

signiﬁcantly between months when migratory prey were present and those in
which ungulate prey were absent (N = 393 hunting subgroups, Mann-Whitney U
test: Z = -0.38, P = 0.703). Overall numbers of hyenas present at kills increased
with the mass of the prey carcass available within each session, even at

scavenged carcasses such as adult giraffe and elephants (Spearman rank
correlation: R3 = 0.98, N = 8 prey types, P < 0.0001, Figure 2.68).

The total numbers of competitors present at kills increased signiﬁcantly
over time within 15 minutes after successful hunts ended (Friedman’s ANOVA:
F113 = 11.27, P = 0.0001, Figure 2.7), but only when multiple hyenas made kills.

During the ﬁrst 5 min, on average, two more competitors arrived at kills made by
pairs than at kills made by solo hunters (Wilcoxon signed—ranks test: T = -1.50, P
= 0.035). On average, 10 minutes after prey capture, more than six competitors
were present at kills made by two hunters, whereas lone hunters virtually always
continued to feed alone (T = -1.00, P = 0.028). Very few new conspeciﬁcs arrived
at any of the kills sampled here more than 10 minutes after prey capture.

Crocuta consumed quantiﬁable amounts of fresh biomass at 41 different
kill sessions lasting an average of 26 i 3 minutes each (range: 6 to 98 min). On
average, each adult female consumed 44,161 at 6,737 kJ (6.4 3r. 1.0 kg) in a
single day. However, per capita energy intake was highly variable at fresh

ungulate carcasses, and decreased signiﬁcantly as the number of adult females
present increased (Spearman rank correlation: R3 = -0.63, P < 0.0001, Figure.

2.8A). The lowest rate of per capita energy gain (527 kJ/hyena/day) was

experienced by 5.5 hyenas feeding on a single juvenile Thomson’s gazelle,

47

 

  
  

-l- Pairs of hunters
' —0— Solo hunters

O\

U!
I .

Feeding subgroup size
co .ts

N
r

 

 

 

 

O 5 10 15
Time after hunt (min)

Figure 2.7. Feeding subgroup size during the ﬁrst 15 minutes after solo hunters
or pairs of hunters killed ungulates of similar size (N = 9 matched pairs of hunts).

48

220,000

Ex 200,000
180,000
1 60,000
140,000
120,000
100,000
80,000
60,000
40,000
20,000
0

Energy gain (kJ/female/d

 

T

I

I

 

(a)

1 J l

O
’00. .

 

0

100
90
80
70
60
50
40
30
20

IIIIIWJ
12345678910

 

1

I

I

% Of females that fed per scan

10

 

(b)

it; It I"-

j i

 

1 I l l l I l l l l l I

 

O r
0 1

2 3 4 5 6 7 8 91011121314

 

11 12 13 14

 

Number of adult females present in subgroup

Figure 2.8. A) Per capita daily energy gain as a function of the number of adult
females present at fresh ungulate kills (N = 41). B) Mean 1 SE percentage of
adult females observed feeding per scan as a function of the number of adult
females present within each subgroup at kills (N = 426 sessions).

49

whereas the highest rate (202,509 kJ/hyenalday) occurred when a hyena fed
alone on an adult wildebeest.
The proportion of scans in which adult females were able to feed at kills

also declined signiﬁcantly as the number of adult female competitors increased
(Spearman rank correlation: R5 = -0.62, P < 0.00001, N = 426 sessions, Figure
2.83). On average, high-ranking females were signiﬁcantly more likely to feed at
kills than were low—ranking females (R3 = -0.60, P = 0.002, N = 24 rank positions,
Figure 2.9A). Females outranking others within their current subgroup also
gained better access to kills than did those with low relative ranks (R3 = -0.70, P

= 0.006, N = 14 relative ranks, Figure 2.98).

DISCUSSION

Fluidity of Spotted Hyena Societies

Spotted hyena clans are dynamic, ﬂuid societies in which subgroup composition
changes frequently over time. Although Mara hyenas spent the majority of their
time with conspeciﬁcs, our data demonstrate that Crocuta clans are atomistic,
individual-based societies (Rodseth et al., 1991). The mean subgroup size for
hyenas (X = 4) was similar to that reported for other species living in individual-
based FF societies including lions (J? = 2 to 5; Packer et al. 1990), chimpanzees
(Pan troglodytes, X e 7 to 9; Symington 1990), spider monkeys (Ateles geoffroyi,
X = 4; Symington 1990), and bottlenose dolphins in Western Australia (X = 4

and 6; Connor et al. 1999) and Sarasota, Florida (I? = 5; Irvine et al. 1981) but

50

100
90 .. (a)
80 -
7O -

6méié¥i { f .
:3 {it i ll , i.i,

 

1

3()_

 

 

10

0 l I I I I I I I I

O 2 4 6 8 101214161820222
Absolute social rank in the clan

 

 

 

100
90
80
70
60
50
40 -
30
20

Ilii *,E{

OIILJJ r.-
01234567891011121314

Relative social rank in current subgroup

 

(b)

% Of scans a focal female fed

I-I-I
I-I-I
I

I

l

 

 

 

 

J I I I

Figure 2.9. Mean :1: SE percentage of scans in which focal females were
observed feeding at kills while holding A) an absolute social rank position within
the clan (N = 24 absolute ranks), or B) a relative social rank position within the
current subgroup (N = 14 relative ranks).

51

not those in Doubtful Sound, New Zealand (X = 17; Connor et al. 1999; Lusseau
et al. 2003). Subgroups lasted longer for hyenas (X = 56 min) than for
chimpanzees (X = 25 min; Lehmann & Boesch 2004), but were shorter than
those observed for Doubtful Sound dolphins (X > 24 hours; Lusseau et al. 2003)
or lions (X = 48 to 75 hours; Packer et al. 1990). Mara hyenas spent roughly the
same amount of time alone (X = 30%) as did spider monkeys (X = 13 to 37%) or
chimpanzees (X = 14 'to 65%; reviewed by Symington 1990), but more time than
Doubtful Sound dolphins (X < 1%; Lusseau et al. 2003) or lions (X = 10 to 15%;
Packer et al. 1990).
Variables Promoting Subgroup Fission and Fusion
Life history stage, sex, social rank, and current activity all inﬂuenced the
likelihood of ﬁnding hyenas of both sexes alone, as did reproductive state among
adult females. Spotted hyenas tended to occur in larger subgroups when they
were active than during hours when they were resting. Intriguingly, this pattern
appears to differ from that observed among certain other mammalian carnivores,
including both brown and striped (Hyena hyena) hyenas and European badgers
(Mills, 1990; Wagner, 2006; Woodroffe and Macdonald, 1993), which are all
usually found in larger subgroups when resting than when active. We also found
that adult spotted hyenas spent more time alone than did younger individuals.
Adult males were generally found alone more often than adult females, but adult
females with neonatal cubs were away from other clan members most often.
Our current ﬁnding that, within each sex, low-ranking hyenas were alone

p more frequently than high-ranking individuals supplements previous research

52

indicating that low-ranking adults spend more time further from the den
(Boydston et al., 2003) and associate less often with same-sexed conspeciﬁcs
(Holekamp et al., 1997a; Smith et al., 2007) than do high-ranking animals. Rank-
related variation in subgroup composition contributes to these patterns. Hyenas
generally prefer to associate with members of their own matriline, and high-
ranking matrilines contain more individuals than do low-ranking matrilines
(Holekamp et al., 1997a; Van Horn et al., 2004a; Wahaj et al., 2004). Among
non-kin, hyenas actively join subgroups containing social companions higher-
ranking than themselves; this gains them social and feeding tolerance from the
dominant animals with which they associate most with often (Smith et al., 2007).

Our ﬁnding that multiple animals (X = 5 hyenas) congregate during
courtship and mating is consistent with previous work showing that numbers of
males observed with females increase as females approach estrus (East at al.,
2003; Szykman et al., 2007). Given the importance of the communal den as a
focal point for social activity (Boydston et al., 2005; White, 2007), including
cooperative and afﬁliative behaviors among kin (Engh et al., 2000; Smale et al.,
1993; Wahaj et al., 2004), we were not surprised to ﬁnd large subgroups (X = 7
hyenas) at these locations.

Overall, our data suggest that cooperative defense of shared resources
during between-group competition (e.g. clan wars, lion-hyena interactions) is a
strong cohesive force in hyena societies, promoting the formation of large
subgroups. Most interestingly, we found that large numbers of hyenas (X = 11 to

17 hyenas) joined forces during intra- and interspeciﬁc between-group conﬂicts.

53

Clan members gathered during cooperative marking and defense of territory
boundaries, during defense of carcasses from either alien hyenas or lions, and in
response to predation attempts by lions.

Crocuta can only successfully defend food from lions when the ratio of
hyenas to lions is high (e.g. 4:1 when adult male lions are absent; Kruuk 1972;
Cooper 1991). Because lions are three to ﬁve times larger than hyenas, the
resource holding power of a single lion exceeds that of a single hyena (Cooper,
1991; Honer et al., 2002; Kruuk, 1972). In addition to being their direct
competitors, lions also represent a leading mortality source for spotted hyenas
(Kruuk, 1972; Mills, 1990; Watts, 2007). Therefore, individual hyenas cannot
effectively compete with lions for possession of a carcass or defend themselves
from predation by lions without aid from conspeciﬁcs. Effective maintenance of
group territories also requires that individuals from multiple matrilines, with low
mean relatedness (Van Horn et al., 2004a), join forces during cooperative
defense against neighboring hyena clans. Loss of a clan war can result in
substantial reduction in the area of a clan’s territory, and repeated losses can
further result in overall loss of the territory to a neighboring clan (K.E. Holekamp,
unpublished data). Cooperative defense of territories appears to offer a similarly
important advantage during intraspeciﬁc between-group conﬂicts in a variety of
other carnivores such as dwarf mongooses (Helogale parvula), meerkats
(Sun'cata sun'catta), and Ethiopian wolves (Canis simensis, reviewed by Creel &

Macdonald 1995), with the larger of two groups typically winning disputes.

54

 

Factors Limiting Subgroup Size

Our data were consistent with all three of the hypotheses suggesting factors
limiting subgroup size in spotted hyenas. First, as predicted by the infant safety
hypothesis (Otali and Gilchrist, 2006), adult females spent the most time alone
with dependent offspring during early lactation, when they stayed near isolated

' natal dens. Mother-cub associations were especially close at the natal den, and
declined as cubs matured. Similarly, lion and chimpanzee mothers are also most
solitary duringearly lactation (Otali and Gilchrist, 2006; Packer et al., 1990;
Symington, 1990). In fact, lionesses keep cubs hidden in thick vegetation for the
entire ﬁrst month of life. Like lions (Packer et al., 1990), female hyenas were also
frequently found alone during late pregnancy, a ﬁnding not predicted by the infant
safety hypothesis. Reproductive suppression associated with attacks by
conspeciﬁcs is unlikely to explain this pattern because female mammals are less
physiologically vulnerable to pregnancy loss resulting from attacks during late
than during early pregnancy (Wasser and Barash, 1983). However, females late
in pregnancy may be least constrained by demands of prior offspring, and
therefore prefer to forage alone to maximize energy intake in preparation for the
substantial energetic demands imposed on them by neonatal cubs.

The immediate threat of aggression disrupted subgroups in the absence
of food, a ﬁnding consistent with predictions of the dispersive conﬂict resolution
hypothesis (Schino, 2000). In contrast, victims of aggression rarely reconciled at,
nor departed from, feeding subgroups. This suggests that, whereas hyenas

sometimes retreat a few steps away from food in response to receiving

55

aggression (Frank, 1986; Kruuk, 1972), individuals rarely leave feeding
subgroups when food of high quality is present. Away from kills, however, targets
of aggression always remained in subgroups after reconciling ﬁghts with former
opponents, and targeted hyenas frequently left their subgroups if ﬁghts were not
reconciled. These data suggest that reconciliation promotes social cohesion by
reducing the potential for escalated aggression among individuals that remain in
their current subgroups. In the absence of conciliatory interactions, subgroup
ﬁssion reduced the risk of continued conﬂict, and the potentially lethal
consequences of escalated aggression. In captivity, hyenas ﬁght intensively, and
severely wound group-mates when denied opportunities to depart (Jacobi, 1975).
However, withinogroup conﬂict rarely leads to mortality of adult hyenas in natural
populations (Kruuk, 1972). Therefore, the ability of Crocuta to resolve conﬂicts by
separating from former opponents appears to provide a second mechanism,
along with reconciliation (Hofer and East, 2000; Wahaj et al., 2001), by which
hyenas reduce the immediate costs of intra-group conﬂict.

Finally, our results were consistent with all predictions of the ecological
constraints hypothesis (Chapman et al., 1995). This hypothesis was able to
explain grouping patterns of all animals in the population, not just those of
reproductive females or hyenas recently attacked, over multiple time scales. Our
data show that feeding competition constrains grouping behaviour in the short-
terrn at kills, and in the long-term during periods of food scarcity lasting several
months. Our ﬁndings, together with data from previous studies (Frank, 1986;

Holekamp et al., 1996; Honer et al., 2005), imply that ecological constraints

56

operate at virtuallyall times in the lives of spotted hyenas. It appears the only
situations that can trump the disruptive force of feeding competition in hyena
societies are those occurring when females have highly vulnerable offspring.

As in other animal societies characterized by FF dynamics (Aureli et al.,
2008; Chapman et al., 1995; Kummer, 1971; Lehmann et al., 2007a; Lusseau et
al., 2004; van Schaik, 1999), the ﬂexible FF structure of Crocuta clans permits
individuals to adjust grouping patterns in response to ﬂuctuations in local
resource abundance. ln Tanzania and Namibia, hyenas redistribute themselves
from less proﬁtable areas to more proﬁtable areas in response to long-term
changes in prey abundance (aner et al., 2005; Trinkel et al., 2004). Similarly,
Crocuta in our study were most gregarious when ecological constraints were
relaxed during periods when prey, particularly large-bodied ungulates, were
abundant. Because hunting subgroup size did not vary seasonally, the beneﬁts of
cooperative hunting could not explain this variation in gregariousness. Instead,
this dynamic pattern may be driven by increased within-group aggression at kills
during the extended period each year in Talek when prey are scarce (Holekamp
etaL,1993)

Over shorter time scales, we showed that hyenas quickly congregated at
kills in numbers correlated with the size and energetic value of captured prey. As
predicted by the ecological constraints hypothesis, reduced feeding competition
permitted the formation of larger subgroups and greater per capita food intake at
large than small carcasses. The relative costs of joining feeding groups varied

with rank such that low-ranking hyenas, which enjoyed little resource holding

57

power, were least likely to feed in large subgroups. Our data are consistent with
earlier research showing priority of access to food is determined by rank in this
species (Frank, 1986; Tilson and Hamilton, 1984). In addition, here we show for
the ﬁrst time that an individual’s relative rank within its current subgroup directly
predicts its immediate rate of food consumption, which likely inﬂuences staying
and leaving decisions in feeding subgroups. Our work, therefore, extends
previous ﬁndings indicating that low-ranking individuals hunt signiﬁcantly more
often, and in smaller subgroups than do high-ranking hyenas (Holekamp et al.,
1997b). More generally, our data support the hypothesis that resource limitation
constrains subgroup size.

On average, adult females consumed 44,161 1 6,737 kJ (6.4 kg) per
hyena per day. This mean daily energy intake value is within the range of values
reported for wild spotted hyenas elsewhere in Africa: 2.5 kg (Kruuk, 1972), 3.6 kg
(Henschel and Skinner, 1990), 3.8 kg (Green et al., 1984), 6.2 kg (Mills, 1990),
7.4 kg (Whateley, 1980), and 9 kg (Gasaway et al., 1991). They also match the
value reported for hungry captive hyenas housed in a group of ﬁve individuals (4
kg per hyena, Henschel & Tilson 1988); here Mara hyenas in subgroup sizes of
4.5 to 5.5 each consumed 27,865 at 7,674 kJ (4.1 kg) per hyena per day.

A number of authors (e.g. Kruuk 1972; Tilson & Hamilton 1984) have
suggested that group living evolved in Crocuta to facilitate cooperative hunting of
large ungulates. Many carnivores including wild dogs, jackals (Canis spp.),
coyotes, and lions (reviewed by Creel & Macdonald 1995) gain some advantages

from cooperative hunting. Similarly, Crocuta enjoy increased hunting success

58

and capture a larger array of prey species when hunting in groups than when
hunting alone (Holekamp et al., 1997b; Kruuk, 1972). However, our current
results, like those from earlier work (Holekamp et al., 1997b), suggest hyenas
typically hunt either alone or in pairs, such that the average subgroup size during
hunting is signiﬁcantly smaller than the mean subgroup size documented in any
other context. i

We suggest that Crocuta often hunt alone because individuals who leave
their group-mates to hunt are likely to be able to feed from any carcass they
acquire for at least a few minutes before other competitors arrive. Here we found
that Mara hyenas often fed alone before additional competitors arrived following
prey capture. Hyenas are able to detect sounds associated with kills from at least
2.4 km away (Mills, 1989). Noise generated by pairs of hunters competing over
kills (e.g. giggles in response to aggression) attracts additional competitors to
kills made by groups; by contrast, hyenas feed silently when alone. Spotted
hyenas can ingest meat and bone at a rate of 1.3 kg/minute (Kruuk 1972) and
lone hunters in our study typically enjoyed much longer periods of solitary
feeding than did hyenas hunting in groups. Each lone hunter should be able to
ingest approximately 6.5 kg of food during only the ﬁrst 5 minutes after making a
kill. This amount is as much or more than the average adult spotted hyena
consumes in a 24-hour period in many parts of Africa (Green et al., 1984;
Henschel and Skinner, 1990; Kruuk, 1972).

Although group hunters are 19% more likely than solo hunters to succeed

in capturing prey, even the addition of a second hunter dramatically increases

59

ensuing subgroup size. Here, on average, within 10 minutes of prey capture,
over six hyenas competed for kills made by pairs of hunters whereas solo
hunters almost always still fed alone. We found that hyenas fed nearly the entire
time they were alone at fresh kills, but hyenas in subgroups of six at kills spent
less than half their time (43%) feeding. Moreover, per capita energy gain
declined rapidly with increasing subgroup size such that the majority of
individuals feeding in large subgroups consumed very little food. Taken together,
these results suggest that hyenas hunting alone enjoy more time feeding, and
thus consume more food mass, than individuals hunting with conspeciﬁcs.
Overall, the initial beneﬁts of increased hunting success are more than offset by
the costs of increased competition in the larger subgroups that form after group
hunts. Rather than functioning as a cohesive force in Crocuta societies, our data
suggest that hunting actually promotes subgroup ﬁssion. In this regard spotted
hyenas differ from societies of wild dogs (Creel, 1997), but are similar to those of
many other gregarious carnivores [e.g. coatis (Gompper, 1996), European
badgers (Kruuk and Parish, 1982), brown and striped hyenas (Kruuk, 1976; Mills,
1990; Wagner, 2006), and kinkajous (Kays and Gittleman, 2001)] in which
individuals reduce feeding competition by leaving group-mates to forage alone or
in small subgroups.

Conclusions

Unlike animals living in cohesive social groups, individuals living in FF societies
are able to make decisions without the consensus of the entire group (Conradt

and Roper, 2005). Our current study demonstrates that Crocuta choose to

60

associate with particular numbers of conspeciﬁcs based upon their own current
state, and in response to ﬂuctuations in the local resource base. Although our
data are consistent with predictions of all three of the hypotheses we tested here,
only the ecological constraints hypothesis can explain variation in grouping
patterns involving all clan members over both short and long time scales.

Extant spotted hyenas apparently descended within the past 900,000
years from a carrion-feeding ancestor with a solitary lifestyle much like that of the
modern striped hyena (Lewis and Werdelin, 2000). Our data suggest that
selection favoring cooperative hunting did not shape gregariousness during the
evolution of this species. However, the ability to capture a larger array of prey
animals more successfully might have emerged as a secondary consequence of
group living favored by other selection pressures. In many different species,
ﬂexible FF lifestyles limit the costs of group living while allowing group members
to aggregate when the beneﬁts of sociality are high or the costs of grouping are
low (Chapman et al., 1995; Wrangham et al., 1993). Here we found that within-
group competition tended to drive individuals apart, whereas intra- and
interspeciﬁc between-group competition was a strong cohesive force within
Crocuta clans. Our data, therefore, suggest that group living might have evolved
in spotted hyenas to permit cooperation among conspeciﬁcs during defense of
shared resources, including bOth space and food. However, constraints imposed
by limited food resources might account for retention of the tendency for Crocuta
to spend large amounts of time alone, rather than the evolution of a more

cohesive social structure.

61

Chapter 3
Smith JE, Memenis SK, Holekamp KE, 2007. Rank-related partner choice in the

ﬁssion-fusion society of spotted hyenas (Crocuta crocuta). Behavioral
Ecology and Sociobiology 61:753-765.

62

Chapter 3
RANK-RELATED PARTNER CHOICE IN THE
FlSSlON-FUSION SOCIETY OF SPOTTED HYENAS
INTRODUCTION
Recognizing the adaptive signiﬁcance of choices made by gregarious animals
with regard to their social partners, behavioral ecologists have recently begun
evaluating the potential ﬁtness consequences of partner choice outside the
context of sexually-selected mate choice (Dugatkin and Sih, 1998).
Primatologists have long known that many species of cercopithecine monkeys
associate most closely with unrelated individuals of similar or higher rank than
their own in the social hierarchy (Cheney et al., 1986; Schino, 2001 ). Other
researchers, however, have paid little attention to partner choice involving non-
kin (Gouzoules and Gouzoules, 1987). Spotted hyenas (Crocuta crocuta) are
highly gregarious carnivores that reside in social groups called clans (Kruuk
1972), which are strikingly similar in their size and hierarchical structure to troops
of cercopithecine primates (Drea and Frank, 2003). This offers a unique
opportunity to compare rank-related partner choice between primates and
carnivores, taxonomic groups that last shared a common ancestor 90-100 MYA
(Springer et al., 2003; Springer et al., 2005). It also offers an important
opportunity to evaluate the power and generality of alternative models of social
partner preference based on dominance rank and related factors.
Like monkeys, Crocuta frequently compete over limited resources and

cooperate by joining forces to direct coalitionary aggression towards conspeciﬁcs

63

(Engh et al., 2005). As in most primates, nepotism is common among Crocuta,
and kin associate with one another more often than do non-kin (Holekamp et al.,
1997a; Wahaj et al., 2004). Among non-kin, high-ranking adult female Crocuta
are more gregarious than low-ranking females (Holekamp et al., 1997a), but we
know very little about which individuals make decisions about partner choice to
generate this pattern, or why such decisions are made. Furthermore, although
adult male Crocuta associate closely with their mates (Szykman et al., 2001), it is
not understood to what extent immigrant males associate with each other, or how
patterns of intrasexual association differ between the sexes. Understanding
these decision-making processes should help elucidate the evolutionary costs
and beneﬁts of social partner choice in Crocuta.

Crocuta clans are ﬁssion-fusion societies in which individuals travel, rest,
and forage in subgroups that frequently change in size and composition (Kruuk,
1972; Mills, 1990), so preferences for social partners among Crocuta are
revealed by how often they occur in subgroups with particular conspeciﬁcs
(Holekamp et al., 1997a; Szykman et al., 2001; Wahaj et al., 2004). Preferences
for certain social partners are likely to emerge in societies where asymmetries in
partner value exist (Dugatkin and Sih, 1998). High-ranking Crocuta can
potentially confer greater ﬁtness beneﬁts to the individuals selecting them as
social partners than can low-ranking animals because dominants can permit
access to monopolized resources such as food and space merely by withholding
aggression (Boydston et al., 2003; Frank, 1986). Feeding competition is

extremely intense among Crocuta and access to ungulate carcasses profoundly

64

affects their lifetime reproductive success (Frank et al. 1995, Holekamp et al.
1996). Dominants can also provide more effective coalitionary support than
subordinates during within-group contests (Smale et al. 1995; Engh et al. 2005).
Thus, Crocuta should be selective when making decisions to join subgroups
containing particular social partners.

Here, we inquired whether social rank or rank distance inﬂuences
intrasexual partner choice among unrelated adult Crocuta of both sexes. We also
investigated the potential beneﬁts of rank-related partner choice among females.
In particular, we asked whether dominant or subordinate females initiate
associations by promoting group formation. Dominant Crocuta beneﬁt from the
presence of other clan members in that they enjoy priority of access to resources
cooperatively obtained and defended by multiple group members (Cooper 1991;
Holekamp et al. 1997b; Boydston et al. 2001; Boydston et al. 2003), but the
beneﬁts of such partnerships to subordinates are unknown. We tested three
hypotheses, each suggesting a potential beneﬁt to be gained by subordinates
from rank-related partner choice. Speciﬁcally, we asked whether subordinate
females beneﬁt from: 1) reduced harassment (or increased social tolerance) by
conspeciﬁcs, 2) increased tolerance during feeding, or 3) increased coalitionary
support during aggressive interactions. The reduced harassment hypothesis
predicts that subordinate females should receive less frequent dyadic aggression
and better access to monopolized space from unrelated dominant females with
which they associate more often. We examined this in situations in which no food

was involved. The feeding tolerance hypothesis predicts that, during feeding

65

competition, females should receive less frequent dyadic aggression from
unrelated dominant females with which they associate more often, they should
be permitted better access to monopolized food, or both. Finally, the coalitionary
support hypothesis predicts that subordinate females should receive more
frequent coalitionary support during aggressive interactions from unrelated

dominant females with which they associate more often.

METHODS
Study site and subject animals

From 1988 to 1999, we studied members of one large clan of Crocuta inhabiting
a territory of 62km2 in the Talek region of the Masai Mara National Reserve,

Kenya (Boydston et al., 2001). Excluding transient males (immigrants remaining
in the clan for less than six months), clan size during the study period ranged
from 45 to 78 residents. Each resident was known individually by its unique
spots, and sexed based on the dimorphic morphology of its erect phallus (Frank
et al., 1990). We estimated birth dates (to i 7 days) using methods described
previously (Holekamp et al., 1996). We classiﬁed natal females as adults at 36
months of age or at their ﬁrst known date of conception, whichever occurred ﬁrst.
Whereas female Crocuta are philopatric, males disperse 1-76 months after
puberty (Smale et al., 1997; Van Horn et al., 2003), which occurs at
approximately 24 months of age in this species (Dloniak et al., 2006; Glickman et
al., 1992). On average, genetic relatedness among natal members of a clan is

extremely low (Queller-Goodnight R = -0.05 :I: 0.007), as is mean relatedness

66

among adult immigrant males (Queller—Goodnight R = 0.009 d: 0.007; Van Horn
et al. 2004). Because we were interested here in partner choices involving
unrelated adult females, we excluded all dyads containing grandmothers and
adult granddaughters, mothers and adult daughters, and adult maternal sisters,
based on known maternal relationships and genotyping. We considered all males
immigrating into the Talek clan to be unrelated adults.

Behavioral data collection

Crocuta clans are structured by linear dominance hierarchies (Frank, 1986;
Kruuk, 1972) like those in cercopithecine societies (Engh et al., 2000; Smale et
al., 1995). Here we determined the social rank of each individual hyena based on
the outcomes of dyadic agonistic interactions; all adult females were dominant to
all immigrant males (Holekamp and Smale, 1993; Smale et al., 1993). We ranked
adult males and females in separate hierarchies, with the highest possible rank in
each being one. Based on these ranks, we calculated rank distance as the
absolute value of the difference in ranks between the members of a same-sex
dyad. We used the terms “dominant” and “subordinate” to refer to the relative
ranks of members of each dyad. We assessed association patterns based on the
co-occurrence of dyad members in observation sessions recorded during each
year of our study. We initiated a session each time we encountered one or more
Crocuta separated from conspeciﬁcs by at least 200 m within the Talek home
range. Sessions lasted from 5 minutes to several hours, and ended when we left
an individual or group. At each session, researchers recorded the location, the

identity of each individual present, and whether or not food was present. We

67

deﬁned sessions as having food when a fresh ungulate carcass was present, and
considered food to be absent only in sessions where no food whatsoever was
present. Sessions not meeting either of these criteria were excluded from the
aggression calculations. At sessions with food, we performed scans every 15 to
20 minutes to record which individuals were present, and whether or not those
individuals fed concurrently.

All aggressive and appeasement behaviors were recorded as critical
incidents using the all-occurrence sampling technique of Altmann (1974). Dyadic
aggression only involved a single aggressor, whereas coalitionary aggression
involved at least two Crocuta directing aggression towards the same target
animal. Aggressive behaviors included head wave, lunge, aggressive posture
(e.g., ears cocked forward with the tail bristled and raised), chase, displace,
stand over, bite, and push. We required at least one minute to elapse between
aggressive interactions within a pair of individuals in order for a second
interaction to be included here; renewed attack within a single minute was
considered a continuation of the original interaction.

Measures of intrasexual association

We estimated the degree of afﬁliation between members of each dyad using the
Twice-Weight Association Index (AI) of Cairns and Schwager (1987), as done
previously (Holekamp et al., 1997a; Szykman et al., 2001). We employed this
method because it was appropriate for Crocuta and because it enabled us to
compare our current data to earlier association data for Cmcuta. We elected not

to use the ﬁssion-fusion decision index (Cross et al., 2005) because Crocuta do

68

not satisfy its key assumption that travel costs are high when individuals move
among subgroups within the territory (Holekamp et al., 2000; Kruuk, 1972). We
calculated intrasexual Als for immigrant males for the ﬁrst time here, and
extended earlier analyses of female-female association by including data
collected from 15 February 1994 through 31 December 1999, not available in our
earlier study (Holekamp et al., 1997a). An Al was calculated for each same-sex
dyad of unrelated adults, hyenas A and B, for each year during which they were

concurrently present in the clan as adults for at least six months. We calculated

A'A,B 333 (A+Btogether) I [(Awithout B) 1' (Bwithout A) + (A+Btogether)] Where (ATBtogether)
represents the number of observation sessions in which A and B were both

present, (Awithout 3) represents the number of sessions in which A was observed

but B was not present, and (Bwithout A) represents the number of sessions in

which B was observed but A was not present.

During the 11-year study, multiple individuals occupied most of the
available rank positions in both male and female hierarchies. Immigrant males
queue for rank, with their social status increasing within the male hierarchy as
they gain tenure in the clan (East and Hofer, 2001; Smale et al., 1997). Females
assume social ranks immediately below those of their mothers, but ranks can
change due to births and deaths of clan members (Engh et al., 2000; Frank,
1986). Therefore, we calculated an overall mean Al for each rank position within
each sex by summing Als across all individuals holding that rank during the
study, and dividing by the total number of Als for that rank position. We also

calculated a mean Al for each rank distance by summing the AIS across all

69

individuals with the same rank distance within each sex, and dividing by the total
number of Als for each rank distance. In another analysis, we controlled for rank
distance by comparing how often a focal adult, hyena B, associated with a same-
sexed adult ranking directly above it in the dominance hierarchy, hyena A, to how

often that focal individual associated with the adult ranking directly below it,
hyena C. That is, we compared AIAB and Al3,c to inquire whether hyena B

associated more often with the dominant or subordinate adjacent-ranking
individual.

High-ranking females have offspring at the communal den more frequently
than do low-ranking females because high-ranking adult females wean their
young faster and enjoy shorter interbirth intervals than do low-ranking females
(Frank et al., 1995; Holekamp et al., 1996). To avoid this factor as a confounding
variable, sessions at the communal den were excluded from all calculations of
Als. Because immigrant males search for mating opportunities by actively joining
subgroups containing sexually receptive females (Szykman et al., 2001), we
conservatively used only sessions in which potentially fertile females, those over
24 months of age, were absent to evaluate patterns of male-male partner choice
while controlling for the effects of mate choice.

Decisions to join subgroups containing other adult females

To determine whether subgroups containing each focal female were more likely
to be joined by subordinate or dominant females during subgroup fusion events
within the clan, we conducted focal animal surveys, totaling at least 2 hours per

animal, on eight adult females from our study clan holding ranks of 7, 8, 9, 11,

70

12, 13, 18, and 21. Adult females ranking above 7 were from the alpha matriline
and those ranking below 21 were from the lowest-ranking matriline. Individuals
within these two matrilines were excluded from this analysis because no
unrelated dominant females were available to join the former and no unrelated
subordinate females were available to join the latter. We ended surveys when the
focal female went out of view for more than 5 consecutive minutes. We recorded
all unrelated females that actively joined each focal female during fusion events
throughout the focal animal survey. To control for kinship, a prospective joining
animal was only counted if none of her kin were present in the subgroup
containing the focal female. We conducted surveys in the absence of food and
away from the den to exclude other factors that might promote group formation,
and possibly confound joining rates.

From these data, we calculated 1) the proportion of subordinate females
that joined each focal female out of the number of subordinate females alive in
the population at the time of the survey and 2) the proportion of dominant
females that joined each focal female out of the number of dominant females
alive in the population at that time. Both proportions were divided by the total
number of focal hours during which each female was observed to correct for
variation in sampling effort among females. We also compared the number of
subordinate and dominant females available to join each focal female to ensure
that our joining rates did not simply result from biases in the number of

individuals available from each category.

71

Tests of hypotheses suggesting beneﬁts of rank-related partner choice

Within each dyad where female A outranked female B, we calculated an hourly
rate of aggression observed in each of two contexts, when food was absent and
when it was present. We calculated rates of aggression within each dyad by
dividing the total number of aggressive acts female A directed towards female B
in a particular context by the total number of hours A and B spent together in that
context. We required that both members of each dyad be observed together in
each context for at least 6 hours during their adult lives. We measured feeding
tolerance by dividing the number of scans in which female B fed when female A
was present by the number of scans in which both A and B were recorded
together during their adult lives in sessions where food was present, and then
multiplied by 100. The only dyads included in this analysis were those in which
both members were present concurrently as adults at sessions with food during
at least 15 scans.

To evaluate rates at which females provided coalitionary support to other
females, we analyzed ﬁghts in which individuals transformed dyadic ﬁghts into
coalitionary aggression by intervening to support particular aggressors. For
example, we analyzed ﬁghts in which female A directed dyadic aggression
towards female B, and female C intervened in the ongoing ﬁght to support female
A. In this case, A was the ‘aggressor’ (recipient of coalitionary support), B was
the ‘target’ of coalitionary aggression, and C was the ‘supporter’ (donor of
coalitionary support). First, we determined how relative rank influenced rates at

which females supported aggressors in coalitionary aggression by comparing the

72

percent of available subordinate and dominant supporters joining each focal
aggressor per hour. For each focal aggressor, we counted the number of ﬁghts in
which a subordinate or dominant supporter intervened on behalf of the focal
aggressor in sessions in which at least one female was present ranking above
the focal aggressor, as well as one ranking below. We calculated the percent of
available supporters joining each focal aggressor per hour in each session by
dividing the number of subordinate supporters by the number of subordinates
present, and the number of dominant supporters by the number of dominants
present during that ﬁght, then dividing these values by the number of hours each
session lasted. For each focal aggressor supported by more than one
subordinate or more than one dominant in multiple sessions meeting our criteria,
we calculated an average rate at which subordinate and dominant females
supported the focal aggressor. We then asked whether Als within dyads were
correlated with hourly rates of coalitionary support provided by the dominant
member. To calculate a rate of coalitionary support for each dyad, we counted all
occurrences of coalition formation in which the dominant supporter intervened to
provide coalitionary support to the subordinate aggressor, and divided by the
number of hours we observed that pair together as adults. Dyads observed
together for less than 20 hours were excluded.

Statistical analyses

We employed non-parametric statistical techniques because the distributions of
our data were statistically different from normal (P < 0.001 in all cases). We

calculated correlation coefﬁcients, Spean'nan’s R, to ascertain whether Als varied

73

with social rank or rank distance, and to test the direction and strength of
relationships between Als or rank distance and 1) hourly rates of dyadic
aggression, 2) feeding tolerance by dominant females, and 3) hourly rates of
coalitionary support provided to subordinates by dominant females. A Mann
Whitney U-test was used to inquire whether mean Als differed between the
sexes. Wilcoxon signed rank tests for matched samples were used to determine
whether Crocuta associated to a different extent with subordinate and dominant
partners of the same-sex that were adjacently ranked. We also used this test to
compare the percentages of available subordinate and dominant females joining
focal females per hour, and to test whether the numbers of subordinate and
dominant individuals that were available to join focal females differed. We used
Wilcoxon signed rank tests to determine whether aggression within dyads
differed between sessions in which food was present and those in which it was
absent, and whether subordinate and dominant females differed with respect to
the rates at which they provided coalitionary support to focal females. We
reported the sum of ranks for the Wilcoxon signed rank tests as exact values (T)
and asymptotic values (Z) for tests with sample sizes of less than 16 and at least
16, respectively, as suggested by Mundry and Fischer (1998). We used SYSTAT
Version 8.0 (Systat Software Inc., Richmond, CA, USA) to perform Mann-
Whitney U-tests and Wilcoxon signed rank tests, and STATISTICA 6.1 (StatSoft,
lnc., Tulsa, OK, USA.) to ﬁt distributions and perform Spearman rank
correlations. We conducted all tests based on two-tailed probabilities, considered

differences to be statistically signiﬁcant when alpha was less than 0.05, and

74

where appropriate, report mean values :I: SE.

RESULTS

' Effects of sex, social rank, and rank distance on association pattems

Unrelated adult females associated signiﬁcantly more closely with each other (AI
= 0.037 1 0.002, N = 50 females) than did unrelated adult males (AI = 0.018 :I:
0.002; Mann Whitney U = 2467.00, N = 57 males; P < 0.001). For both sexes,
mean intrasexual Als increased with social rank (Figure 3.1A). This relationship

' was statistically signiﬁcant among females but not among males (Spearman rank

correlation for females: R3 = -0.88, N = 24 rank positions, P < 0.001; for males:

R3 = -0.39, N = 18 rank positions, P = 0.098). Both sexes of Crocuta associated
most often with non—kin holding social ranks similar to their own, as indicated by
signiﬁcant negative relationships between intrasexual Als and rank distance,

(Spearman rank correlation for females: R3 = -0.97, N = 24 rank distances, P <

0.001; males: R3 = -0.86, N = 18 rank distances, P < 0.001; Figure 3.18).

When we controlled for rank distance, females associated signiﬁcantly
more often with females occupying rank positions immediately above them than
they did with females occupying rank positions immediately below them
(Wilcoxon signed ranks test; 2 = -1.99, N = 33 females, P = 0.046; Figure 3.2).
However, the relationship between rank and Als among immigrant males was not
statistically signiﬁcant after we controlled for rank distance (Wilcoxon signed
ranks test; 2 = 0.16, N = 49 males, P = 0.87; Figure 3.2). That is, the AIS of focal

immigrant males with adjacent-ranking, subordinate males did not differ from the

75

Association index

Association index

0.06 1

 

 

 

 

 

 

A) lFemaIe-female
QM— I¥¥¥¥f¥ ¥¥¥¥**¥; ¥ T
0.03- O &
:::::HHM lift 11“? ¥ El'..
:::::1 EEEEEEEEEEEE {Male-ma...
dm4§§ }{

H l

:22. % ”Willi? ..

Figure 3.1. Mean 1 SE intrasexual association indices plotted as a function of A)
intrasexual social rank and B) intrasexual rank distance for unrelated adult
Crocuta. Squares represent adult females and circles represent adult males that
held a particular rank while observed with an unrelated adult of the same sex. By
convention, the highest rank possible is 1.

76

 

 

 

 

 

 

 

 

1:: I Female-female
l _ l ElMale-male
g 0.05 - N ' 33
'0
.E
5
g 0.04 J N.S.
-- r
3 ' N-49 '
3 _
< 0.03 - | __
0.02 - I “ I
With adjacent-ranking With adjacent-ranking

Subordinate Dominant Subordinate Dominant

Figure 3.2. Mean t SE association indices of Crocuta of each sex with animals
of the same sex occupying rank positions either immediately above, or
immediately below, them in the hierarchy of the clan. The asterisk over the
bracket indicates P < 0.05, and NS. indicates P > 0.87.

77

Als of focal immigrant males with adjacent-ranking, dominant males. We,
therefore, focused our subsequent analyses exclusively on Als among unrelated
adult females, the more selective sex in the context of rank-related, intrasexual
partner choice.

Decisions to join subgroups containing other adult females

The total time of focal animal surveys for each of the 8 focal females ranged from
2.13 hours to 6.30 hours, averaging 4.39 :l: 0.46 hours per female. During these
surveys, unrelated adult females joined subgroups containing one of the eight
focal females a total of 18 times such that unrelated adult females joined each
focal female at an average rate of 0.54 :I: 0.11 times/h. Subordinate females
joined focal females 15 times (83.3% of fusion events), whereas dominant
females joined focal females only three times (16.7% of fusion events). After
correcting for the number of potential subordinate and dominant joiners available
in the population for each focal female, we found that the percentage of available
subordinate females joining focal females per hour was signiﬁcantly greater than
the percentage of available dominant females joining focal females per hour
(Wilcoxon signed ranks test, T= 36, N = 8 females, P < 0.008; Figure 3.3). This
result was not simply an artifact of the number of unrelated subordinates (N = 9 :I:
2) and dominants (N = 11 :t 2) available to join focal females because numbers of
individuals from each category did not differ signiﬁcantly (Wilcoxon signed ranks
test; T- = 19, N = 8 females, P = 0.95). Overall, subordinate females selectively
joined subgroups containing dominant females signiﬁcantly more often than vice

versa.

78

 

0.12

—:r-

 

0.10 ‘ I I
N 8

0.08 -

0.04 a

% available females joining/hr
O
O
O)

 

   

0.00 -
Subordinate females Dominant females

Non-kin joining subgroups containing focal

Figure 3.3. Proportion of unrelated subordinate and dominant adult females
available to join that actively joined subgroups containing focal females, divided
by the number of hours during which we monitored each focal female. The
asterisk over the bracket indicates P < 0.05.

79

 

Tests of hypotheses suggesting beneﬁts of rank-related partner choice

Among dyads of unrelated adult females, the majority (95.2%) of all aggressive
acts (N = 2374) were directed by dominants towards subordinates. Therefore, all
rates of aggression reported here refer to aggressions directed by dominants
towards subordinates within dyads. The mean rate of dyadic aggression was
slightly higher at sessions in the presence of food (0.06 i 0.005
aggressions/hour) than in its absence (0.05 :t 0.003 aggressions/hour), but this
difference was not statistically signiﬁcant (Wilcoxon signed ranks test, 2 = 1.51, N
= 367 dyads, P = 0.13). The beneﬁts received by subordinates from dominants
were more strongly related to Als than to rank distance for all measures of
aggression when we considered Ale and rank distance as explanatory variables
(Table 3.1). In every case, the effect of rank distance was substantially weaker
than that of Als. As predicted by the reduced harassment hypothesis, rates of

dyadic aggression were significantly negatively related to Als within dyads when
food was absent (Spearman rank correlation, R3 = -0.33, N = 367 dyads, P <
0.001; Figure 3.4A). The same relationship emerged when food was present,
supporting the ﬁrst prediction of the feeding tolerance hypothesis (Spearman
rank correlation, R3 = —0.12, N = 367 dyads, P < 0.001; Figure 3.48). Tolerance
by dominants of subordinates during feeding was signiﬁcantly higher within

dyads that associated more often than within dyads that associated less often,

supporting the second prediction of the feeding tolerance hypothesis (Spearman

rank correlation, R3 = 0.33, N = 37 dyads, P = 0.045; Figure 3.4C).

80

Table 3.1. Spearman rank correlations between association indices or rank
distances and hourly rates of A) non-food and B) food-related dyadic aggression
directed by dominants towards subordinates, C) feeding tolerance of
subordinates by dominates, and D) hourly rates at which dominants provided
coalitionary support to subordinates within dyads of unrelated adult females.

 

Dependent variable Independent variable
Association index (Al) Rank distance
RS P-value Rs P-value

 

A) Non-food aggressions/hour -0.33 <0.001*** 0.14 < 0.006***
B) Food-related aggressions/hour -0.12 <0.001*** -0.003 0.96

C) Feeding tolerance (% scans) -0.33 0045* -0.13 0.44

D) Fights supported/hour 0.093 0.078 -0.033 0.53

 

Asterisks indicate signiﬁcant relationships at P < 0.001‘***’ and P < 0051*).

81

 

O 0.02 0.04 0.06 0.08 0.1 0.12

Association index

 

 

 

0.8
.E
7: 0.6
C
.2
in 0.4
t,
u,o.2
<

o :::,_.n_ , , .11.;I ,.-.....
0 0.02 0.04 0.06 0.08 0.1 0.12
Association index

5 so — C)
8 50 — ° °
3 40 O o
7; O o 00 B 8
g 30 . 0 <9 (9 o o ‘6 o
(I!
E 20 . o o oo o @o oo
8 o 0
m 10 O
.E o
g 0 l l I l l I
'L 0.00 0.02 0.04 0.06 0.08 0.10 0.12

Association index

Figure 3.4. Hourly rates of dyadic aggression directed by dominant females
towards subordinate females when food was A) absent and B) present (N = 367
dyads for both). C) Feeding tolerance, measured as the percent of scans in
which a dominant female tolerated a subordinate female feeding at a kill when
both were concurrently present, plotted against association indices within 37
dyads.

82

Unrelated adult females provided coalitionary support to focal female
aggressors during a total of 28 ongoing dyadic ﬁghts involving 12 adult female
aggressors that started ﬁghts in sessions where both subordinate and dominant
unrelated females were present. As expected, dominant females supported
aggressors at a signiﬁcantly higher mean hourly rate than did subordinates
(Wilcoxon signed ranks test, T= 66, P = 0.034; Figure 3.5A). However, our data
did not support the main prediction of the coalitionary support hypothesis; we
detected no signiﬁcant relationship between Als and hourly rates at which

dominant females provided coalitionary support to subordinate females within
dyads (Spearman rank correlation, R3 = 0.093, P = 0.078, N = 367 dyads; Figure

3.5B).

DISCUSSION

Sex differences in rank-related partner choice

Overall, high-ranking animals of both sexes were more gregarious than low-
ranking individuals. Our data suggest that Crocuta assess the ranks of same-
sexed conspeciﬁcs and associate most often with the adjacent-ranking
individuals based on those assessments. Females associated more often with
same-sexed adults occupying rank positions directly above them than they did
with individuals ranking immediately below them in the clan’s dominance
hierarchy. Unrelated adult female hyenas also generally associated with one
another more closely than did immigrant males, reﬂecting the same patterns as

those found during greeting ceremonies of Crocuta (East et al. 1993). Greeting is

83

 

9"

c
E:
x.

 

N
0|
1

n=12

.N
c
4

-L
O
1

.°
or
I

 

Subordinate females Dominant females

Non-kin providing coalitionary support

% available supporters joining/hr
A
0|

.°
0

 

 

|_ 0.06 — B)
.C
E 0.054 0
F 0.04 —
a 0
$0.03 ~ 0
9 0.02 -
50011 00% 0000 o
r: ' o 0
0-00 {50. {Be 9

0.00 0.02 0.04 0.06 0.08 0.10 0.12
Association index

 

 

Figure 3.5. A) Proportion of subordinate and dominant females present during
ongoing ﬁghts that intervened to provide coalitionary support to focal female
aggressors per hour. The asterisk over the bracket indicates P < 0.05. B) The
relationship between association indices and rates at which dominant females
provided coalitionary support to subordinate females within 375 dyads.

an afﬁliative behavior exhibited during subgroup reunions (Kruuk, 1972); females
initiate more greetings with same-sexed partners than do males, and
subordinates initiate greetings more often than do dominants (East et al., 1993).
Similarly here adult females associated more often with same—sexed partners
than did adult males, and subordinates initiated associations more often than did
dominants. Adult males in our study also associated less often with one another
(AI = 0.018 :I: 0.002) than with adult females (AI = 0.062 :I: 0.002: reported by
Szykman et al. 2001). The association patterns we observed among hyenas
were generally consistent with those in female-bonded groups of cercopithecines
(lsbell and Young, 2002; Sterck et al., 1997a; Wrangham, 1980), suggesting that
sexually dimorphic patterns of reproduction and dispersal generate sex
differences in association in hyenas as they do in primates.

Adaptive partner choice in ﬁssion-fusion societies

Dominant hyenas beneﬁt from cooperative partnerships with subordinates (Kruuk
1972; Cooper 1991; Boydston et al. 2001), but our current analysis suggests that
low-ranking females actively join subgroups containing dominants. At ﬁrst, this
surprised us, given that dominants often usurp food from subordinates, and that
animals are more likely to be targets of dyadic aggression in the presence than in
the absence of dominants (Frank, 1986; Mills, 1990). However, the beneﬁts
gained by subordinates from joining subgroups containing dominant females may
outweigh these potential costs. Subordinate females who associated more often
with dominant females were the recipients of relatively low hourly rates of

aggression both away from food and at kills, and they also gained relatively good

85

access to food. These data suggest that females minimize the costs of group
living by initiating social interactions with conspeciﬁcs that outrank them, and by
forming relationships likely to offer them return beneﬁts in a variety of ways. By
refraining from driving close subordinate associates away from food, dominants
allow subordinates to access a highly valued resource, which should enhance
subordinates’ ﬁtness.

Life in a ﬁssion-fusion society appears to permit Crocuta to be more
ﬂexible in their decisions about partner choice than animals living in more
cohesive social groups, including most cercopithecines (Conradt and Roper,
2000; Couzin, 2006). The salient features of ﬁssion-fusion societies likely to
affect decisions about partner choice are that subgroups are highly labile, and
that opportunities exist for dispersive conﬂict resolution (Wahaj et al., 2001).
Patterns of social tolerance in Crocuta resemble those documented for other
mammals living in ﬁssion-fusion societies. For example, African elephants
(Loxodonta am’cana) avoid one another to minimize conﬂict (de \frlliers and Kok,
1997). Moreover, fed deer (Cervus elaphus) stags living in ﬁssion-fusion
societies structured by stable linear dominance hierarchies (Clutton-Brock et al.,
1982) direct the lowest rates of dyadic aggression towards close associates
(Appleby, 1983a). Low rates of aggression result when subordinate stags actively
avoid dominants, but low rates of aggression within Crocuta dyads appeared to

be the product of partner solicitation, not avoidance, by subordinates.

86

Altemative theoretical ﬁameworks in which to View partner choice

Three types of models have been developed to explain patterns of partner choice
among gregarious animals outside the context of mating. These models include
those derived from optimal reproductive skew theory, Seyfarth’s (1977) rank
attractiveness model, and models derived from biological market theory. Here we
consider the extent to which predictions from each of these classes of models
are consistent with our data documenting partner choice in Crocuta.

Early skew models (e.g., Vehrencamp 1983) were developed to predict
patterns of reproductive partitioning, but more recent skew models also predict
patterns of group size and resource partitioning within groups of social foragers.
Patterns of partner choices among Crocuta are partially consistent with
predictions of a transactional skew model (Clutton-Brock, 1998; Reeve et al.,
1998). Speciﬁcally, Hamilton’s (2000) model predicts when subordinates should
choose to join a foraging group in which a dominant animal controls resource
access. This model predicts how high-ranking individuals who control resource
division should adjust their monopolization of resources according to the costs
and beneﬁts of grouping, and the extent of theiroability to control resource
partitioning within the group. It also predicts that dominants should permit
subordinates to feed as an incentive to stay when the presence of additional
group members at a feeding site increases the ﬁtness of dominants (Hamilton,
2000)

Hamilton’s (2000) model assumes that the presence of subordinates

enhances the ﬁtness of dominants, and this is true among Crocuta; dominants

87

need subordinate allies during cooperative defense of the clan’s territory and
carcasses against lions or other hyenas (e.g., Boydston et al. 2001), and when
hunting certain types of ungulates (Kruuk 1972; Holekamp et al. 1997b.).
Furthermore, Hamilton’s model correctly assumes that subordinate Crocuta will
be less likely to leave a group if offered a proportion of the resources obtained by
the group, and that dominants can control the division of at least some of the
resources obtained by the group. However, this model further assumes that the
subordinate’s only choice to joining the group is to forage alone, and that
subordinates cannot choose among alternative groups; neither of these
assumptions hold in Crocuta. Most importantly, the skew model assumes that all
subordinates will receive an equal share of the dominant’s leftovers, an
assumption not met in hyenas because subordinates themselves vary in rank.
This highlights the more general inability of skew theory to predict patterns of
association and partner choice in Crocuta. That is, skew models ignore individual
variation within dominant or subordinate classes, and assume that all members
of eaCh class are identical. Whereas Hamilton’s (2000) model successfully
predicts that subordinates will join groups containing dominants, and that
dominants will offer subordinates incentives to stay by permitting them to feed, it
fails to predict that subordinates will choose to associate with dominants ranking
just above them in the clan’s social hierarchy. One type of model derived from
skew theory that takes intra-class variation into account is referred to as a
“bidding game (Reeve, 1998; Reeve and Emlen, 2000)." However, this is in effect

a “biological market” model (Noe and Hammerstein, 1994; Noe and

88

Hammerstein, 1995; N05 et al., 2001). We suggest biological market and rank
attractiveness (Seyfarth 1977) models are more consistent with our observations
than are skew models.

Originally proposed to explain rank-related association and grooming
patterns in primates, Seyfarth’s (1977) rank attractiveness model predicts that
rates at which dominant animals offer protective coalitionary support to
subordinates during ﬁghts should be proportional to the rates at which
subordinates initiate afﬁliative interactions with dominants. This model thus
assumes that multiple types of cooperative behaviors are involved in the
formation and maintenance of long-term social bonds, and that any particular
helpful behavior might be exchanged, even after a substantial lapse of time, for
cooperative behavior of a different sort. Similarly, biological market theory
assumes individuals can exchange various types of goods and services, such
that cooperative interactions may involve use of multiple “currencies” traded over
extended periods of time (N06, 2001; Noe, 2006; Not: and Hammerstein, 1994).
Through these exchanges, market theory predicts that individuals should make
decisions to gain immediate beneﬁts from social interactions in the short-term,
net beneﬁts at the end of a series of social interactions in the long-term, or both
(N05 and Hammerstein 1994, Noe 2001, 2006). For example, baboons gain
immediate hygienic and hedonic beneﬁts from grooming (Barrett et al. 1999;
Barrett and Henzi 2001) as well as the net beneﬁts associated with Iong-terrn

social bonds (Silk et al., 2003).

89

Rank attractiveness and biological market models assume that
asymmetries exist between the goods or services offered by dominants and
subordinates within hyena dyads because dominants control access to
resources. A key prediction made by these theories is that subordinates living in
despotic societies should trade services in exchange for tolerance by dominants
(N06, 2006). These theories, however, relax the assumption of most
transactional skew models that dominants have complete control over resources
shared by the entire clan, and assume instead that dominants only control
commodities within their immediate subgroup or current marketplace. Dominant
hyenas can facilitate subordinates’ access to commodities such as food and
space (Frank 1986; Boydston et al. 2003), and they can also provide effective
coalitionary support to subordinates during within-group contests (Frank 1986;
Smale et al. 1995; Engh et al. 2005). In exchange for these commodities,
subordinates can offer services to dominants with which they associate, including
help with prey capture and defense of resources from conspeciﬁcs and lions.
Association provides a proxy measure of the potential aid of these sorts that
subordinates can offer to the dominants with which they associate. Here
dominant hyenas differentially permitted access to monopolized food and space
by the subordinates with which they associated most often.

Both rank attractiveness and market models predict that the rates at which
dominant animals offer social tolerance, feeding tolerance, and protective
coalitionary support to subordinates during ﬁghts should be proportional to the

rates at which subordinates initiate social interactions. Among Crocuta, dominant

90

females participated more frequently in coalitionary aggression than did
subordinates, but our data were inconsistent with the coalitionary support
prediction because rates of support provided by dominants to subordinates were
not correlated with Als. However, our data were consistent with the other two
predictions: rates of both social and feeding tolerance were correlated with Als
among unrelated adult females.

In contrast to skew theory, models of rank attractiveness and biological
markets emphasize the importance of individual variation within dominant and
subordinate trading classes. That is, when conspeciﬁcs within trading classes
' vary in their relative ability to provide commodities, individuals should be able to
assess the value of each potential social partner, and compete for partners of the
highest relative value based on those assessments. The rank attractiveness
model suggests that priority of access to social partners is largely determined by
an individual’s social rank. By virtue of their status, high-ranking animals can
provide greater beneﬁts to other individuals than can middle- or low-ranking
animals. This ability to provide more beneﬁts makes high-ranking animals the
most attractive social partners, but only other high-ranking individuals have
sufﬁcient free choice to interact frequently with them. Similarly, market theory
predicts that this outcome results from market forces favoring those individuals in
weak bargaining positions to become less selective by loWering their demands,
but allowing individuals in strong bargaining positions to become more selective

and demanding.

91

Unlike traditional skew theory, these other models predict that low-ranking
individuals cannot compete effectively for access to the highest-ranking
conspeciﬁcs. Speciﬁcally, market forces associated with Competition are
expected to generate patterns of assortative, mutual partner choice involving
evolutionary trade-offs between the demands that an individual would prefer to
ask of a particular partner in the biological marketplace and the demands
imposed by other animals competing for access to that same partner. This
occurs in human markets (Pawlowski and Dunbar, 1999), and in the societies of
various non-human primates (Barrett and Henzi, 2001; Barrett et al., 1999; Henzi
and Barrett, 2002; Henzi et al., 2003; Manson et al., 2004). Our joining data are
consistent with these expectations because female Crocuta prefer partners that
outranked them. Although the highest-ranking animals in the clan could
potentially offer more goods and services, we found that subordinates associated
most closely with the animals holding ranks immediately above them rather than
with the highest-ranking females in the clan. Skew models cannot account for
this result. However, if the highest-ranking females pair off ﬁrst with other high-
ranking individuals, then our result could be the outcome of competition in which
females must settle for social partners ranking just above them in the overall

hierarchy.

Dominant Crocuta tolerated subordinates based on social relationships
lasting from one to eleven years. Our ﬁndings suggest that dyads of female
Crocuta make repeated “cooperative investments” (Noe, 2006) in long-term

partnerships, as predicted by both rank attractiveness and biological market

92

models. Biological market theory, in particular, is consistent with the large
amount of unexplained variation we observed in rates of aggression and feeding
tolerance during our study, which suggests that the values of services offered by
social partners vary dynamically within the life span of these relationships and
that a Crocuta clan might indeed be a biological marketplace. If this is the case,
then decisions made by Crocuta about partner choice should also be inﬂuenced
by immediate circumstances encountered within shorter portions of our study
based on the socio-ecological conditions prevailing during those periods, as
predicted by market theory. Thus, future work should inquire how variation in
these larger-scale market forces shapes patterns of social interaction and partner
choice among spotted hyenas. To more fully understand the rules governing
decision-making with respect to partner choice, it will be necessary to identify the
cues used in assessment of partner value, and to determine how the dynamics of
social behavior at the individual level vary with larger scale socio-ecological

conditions.

93

Chapter 4
Smith JE, Van Horn RC, Powning KS, Cole AR, Graham KE, Memenis SK,

Holekamp KE, 2010. Evolutionary forces favoring intragroup coalitions in
the spotted hyena (Crocuta crocuta). Behavioral Ecology 21 :284-303.

94

Chapter 4
EVOLUTIONARY FORCES FAVORING INTRAGROUP COALITIONS
AMONG SPOTTED HYENAS AND OTHER ANIMALS
INTRODUCTION
As originally proposed by Charles Darwin (1859), the theory of natural selection
fails to explain seemingly altruistic acts in which animals reduce their individual
ﬁtness to help increase the ﬁtness of others. In light of this problem, behavioral
ecologists have invoked the theoretical constructs of kin selection (Hamilton,
1964), reciprocal altruism (T rivers 1971), direct beneﬁts (also called by-product
mutualisms, Brown, 1983; Connor, 1995; West-Eberhard, 1975), and group
selection (Wilson, 1975a; Wilson and Wilson, 2007) to explain the evolution of
cooperation (reviewed by Clutton-Brock, 2009; Dugatkin, 2002; Noe, 2006;
Nowak, 2006; Queller, 1985; Sachs et al., 2004; West et al., 2007a; West et al.,
2007b). Agonistic aiding, also called intervention or coalition formation,
represents a cooperative act; intervening in a ﬁght is potentially costly to the
donor, who risks physical injury, expends energy ﬁghting, and allocates time to
this behavior that might otherwise be devoted to other activities. Agonistic aiding
is beneﬁcial to the recipient because it increases the recipient’s likelihood of
winning the ﬁght.
Agonistic aiding occurs when group members join forces to attack either

members of their own social group (intragroup coalitions), or members of a
different group (intergroup coalitions). A social group is “any set of organisms

belonging 'to the same species that remain together,... interacting with one

95

another to a distinctly greater degree than with other conspeciﬁc organisms” (p.
8, Wilson, 1975b). Whereas researchers have investigated patterns of intergroup
coalition formation in many taxa (reviewed by Creel and Macdonald, 1995;
Fashing, 2001; Harcourt, 1992), research on intragroup coalitions has historically
focused on non-human primates. In fact, early workers suggested that complex
patterns of intragroup coalition formation might be unique to primates. For
example, Harcourt (1992) proposed that primate coalitions are more complex
than non-primate coalitions. Speciﬁcally, he posited that intragroup coalitions
among primates might occur at higher frequencies, involve larger numbers of
coalition partners, and involve adult beneﬁciaries of support more often than do .
non-primate coalitions. Further, because the outcomes of coalitions can affect
the social ranks of individual group members, and because rank often inﬂuences
reproductive success, agonistic aiding should theoretically have profound ﬁtness
consequences for all participants (lsbell and Young, 2002; Seyfarth, 1977; Sterck
et al., 1997a; Wrangham, 1980). Nonetheless, most studies have investigated
adult male coalitions or interventions by adult females on behalf of their immature
offspring (reviewed by de Waal and Harcourt, 1992; Olson and Blumstein, 2009).
As a result, we currently understand little about the evolutionary forces promoting
intragroup coalitions among adult females, especially in non-primates (Silk,
2007a;b).

In light of recent advances in our understanding of social complexity
among non-primates, we ﬁrst perform a comprehensive review of 49 species to

evaluate the notion that intragroup coalitions formed by primates differ from those

96

formed by non-primates. We then conduct a detailed investigation, documenting
patterns of coalition formation in a non-primate, the spotted hyena (Crocuta
crocuta). Spotted hyenas are gregarious carnivores that live in female-bonded
groups (Holekamp et al., 1997a; Smith et al., 2007), called clans (Kruuk, 1972).
Because hyena clans are strikingly similar in size and hierarchical structure to
troops of cercopithecine primates (Drea and Frank, 2003; Holekamp et al., 2007),
this provides a unique opportunity to compare the evolutionary forces favoring
coalition formation in primates and carnivores, taxonomic groups that last shared
a common ancestor 90-100 MYA (Springer et al., 2003). Speciﬁcally, here we
describe patterns of coalition formation among various age-sex classes of
spotted hyenas belonging to a single, large social group. We then explain these
patterns by inquiring in particular why natural selection favors coalitionary
support among adult female spotted hyenas.

Like troops of cercopithecine monkeys (Drea and Frank, 2003; Holekamp
et al., 2007), hyena clans are multigenerational groups that usually contain
multiple matrilines of adult females and their offspring, as well as multiple adult
immigrant males born elsewhere (Kruuk, 1972). Further, as in many
cercopithecines (Chapais, 1992), adult female spotted hyenas provide
coalitionary support to their juvenile offspring to aid them in rank acquisition (East
et al., 2009; Engh et al., 2000; Smale et al., 1993). Although it has been shown
that hyena siblings generally form coalitions at higher hourly rates than do
unrelated individuals (Wahaj et al., 2004), it is unknown to what extent kinship

and other factors inﬂuence aiding decisions made by adult spotted hyenas.

97

Because the ultimate goal of this investigation is to elucidate the evolutionary
forces shaping this decision-making process, here we test hypotheses
suggesting that kin selection, reciprocal altruism, and direct beneﬁts,
respectively, drive intragroup coalition formation among adult female spotted
hyenas, as detailed below.

Predictions based on kin selection

Kin selection theory (Hamilton, 1964) predicts that if individuals possess the
ability to discriminate on the basis of kinship, then they should gain inclusive
ﬁtness beneﬁts by biasing helpful behavior towards relatives, and harmful
behavior away from them. Because spotted hyenas can discriminate among
conspeciﬁcs based on both maternal and paternal kinship (Van Horn et al.,
2004b; Wahaj et al., 2004), kin selection theory may explain patterns of coalition
formation among adult females. If this theory is correct, then females should
intervene to support females to which they are most closely related, and direct
coalitionary attacks at lower frequencies or intensities towards relatives than
towards non-relatives. Morebver, when individual ﬁtness costs associated with
ﬁghting are context-dependent (Maynard Smith and Price, 1973), Hamilton’s rule
(1964) predicts females should help kin most often when costs to donors are low,
or when beneﬁts to beneﬁciaries are high. Because clans are ﬁssion-fusion
societies in which individual members travel, rest and forage in subgroups that
frequently change composition (Kruuk, 1972; Mills, 1990; Smith et al., 2008), the
immediate social environment in which cooperation occurs is variable. Therefore,

because hyenas typically direct attacks towards subordinates (Frank, 1986), the

98

number of dominant bystanders present at any particular ﬁght should represent a
proxy for the risk of retaliation for joining that ﬁght. Thus, if females act to
minimize their own risks, then they should intervene most often when ﬁght
intensity is low or when few dominant bystanders are present. However, if
hyenas base decisions solely on the projected beneﬁts to kin, then donors should
intervene most often during ﬁghts of high intensity when kin are most at risk.
Predictions based on reciprocal altmism

Natural selection might favor interventions on behalf of non-kin via reciprocal
altruism if the projected future beneﬁts to the donor outweigh the immediate
costs (T rivers, 1971). Although ﬁrm evidence of reciprocal altruism among non-
kin in animals societies is rare (Clutton-Brock, 2009), reciprocity is most likely to
evolve in long-lived animals with low dispersal rates, frequent social interactions,
and cognitive abilities permitting individuals to remember earlier interactions
(Ridley et al., 2005). Female spotted hyenas satisfy these conditions. They live
up to nineteen years in the wild (Drea and Frank, 2003), they are philopatric
(Smale et al., 1997) and highly gregarious (Holekamp et al., 1997a; Smith et al.,
2007), and they demonstrate many of the same cognitive abilities as monkeys
(Holekamp et al., 2007).

If the reciprocal altruism hypothesis is correct, then female hyenas should
offer other females either immediate coalitionary support (direct reciprocity) or
access to other commodities (interchange trading) in exchange for future
coalitionary support from non-kin (Hemelrijk and Ek, 1991). However, partner

choice mechanisms (Schino and Aureli, 2009) based on the threat of switching

99

partners within the ‘biological marketplace’ (Noe and Hammerstein, 1994) failed
to explain patterns of coalition formation among unrelated adult female hyenas in
our previous study (Smith et al., 2007). Nonetheless, it remains possible that
females might exchange coalitionary support within a ‘service economy’ (e.g. de
Waal, 1997) for access to commodities such as help during hunting, afﬁliative
greetings, social tolerance, or establishment and maintenance of stable alliances.
Aid during hunting is valuable because pairs of hyenas are 19% more likely to
capture prey than are lone hunters, and because hyenas can only capture some
species of prey when hunting in groups (Holekamp et al., 1997b; Kruuk, 1972).
Greetings occur when two females stand parallel to each other and sniff each
other’s anogenital region, each exposing its fully erect ‘pseudo-penis’ (East et al.,
1993; Kruuk, 1972; Wahaj et al., 2001). Greetings are valuable in that they
reinforce social bonds, but they involve risk of genital injury. Social tolerance
occurs when one female receives low rates of dyadic aggression from another,
thus permitting the former better access to resources (Smith et al., 2007).

Females might maintain alliances based on repeated acts of support.
Whereas coalitionary support indicates short-term cooperation, alliances are
relationships between two individuals who repeatedly join forces over long time
periods (Noe, 1992; Seyfarth, 1977). Therefore, if unrelated adult females form
stable alliances, then donors should intervene repeatedly on behalf of the same .
beneﬁciaries.

Finally, indirect reciprocity might explain patterns of support if members of

the audience observe a donor behaving altruistically and later help that same

100

donor even though the audience members were not themselves direct
beneﬁciaries of the original support (Nowak and Sigmund, 1998; Nowak and
Sigmund, 2005). Because high-ranking adult female hyenas provide the most
effective support (Smith et al., 2007), if females intervene to improve their own
reputation, then we should detect an “audience effect” (Zuberbtihler, 2008) in
which females donate the most support when large numbers of higher-ranking
group members are present in the audience.

Predictions based on direct beneﬁts

Direct beneﬁts result from cooperative acts in which a donor gains immediate
individual ﬁtness beneﬁts by helping, and the recipient beneﬁts as a by-product of
his or her partner’s selﬁsh behavior (Brown, 1983; Connor, 1995; West-Eberhard,
1975). We tested two hypotheses suggesting direct beneﬁts that females might
gain by intervening in ﬁghts.

Food access hypothesis

Feeding competition is intense among spotted hyenas; these animals typically
feed at fresh carcasses that are energetically rich but highly ephemeral, often
persisting for only a few minutes (Frank, 1986; Kruuk, 1972; Mills, 1990; Smith et
al., 2008). Thus, adult female hyenas might directly beneﬁt from joining ﬁghts if
doing so provides them with immediate opportunities to feed (Mesterton-Gibbons
and Sherratt, 2007). This hypothesis predicts that females should intervene most
often during ﬁghts over food and during times of year when feeding competition

is most intense, as indicated by low prey abundance (Holekamp et al., 1996).

Although recently attacked hyenas rarely depart altogether from kills in response

101

to conspeciﬁc aggression (Smith et al., 2008), coalition partners might be more
likely to gain access to food if coalitionary aggression more effectively displaces
competitors from kills (e.g. the victim of aggression retreats from the carcass)
than does dyadic aggression.

Status guo hypothesis

Because high-ranking female hyenas enjoy greater reproductive success than
low-ranking females (Frank et al., 1995; Hofer and East, 2003; Holekamp et al.,
1996), females might directly beneﬁt by providing support if doing so either
reinforces or improves their rank positions in the dominance hierarchy. If support
reinforces the status quo, then high-ranking females should intervene most often
and direct most coalitionary attacks towards subordinate targets. Conversely, if
females generally intervene to improve their own social status, then low-ranking
females should intervene most often, and direct most attacks towards dominant

targets.

METHODS

Study site and subjects

The study area, in the Masai Mara National Reserve, Kenya, is characterized by
open, rolling grasslands. Prey availability varies seasonally and feeding
competition is reduced annually when migratory ungulates join resident herds
(Holekamp et al., 1996; 1997b; 1999; Smith et al., 2008). Our {subjects belonged
to a single large clan of spotted hyenas that defended a common territory in the

study area (Boydston et al., 2001). Although this clan has been continuously

102

monitored since May 1988, limited paternity data were available prior to 1996
(Engh et al., 2002) and the clan permanently split into two clans by 2001 (Smith
and Holekamp, unpublished data). Thus, all coalition data reported here were
collected from January 1996 to December 2000 to ensu‘re that subjects belonged
to the same social group, and that they were of known maternal and paternal
relatedness.

We identiﬁed individual hyenas by their unique spots, and we sexed them
based on the morphology of the erect phallus (Frank et al., 1990); we estimated
the ages of cubs to i 7 days (Holekamp et al., 1996). We classiﬁed females as
adults at 36 months of age or at their ﬁrst known date of conception, whichever
occurred ﬁrst. Males disperse 1-76 months after puberty (Smale et al., 1997; Van
Horn et al., 2003), which occurs around 24 months of age. We considered natal
males older than 24 months and all immigrant males to be adults, and all natal
animals that had not yet reached adulthood to be juveniles.

All adult female hyenas are socially dominant to all adult immigrant males
(Frank, 1986). We ranked adult females in a linear dominance hierarchy for each
month of the study (Holekamp and Smale, 1993; Smale et al., 1993). Adult
females generally maintained stable dominance relations, and most ranks
changed gradually over time due to maturation and deaths (Frank, 1986;
Holekamp and Smale, 1993). However, two brief periods of social instability
occurred during this study (Van Meter et al., 2009).

Terminology

We deﬁned a ‘dyadic ﬁght’ as any agonistic interaction involving only two

103

contestants. The act of a third individual (here called the ‘donor of support’)
joining an ongoing dyadic ﬁght transformed it into a triadic interaction, called a
‘coalitionary intervention’ (de Waal and Harcourt, 1992). The donor intervened on
behalf of one of the. original contestants (here called the ‘beneﬁciary of support’),
but behaved aggressively towards the other original opponent (here called the
‘target of coalitionary aggression’). The donor and beneﬁciary of support were
‘coalition partners’. ‘Coalitionary aggression”, or ‘coalition formation’, referred
more broadly to all agonistic acts involving at least two aggressors
simultaneously joining forces to direct aggression towards the same target.
‘Coalition size’ was the total number of coalition partners (e.g. donors of support
plus the beneﬁciary of support) that joined forces to direct coalitionary aggression
towards the same target.

We characterized the intensity of each aggressive act based on the
highest level of threat involved, and thus the potential risk of injury, as low or high
intensity (Van Horn et al., 2004b). Based on the events immediately before the
conﬂict, we assigned each agonistic interaction to one of ﬁve contexts.
‘Scapegoating’ occurred when a target redirected aggression towards an
individual not previously involved in the ﬁght. Aggressive encounters directly
related to fresh meat occurred in the context of ‘food’. Mothers directed ‘maternal
interventions’ towards targets in defense of their offspring. Aggressive acts were
considered ‘pesky’ when the aggressor attacked the target in response to the
target disrupting ongoing behavior, as for example, when a target involved in play

inadvertently bumped into the aggressor, who was sleeping. ‘Unprovoked’

104

aggressions occurred in the absence of any stimuli obvious to observers, and in
situations in which the aggressor and target had little or no contact during the
minutes preceding the attack.

Behavioral data collection

We observed hyenas daily using our ﬁeld vehicles as mobile blinds. We initiated
an observation session each time we encountered one or more hyenas in a
subgroup separated from other subgroups by at least 200 meters, as detailed
elsewhere (Smith et al., 2008). Upon arrival at each session, and during
subsequent scans performed every 15-20 minutes, we recorded the identity and
activity of every hyena present. We also recorded all occurrences of hunts,
greetings, aggressive acts and appeasement gestures as critical incidents
(Altmann, 1974). We recorded the identity of lone hunters, and when multiple
hunters cooperated during a hunting attempt, we recorded the order in which
other hunters joined the original hunter in pursuit of prey. Because a hyena
solicits a greeting by lifting its leg, we recorded which member of each dyad lifted
her leg ﬁrst during greetings.

When multiple females joined the same ﬁght, we recorded the sequence
in which individuals intervened, and considered this to be a single coalitionary
attack. We tabulated the hourly rate of coalition formation for each hyena
belonging to a particular age-sex class category. To correct for variation in the
amount of time we observed each hyena in situations during which it had
opportunities to form coalitions, we based rates on sessions in which each focal

animal was concurrently observed with at least one potential coalition partner

105

and at least one potential target. Speciﬁcally, we calculated the rate at which
hyena A engaged in coalitions as [(number of coalitions hyena A formed to attack
any clan member)/(number of hours hyena A was observed with at least two clan
members)]. We also calculated a rate of attack directed towards adult females as
[(number of coalitions hyena A formed to attack any adult female)/(number of
hours hyena A was observed with at least one adult female and at least one
other clan member)].

Because the intended beneﬁciary of support is unclear in cases involving
multiple aggressors (Silk, 1992), we limited our model selection analyses to
interventions by individual adult females in ﬁghts between two other adult
females. Following Widdig et al. (2002; 2006), only when more than one
supporter simultaneously intervened in the same dyadic conﬂict did we divide the
event into multiple triads that included the same target and beneﬁciary but
different donors. When multiple acts of support involving the same donor,
beneﬁciary, and target of coalitionary aggression occurred within the same
minute, we considered these interactions to be part of the original ﬁght (Engh et
al., 2005). We, therefore, considered potential donors to have only one
opportunity per minute to join an ongoing ﬁght between the same contestants.
Testing the kin selection hypothesis
To assess the effects of kinship on agonistic aiding, we assigned coefﬁcients of
relatedness to pairs of adult females based on demographic records combined
with genetic data indicating paternity (see Engh et al., 2002; Van Horn et al.,

2004a). Information on maternal and paternal lineages included up to ﬁve

106

generations for each adult female whose presence as an adult in the clan
overlapped with that of at least one of her surviving adult female relatives for
some part of the ﬁve year study period (N = 31 adult females). Here, we assigned
a coefficient of relatedness of zero only to those dyads for which 1) no genetic
data indicated shared paternal descent, 2) females belonged to different
matrilines, and 3) the pairwise Queller—Goodnight R-values for the dyad was less
than -0.118, a value that was less than that of 95% of the known half siblings (N
= 244 pairs of half siblings) in our study clan.

We entered coefﬁcients of relatedness directly into our statistical models
as continuous predictor variables, but we also assigned pairs of adult females to
one of the three following kinship categories to graphically represent our results
and to perform social network analyses: Close kin (coefﬁcient of relatedness (r) =
0.5; mother-daughter or full sisters), Distant kin (r = 0.125 to 0.25; grandmother-
granddaughter, maternal or paternal half sisters, aunt-niece), and ‘Non-kin’ (r ~
0.00). On average, R-values within female dyads examined here were 0.462 :I:
0.028 for close kin (N = 25), 0.279 :I: 0.040 for distant kin (N = 36), and -0.228 t
0.006 for non-kin (N = 161).

To simultaneously investigate patterns of support among multiple females,
we constructed a representative, weighted network in NetDraw Version 2.064 for
each kin group of adult females concurrently alive from January 1997 to
December 1998. First, we constructed a directed graph to represent supportive
interventions. Here arrows connecting females were called ‘arcs’. Arcs originated

from each donor of support, and terminated at each beneﬁciary of support.

107

Members of each dyad were connected by: 1) two arcs (reciprocal support), 2)
one are (unilateral support), or 3) zero arcs (no support). For each kinship group,
we calculated the “density” of its support network as [(number of realized
arcs)/(number of potential arcs)]. Density is a measure of network cohesion that
describes how well connected group members are to one another (Wasserrnan
and Faust, 1994; Wey et al., 2008). Second, we applied similar methods to
construct an agonistic network for this same subset of adult females. In the
coalitionary attack network, arcs originated from intervening females and
terminated at victims of coalitionary attacks.
Testing the reciprocal altruism hypothesis
We used partial matrix correlations to test the null hypothesis of no
correspondence between support given and future commodities received among
unrelated adult females (de Vries, 1993; Hemelrijk, 1990a; Hemelrijk, 1990b).
This method accounts for nonindependence among dyads. Following Hemelrijk
(1990a), we limited this analysis to those cases in which each female was
observed providing support or receiving a monitored commodity at least once,
and both members of each dyad had the opportunity to intervene on behalf of the
other. On average, members of non-kin dyads had 13 i 1 opportunities to
support one another (Range: 2 to 24 opportunities, N = 49 dyads).

We corrected for the number of opportunities available to females within
each dyad to donate support as [(number of times female A joined to support
female B in dyadic ﬁghts)/(number of dyadic ﬁghts in which A observed B ﬁght

with another adult female)] or to help during hunting as [(number of hunts in

108

which A aided B)/(number of hunts in which A observed B hunting)]. We
calculated hourly rates of greeting initiation by A towards B as [(number of
greetings A solicited from B)/(number of hours A and B spent together)]. To
measure social tolerance, we calculated hourly rates of dyadic aggression
received as [(number of dyadic aggressive acts A directed towards B)/(number of
hours A and B spent together)]. We initially calculated separate rates for
tolerance at food and away from food, but because both analyses produced
equivalent results, we simply report a combined measure of tolerance.

Testing the direct beneﬁts hypotheses

First, we tested the food access hypothesis by inquiring whether females were
most likely to intervene in ﬁghts occurring over food, and during months of prey
scarcity. We also asked if coalitionary aggression was more effective at
displacing competitors from kills than was dyadic aggression by comparing the
proportion of ﬁghts in which each female retreated from carcasses after receiving
dyadic or coalitionary aggression from a particular female aggressor. Second, we
tested the status quo hypothesis by comparing the relative ranks of intervening
females and targets of coalitionary attacks. We also compared the proportion of
interventions in which one adult female intervened on behalf of the dominant or
subordinate contestant in dyadic ﬁghts.

Statistical analyses

We tested the effects of kinship, and variables indicating direct beneﬁts, on
intervention decisions using the generalized estimating equations (GEE)

approach to logistic regression for ﬁtting marginal generalized linear models

109

(Faraway, 2006; Hardin and Hilbe, 2003; Liang and Zeger, 1986). This procedure
is particularly good at handling binary data, while including both ﬁxed and
random effects, when model residuals are non-independent and non-normally
distributed (Hardin and Hilbe, 2003; Liang and Zeger, 1986). GEE models for
binomial data are similar to, but extend, traditional logistic regression by
controlling for pseudoreplication resulting from repeated measures (Carlin et al.,
2001; Williams, 2000). We implemented these models using geepack (Halekoh
et al., 2006) in R Version 2.6.2 (The R Foundation for Statistical Computing
2008).

Sample sizes, limited by knowledge of paternal kinship, did not permit us
to build a single model. Instead, as done in similar studies (e.g. Schino et al.,
2007b), we built two separate models to evaluate the effects of the explanatory
variables on decisions made by females to 1) support beneﬁciaries and 2) direct
coalitionary attacks towards targets when given the opportunity to do so. We
used an information-theoretic approach to identify the most parsimonious models
(Burnham and Anderson, 1998). We sequentially entered and dropped all
potential explanatory terms, including all 2-way interactions, and deemed the
candidate model with the smallest Quasilikelihood Information Criterion (QICu)
value to the be the best model (Pan, 2001).

We used STATISTICA 6.1 (StatSoft, lnc., Tulsa, OK, USA.) to conduct
post-hoc comparisons. Whenever possible, we used GPOWER (Faul et al.,
2007) to perform post-hoc power calculations for non-signiﬁcant, univariate

analyses that were based on small sample sizes. We compared means for two,

110

or among more than two, independent groups using the Mann-Whitney U and
Kruskal-Wallis tests, respectively. We used matched pairs for post-hoe
comparisons whenever this was permitted by our sample sizes. We used
Wilcoxon-signed rank tests and Friedman’s ANOVA for repeated measures when
comparing the means of two, or more than two, dependent groups, respectively.
We compared frequencies using the Chi-square test and tested correlations
using Spearman’s R. We applied the sequential Bonferroni adjustment to correct
for multiple testing, and report p-values in their corrected form (Rice, 1989).
Where appropriate, we report mean values :I: 1 SE and sample proportions :I: 1
SD for binbmial trials (Agresti and Coull, 1998). We based Kendall’s (Kr) matrix
partial correlations on 2,000 random permutations and one-tailed probabilities
because the reciprocal altruism hypothesis makes clear, directional predictions
(de Vries, 1993; Hemelrijk, 1990a). All other tests were based on twoJailed

probabilities and considered signiﬁcant at alpha less than 0.05.

RESULTS

Coalitions formed by primates and non-primates share similar characteristics
Overall, the results from our literature review suggest that primates and non-
primates fail to differ signiﬁcantly with respect to the proportion of agonistic
interactions that involve coalitions, rates at which individuals form coalitions,
mean coalition size, or the proportion of coalitions involving only two allies (Table
4.1). The available data suggest that primates of both sexes are signiﬁcantly

more likely to form intragroup coalitions (87.5 % of 32 species studied) than are

111

 

 

 

 

 

 

€00an 00:59

 

s. -l- l: :3 to $2 mm mm 02488 033 .1 6V
G_Eoonmg 02.8
i: ..> i: -l. ii i: l: ..... as 8 22 << 36 :4 «3:5 page as
EEom 53:30.5 35530
-i- ..> > 0.2 s R -l- l: ..... s 4 “_m << 23...”: 9-: .0 «560 As
Aonocoav
m> .> l: as B 6-31m to 0.0 as 3 mm << 3% o «sesame .c as
$8 5-8mm mod -l- $8 ”8623898 :3
m> ..> z _2 i... i.-- cod no -l. _>_: << 382; vm--
m> .33 8-8.3 So to s8 .22 << 32.;
«£3 l: 8.9.. So to $2 2.2 5. 2,3me
-l. .> i: _>_ -l. Nod no la a”. 5.0838 2
..l .> I: _2 I: l... 3.0 to ..l 2.2 << 0633 a «$828523
«.5», > > 5. some 5-00.” mod 5 $2. mm mm 08-8 8:832. 53$
.II II. It. md Nm .xs 9 ”@2030 $900 cams.
-l- -.l ii i: .1 mm mm .l- access
m> .> r-.. -l- -l- i: no mm as 2 mm 2 :8 283mm 0:6: E
g
«ﬂeeces «00:0 03:0 «:5 .095... .
..o 5.00.309. 0:03 cams. «00.330 0:05 800 03 xcw 0a< onion?
60:00.00 Ex 3:0 .530 3n.» .2. .80 .3230 Baum—0:00 0:20.
3.8525. -5. .0 e. 52.80 5332.30 3.x. 0.6.50 383:» 3.8%

.muoecoo 0305055 @550 cozmctoh 53:08 *0 0530.“. ._..v 030...

112

 

 

 

 

 

0:0000E 000E083

 

m> -l- --l 5.0 0.0 as mm mm << 00.40. 00 0.002500... .3. .0:
0000003. __0.0E30v
m> > -l- l: -l- -l- l: --l 00 << 0.0m. 4. 0000.020
m> 2 -l. l: ..... 3.0 ..... as 4 mm 43. 00.00. 0 000.00.... c:
A___.cc0Ev 03.500
.40 N. 8.0. 0.0 .0 0.. as 2 mm mm mm. 3.3.000... a:
I... ll. md l--- 03 5 ”00.030 000.00 c002 €02.00 00009
m> .> > l: E: ..... -l. u: S. 30.0 0.00.00.
ii 0.0 4.4 as .0 mm mm 04. 00005000... .0.
0300000 00>.00E0cv
m> i: is 00.0 «.0 4.00 0. .4... 0.0: 0 0000050.. .0 a.
as 0.. 8.0 4.~ «0.0 -l- as 4. ”00030 000.00 000:
m> z > --l 0.0. hm i: ii 8.9 as 0 u: 5. 3.0.0... .7.
10> .2 l: as 00 3-0. .0 80.0 8.0 as t .23. .3. 380-00 N.
40> .2 i: -l. -l- 8.0 no as 0 .25. << 3.0-5:-..
m> ..l > -l- i... No... .0 as N... H. mm 3.0.14. 0 4.0300000 30:03
.l. .> ..l l: --l as . 0.“. << 348 0 0500000000 .0 E
002.0000 «00.5 002.0 «.3 09.0..
.0 $3020.00. 0003 0002 «00300 000.0 £00 03 x0m 00< 85.000
60:00.00 5.. -=000 "00.0 .00 .00 000.000 E05300: 030.0.
.0. 0000.030 .0 .4. 003.000 5.200.000 .0 .4. 0 .0000 4 000.000 00.030

€85.80. .3 0.04»

113

-.-. .0 0.0 E: 0\0 00 .llll00_0:.0 000.00 0005. .0:0000E

 

 

0> > > s. ..... is i: -...- i- .. .2. 0.0.00. .0 00.0.0000
i: E: ..-0. .0 0.0 0.0 .\. 00 00 mm 0..-... 000.300... 00.0..
--- --- mod --- --- ”00:03.0 000.00 :mm—Z .
000> is > 0.5. -...- i: 00.0 ... 0. .3. 3.0... 00
i: z > s. i: .0 0.0 --..- 00 «.0 0.00.0. 00
0> z > s. E: i: i: E: u... << 3... 0. 0.0000000.
I0> .5. E: s. -.... i: .000 .0 .5. u... 00 00.00... 0. 00000..
0> > > .2 i- i: E: E: 00 << 0.00-00. 00-00 000.0... .s. .0..
0.. 00 -I- 00.0 :I- 0\. 0. ”00.030 000.00 000.).
0> > > s. 0.00 is .00 .0 0.0. mm mm 0..0. .0
i... z > s. E: E: .00 ..0 £- ... << 3.00-00. 00-0
..... ..... ..... s. -..-- E: 0000 00.0 .\. 0. mm 00 0.00.00 0.0000000.
I00> I: I... s. .i- i: ..0 0.. .\. 0 u... << 0...... 0. 0000000...
000> z > s. ..... is .00 ..0 i- 00 mm 000-... 0.000:......0..
.\. 0. .00. 0.0 00.0 ..... .\. 00 00.030 000.00 000;.
0> > > .2 .00. .0.0.0.0 000 0.. n0.00 .2... <0. 0.00.0. ..0000000:00000.
0> z > s. I: i- 00.0 0.0 is n... <<0.00-0..0. 0.0.00. 00.0..
300000.02... 00...... 0.3 .0002.
.0 $3020.00. 000.. 0002 «003:0 0:05 800 02 x00 0m< 0.5.00.0
£260.00 c... 030 ..__000 "00.0 .00 .00 000.000 E03300: 0:26.
.0. 0000030 -00. .0... 00.0.30 3000000000 .0... 0.0000 0.00.300 00.0000

60000000. ...0 0.00.

114

 

 

 

 

-.... -...- Foo ..... ﬁx. m ”ﬂag ﬁgmocmE boomv
m> ..... > 2 E: i: 85 no ..... mm 8 2,838 3 now 3.2.98
«3 --..- ---.. vooo moo axe m u.”— << 383.3 cm 2.30330 88
ii ..... ﬁo ..... £0 5 Hag u.._A>9_coE “mama
-..-- .> i: E: E: -.... ..... --..- E: ”E << 353: mm 23.5% 38
m> > > n22 -.... -.... to me .x. 5. mm mm 38:: .Sﬁoogwoﬁmv
03595.
-.... 5.. wood -..-- co or ”ﬂag :mEachv
----- ----- > _>_ ---.. ----- Nooo Nod ..... “E << 388 or «3:35
..... ..... mood voo $3 5:2 << ammo: m «Bannogl :8
.x. B aé o...“ So i: E: ”38.2 .3383
m> > > .2 $5 $-83 Nod no E: 2.2 << 8:8 5 258:
56> -.... > s. -.... ..... mood to --... “E << 88“.“sz 5 2:62am .E 68
Eugene 8598
s. i: 3.0 No $2 mm mm 0838 $92.28:
$36ng 59.5:
m> -...- > s. I... -..-- moo vo $9 mm mm mvmow «53:0».283
wmacmomE 3:393 «Eta
m> z > s. ..... -.... moo Nd gov mm mm cow .2535:an
«8:23.626 8:3 «E was“:
.3 $302902 «can :35. «33.5 95.5 ﬂow 03 xow am< 2533»
60300.3 Ex 9:» -230 "an? .2. 3a -3059“ bug—2.3 9.29
.5. 3525 .Ex 3.x. 55:30 23:05:30 3.x. “3:00 #6636 wmammm

62.5on .5. 2%...

115

 

 

$9.55 5.30; umzcmEv

 

m> E: > 2 $8 aé QN mood mod ---.. mm << as; 8 83:3 3.23;. :8
€32.58 338-3,“.95
.> i: _>_ ..... i: i: .......... u: << ammo 3” «:8on .o 38
g 8 ad 3 mod E: .x. or 52.8% $93 $22
> E: .5. i: i: <0 2 mm mm 28-8 33538
m> -..-- ...-- Nod 8.0 E: .22 << 33$ v 85-925
m> .z E: E: «a 8 ova 3 mod to E: “E << 3:3 0 2.53%... .o 6%
E5038 omtatssoﬁv
..... z > s. .x. E AEEN «0.0 no .32.. mm mm 02 «=me 23888
Am>mxcoE .9:ng
---.. --... > _2 ---.. ----- ..... ---.. -..-- mm mm 23.3 w..m=m.$.=on .m Ammo
Among-c9: _mtSUmv
-..-- --..- > _2 ---.. ANTNV m. E: -..-- -.... mm mm 3mm .imuﬂwo 353% 63
utoEQ umEotémc «EB
.z > _2 i: i: to ad «a v mm << 86: m 2:5. 5.53m $3
$552 3.56:3
--..- ..... > s. E: i: Nod No 0: mm . << 86% S 3.8 353 9.3
«8:28.. 32% «2...... «:5 395.:
.3 539.300.. 0:03 :35. ~33...» 9.0.3 ﬂow m3 xww ¢m< 85.35"...
£03023 Ex 3:» -230 "on...“ .2. .2. -393? 3.33.9.3 9.9.9
.8 coco—USN -c_¥ 3.x. F52:25 23:02.30 m0 .x. “3:00 uuuwﬁam mmemw

smaczcooo .3 22.:

116

 

 

 

 

 

 

 

 

0%. 3-3 mm no ..... .x. m ”Ev-3|“... $93.: 8.2
m> E: i: n_.s_ E: 5.. od 3 «av mm << 02
a: 3-33 80.0 mod 9: mm mm owl: gigg
m> --- .-..- i: .5. ..... mm mm 0mm... was: £50 £8
. €000 nmwocézczo
i: > as). i: 5.. 5.. :i <0 mm mm mm gum-o 3.5.: .2 as
.x. B 6.0 3 5o ..... <0 mm ”863m 828 :82
..... -> i: i: $5 3 86 ad $8 "E << 02 3380 8:205:
.x. 8 $8 3 i: .x. 0 mm mm Bow-m «:3: «:32 as
ea: 505$
..... i: > _>_ g. o m i: .5. ”E << 3va m 8. 20550 0.8
.x. mm 8-8 3 No a» i ”-85-: $4815.» ilcmmIs. E223 8:80
mxo> z > E). .x. K AEvva to to $3 mm mm 2,80,.“ 332...
m> #2. 3.83 no 3. i.- mm 2. US 3320 88
3
.x. .2 6-8 3 E: i: i: ”ad-mg.- 39.85523
I0> <0 mm 6-0 in i: i: .3. 2.2 << 38$ m Rats-0
10> .x. 0 6i 3 .5. E: i: 22 mm asm-omv :4. 833 a8
3:23.026 8:.“ a? $9.2.
._0 $3095.09. 0:0: 2002 «00330 030.6 £00 03 xam 0m< 25.350
£03020» :3 030 .__000 EN.» .00 .00 .00-230 P530989 0.3-.0.
.qucoaﬁ .50. 3 .x. 52.30 2355.80 3.x. 0.2.8 383:0 $.09?-

szczcoov .3 asap

117

 

 

230% 92299

 

¢> i: > n...2 $8 6&3 no 8 $2 mm mm 882me (3:38:13.
335 50:05. «.3th
m> ..... > n..s. $8 8-3 «a E: i: .x. 2 .2.2 mm. 2.60-8. wmém .38 .o a...
«335 :20:
i: E: no... to i: .22 << 8.5. m 8.5.. use .32
.53 26.3..
m> 53. «o... .x. a .22 << 38$... «is. «Eu... 2:8 5..
.9390; 2.2,.
> > z n...2 $2: 3:” 88... So a: .22 << 8.3.2 «émﬁomas
uAmﬁmN 9.5.3
> i: > .2 ..... i: 02 3 mm mm 83-8 «8%.. 9:3 c...
c.2286 c822.
..--- ..... > n...2 0» «m 21.3 i: E: <2 “r. << 253: NE $8.5m
E: i: > .2 i: i: ---.. --..- -.... m“. ..< 2m~-~ 353318.
o.mE_ca_ou mmocmﬁonv 305:3
> 5.. > n...2 6.. 8 6-8 mm to no .5. .22 << 2.88... a 396:: am.
so... 2.; c8222
m> i: is $2 .2 on --... mood No 5.. mm. mm 2.3 «595532.18.
«£23326 8...“ «.3 39.9.
.5 5.02902 «no: :35. 393:» 93.5 80w. m>_ xmm wm< 85.33
6033.0» Ex aim .530 "mu? .2..— .an .3235 2.5.3.289 9.5.5.
3.8526 -5. 3.2. 82.30 £352.30 3.2 32.8 3833 3.3%

$2.28. .3. «amp

118

 

Quoﬁhm

Nd H QN ch H mod

$nﬂmw

_.0.0>O

 

OF v.0 ﬂ .0>>On_
800 ”S... 002m.

V506 ﬂ .0>>O&
00—20 ”050 .00..w
mud n n. .ON. .3 u N

mmmd n .0260
53.0 650 5th
mod n n. .36 u N

«B... u .250.
v2.0 .8... 602m.
2... no. .5. "N

A00.0E..0-coz .0>

0905.... 0.03....

$523.50....

 

E u 2 .873 .2 u 2 .30. a u 2 0.0-88.3 .3 u 2 a2 03...
.2 S a 8 n... a .0 3... H P to 2. 0 H t 0305:9202 .0
E u 2 .20. an n 2 0.0-08.3 .00 u 2 .2. $0.
.2 u n." E a... a a." S... a B... .2 n a 2 0205... 2
.0:02.000

0020 N 2:3 .3

 

«003:0 .2. .00...

0.0.5 .05. 300

r5022, 00.0000 u 2.

 

0:02.000 .0 .2. 00.0 :02.000 .00 0:02.000 02000.80 .0... .mmc0..mwﬂc00s.
"2.0:.Eam
€09.00...

10> E: E: E: 5.. E: 0.. << 060. 00 03.85:. .0 a...
£00.. 020.0292.
10> ..> z 0.5. :I. i: :I. -.... .3: mm 2. 0w. 03:00 8...
.52.... 0.0.0.38. .2233
m2 ..... > 0.2 i: 6-0. . -...- --..- 2.3” mm 2 2,88. 80... 3:020 E.
302332... 8....“ «.3 .35..
.0 $2005.00. 000.. 000.... 30.300 0:05 0.00 0>. xam ou< 8.3.00.0
£033.00 20. 020 ....000 "00.0 .00 .00 000.30 2.0.2.0000 000.0.
.0. 00:02>m .2! .02. P82.000 2.052.000 .02. 0.0:0o 2.00330 mm.omn.w

.82....50. .5. 220...

119

.000 00.0.0 05 2.0500000. 0002.000 .0 00.0.50. .05.0 0. 0.005500 .0 00.050.0000

000.0000 0.05.00 .0 000. 05 5 000.000 00.0000 .0050 .05 00.00.05 10 .0 m. .0 5.00.0000 < 0.02.000 2.00.050 05 5 0000.00.
0.02, 0.0000. 00.0.00. .0. 00:00:00 .05 00.030 00052.00 0.0000. 00.0.00. .0. .00000 0. 00:0. 50.5.0 .000.0.00. .0 00.030 .05
00.00.05 « 0< 200000805. 00.000000 .050 .0. .0 25020.09. 200.. .0. 000000 .0 00000098 050000. 0.00 0000.05 0.0050020
0.0.05.0 .000.0.00. 05 .0. 00000_>m. .00.00.0>0 .00 002. 0.0050020 u I... 000 0.0050020 05 005000 0000030. 00 u z 20.005002:
0505000000 0000030. 002 u > "002.0000 .0000 .0 50.020 .000.0.00. 60000.00 0... .0 000.0. 2.000..0.0>m .000000.
0.0000. 50.0.00 u 0 000 0.0000. 00.0.00. u 5. ”0.0005. 2.000....000 .0 .00050 .05....0005. 0550.. 0.00000 020 2.00 5.2. 0005.000
.0 .00050...oo.. u 000.0 N 2.00 5.2, 0002.000 .0 .2. 20200.0. 0.000 05 00.026. 00.000.000 2.000....000 .0050 0. 000.0. 0.0.. .05
0.00000 005.000 .0 .00050 000.0>0 n .0000: 00.0 002.000 000.... 2020.. .00050... 0.00>0 00000.00000 000 20.00 0.0 0005.000
0000000 0005.000 .0 0.000. 0.0.0500 0 00.000050. 2:00.02. 000.0000. .000. 05 05.00.00 E020... 00 .0 500.05 .0050 00 00000
00.0. 2.50... 20020000 .0 000.0 5 0.00000 .0 .00050 0000....0000000 2500000000 0.02. 000.000 00.5 .000. .0 00.03 5 0.000

.0 500.002.000.508 .0 .00050. n .002000 .00 .000 .00 0002.000 20020000 2000000000 0.02. 000.0. 0000.00 000 .000. .0

000 2.0.9.0000 0000.00 000 .000. .0 .5000 5000.00 000 .000. .0 00.02. 5 0.000 .0 50200022050505 .00....0 00 000.000. 0005.000
.0 .00050. n 000.0 .00 .000 .00 0002.000 2200000055 0>.000.000 =0 .0 500.002.055.000 .0 50800.30... u 20002-000
0.02. «05. 0«00 0>.000.000 .0 .x. ”000.00 500 u m .0 00.050. u ". .0055 u 5. ”0.500 000 0050020.. 500 u m .0 2.0500 05.08
2.0550200. 200. 0.500 n < 2.0.0.00 05.0.0.0. 200. 00:005.. u .. 000.00 000 0000 2.0.9.0000 000 .0000 000.0 200.0 0.2. u >2

.0 20000.0 05000.00... 00050.55 0500.05. 020009.800 H Um .05000 n 0 .002... 0000.50 00.000 0.0.. 50.0.50 ._ .0020 200.0
5 0.00500 :0 .0 .00500 .08. u 0N.0 000.0 .000000 2.000....000 0555 2.000060 0.000 .05 0.0500 .0 .00050 n .00500 «00.000

 

 

 

 

2. o a 2: 2. u. a .0 2. 0 a «0 .2026
So... u 026.. 00.3 n .0260
.Sz. 08... 0N.» 0000 000... 00... 00.0. .0000500-coz .9 00.05....
«0... u 0 .2... u N .0... u 0 .00.. u N 0.03.0 205.22.00.02
.0. "2.2.008. .02. 2.00000 .3 "2. 2.3000 030500.52 .0
.ouu2..2o«8. .3...2..220~0 .omu2..2~«00 00.0800...
02.0000 «00.... 0.0.0.00 .000.0.00m. 00000.00 0.x
".0. 00000.>0 .22. 00.0000 .00 00.030 .0 .2. 22800.2. 00.0000 u 2.

€02.28. .5. 0.00.2

120

2666. 66.66.062.166.

2.666 .._6 .6 2.66.0. .66. 2666. .6660. .00 26660 .6666 .._6 .6 66.666016... 26660 066266.. .66. 2666. .6260 666 .6260. .66.
26.66 6666566666 .__66.6.60 666 666666626660 .._6 .6 666.666... .66. .666. .66.. .06. .666. 666.660. :6. 26.66 666066666
0666.... 666 666.2 6.6622266. .66... .66. 26660 .._6 .6 6666.60 .6666 .._6 .6 6606.2 066. .._6 .6 .66660 0666 ...6 .6 66660. .66.
26666 .._6 .6 26...... 66. .66. 26.6. .560 K66. _._6 .6 666266. .666. .._6 .6 .6660. .06. 2.66. .._6 .6 66666.60. .66. .6660 .65..
666 6.6660 0660 .6665... .66. 26.6. 66.6660. .66. .666. .._6 .6 .666N .6666 ...6 .6 .6662. .656 .._6 .6 6.660 .6666 .._6 .6 666m.
.66. .0666 .._6 .6 6.6_6> .6666 .__6666.60. .66. .666. .6666: :6. .666. 6660.0. .66. 2.666 .._6 .6 _666> .666. .566 .066.
.2666. .60. .6666 .._6 .6 66.660. .66. 2666.6. .._6 .6 6.66.60. :6. 20.6. 6.2.660 666 6.2.6.60. .60. 2.66. 662.06.... 666 656.66
.266. 6266662 666 656.66. .66. 2.66. 66.66662 666 656.66. .66. .0666 .662 666 66660 .6666 .6666... .66. .666. 666.260
”.66. 00.6666: 666 6.66:. .06. 2606. 66.66.60 .666. 066.60. :6. 26666 .._6 .6 6.6652 .666. 6.66660 666 66.666626. .60.
2.66. .._6 .6 .660. .6: .666. .56.... 666 .666. .6: .56. 06666.2. E. 2666. 0.6.6661 0.6. .662. 66. .6: .666. .65....
666 _662. 66 .066. .650... 666 .662. 66. .6: 2.66. .._6 .6 6.66660 .666. 6.66660. .3. 2666. .000 .666. 0.50 066. 5.0. .6:
26660 .0 .6 .66660 .6666 6.6.6560 666 .66660. .0: 26666 .._6 .6 6.66.0 .066. .65.... 666 _662. 66. c 0 26.66 6666.02.66
66.66016: .866. 666660 ”6.6. 06.66.6016. 26.6. 066666.601... .666. 0.6.62. .6666 ...6 .6 5.0 6.6. 066.260

.666. .662 .666. .662 .666. 06.08.6010 2.66. .6 .6 .66... .6. .56. 060.666. .6. 2.66. 6662.. .6. 26660 .._6 .6 6.666).;
.6660 .._6 .6 6696.0. .6. 26660 .666; .666. .._6 .6 6.6660 .666. 6.666.. 666 66.602 0666 .._6 .6 .66.... .6666 .._6 .6 .6622 ”.66.
.20 666 2.6.6660 ”.66. .066: 66> 666 .662. 66 .666. 5.6.63 666 .663 66 .666. .662. 66 .6660 .666E666<.666660 666 666660.
20. 2000. ..00.0.0.0 .0000 .._0 .0 05.6.. ”00005... 2... "000020.00. .0.00_.0>0:0 0.02. 0.00 030.0058 .00 .0502. 500. 5 0050000000
00.0.50. co_mmm.mmm 30:03:00“. .00.... 0. 000.0. .50.— Om_m 6mg. 600.0..ch Ucm 950.30.— .m..>... 0.32.300. 000090 x00. Dmuconu0mn—x

.00... 00.000. 00000 000.000. 2.00 000.000.0042. 00.080. 5000 00 00.02.00 0.02. 0.00 .050 _.0 000.002. 200.0 05 00000 0.0.0.005
002. 0.00.0. 0:0. 0.05.000 000.0005 ..050. 2.0.0550 0:..00 00.00. 20 050000 2.000....000 :0 00000 0.00: 005.50 02.. 0002000
00500.0 00.0000 0:..00 0:0..:0>.0.5 00 2.00 00000 0.000 005.506. .00.0...0 0. 05000.00 0.000.200. 000.00 0.0...000 0:..00
000..:0>.0.5 0. 0055.. 0.00. .0.00.00.0 000.0 .050 02.. 0002000 002. 2:0 .0 0:0..00.0.5 00.0000 0:..00 0:0..:0>.0.5 0>.000.000
000 5.00000 .0 0000000000 :0 :0 00000 0.000 0000.0 200.0 0.05.05 50.. 00.0.0000 0.006 .00 .00 00.0000..0.0. v.00. 05000000.

.0. 00.5000 0. .0 .000. .050. 00...:000000 05.0.0 .0 00.000. 0.0.0055 0. 000000 00000.05 .0 000.050. 06.0000 .0050 0.00 0.0000
.00. .000000 0.000 0.0000 0.0.. 0.00 0:...0..00 00.00. ..000 0. 0000000. 5 00500.0 5 0:0..:0>.0.5 :0 2.00 00000 0.00.. 0.00.0.
.0 0.00000 00....000 .0..:0.00 00 0.00..0>0 2.0000:0..0050 0.02. 2.0.000 00.00.00.000 05.0 0.00.000. =0 .0: 0000000 0.000 .00.000

.00 0 :0 00.0. 0.005.00 0. 0.50 000.0000 000:. 20 000.20 0.02. 00.0. 2.5016 "00.030 0000.0 0000..0> .0000 00.00 0.00000

86666660. .5. 6.66.

121

non-primates of both sexes (58.8 % of 17 species, Chi-square test: 12 = 5.24, d.f.

= 1, P = 0.022). However, this pattern may be attributed to the literature bias,
favoring studies on intragroup coalitions in primates (72 references available).
Only 26 references on non-primates were found. Nonetheless, adult animals
joined forces during intragroup conﬂicts in virtually all species reviewed, except
for some species of pair-bonded birds for which limited data are currently
available. Finally, we failed to detect any differences in the evolutionary forces
favoring coalition formation between primates and non-primates. In fact, although
evidence of reciprocal trading of coalitionary support among non-kin was
generally rare, most species reviewed here gained both direct and indirect
beneﬁts from participating in coalitionary interventions.

Size, composition, and context of intragroup coalitions among spotted h yenas
Qgglition fonnatiorgmorﬂyenals of all age-sex classes

On average, from 1996 to 2000, our study clan contained 75 i 1 hyenas (Range:
58 to 95), including 23 i 1 adult females, for each month of the study (Range: 21
to 25, Table 4.2). We observed 11,194 aggressive interactions and 6,944
greetings among all clan members. Coalitions formed during 14 % of aggressive
interactions (N = 1,589 coalitions). Whereas afﬁliative greetings occurred at a
rate of 1.7 greetings per hour, intragroup coalitions formed only once every 2.7
hours of observation (0.38 coalitions/hour). Based on these 1,589 coalitions, we
calculated the frequency, percentage, hourly rate and mean size of

coalitions for each clan member (N = 185 subjects, including 37 adult females).

122

6-N ”omcm: oAmmdé ”omen: AWN ”coco: wand-.. ”coca: 063 u 2. 520

 

 

 

Er 3.0 H QN mood H 36 mod H QN Nto H Ed . mace? mu u N 05cm
3.. a $0 2. .8 u 2. 8.3 .38
£6. o 2.0 H 3 48.0 H No.0 8 S... H 3 So H mod 8m 3.. SN. t u H “$5.55.
$0 2. § 8. 3 u 2. Home
§ 8 .2 Ed H 3 mod H mod 8? 3o H mm Nod H 2.0 mom as mm. H u H :28 .smz
as 9.. cs 9. an n 2. moss
3.. 3: mm mod H 3 wood H 36 Km 86 H 3 mod H a; 5m 3.. S: S u H 2.52...
3.. we. 3.. me. E. u 2. _ 8.38..
3.. 9w 3 mm nod H wN mood H vod mwm mod H mN No.0 H 9.0 in g mm: .1. u N 2.52...
as 8. as mm. an n 2. 8.38..
as E. 5 mod H 3 35 H mod 84 8.0 H 3 86 H m; vmw es mom. mm u H =33.
«Eco Soﬁa 6933. «Eon @833 u 3
aAmcozcoZQE «8.35 .3 60.33 .8 9.935% .3 mum .3 93 96868 330 Ho as. $.20
>tma-2m =m .6 as. mum 5.580 99 25o: =m Ho ox... 5538 2a. 259. =m we as. 5.8:. xom
s. 5.29.25 wchmoE mecmos. #coEﬁoo mecmos. mecmos. #coEﬁoo Bozo? N -om<
umEm: oﬁEoWQmEﬁ 0:503 2mth :33 En n...=m_o o...“ ho .mnEoE 28 «:30 33m
Ho 5 9035295 2959. 8696 9.095 b22580 2352 869:8 mxomtm EmcoEEoo 3350550

968. 6233. _mcozmz Ems. 6me 9: c. 5.0 09m. .29.?
m 55.3 momma? xomomm 223% ow 96:23 mace? 3..on 9.25 5.38.8 coaﬁoo Ho accormn. .ué snub

123

.000
8 000....000 0000 .0. 0.0.0000 00.5 .. 00.02, 0. 00080000 0.0..; .0500 03.0 + 0:00:08 :0. 0E..0.> 0.080. 0.08.0 0”
.000 .00 :08 0:00:08 00.0. 00.000 03.0 + 000....000 .00. S 00.0 00. .0 .00E0E >00 0.0000 8.000 .00 008 000....000
00.0 0020000 02, .000.0>0.0O.. 00000.0 x00|000 .00.0...0 00. o. 00.000.00 0000>0 .0 0.00 00. 000. 000. 00.5 0020000
0008.0 .88 00. $00.0 0.0. 00000 00000.0 000 0.00.00. 8 00000.00 0.000.200. _0.0>00 0000009. .0000280. 000.0
x00l000 .00>0_0. 00.800.000.00 000>0 0.00.0 0 200 00.0.5 0. 000000.80. 0000.0. 08.0.... 0.02, 000000280. 2.00-020.
0000000 00. 0. E00 0000.08.00. .00. 000098 000800800 200800 :0 .o 0.00 00. 20.0.00 .8 08 000. 0...0.> 0.000
00. .00.0. 8 000.8 00.0.0. 00000.0 x00l000 .00.0...0 E0... 0.000.200. 0.0...00. 00>.o>0. 000....000 .000. 00000000 00.0.00.
..000 02,. 0003.00 08000.0 00000 0000280. .000.>.00. 000. 0 00.03 0. 000000280. 2.00-0..0. .0 008000.00 000
500000.“...v .E..0.> 0.00.0. ..000 00 00.030. 0808.0 0.03 .00. 000.0 x00..000 000.0000 00. o. 00.000.00 .00000 000 .000.
.0 0000.00. .00. 000....000 .0 02.000.00.000o 00.0 00. .0 .0080... >00 00.0.50. 0808.0 0.0.5 .00. 000.0 x00-000 00....0000
00. 0. 00.000.00 .00000 000 .000. .0 0000.00. .00. 0:00:08 .0 000008080000 0.00 0800080 00~.0 0.00.00 0.000

00. 00 00000 000.0 x00-000 0000 8 00.000.00 0.000.200. .0. 080.0200 0.0.... 00N.0 00....000 000 00.0. 2.000 mm H 000.20

€02.08. .0... 0.00»

124

Unless stated otherwise, sample sizes were the same as the number of subjects
belonging to each age-sex category represented in Table 4.2.
Overall, the rates at which hyenas formed coalitions varied among age-

sex classes (Kruskal-Wallis test: H4203 = 36.50, P = 0.023, Table 4.2). Adult

females (N = 37) directed coalitionary attacks towards clan members of any age-
sex class at higher hourly rates than did immigrant males (N = 38, Mann-Whitney
U tests: Z = -2.84, P = 0.022), but at rates similar to those of juvenile males (N =
52) or females (N = 57), and adult natal males (N = 24, Z = 0.07, 0.62, and -1.30,
respectively; P 3 0.582 for all cases). Modal coalition size (71 % of coalitions)
was two hyenas, and coalition size did not vary signiﬁcantly among age-sex
classes (Kruskal-Wallis test: H4179 = 7.65, P = 0.11, Table 4.2).
_C_oa_lﬂon formation a_mong_adu|t fem_a_le hvengs

Adult females (N = 37) were the victims of 51 "/0 of all coalitionary attacks;
this was disproportionately high based on their representation in the clan (31 %
of clan). In fact, adult females received a greater proportion of attacks (27 i 2 %,
N = 37 victims) in the form of coalitionary aggression than did other members of
the clan (10 i 1 % of attacks, N = 148 victims, Mann-Whitney U-test: Z = 6.941,
P < 0.00001). Moreover, the hourly rates at which hyenas directed coalitionary
attacks towards adult female victims varied among age-sex classes of
aggressors (Kruskal-Wallis test: H4203 = 36.5, P < 0.0001 , Table 4.2). Natal
animals were signiﬁcantly more likely to direct coalitionary attacks towards adult
females than were immigrant males (Mann-Whitney U-tests: U 5 -2.769 and P <

0.0006 for all comparisons, see Table 4.2 for sample sizes). On average, adult

125

females directed coalitionary attacks towards other females more often than did
individuals belonging to any other age-sex category, but we detected no sex
differences in the rates at which juveniles or natal adults directed coalitionary
attacks towards adult females (Mann-Whitney U-tests: U = 1468.5 and 352.0, P =
0.934 and 0.528, respectively).

Overall, adult females participated in 480 coalitions attacking other adult
females (Table 4.2). Of these attacks, 57 % involved juveniles joining forces with
adult females and 29 % involved two or more adult females joining forces to
attack a third adult female. Adult females never joined forces with immigrant
males to attack adult females. Most (86 %) all-female coalitions contained two
partners, 12 % contained three partners, and only 2 % contained more than three
partners. Fifty four percent of third-party interventions in disputes between two
adult females were by lone adult females. Thus, adult females intervened more
often than expected based on their representation in the clan (31 °/o of clan).
Adult females intervened in ﬁghts between adult females about three times more
often than did juveniles of both sexes and almost six times more often than did
adult natal males (Table 4.2).

Social and ecological contexts of interventions by adult female h yenas

Lone adult females intervened in 81 of 1171 (7 %) ongoing dyadic ﬁghts between
other adult females, and a context could be assigned to 63 of these interventions.
Overall, these interventions were more frequent during unprovoked ﬁghts (56 %)
than during ﬁghts in the contexts of food (22 %), maternal intervention (16 %),

pesky (6 %), or scapegoating (0 %). After correcting for opportunities to join, we

126

found that the tendency for adult females (N = 24) to intervene in ongoing ﬁghts
varied signiﬁcantly with the context of the original ﬁght (F riedman’s ANOVA: F424

= 32.7, P < 0.00001, Figure 4.1). Speciﬁcally, females were more likely to
intervene in unprovoked dyadic ﬁghts than in ﬁghts involving scapegoating or
food (Wilcoxon signed-ranks tests: 2 _>_ 3.24 and P 5 0.005 in both cases).
Females also intervened to a greater extent over food than during scapegoating
(Z = 3.059 and P = 0.016).

In those ﬁghts (N = 744) for which detailed information was available on all
behaviors displayed by the original attacker, we investigated the possibility that
attackers bristled their tails to solicit support from bystanders. Surprisingly,
females were signiﬁcantly less likely to donate support in ﬁghts during which
original attackers bristled their tails (2.3 i 1.2 % of opportunities) than when
attacking females failed to do so (2.6 i 0.3 % of opportunities; Wilcoxon Signed
Ranks Test: N = 25 donors, 2 = 2.224, P = 0.026). It remains possible that
females solicit help using subtle forms of solicitation not obvious to human
observers.

Modeling factors to explain interventions by adult female h yenas

We used model selection to assess the effects of kinship, and variables
indicating direct beneﬁts, on the tendency for females to intervene in disputes
between adult females. We only included those pairs of adult females for which
we could unambiguously assign a coefﬁcient of relatedness in our statistical
models. Multiple females were often simultaneously available to intervene at

each ﬁght; each focal female (N = 31) represented in the statistical models, on

127

 

0.06

 

 

 

C
0.05 -
>.
.0
t:
g 0 04 -
o .
a.
o.
o
g. 0.03 -
m
C
.9
§ 0.02-
L-
8
E
0.01 .
0.00— S M al
cape- atern
goating Food Intervention Pesky Unprovoked
Context

Figure 4.1. Mean :l; SE interventions per opportunity in which the same adult
females (N = 24) intervened in ﬁghts between two other adult females based on
the context of the original ﬁght. Different letters above bars indicate statistically
signiﬁcant differences after correcting for multiple testing.

128

average, provided support or directed coalitionary attacks, respectively, in 2.7 i
0.7 % and 1.9 i 0.5 % of the cases in which she was available to do so. The
same adult females (N = 31) were included in both models, and the extent to
which females intervened in ﬁghts was statistically equivalent for both data sets
(Wilcoxon Signed-Ranks Test: N = 31 females, 2 = 1.153, P = 0.249, Effect size:
0.257, Power = 0.403). Final models identiﬁed the subset of candidate predictor
variables that most strongly structure patterns of coalitionary support and attacks
(Tables 4.3 and 4.4). Neither model retained strongly intercorrelated variables (r?
_<_ 0.10).

Kin selection hypothesis

Negotistic coalitionary support among adult females

As predicted by the kin selection hypothesis, interventions by adult females were

 

highly nepotistic (Table 4.3, Figure 4.2). Females biased support towards kin,

and did so regardless of the intensity of the original ﬁght (Kinship * Intensity

interaction: -0.019 :1: 0.150, Wald-Statistic = 0.015, P = 0.902). This pattern is
best illustrated by placing a subset of female dyads, all of known kinship and all
present together as adults in the clan, into a weighted cooperation network
(Figure 4.3). Network density increased with the level of relatednessamong
kinship groups. The density of the close kin network (realized arcs/total possible
arcs = 0.38) was over three times that of distant kin (0.12) and over seven times
that of non-kin (0.05, Figure 4.3). In addition to providing a concrete example of

inter-individual dynamics and coalition formation within one "cohort"

129

Table 4.3. Independent variables predicting patterns of coalitionary support
donated by adult females on behalf of adult female beneﬁciaries during
interventions in dyadic ﬁghts.

 

Coefﬁcients: Estimate t SE Wald-Statistic P-value
(Intercept) -3.317 1: 0.446 55.370 0.00001
Higher-ranking (HR) 0248 i 0.043 32.769 0.00001
bystander number

Kinship 3.472 i 1.043 11.081 0.00087
Beneﬁciary subordinate 0.538 i 0.389 1.917 - 0.16620
to potential donor

Food 0069 i 0.081 0.738 0.39018
HR bystander number 0.435 :I: 0.126 12.009 0.00053
* Kinship

Beneﬁciary subordinate 0.392 :I: 0.099 15.559 0.00008
* Food

 

Comparison of the candidate models ruled out the following additional factors as
predictors of whether or not females provided coalitionary support: absolute
social rank of the potential donor (Estimate :l: S.E.: 0.024 0 0.041, Wald-Statistic
= 0.337, P = 0.562), intensity of the original ﬁght (-0.053 1: 0.057, Wald-Statistic =
0.844, P = 0.358) and prey abundance (0042 i 0.071, Wald-Statistic = 0.346, P
= 0.556). The overall ﬁt of the model yielded a Wald-type statistic of 375.8,
whose distribution is approximately Chi-squared. Model results are based on
adult females intervening in 45 out of 1477 ongoing ﬁghts (N = 241 dyads). On
average, each adult female (N = 31) intervened to donate support in 2.7 0 0.7 % -
of opportunities that she was available to do so.

130

Table 4.4. Independent variables predicting patterns of coalitionary attacks
directed by adult females towards adult female victims during interventions in
dyadic ﬁghts.

 

Coefﬁcients: Estimate 0 SE Wald-Statistic P-value
(Intercept) -5.274 0 1.040 25.734 0.00001
Food -2.219 i 0.975 5.180 0.02284
Victim subordinate 2.263 :t 1.048 4.665 0.03077
to potential attacker

Intensity of -1.203 0 0.557 4.664 0.03080
original ﬁght

 

Model selection excluded the following variables as useful predictors of whether
or not females initiated coalitionary attacks: absolute social rank of the potential
attacker (0108 i 0.068, Wald-Statistic = 2.545, P = 0.111), prey abundance (-

. 0.229 0 0.479, Wald-Statistic = 0.229, P = 0.632), the number of higher-ranking
bystanders present (0100 i 0.223, Wald-Statistic = 0.202, P = 0.653), the
absolute rank distance between the potential attackers and victims (-0.020 t
0.053, Wald-Statistic = 0.142, P = 0.706), and the kinship between potential
attackers and victims (0.074 0 1.155, Wald-Statistic = 0.004, P = 0.949). The
overall ﬁt of the model yielded a Wald-type statistic of 257.4, whose distribution is
approximately Chi-squared. Model results are based on adult females intervening
in 22 out of 1477 ongoing ﬁghts (N = 31 females). On average, each adult female
(N = 31 ) intervened to direct coalitionary attacks towards other adult females in
1.9 :L- 0.5 % of opportunities that she was available to do so.

131

 

 

0.16

0.14 ~ :1 "°"4‘i"
- Distmt kin

133
”-12 ' I Close kin

Non-food related conﬂict

 

 

 

 

   

 

 

 

 

 

.E‘
c
3
1:
o
0.
o.
o
I-
0
0.
E 0.10 -
g
'5
5
0 0-08 ' Food-related conflict
‘0:
a
3 0.05 ‘
w
I? 34 114
.E 0.04 -
.2
g 61
_g, 0.02- -- 137
IL
0.00 I

Subordinate Dominant Subordinate Dominant

Rank of potential beneﬁciary relative to donor

 

 

 

Figure 4.2. Proportion of fights in which adult females supported beneﬁciaries
subordinate or dominant to themselves, out of all opportunities to intervene in the
presence or absence of food. Bar color indicates kinship of each dyad as non-kin
(white), distant kin (gray), and close kin (black). Sample sizes over each bar
indicate the number of opportunities potential donors had to suppbrt potential
beneﬁciaries. Error bars represent 0 1 standard deviation for binomial trials.

132

A) Non-kin:
Density of support = 0.05

__,...klp 9C7 ............. .wr

.ilg

   

...Dgo|

6jab

     
 

. . .
=\._\

0

.bailﬁj‘i‘.

    
 

 

K
' 1
‘
.._. \
.' ‘

l
n
.
n

‘, n
.

   
   
 

Figure 4.3. Cooperation networks among females present concurrently as adults
from 1997 to 1998 that were A) non-kin, B) distant kin, and C) close kin. Each
circle (node) represents an adult female. Node size is proportional to the social
rank held by that female (e.g. alpha female is largest). Connected dyads within
each network represent dyads belonging to a particular kinship group based on
maternal and paternal kinship. Whereas dyads connected by solid black arcs
formed coalitions, dyads connected by dashed gray lines failed to support one
another in ﬁghts despite the fact that both dyad members had opportunities to do
so. Solid, black arrows (arcs) originate at donors of support, and arrowheads
point towards beneficiaries of support. Double-headed arcs represent reciprocal
support within dyads, whereas single-headed arcs represent unilateral support.
Densities reﬂect the number of realized arcs (number of solid black lines) divided
by the total number of possible connections (number of dashed gray lines and
solid black lines for each kin group). Line thickness (weighted edges) indicates
the proportion of times adult females intervened out of all opportunities to do so

133

Figure 4.3. (Continued)

B) Distant kin:
Density of support = 0.12

 

 

C) Close kin:
Density of support = 0.38

Olg

 

134

of adult females, these diagrams show how female dyads are embedded in the
larger network of female social relationships. They also highlight our particularly
striking ﬁnding that, despite the relative paucity of close kin as potential allies,
close kin were clearly the likeliest allies.

Overall, the extent to which adult females intervened on behalf of their

sisters varied with their genetic relationship to those sisters (KruskaIl-Wallis

ANOVA: H234 = 7.025, P = 0.030). First, adult females supported their full sisters

(10.1 :I: 3.6 %, N = 8 donors) three or four times more often than they supported
their maternal half sisters (2.9 i 1.3 %, N = 9 donors) or paternal half sisters (1.5
i 1.0 %, N = 17 donors), respectively. This bias in support towards full sisters
remained statistically signiﬁcant for paternal half sisters after correcting for
multiple testing (Mann-Whitney U-test: Z = 2.462, P = 0.014, Effect size: 1.093,
Power: 0.859). However, the difference between full siblings and maternal half
sisters did not remain statistically signiﬁcant after the correction (Mann-Whitney
U-test: Z = -1.000, P = 0.317, Effect size: 0.925, Power: 0.608). Although females
were twice as likely to donate support to maternal than paternal half sisters, this
difference failed to reach statistical signiﬁcance (Mann-Whitney U—test: Z = 1.812,
P = 0.070), perhaps because of our low statistical power (Effect Size: 0.080,
Power: 0.056). On average, mothers supported their adult daughters roughly
twice as often as those same adult daughters supported their mothers in ﬁghts,
but this difference was not statistically signiﬁcant (9.6 i 4.2 % and 4.7 i 2.6 % of
opportunities, respectively; Wilcoxon signed-ranks test: T= 19, P = 0.678, N = 13

matched pairs, Effect size: 0.291, Power: 0.231). Interestingly, however, females

135

supported their full sisters (N = 8 donors, 10.1 i 3.6 % of opportunities) more
often than they supported their mothers or daughters (N = 18 donors; 6.3 i 2.4 %
of opportunities; Mann-Whitney U—test: Z = 2.060, P = 0.039).

Audience effects modulated by kinship

Donating support was costly in the presence of higher-ranking bystanders.
Donors of support were attacked by an adult female bystander immediately after
14 d: 6 % of their interventions occurring in the presence of at least one higher-
ranking adult female bystander (N = 30 donors), whereas none of the 18 donors
were subsequently attacked when higher-ranking bystanders were absent
(Mann-Whitney U-test: Z = 2.186, P = 0.029). The effects of higher-ranking
bystanders were modulated by kinship, as predicted by Hamilton's rule (Table
4.3). That is, adult females were signiﬁcantly less likely to intervene on behalf of
distant kin or non-kin victims of aggression as the number of higher-ranking
bystanders increased (Spearman rank correlations: R3 = -0.719 and -0.840, P <
0.004 and 0.005, N = 14 and 9 audience sizes, respectively). Interestingly,
however, females intervened on behalf of close kin regardless 0f the increased
risk of doing so near large numbers of dominant bystanders (R3 = -0.431, P =
0.213, N = 9 audience sizes). Because the number of higher-ranking bystanders
was correlated with the absolute number of adult females present (r2 = 0.25), we
investigated the prospect that the latter variable explained the apparent
bystander effects. However, our data allowed us to rule out this possibility
because the ﬁt of our best statistical model of coalitionary support was not

improved by either the addition of the absolute number of adult females present

136

(-0.019 :t 0.031, Wald-Statistic = 0.374, P = 0.540) or the interaction between this
number and kinship (-0.044 is 0.074, Wald-Statistic = 0.352, P = 0.553).

Kinship fails to protect females from coalitionau attacks

Although adult females strongly biased their supportive interventions towards kin,
kinship failed to protect potential victims from becoming targets of coalitionary
attacks (Estimate: 0.074 :t 1.155, Wald-Statistic = 0.004, P < 0.949). Our best
model revealed that coalitionary attacks were signiﬁcantly more likely when the
intensity of the ﬁght was low than when it was high (Table 4.4), independent of
kinship (Intensity * Kinship interaction: 1.693 0 1.970, Wald-Statistic = 0.738, P =
0.390). For the subset of dyadic ﬁghts in which triadic genetic relationships were
known, adult females (N = 25) were slightly more likely to intervene in ﬁghts
when their kinship to each of the contestants differed (3 i 1 %) than when
females were equally related to both contestants (1 i 1 %), but this difference
was not statistically signiﬁcant (Wilcoxon signed-ranks test: T: 5, P = 0.500,
Effect size: 0.143, Power: 0.136). When intervening in ﬁghts in which they were
more closely related to one of the two contestants, female aggressors (N = 8)
were signiﬁcantly more likely to attack the more distantly related (94 i 6 %) than
the more closely related of the contestants (6 i 6 %, Wilcoxon signed-ranks test:
T = 0, P = 0.018). Although kinship generally failed to protect females from
attacks, this ﬁnal result is consistent with predictions of kin selection theory

because females biased attacks away from the more closely related contestant.

137

Reciprocal altruism hypothesis

No evidence of reciprocity or interchange trading among non-kin

To test for reciprocal trading, we examined the correlation between support given
and services received within dyads of non-kin. After controlling for effects of
kinship, no signiﬁcant relationship emerged between support donated and
received by adult females (Partial rowwise matrix correlations: TauKr = -0.155, P
= 0.860). We also found no evidence of interchange trading. Non-kin failed to
trade support in exchange for help during hunting (TauKr = 0.012, P = 0.420),
opportunities to greet (TauKr = -0.079, P = 0.861), or social tolerance (TauKr =
0.327, P = 1.000). Our data also fail to indicate that natural selection favors
agonistic aiding on the basis of indirect reciprocity; females provided the least
support when the number of dominant bystanders in the audience was high
(Table 4.3).

No evidence of stable alliances among non-kin

Unrelated adult females failed to form stable coalitionary alliances. This could be
most clearly seen in the support networks, where double-headed arcs,
representing reciprocal support, were limited to kin (Figure 4.38 and 4.3C).
Moreover, the support network among non—kin was extremely sparse, and non-
kin capitalized upon only 5 % of the possible network connections (Figure 4.3A).
When we focused on all 106 dyads in which kinship category could be assigned
and in which both members had opportunities to provide support, dyads
engaging in reciprocal support (6 %) were far less common than dyads providing

unilateral support (25 %) or no support (69 %) on behalf of either member of the

138

 

dyad (Figure 4.4). Reciprocal support was only observed between genetically
related adult females (Figure 4.3B and 4.3C) such that dyads of kin participated
in reciprocal support more often than did non-kin dyads. Repeated interventions
by the same donor on behalf of the same beneﬁciary were rare. No adult female
was ever observed supporting the same unrelated adult female in more than one
ﬁght (N = 31 females involved in 115 dyads). Even among kin, multiple instances
of interventions by the same donor on behalf of the same beneﬁciary only
occurred in 2.2 % of 90 distant kin dyads and 11.1 % of 36 close kin dyads.
Direct beneﬁts hypothesis

Coalitionam interventions fail to increase immediate access to food

Three lines of evidence indicate that intervening females do not accrue direct
beneﬁts in the form of immediate, enhanced intake of food. First, females were
signiﬁcantly less to likely to intervene during ﬁghts involving food than in ﬁghts
unrelated to food (Table 4.3 and 4.4, Figure 4.2 and 4.5). Whereas females were
signiﬁcantly more likely to support subordinates (6.2 i 1.7 %) than dominants
(1.7 :t 0.5 %) in the non-food context (Wilcoxon signed—ranks test: 2 = 2.20, P =
0.028, N = 22 females), females were unlikely to provide support during ﬁghts
over food, regardless of their rank relative to that of potential beneﬁciaries (Z =
1.05, P = 0.294, N = 21 females, Effect size: 0.169, Power: 0.151). Second,
coalitionary aggression was no more effective in displacing competitors from kills
(52.6 i 11.8 %) than was dyadic aggression (51.6 i 9.2 %) by the same
aggressor (Z = 44.50, P = 0.944, N = 19 females, Effect size: 0.059, Power:

0.606). Finally, local prey abundance failed to predict female interventions in

139

 

 

1 00
- Reciprocal support

43
9° ' _ - Unilaterd support
80 _ [:1 No support
70 _ 26
———- 10
60 _

% of dyads in caegory

 

 

  

 

 

 

 

 

 

Non-kin Distant kin Close kin

 

Kinship

 

 

Figure 4.4. Variation among kinship categories regarding the proportion of dyads
in which both members supported one another (reciprocal support: black), only
one member supported the other member (unilateral support: gray), or neither
member provided support to the other (no support: white). Sample sizes over
bars indicate the number of dyads in which both members had opportunities to
provide reciprocal support (N = 106 dyads).

140

 

 

 

 

 

 

 

 

 

0.10
- Low-intensity
D High-intensity
2‘ 0.08 - Non-food related conflict »
.5
o
E
0 0.06 - 249
Is
0.
Ill
.2
35
g 0.04 -
C ' t .
% Food-related conﬂrc 361
'0
- ' 86
138 112
225 [
0.00 - I - I -
Subordinate Dominant Subordinate Dominant
Rank of potential victim relative to attacker

 

 

Figure 4.5. Proportion of ﬁghts in which adult females attacked victims
subordinate or dominant to themselves, given the opportunity to intervene in the
presence or absence of food. Bar colors indicate whether females intervened in
ongoing dyadic ﬁghts of low (black) or high (white) intensity. Dyads belonging to
all kinship categories are pooled because kinship failed to protect victims from
attacks. Sample sizes over bars indicate the number of opportunities females
had to attack targets already under attack by another adult female. Error bars
represent 1 1 standard deviation for binomial trials.

141

ﬁghts (Tables 4.3 and Table 4.4).
_Q_o_alitionarv interventions reinforce the status guo

Our data are generally consistent with the idea that females beneﬁt
directly from reinforcing the status quo. A univariate approach revealed that high-
ranking females, with comparable opportunities to intervene, on average, did so

signiﬁcantly more often than did low-ranking females (Spearman rank
correlations: R3 = -0.733, P < 0.0001, N = 25 rank positions, Figure 4.6).

Nonetheless, relative ranks within dyads (e.g. beneﬁciary subordinate to potential
donor in Table 4.3 and victim subordinate to potential attacker in Table 4.4) were
better predictors of whether adult females intervened than were the absolute
ranks of potential donors (Table 4.3) or attackers (Table 4.4). Females directed
the majority of coalitionary attacks towards subordinate victims (Table 4.4, Figure
4.5 and 4.7), and in most ﬁghts (93.9 i 2.5 %) supported the higher-ranking of
the two contestants (Vlﬁlcoxon signed-ranks test: Z = 5.012, P < 0.00001, N = 36
donors). Cooperation and agonistic networks reﬂect these rank effects. High-
ranking females, indicated by large node sizes (Figure 4.3), provided the most
support in cooperation networks. Because females directed most attacks down
the hierarchy towards victims subordinate to themselves, arrows originated at
nodes larger than those at which they terminated in the agonistic network (Figure
4.7). The density of the agonistic network (0.08 realized arcs/potential arcs) was

low in part because these females only attacked subordinates (Figure 4.7).

142

 

 

0.06

P

o

ow
l

'2

P

o

w
l

0.02 -

Interventions per opportunity

0.01 .

 

 

 

 

Female social rank

 

 

 

Figure 4.6. The effect of absolute social rank on the proportion of ﬁghts in which
females intervened on behalf of other adult females out of all opportunities in
which females holding a particular social rank might have supported potential
beneﬁciaries. By convention, the highest possible rank is one (e.g., the alpha
female). *

143

 

bail

 

 

 

Density of attacks = 0.08

Figure 4.7. Agonistic network: A directed, weighted attack network structured by
the same adult females as in Figure 4.3. Solid, black arrows (arcs) originate at
intervening aggressors, and arrowheads terminate at victims of coalitionary
attack. Line thickness (weighted edges) indicates the proportion of times adult
females targeted the other member of the dyad out of all opportunities to do so.
All dyads are shown on a single network because kinship failed to protect
females from coalitionary attacks.

144

DISCUSSION

Intragroup coalitions in spotted h yenas compared to those in other taxa

Overall, we failed to detect any differences in the salient characteristics of
intragroup coalitions formed by primates and non-primates. Moreover, the
evolutionary forces favoring intragroup coalitions in spotted hyenas were
strikingly similar to those found for many cercopithecines. Indeed, many monkeys
simultaneously direct coalitionary support towards kin while limiting their
individual costs by directing attacks towards subordinates to reinforce the status
quo. For example, this pattern occurs among female baboons (Papio
cynocephalus) as well as male stumptail (Macaca arctiodes), bonnet (M. radiata),
and barbary (M. sylvanus) macaques (see Table 4.1 for references). In fact, our
ﬁndings that nepotism and direct beneﬁts shape intervention decisions among
adult female hyenas parallels ﬁndings from most vertebrate species for which
comparable data are available.

Despite growing intereSt in the topic, our review revealed that the number
of studies on intragroup coalitions in primates and non-primates remains
unbalanced. This suggests the need for additional comparative data. Indeed,
future work might yet expose other aspects of coalition formation, such as the
extent to which adults solicit agonistic aid from bystanders (e.g. de Waal and van
Hooff, 1981; Packer, 1977; Perry, 1998; Slocombe and Zuberbuehler, 2007) or
trade support with non-kin in exchange for other currencies (reviewed by Schino,
2007), that distinguish primate coalitions from those formed by non-primates. We

found no evidence here that adult female hyenas or other non-primates solicit aid

145

or trade support for other commodities.

As in most of the species reviewed here, coalitions were generally rare
among spotted hyenas and most often involved only two aggressors.
Interestingly, adult female spotted hyenas directed coalitionary aggression
towards other females at higher rates than did any other age-sex classes, and
did so at rates four times higher than those of immigrant males. Furthermore,
adult females and immigrant males never joined forces to attack adult females. In
this respect, coalitions formed by spotted hyenas differ from those of many other
mammals. Unlike most mammals, spotted hyenas live in female-dominated
societies (Frank, 1986) in which adult females compete most intensely with one
another (Van Meter, 2009). Our current results demonstrate that adult females
are also each other’s most important allies, and that females with the most allies,
such as those in the alpha matriline, are the most powerful.

Nepotism among adult female spotted h yenas

As in most adult mammals reviewed here (Table 4.1) and earlier (e.g. Silk, 2002;
Widdig, 2007), intragroup coalitions among adult female spotted hyenas were
generally nepotistic. Consistent with the predictions of kin selection theory
(Lehmann et al., 2007b), the density of cooperation networks clearly increased
with the degree of genetic relatedness among adult female hyenas. Moreover,
mothers received support from their adult daughters as often as adult daughters
received support from their mothers. These data support the hypothesis
proposed by Holekamp and Smale (1995) suggesting that, in addition to gaining

indirect beneﬁts by helping their young daughters acquire their dominance status,

146

mothers directly beneﬁt by gaining new adult allies when their daughters mature.

Females preferentially supported full sisters over half sisters, biasing more
support towards maternal than paternal half sisters. Although the latter result was
not statistically signiﬁcant, we suspect that our low statistical power may have
kept us from detecting a biologically meaningful difference. This proved to be the
case, for example, when Silk et al. (2006) discovered that the inability of Smith et
al. (2003) to distinguish between nepotism directed towards maternal and
paternal kin in baboons was an artifact of small sample sizes. More broadly, our
results indicate that nepotism directed by adult females towards both maternal
and paternal kin resemble patterns observed among adult female mountain
gorillas (Gorilla g. ben'ngei, Watts, 1997), adult female rhesus macaques (Widdig
et al., 2006), and juvenile spotted hyenas (Wahaj et al., 2004). Despite its
seemingly important inﬂuence on coalition formation, paternal kinship was rarely
considered in the vast majority of studies reviewed here, underscoring the need
for future studies that evaluate the role of paternal kinship in structuring
cooperative decisions in animals.

As predicted by kin selection theory, adult female hyenas supported
relatives most often when the cost of providing support was low. In particular, the
mere presence of dominants inﬂuenced the tendency for focal females to donate

support, suggesting an audience effect (Zuberbtihler, 2008). Interestingly,
1 whereas adult female hyenas reduced their tendency to donate support to distant
kin and non-kin as the number of higher-ranking bystanders increased, donors

continued to support close kin independent of the increased risk of doing so. Our

147

 

results are consistent with the idea that, unless they are helping close relatives,
females generally avoid counterattacks by refraining from involving themselves in
disputes when dominants are present. In particular, female hyenas monitOred the
composition of their current subgroup, assessed their relatedness to potential
beneﬁciaries, tracked the number of dominant bystanders in the audience, and
modiﬁed their level of cooperation based on this knowledge. These ﬁndings
extend recent experimental evidence from captive spotted hyenas in which pairs
of hyenas were most likely to solve a cooperation task when additional
conspeciﬁcs were present in the audience (Drea and Carter, 2009). Moreover,
our work provides another key example of how the mere proximity of dominants
inﬂuences decision-making in animals. Proximity to dominants also inﬂuences
food calls by brown capuchins (Cebus apella; Pollick et al., 2005), caching by
Western scrub-jays (Aphelocoma califomica; Dally et al., 2006), recruitment
screams by chimpanzees (Pan troglodytes; Slocombe and Zuberbuehler, 2007),
and maternal care by rhesus macaques (M. mulatta; Semple et al., 2009).

In contrast to the predictions of kin selection theory, however, kinship
failed to protect adult female hyenas from coalitionary attacks here, as is also
true for immature hyenas (Wahaj et al., 2004) and for many primates (reviewed
by Widdig, 2007). This ﬁnding might be explained by the direct beneﬁts gained by
intervening females. That is, insofar as coalitions help to maintain the status quo,
it is just as important to an adult female’s reproductive success for her to keep a
lower-ranking sister or daughter in her place as it is for her to maintain her

dominance over unrelated adult females.

148

More broadly, coalitions forming during intragroup conﬂicts among spotted
hyenas were more nepotistic than those forming during intergroup conﬂicts with
alien hyenas during ‘clan wars’ or with lions (Panthera leo, Boydston et al., 2001;
Cooper, 1991; Honer et al., 2005; Kruuk, 1972). In fact, intergroup interactions.
promote clan-level cooperation such that hyenas from multiple matrilines with low
mean relatedness join forces (Van Horn et al., 2004a). On average, coalitions
forming during intragroup interactions here contained only 2.4 i 0.01 hyenas,
whereas those forming during conﬂicts with alien hyenas or lions contain 14 i 1
and 16 :I: 2 hyenas, respectively (Smith et al., 2008). We suggest that unrelated
hyenas form large coalitions during intergroup conﬂicts because coalition size
determines outcomes in these conﬂicts, and thus affects access to resources
shared by all group members.

No evidence of reciprocity among unrelated adult female hyenas

We found no evidence of direct reciprocity or interchange trading among adult
females. Speciﬁcally, we failed to detect a correlation between support given and
services received among unrelated dyads of adult females. Further, females
were generally least likely to donate support when many higher-ranking
bystanders were present. This ﬁnding is inconsistent with reputation-based
models of indirect reciprocity (Nowak and Sigmund, 1998; Nowak and Sigmund,
2005) and models based on coercive tactics such as harassment or punishment
in which dominants force subordinates to cooperate (Clutton-Brock and Parker,
1995). Finally, we found no evidence that female spotted hyenas establish

stable, enduring alliances with non-kin based on repeated acts of unilateral or

149

reciprocal support. Unlike stable male alliances among lions (Packer and Pusey,
1982), baboons (Noe, 1984), Cheetahs (Acinonyx jubatus, Caro and Collins,
1987), dolphins (Tursiops aduncus, Connor et al., 1992, 2001), or wild horses
(Equus caballus, Feh, 1999), coalitions were temporary among adult female
spotted hyenas.

Feeding competition limits coalition formation among adult female h yenas

In contrast to the predictions of the food access hypothesis, females intervened
least often in disputes over food, and coalitionary aggression was no more
effective than dyadic aggression in displacing competitors from carcasses.
lntragroup coalitions often provide improved access to food in species that are
primarily herbivorous such as rock pigeons (Columbia livia, Lefebvre and
Henderson, 1986) and (Cebus capucinus, white-faced capuchin, Vogel et al.
2007). However, extremely high opportunity costs appear to prevent such
immediate beneﬁts from accruing for spotted hyenas. Interventions may be
particularly costly in these contexts because intervening females must allocate
time to cooperating that could otherwise be spent consuming fresh meat at a rate
of up to 1.3 kglminute (Kruuk 1972). Moreover, escalated aggression reduces
per capita energy gain by attracting additional competitors (Mills, 1989; Smith et
al., 2008). Overall, these lines of evidence suggest that adult females likely gain
more energy from directly allocating time to feeding than from forming coalitions
when food is present. ‘

Adult female hyenas gain direct beneﬁts from reinforcing the status quo

Although females forming coalitions at a carcass apparently do not improve their

150

immediate access to food, females that direct coalitionary attacks towards
subordinates away from food should nevertheless beneﬁt directly in feeding
situations. That is, because dominance relationships are stable across contexts
in this species (Frank, 1986), by using coalitions to reinforce the status quo away
from food, females guarantee their priority of access to resources during
subsequent competitive interactions at carcasses. Indeed, coalitions appear
central to the maintenance of rank relations long after adult female spotted
hyenas establish their ranks as juveniles (Holekamp and Smale, 1993; Smale et
al., 1993; Engh et al., 2000). Consistent with our ﬁndings here, a number of other
mammals living in despotic groups similarly use coalitionary attacks to reinforce
the status quo (see Table 4.1 for details and references).

Females occasionally directed coalitionary attacks towards dominant
hyenas in this study. Infrequent challenges of the status quo in the form of
revolutionary coalitions can have profound ﬁtness consequences for spotted
hyenas when they result in permanent rank reversals (Hofer and. East, 2003;
Holekamp et al., 1993; Mills, 1990). Coalitions from low-ranking matrilines are
known to have overthrown higher-ranking matrilines in three different study
populations of spotted hyenas (Serengeti: Hofer and East, 2003; Mara:
Holekamp et al., 1993; Kalahari: Mills, 1990). Thus, as in many primates
(reviewed in Kummer, 1967; Silk, 2002; Silk, 2007a; Silk, 2007b), even rare
challenges of the status quo can have important effects on individual ﬁtness
among hyenas.

Do members of highly cooperative groups enjoy enhanced ﬁtness?

151

The group selection hypothesis has recently resurfaced as a potential
explanation for the evolution of altruistic acts, including agonistic aiding (Wilson
and Wilson, 2007). Our current data are only consistent with the group selection
hypothesis if we consider family groups to be the relevant targets of selection.
Whereas fully addressing the group selection hypothesis at the level of the clan
is beyond the scope of this study, members of hyena clans with more stable
cooperation networks might enjoy higher ﬁtness than those belonging to clans
with less stable networks (Kun and Scheuring, 2009; Nowak, 2006). Flack et al.
(2006) demonstrated the stabilizing function of third-party interventions among
pigtailed macaques (M. nemestrina); their experimental removal of intervening
animals destabilized the social network within a single generation. Our data
suggest that third-party interventions by high-ranking female spotted hyenas
might similarly stabilize social relationships among clan members. Future work
should therefore inquire whether clans with the most stable coalition networks
enjoy the greatest ﬁtness.

To cooperate or not: a complex decision

Because of the polyadic nature of intragroup coalitions, many argue that
agonistic aiding represents a decision-making process that is particularly
cognitively demanding (e.g. Connor, 2007; Harcourt, 1992; Kummer, 1967).
Indeed, our data suggest that adult female spotted hyenas base such decisions
on multiple factors. Our results extend earlier work demonstrating that spotted
hyenas recognize third-party rank relationships; hyenas support the higher-

ranking of two contestants when intervening in ﬁghts, even when the dominant

152

 

individual is losing (Engh et al., 2005). Although social facilitation appears to
promote coalition formation among captive juveniles (Glickman et al., 1997;
Zabel et al., 1992), wild adult spotted hyenas in this study were selective when
donating support to social partners, and generally adopted a strategy that
reduced their personal costs of intervening. Here adult female hyenas made
ﬂexible decisions about whether or not to cooperate based on multiple forms of
information including dyadic rank and kin relationships, but also based their
immediate ecological and extradyadic social circumstances. More broadly, the
results from both our literature review and our current study of spotted hyenas
are consistent with the emerging view that, although evolutionary explanations
for cooperation are often proposed as mutually exclusive options, multiple factors
typically shape complex patterns of cooperation found in nature (CIutton-Brock,
2009; West et al., 2007a). Therefore, although progress is being made in solving
the evolutionary puzzle of cooperation, our work emphasizes the need for novel,

integrative theoretical frameworks in which to view complex forms of cooperation.

153

Chapter 5

Smith JE, Powning KS, Dawes SE, Estrada JR, Hopper AL, Piotrowski SL,
Holekamp KE, To be submitted, June 2010. Gestural communication
reinforces alliances among greeting partners in the spotted hyena.
Animal Behaviour.

154

Chapter 5

GESTURAL COMMUNICATION REINFORCES ALLIANCES
AMONG GREETING PARTNERS IN THE SPOTTED HYENA

INTRODUCTION
Societies characterized by ﬁssion-fusion dynamics consist of subgroups of
variable size and composition in which group members regularly join (fusion) or
separate from (ﬁssion) each other (Kummer, 1971). This ﬂexible lifestyle
characterizes the societies of humans, chimpanzees (Pan troglodytes), spider
monkeys (Ateles spp.), elephants (Loxodonta spp.), many cetaceans (e.g.
bottlenose dolphins, Tursiops truncatus), and most terrestrial carnivores. This
social structure permits individuals to separate temporarily from one another
when the costs of grouping are high, and to aggregate when the costs of
grouping are low or the beneﬁts of sociality are high (Aureli et al., 2008; Lehmann
et al., 2007; Rodseth et al., 1991; Schino, 2000; Smith et al., 2008; Wrangham et
al., 1993). Although life in a ﬁssion-fusion society permits individuals to reduce
conﬂicts of interest (Conradt and Roper, 2005), this lifestyle imposes 'a unique set
of challenges upon group members that are often separated from one another for
long durations (Aureli et al., 2008). As a result, individuals must cope with
uncertain relationship status after temporary separations (Barrett et al., 2003).
Theory predicts that ritualized signals should evolve that quickly
communicate the intent of senders to receivers when relationship status is
uncertain (Endler, 1993; Maynard Smith and Price, 1973; Zahavi, 1980).
Consistent with this prediction, many animals use visual displays, called

gestures, to communicate their intent to conspeciﬁcs (Hewes, 1973; Pollick and

155

de Waal, 2007). Greetings, or meeting ceremonies, are important non-aggressive
gestures that often involve risky and intimate contact. Ritualized greetings can
function to reconcile ﬁghts (Aureli and de Waal, 2000), signal acknowledgement
of relative social status (de Waal, 1986; de Waal and Luttrell, 1985; Preuschoft
and van Schaik, 2000), reduce tension among individuals with insecure social
relationships (Aureli and Schaffner, 2007; Dias et al., 2008; Kutsukake et al.,
2006), or reinforce social bonds (Smuts, 2002; Smuts and Watanabe, 1990).

Spotted hyenas (Crocuta crocuta) engage in greetings when two partners
stand parallel to each other and face in opposite directions to sniff each other’s
anogenital region (East et al., 1993; Glickman et al., 1997; Kruuk, 1972). The
importance of the erect phallus in both sexes makes these greetings particularly
intriguing. Females erect their penile clitoris and males erect their penis during
greetings. Symmetric greetings occur when both members engage in the same
set of behaviors, such as both lifting their leg during mutual investigation of the
genitalia (East et al., 1993). In asymmetric greetings, only one partner exhibits
the behavior. Greetings that serve as a form of reconciliation only account for
roughly 8-9% of all greetings in this species (East et al., 1993). Although previous
studies (Hofer and East, 2000; Wahaj et al., 2001) demonstrated that these
conciliatory greetings are useful in preventing escalated aggression between
former opponents, the vast majority of greetings occur in other contexts,
suggesting that these signals may serve other important functions.

Early workers found that spotted hyenas in the Serengeti National Park

typically initiate greetings with social partners dominant to, or older than,

156

 

themselves (East et al., 1993; Kruuk, 1972). Although both of these earlier
studies agreed that low-ranking hyenas were most likely to initiate greetings, the
interpretation of these results differed. Kruuk (1972) hypothesized that greetings
likely function to reestablish social bonds at reunions because hyenas “expose
the most vulnerable area of their bodies to the teeth of their opponents (p. 229)”.
In contrast, East et al. (1993) concluded that “greetings are a ritualized, active
form of submission that conﬁrm asymmetries in status between greeting partners
(p. 364)”, and referred to the erect phallus as a “ﬂag of submission”.

Although greetings might signal submission, East et al. (1993) failed to
rule out alternative hypotheses that appeared in the literature after 1993, or to
use contemporary multivariate statistics to account for correlations among
potential predictor variables. Here, we take advantage of modern conceptual
frameworks and quantitative methods to extend earlier work, and to resolve
discrepancies in the interpretation of early studies. Adopting the methods of East
et al. (1993), we ﬁrst replicate their work by documenting the occurrence of
greetings among members of a single, large social group of spotted hyenas in
the Masai Mara National Reserve. Next, we conﬁrm that reconciliation only
accounts for a small fraction of greetings in our population and, for the ﬁrst time,
reveal how conciliatory and non-conciliatory greetings differ. Finally, we test three
competing hypotheses, each of which aims to elucidate the type of information
communicated during non-conciliatory greetings among spotted hyenas.

Spotted hyenas are long-lived carnivores that reside in complex ﬁssion-

fusion societies, called clans, containing up to 80 individuals that defend a

157

common territory (Kruuk, 1972). Individual members travel, rest, and forage in
subgroups that change membership roughly every hour (Kruuk, 1972; Mills,
1990; Smith et al., 2008). Clans are structured by linear dominance hierarchies
(Frank, 1986), and contain one to several matrilines of adult females and their
offspring, as well as multiple adult immigrant males. \ﬁrtually all males
permanently disperse from their natal clans after puberty, but females are
philopatric (East and Hofer, 2001; Mills, 1990; Smale et al., 1997).

Our main goal here was to investigate the function of non-conciliatory
greetings among adult females. Hyenas belonging to this age-sex category greet
each other at the highest frequencies (East et al., 1993). Moreover, although
rank relationships are extremely stable among adult females (Engh et al., 2000),
those of juveniles are often not yet ﬁrmly established (Holekamp and Smale,
1993; Smale et al., 1993). Further, although adult females maintain long-term
social bonds (Holekamp et al., 1997a), associations among adult males are often
weak (Smith et al., 2007) or short-lived (Van Horn et al., 2003). Adult females
make active decisions to join temporary subgroups containing their kin
(Holekamp et al., 1997a). Among non-kin, adult females associate most often
with females ranked directly above them in the dominance hierarchy and, by
doing so, gain enhanced tolerance from dominants (Smith et al., 2007).
Predictions based on the submission hypothesis
To minimize the costs of competition, dominance hierarchies structure societies
in which individuals use transient signals to communicate their knowledge of

power asymmetries among group members (Lu et al., 2008; Preuschoft, 1999;

158

Preuschoft and van Schaik, 2000). Because spotted hyenas use multiple status
indicators to reliably signal submission in a variety of contexts (Frank, 1986;
Kruuk, 1972), the initiation of greetings might represent another formalized status
signal. Although this hypothesis generally predicts that low-ranking females
should solicit greetings more often than high-ranking females, as found by East
et al. (1993), it importantly predicts that initiation of greetings should be strictly
unidirectional within dyads across ecological contexts (de Waal and Luttrell,
1985; de Waal and Luttrell, 1989). Therefore, the degree of directional
consistency and transitive properties of greeting initiation should coincide
precisely with those found in a dominance hierarchy based on ﬁght outcomes.
Moreover, because animals closest in rank possess the greatest need to clarify
dominance (de Waal, 1991), females should greet most often and engage in the
least symmetric greetings with females holding ranks similar to their own.
Predictions based on the tension reduction hypothesis

The tension-reduction hypothesis posits that, to reduce the probability of costly
ﬁghting, natural selection should favor the evolution of ritualized gestures that
signal peaceful intent, and, thus, reduce ﬁghting among individuals with insecure
relationships, especially in contexts in which tensions might otherwise be
elevated (Aureli and Schaffner, 2007; Colmenares et al., 2000; Dias et al., 2008;
Hohmann and Fruth, 2000; Kutsukake et al., 2006). If greetings evolved to
reduce tensions, then adult females should greet most often, and engage in the
most symmetric greetings, with those females with whom their social

relationships are least secure, such as non-kin, distantly associating kin, or

159

hyenas with which they rarely form coalitions. Further, feeding competition is
intense in this species; hyenas feed at kills that are energetically rich and highly
ephemeral (Frank, 1986; Kruuk, 1972; Mills, 1990; Smith et al., 2008). Thus, if
greetings reduce tensions over food, then females should greet most often per
opportunity when kills are immediately present, especially during those months
when prey abundance is locally scarce. Finally, if greetings reduce immediate
tensions, or substitute for aggression in tense contexts, then greeting partners
should direct less aggression towards one another in the minutes directly after
reunions than do non—greeting partners.

Predictions based on the social bonding hypothesis

The social bonding hypothesis posits that group-living animals should use risky
interactions to routinely test and reinforce afﬁliative relationships (Kummer, 1978;
Smuts, 2002; Zahavi, 1977b). Speciﬁcally, senders and receivers should
theoretically use costly signals to exchange honest information about social
bonds (Gintis et al., 2001; Zahavi, 1977a). Because hyena greetings are
potentially costly if they result in severe genital wounding (Kruuk, 1972),
greetings might facilitate the development and maintenance of social bonds. If
greetings reﬂect afﬁliative ties, then hyenas should greet their preferred social
partners most often, and these greetings should be the most symmetric. Further,
if greetings reinforce bonds, then they should occur most often in contexts in
which tensions are reduced (Smuts and Watanabe, 1990), Gas in the absence of
direct feeding competition, and might facilitate cooperative hunting (Creel and

Creel, 2002) or coalition formation (Smuts, 2002).

160

METHODS

Study site and subjects

From June 1988 through December 2004, we monitored spotted hyenas from a
large clan that defended a stable group territory in the Masai Mara National
Reserve, Kenya (Boydston et al., 2001). Spotted hyenas feed primarily on
antelope they hunt themselves, but local prey abundance varies seasonally
(Holekamp et al., 1997b). Feeding competition is reduced and social cohesion is
enhanced among clan mates when migratory ungulates are present with resident
herds (Holekamp et al., 1996; Smith et al., 2008).

We ident'rﬁed individual hyenas by their unique spots, and sexed them
based on the morphology of the erect phallus-(Frank et al., 1990). Mother—
offspring relationships were established based on nursing associations
(Holekamp et al., 1993), and paternal kinship was determined based on
genotyping (Engh et al., 2002). We estimated (to :I: 7 days) the ages of cubs
upon ﬁrst observing them above ground at dens (Holekamp et al., 1996). We
considered cubs found more than 200m from the den on at least four
consecutive occasions to be den-independent; this occurred when cubs were 8-9
months old (Boydston et al., 2005). Den-dwelling cubs and den-independent
subadults were considered juveniles. Females were classiﬁed as adults at 36
months of age, or at their ﬁrst known date of conception, whichever occurred
ﬁrst. We considered all immigrant males to be adults (Van Horn et al., 2003).

Here we determined the social rank of each individual hyena based on the

outcomes of dyadic agonistic interactions; all adult females were dominant to all

161

immigrant males (Holekamp and Smale, 1993; Smale et al., 1993). We ranked
adult males and females in separate hierarchies, with the highest possible rank in
each being one. We calculated rank distance as the absolute value of the
difference in intrasexual ranks between the members of each dyad.

Behavioral data collection

Using our ﬁeld vehicles as mobile blinds, we observed hyenas daily. We initiated
an observation session each time we encountered one or more hyenas
separated from other clan members by at least 200 meters; hyenas in different
sessions were typically separated by at least 1 km (Smith et al., 2008). Upon
arrival at each session, and during subsequent scans performed every 15-20
minutes, we recorded the identity and activity of every hyena in that focal
subgroup. We recorded as critical incidents all occurrences of agonistic
interactions, greetings, and hunting (Altmann, 1974). Aggressive behaviors
included head waves, lunges, chasing, displacements, standing over, biting,
pushing, and aggressive postures. .

Following Wahaj et al. (2001), we considered a greeting to serve a
conciliatory function between former opponents if it occurred within 10 minutes
immediately after a ﬁght. Based on this deﬁnition, we categorized each greeting
as either conciliatory or non-conciliatory. Dyadic aggression involved only one
aggressor, whereas coalitionary aggression involved at least two hyenas joining
forces to direct aggression towards the same target animal (Smith et al., 2010).
lntragroup coalition partners were those females that formed at least one

aggressive coalition directed towards one or more members of their clan.

162

 

Resident females also joined forces to attack alien hyenas during intergroup
conﬂicts at territory boundaries, called clan wars, or to attack lions during
conﬂicts over food (Kruuk, 1972). Hunting partners were those females that
hunted together. Thus, within each session, each pair could potentially cooperate
as intragroup coalition partners, as hunting partners, as participants in a clan
war, and/or as coalition partners in joint attacks directed towards lions.

Kruuk (1972) identiﬁed the initiators of each greeting based on the role of
each partner in the leg-lifting part of the display. East et al. (1993) extended this
deﬁnition to identify a hyena as initiating a greeting if it lifted its leg ﬁrst,
approached ﬁrst, or erected its phallus ﬁrst. They generally found equivalent
results for all three measures, and, in some cases, rank-related asymmetries in
leg lifting were even more pronounced than those based on phallic erections
(East et al., 1993). To conﬁrm that Kruuk’s (1972) leg-lifting measure provides
equivalent information to that conveyed by the erect phallus in our study
population, we recorded the characteristics of erections for a subset of greetings
(N = 855 greetings). As in East et al. (1993), we found high concordance
between leg lifting and phallic erections. When a focal hyena (N = 135) initiated a
greeting by lifting its leg ﬁrst, it was also signiﬁcantly more likely to erect its
phallus ﬁrst (88 :I: 1.7 %, Wilcoxon-Signed Ranks Test: 2 = 9.518, P < 0.00001).
Thus, in the current study, we used the leg-lifting display to identify the initiators
of greetings. This measure was the most conspicuous to human observers and
was consistently used in both earlier studies (East et al., 1993; Kruuk, 1972).

Here we deﬁned symmetric and asymmetric greetings as those greetings during

163

 

which both members, or only one member, respectively, engaged in leg lifting.

We also timed a subset of greetings using a stopwatch, starting when the
initiator lifted its hind leg and ending when the terminating hyena put its hind leg
back down on the ground. Most of these greetings were timed during a single
year of our study. From these data, we calculated the duration of each greeting
for which a clear initiator, start time, and end time could be discerned.

Whenever possible, in addition to recording the stereotypic behavior of
leg-lifting, we also recorded all well-recognized submissive signals (e.g. head-
bobbing, submissive posture, groveling (carpel crawling), and open mouth
appeasement), including unsolicited appeasements, and all afﬁliative behaviors
(e.g. nuzzling, rubbing-against, snifﬁng, friendly approaches or presenting of the
hindquarters) emitted by hyenas during greetings.

Following Van Meter (2009), we calculated the hourly rate of greetings for
each hyena while controlling for variation among sessions with respect to
opportunities for each hyena to greet other clan members. Here we calculated an
hourly rate of greeting for each individual present in a given observation session
with at least one potential partner as: (number of greeting interactions involving
that individual/number of potential greeting partners present/number of hours in
that observation session). We then averaged the rate per session for each
individual during which that animal belonged to a particular age-sex class.
Testing the submission hypothesis
First, we conﬁrmed that juveniles and lower-ranking hyenas in our study clan

initiate greetings most often, as observed by East et al. (1993). Then we

164

extended this ﬁnding by testing the null hypothesis that the hierarchical orders of
adult females winning ﬁghts and receiving greetings were equivalent (de Vries et
al., 1993). We also calculated the directional consistency index (DC, van Hooff
and Wensing, 1987), linearity index (h’, de Vries, 1995), and Kendall’s coefﬁcient
of linearity (K, Appleby, 1983) for dominance and greeting matrices constructed
for the same adult females, all of which had the opportunity to interact with each
other as adults both at and away from kills.

DC was based on the number of times that a behavior was performed in
the direction of higher frequency within each dyad (H) minus the number of times
it occurred in the direction of the lower frequency within each dyad (L), divided by
the number of times it was performed by all individuals: DC = (H - L)I(H + L). DC
ranges from zero, for completely bidirectional exchanges, to one for completely
unidirectional exchanges (van Hooff and Wensing, 1987). Because this measure
is a proportion, it allows for meaningful comparisons between matrices containing
unequal numbers of interactions. It also provides equivalent information to
indices used by previous authors (e.g. Noe et al., 1980; Rowell, 1966).

We evaluated the linearity and transitive properties of both matrices using
two methods. Linearity is a measure of how consistently individuals positioned
higher in the hierarchy outrank all individuals ranked lower than themselves
(Whitehead, 2008). If such relationships are transitive, then when A outranks B,
and B outranks C, A must also outrank C. We used an improved version of
Landau’s h index of linearity (Appleby, 1983b; Landau, 1951), called h’, which

corrects for unknown relationships. This corrected measure allowed us to make

165

comparisons between matrices in which relationships within some dyads were.
unknown. This measure ranges from zero, for a completely non-linear system, to
one for a completely linear system (de Vries, 1995). Second, we calculated K to
assess linearity based on transitivity of triads in each matrix (Appleby, 1983).
Testing the tension reduction and social bonding hypotheses
Because greetings may occur either at subgroup fusion when individuals reunite
after being separated, or among hyenas that have been present together in a
subgroup for several hours (Kruuk, 1972), we ﬁrst quantiﬁed the distribution of
greetings over time after reunions. Because the vast majority of greetings
occurred within the ﬁrst 10 minutes after fusion events (see Results), we
constructed statistical models to assess the effects of social and ecological
variables on the propensity for adult females to greet, given the opportunity to do
so, within 10 minutes after fusion. Models based on greetings occurring within 5
minutes after fusion produced equivalent results. Thus, for the sake of brevity, we
report only the former results. We entered subgroup size at fusion (number of
possible greeting partners available to each arriving female) as a covariate to
control for the possibility that greeting queues might form at fusion events.
Although the mean genetic relatedness among natal members of our clan
is extremely low, adult females belonging to the same matriline are all closely
related to one another (Van Horn et al., 2004a). Here, we modeled the effects of
kinship on greeting interactions by entering two measures of relatedness directly
into our models as continuous predictor variables. First, following Wahaj et al.

(2001), we assigned coefﬁcients of relatedness based only on known maternal

166

relationships from pedigree data. Second, following Smith et al. (2010), we
assigned coefﬁcients of relatedness to female pairs based on knowledge of
maternal pedigrees and genetic assignment of paternal relationships.

Statistical models explaining the tendency to greet at fusion were based
on detailed records extracted for all fusion events from 1996 to 2000. Although
data on greeting symmetry were available throughout our longitudinal study, we
limited our statistical models explaining symmetry to those greetings observed
before 2001 to ensure that social ranks were stable; our study clan permanently
split into two clans by 2001 (Smith and Holekamp, unpublished data).

Because patterns of association reﬂect social preferences among hyenas
(Holekamp et al., 1997a; Smith et al., 2007; Wahaj et al., 2004), we calculated
the Twice-Weight Association Index (AI) of Cairns and Schwager (1987) to
investigate the strength of social bonds within dyads. An Al was calculated for
each pair of females, hyenas A and B, during which they were concurrently
present in the clan as adults. We calculated AIAB as: (A+&ogethe,) l [(Awm,out a) +
(memt A) + (A+Btogem.r)] where (A+Btogethe,) represents the number of sessions in
which A and B were both present, (Amhout 3) represents the number of sessions in
which A was observed but B was not present, and (Bwithom A) represents the
number of sessions in which B was observed but A was not present.

Statistical analyses
We implemented all matrix analyses in MatMan 1.0 (Noldus lnforrnation
Technology, Wageningen, The Netherlands). For measures of linearity, we used

a two-step test with 10,000 randomizations (de Vries 1995). Matrix analyses

167

 

were based on one-tailed probabilities because these hypotheses make clear,
directional predictions (de Vries, 1993; Hemelrijk, 1990a). All other tests were
based on two-tailed probabilities. Differences were considered signiﬁcant at
alpha less than 0.05. We applied the sequential Bonferroni adjustment to correct
for multiple testing, and report all P-values in their corrected form (Rice, 1989).
Where appropriate, we report mean :I: 1 SE and sample proportions :I: 1 SD for
binomial trials (Agresti and Coull, 1998).

We built generalized linear mixed models (GLMM) to evaluate the effects
of the predictor variables using Ime4 (Bates and Maechler, 2010) in R Version
2.6.2 (The R Foundation for Statistical Computing 2008). We entered the identity
of each hyena as a random effect to avoid potential pseudoreplication, and
tested the signiﬁcance of its inclusion in each model using likelihood-ratio tests
(Pinheiro and Bates, 2000). For each data set, we sequentially entered and
dropped all potential explanatory terms, including all 2-way interactions, and
deemed the candidate model with the smallest Akaike‘s information criterion

(AIC) value to the be the best model (Burnham and Anderson, 2002). No strongly
Intercorrelated variables were retained in any of the ﬁnal models (I2 5 0.15). We

obtained statistics for terms removed from our best model by adding each term
individually to the minimal model. We modeled the duration data assuming a
Gaussian family because these data were normally distributed. To predict
decisions made by adult females to: 1) greet with potential adult female partners
within 10 minutes after fusion events and 2) engage in symmetric gestures when

greeting other adult females, we modeled data using binomial response

168

 

variables. Otherwise, we used STA TISTICA 6.1 (StatSoft, lnc., Tulsa, OK,
USA.) to analyze data failing to meet assumptions of normality and/or
homoscedasticity of variances. We compared means for two, or more than two,
independent groups using Mann-Whitney U and Kruskal-Wallis tests,
respectively. We compared means from dependent groups using Wilcoxon-
signed rank tests and tested correlations using Spearman’s R.

RESULTS

Distribution and patteming of conciliatory and non-conciliatory greetings

We recorded a total of 15,852 greetings involving 414 individual hyenas during
15,288 observation hours. Only 0.3 % (N = 52) of all greetings involved more
than two partners, such that the mean number of hyenas involved in each
greeting was 2.0 i 0.0 hyenas (Range: 2 to 7 partners). Greetings were generally
spontaneous, and rarely occurred in response to overt aggression. In fact, only
8.9 % of greetings were conciliatory and the rest were non-conciliatory (91.1 %).
Interestingly, focal hyenas (N = 387) were more almost twice as likely to engage
in one or more affiliative behaviors (35.9 i 2.0 % of greetings) than to display
submissive signals when soliciting greetings (18.2 i 1.0 % of greetings, Wilcoxon
Sign-Ranks Test: 2 = 11.00, P < 0.00001). Similarly, recipients (N = 395) of
invitations to greet were twice as likely to engage in afﬁliative behaviors (21.9 i
1.1 % of greetings) than to display submissive signals (10.6 i 0.9 % of greetings,
Wilcoxon Sign-Ranks Test: 2 = 10.13, P < 0.00001). Hyenas never attempted to
greet with heterospeciﬁcs, and natal hyenas only greeted with clan members.

Only on 7 occasions did immigrant males greet with males from a neighboring

169

 

clan; it is possible that these males previously belonged the same clan.

Duration of all greeting ceremonies

Greetings were similar in duration to those measured by East et al. (1993),
lasting, on average, 20.9 :I: 0.7 sec. (range = 1 to 95 sec., N = 283 greetings). We
ﬁrst conﬁrmed the singular ﬁnding on duration from this earlier study. Speciﬁcally,
when we used the methods of East et al. (1993), we also found that intrasexual
greetings were longer between adult females (24.6 :t 1.1 sec., N = 27 adults)
than adult males (18.7 i 2.0 sec, N = 17 adults, Mann-Whitney U—test: Z = 2.194
and P = 0.028). We then extended this result by modeling other factors that might
inﬂuence the duration of greetings among different age-sex classes. Interestingly,
adult females (N = 31, 22.8 i 1.0 sec.) participated in greetings that lasted longer
than those involving juvenile females (N = 20, 21.0 i 1.1 sec.), juvenile males (N
= 32, 17.4 i 1.1 sec), or immigrant males (N = 18 males, 19.4 i 2.0 sec., Table
5.1). The duration of greetings was also inﬂuenced by the relationship between
the hyenas involved. That is, our best model revealed that greetings between
adults (24.7 i 1.4 sec.) lasted longer than those involving one or more juveniles
(19.2 i 1.4 sec., Age composition: Table 5.2). Further, greetings initiated by
females (23.4 i 1.1 sec.) lasted longer than those initiated by males (18.4 i 0.9
sec., Sex of initiator: Table 5.2). Interestingly, conciliatory greetings lasted
roughly 50 % longer (30.7 i 4.8 sec.) than did non-conciliatory greetings (20.2 i
0.7 sec., Table 5.2). Finally, greetings initiated by dominants (22.4 i 1.4 sec.)

lasted longer than those initiated by subordinates (20.1 i- 0.8 sec., Table 5.2).

170

 

Table 5.1. Independent variables predicting the duration of greetings in which
focal hyenas participated.

 

Coefﬁcients: Estimate i S.E. Z-value P-value
(Intercept) 18.517 :I: 0.807 22.947 <0.000001
Age (adult) 2.584 d: 1.122 2.303 0.022
Sex (female) 1.720 :I: 1.144 1.503 0.134

 

The effects of age and sex were additive (Interaction: -0.769 :I: 2.344, 2 = -0.328,
P = 0.743). Neither the main effect of social rank (0005 :t 0.032, Z = -0.141, P =
0.888), nor its interaction with age (0.052 :I: 0.063, Z = 0.828, P = 0.409) or sex (-
0.056 :I: 0.064, Z = -0.877, P = 0.382) improved the ﬁt of the model. Including the
random effect of hyena identity improved the best model (Likelihood ratio test: 12
= 4.3, d.f. = 1, P = 0.038), which was based on 283 greetings involving 32

juvenile males, 18 adult males, 20 juvenile females and 31 adult females.

171

Table 5.2. Independent variables predicting the duration of greetings based on
the relationships between initiators and recipients of greetings.

 

Coefﬁcients: Estimate 0 SE. Z-value P-value
(Intercept) 16.706 :I: 1.100 15.193 < 0.0000001
Conciliatory greetings 10.254 1 2.667 3.844 0.0001
Age composition (both adults) 3.793 t 1.589 2.388 0.018

Sex of initiator (female) 2.820 t 1.478 1.908 0.058
Initiator outranks recipient 2.770 :I: 1.412 1.962 0.051

 

The inclusion of the following additional factors as predictors of the duration of
greetings within female pairs failed to further improve the ﬁt of our best model:
kinship (0.876 0 1.4980, 2 = 0.585, P = 0.559), rank distance (0045 :I: 0.065, Z =
-0.697, P = 0.487), absolute rank of initiator (0.026 t 0.047, 2 = 0.546, P =
0.586), and absolute rank of recipient (-0.007 :I: 0.047, 2 = -0.159, P = 0.874).

Including the random effect of hyena identity improved the model (Likelihood

ratio test: 12 = 15.1, d.f. = 1, P = 0.0001), which was based on 283 greetings

involving 32 juvenile males, 18 adult males, 20 juvenile females and 31 adult
females. Conciliatory greetings involving former opponents lasted longer than
non-conciliatorygreetings. Age composition inﬂuenced duration; greetings
between adults lasted longer than greetings involving one or more juveniles.

172

 

Hourly rates of greetings vag with life histogy stage and context
East et al.(1993) reported the frequencies of greetings (but not hourly rates)

between different age and sex categories observed exclusively at the communal
den. By contrast, we report here the hourly rates at which different classes of
animals participated in greetings not only at dens but also at kills and locations
away from kills and away from dens (called “other contexts”). Overall, the hourly
rate at which hyenas participated in greetings varied among life history stages
(Kruskal-Wallis test: H2549 = 102.7, P < 0.0001). In general, adult females (N =
75) and subadults (N = 209) participated in greetings at similarly high hourly rates
(Mann-Whitney U-test: Z = 0.276, P = 1.0000). Both adult females and subadults
greeted at signiﬁcantly higher rates than did den cubs (N = 289 cubs, Z 3 8.101,

P 5 0.000001). We detected no sex differences in the rates of greeting at which

den cubs (Z = -0.690, P = 1.0000, N; = 137, NM = 152) or subadults (Z = 0.270,

P = 1.0000, N): = 97, NM = 112) greeted. Immigrant males (N = 76) greeted with

clan members at lower rates than did subadults or adult females (Z 3 2.979, P 5
0.014), but at rates similar to those of den cubs (Z = 0.702, P = 1.0000).

Next we inquired whether hourly rates of greeting varied among contexts.
Here we combined subadults and adult females into a single category, called
den-independent hyenas, because these natal hyenas participated in greetings
at indistinguishable rates within each context. Den cubs were only observed at
the den, but we compared rates at which den independent hyenas (N = 232) and
immigrant males (N = 76) greeted in all three contexts. Overall, the hourly rates

_ at which hyenas participated in greetings varied signiﬁcantly among life history

173

 

stages and among contexts (Kruskal-Wallis test: F6,1299 = 234.3, P < 0.0001,

Figure 5.1). In contrast to the predictions of the submission and tension-reduction

hypotheses, hyenas participated in greetings at the lowest hourly rates at kills.

Initiation ofgreetings by juveniles and subordinates

 

Adopting the methods of East et al. (1993), we analyzed patterns of initiation for
all greetings such that our initial analysis included both conciliatory and non-
conciliatory greetings. As found by East et al. (1993), the younger or socially
subordinate of the two partners here typically solicited greetings by lifting their leg
ﬁrst. The extent to which juveniles initiated greetings with adults depended upon
on the age-sex category of the partners involved (KruskaI-Wallis ANOVA: H3510
= 10.873, P = 0.0124, Figure 5.2). When greeting partners differed in age, the
younger hyena lifted its leg ﬁrst signiﬁcantly more often (76.6 i 2.2 %) than did
the older partner (N = 304 focal hyenas, Wilcoxon Signed-Ranks Test: Z = 8.25,
P < 0.000001). Both den cubs (N = 259) and subadults (N = 178) initiated a
signiﬁcant majority of their greetings with adult females (Z 3 6.921 and P 5
0.000001). However, the initiation of greeting was generally symmetric when den
cubs (N = 85) or subadults (N = 108) greeted with adult males (Z 5 0.856, P 3
0.398). Focal den cubs (N = 79) and subadults (N = 103) initiated a greater
proportion of greetings involving adult females than they did when greeting adult

males (Z < 3.922, P 5 0.007). However, we found no sex difference in the extent

to which den cubs initiated greetings with adult females (N; = 111, NM = 129) or
males (NF_= 40, NM = 44, Mann-Whitney U-test: Z 5 0.994, P 3 0.960 for both).

Strikingly, although both sexes of subadults (NM = 58, N; = 50) initiated greetings

174

 

ty
5:
M
o

0-18 “ [:1 Kill scenes 226
_ D
0'16 - Communal den
0.14 —
- "Other" contexts 224 74
0.12 ~ C

A,B,C,F

 

 

Hourly rate of greetings per opportuni

 

Den cubs Den-independent Adult immgrant
natal hyenas males

Life history stage

Figure 5.1. Mean 0 SE hourly rates at which hyenas in each life history stage
participated in greeting ceremonies at kills, dens, and “other" contexts (both away
from kills and away from dens). Because we detected no sex differences in the
rates at which den cubs or den-independent natal animals greeted, we pooled
the values within each life history stage. Sample sizes shown above each bar
represent numbers of individuals in each category. Letters above bars Indicate
statistically signiﬁcant differences after correcting for multiple testing at P < 0.05.

175

90 - 91

 

 

 

 

   

Juvenile female
2 85 ' Juvenile male C
.E 129 87
g 80 - 111
.a B B
3‘ 75 — 40 B
8
E 50 58
.0
IS A A
1‘0
0
.E
ii
0
8
0
I.-
o
e\°

Den cub- Den cub- Subadult- Subadult-

Immigrant male Adult female Immigrant male Adult female
Partner type

Figure 5.2. Percentage of greetings with adults initiated by juveniles. Den cubs
were those juveniles still residing at the den whereas subadults were those
juveniles that were independent of the den but not yet adults. Sample sizes
shown above each bar represent numbers of individuals involved in greetings
with partners in each adult category. Letters above bars indicate statistically
significant differences after correcting for multiple testing.

176

with adult males to similar extents (Mann-Whitney U-test: Z = -0.231, P = 1.000),
male subadults (N = 91) were signiﬁcantly more likely than female subadults (N =
87) to initiate greetings with adult females. (Z = -2.903, P = 0.0185). Among
same-sexed dyads of adults, the subordinate partner was signiﬁcantly more likely
to lift its leg ﬁrst in both sexes than was the dominant greeting partner (NM = 42,
N; = 59, 75.1 i 4.9 and 73.7 i 3.2 %, Z 3 4.19, P 5 0.00003). Therefore, as
before, when we used the same methods used by East et al. (1993), we
observed the same patterns they did.
Linearity and directional consistency among adult females
Although the results reported above are consistent with those of East et al.
(1993), to qualify as a formal status indicator, non-conciliatory greetings should
exhibit a degree of linearity and directional consistency similar to that produced
by ﬁght outcomes. To test the submission hypothesis, we focused on 8 years of
painrvise interactions among a subset of adult females (N = 19), all of which had
opportunities to greet and ﬁght one another as adults, both at kills and away from
kills. A dominance matrix containing these 19 females was generated based on
outcomes of dyadic ﬁghts; the “winner” was the hyena being appeased and the
“loser” displayed appeasement when the ﬁght ended (Engh et al., 2005). A
greeting matrix involving those same 19 females was based on interactions in
which the “initiator” lifted her leg ﬁrst to solicit greeting from a “recipient”.

After correcting for the number of interactions observed for each dyad
(TauKr = 0.329), a partial rowwise matrix correlation revealed a weak, but

signiﬁcant, tendency for losers of ﬁghts to also be the member of each dyad

177

responsible for initiating non-conciliatory greetings (TauKr = 0.186, P < 0.05,
Figures 5.3 and 5.4). Nonetheless, in contrast to the results predicted by the
submission hypothesis, we found multiple striking discrepancies between
dominance and greeting matrices. First, only 15.8 % of the rankings generated
by ﬁght outcomes and those generated by the initiation of greetings were in
agreement. Second, the directional consistency of ﬁght outcomes was high
overall and nearly perfectly asymmetric, indicating an extreme imbalance in
competitive ability within dyads of adult females (DC = 0.97). This value
remained virtually the same when calculated for these 19 adult females at food
(DC = 0.98) and away from food (DC = 0.97). In stark contrast, the directional
consistency of greeting interactions was low overall (DC = 0.65) and varied
among contexts. Although greetings were generally rare at kills, when they did
occur, the DC (0.83) at kills was greater than the DC (0.66) away from kills. Thus,
contrary to the predictions of the submission hypothesis, the initiation of
greetings is neither unidirectional nor is the extent to which initiation is
unidirectional ‘context-free’.

We also found another striking discrepancy between the two matrices.

Fight outcomes produced a rigid and signiﬁcantly transitive linear dominance
hierarchy (Improved Linearity Test: h' = 0.59, P = 0.0001, Figure 5.3). This
structure remained statistically signiﬁcant when based on ﬁghts occurring only
over food (h' = 0.35, P = 0.014) or only in non-food contexts (h' = 0.47, P =
0.0001). In contrast, greeting initiation failed to produce a linear rank order (P =

0.136, Figure 5.4), with a linearity index (h' = 0.26) that was less than half of that

178

Figure 5.3. A dominance matrix based on the outcomes of 717 dyadic ﬁghts,
each of which had a clear winner and loser. Each row of the matrix represents a '
different adult female (N = 19 hyenas), all of which were present together as
adults both at food and away from food. At the intersection of each row (the
winner) and column (the loser), a cell shows the number of ﬁghts won against the
loser. We listed individuals based on their rank order, with the alpha female (bsh)
represented in the top row and the left most column of the matrix. Similarly
shaded adjacent cells within the ﬁrst row and the left most column represent
adult females belonging to the same matriline. For example, the alpha matriline
contains bsh, mrph, sein, gil, bb, kip, who, and mali. A total of ﬁve matrilines are
represented within the matrix. Dyadic relationships were either unidirectional
(71.9 %, N = 123), bidirectional (4.1 %, N = 7), or unknown (24.0 %, N = 41).
Black squares indicate that one adult female member of the dyad won the
majority of ﬁghts within that dyad. White squares indicate dyads for which an
equal number or no agonistic interactions were observed.

179

Figure 5.3.

 

 

Losers of ﬁghts

 

swﬁu :0 SJauurM

 

56 71 35 46 59 67 71 717

55 63

3O 14 21 74 19

10

0

Losses

 

180

 

Figure 5.4. A greeting interaction matrix based on 354 greetings, each of which
had a clear initiator and recipient. Because the individuals included here (N = 19
hyenas) were the same adult females shown in Figure 5.3, and because the act
of initiating greetings failed to produce a signiﬁcantly non-random ordering of
relationships, we ranked females based on the dominance rank order established
in Figure 5.3. As before, adult females belonging to the same matriline are shown
in adjacent cells in the top row and left-most column are shaded in the same
color. At the intersection of each row (the recipient of greetings) and column (the
solicitor of greetings), a cell shows the number of greeting interactions received
by a particular individual from the initiating hyena. Dyadic relationships were
either unidirectional (39.7 %, N = 68), bidirectional (21.1 %, N = 36), or unknown
(39.2 %, N = 67). Black squares indicate that one member of the dyad
preferentially received greetings more often than it initiated greetings with the
other adult female within that dyad. White squares indicate dyads for which an
equal number of greetings or no initiations were observed.

181

 

 

 

vmn or

002000”.

Hr

on

RN

mm 0.. Fr Fr 3.

005.080 .0 0.0.00.0.

00.

NF

mm

NM

FN

m

o

020......

 

sBupanS jo stuegdgoea

 

 

.40 050..

182

produced by ﬁght outcomes. Moreover, greeting interactions failed to produce a
linear structure within a single context (Food: h' = 0.18, P = 0.967, Non-food: h' =
0.23, P = 0.397). Kendall’s coefﬁcient of linearity further conﬁrmed these results;

ﬁght outcomes (Overall: K = 0.55, X2 = 87.2, Food: K = 0.27, x2 = 44.5, Non-food:
K = 0.42, X2 = 67.4, P 5 0.022 in all cases), but not greeting interactions (Overall:
K = 0.20, x2 = 34.1, Food: K = 0.03, x2 = 14.9, Non-food: K = 0.16, x2 = 27.6, P3
0.201 in all cases), produced signiﬁcant, transitive rank relationships (25.8 d.f. for
each test). Whereas all dyads for which one member wonthe majority of ﬁghts
(black squares) were above the diagonal in the dominance matrix, the greeting
matrix contained multiple dyads for which the member of the dyad that received
the most greetings (black squares) was the female situated below the diagonal.
Overall, the majority of thesedata fail to support the submission hypothesis.

' Most greetings occurred directly after fusion events

Timing relative to the proceeding fusion. event was known for 13,074 greetings,
involving 449 different hyenas. Hyenas generally initiated greetings immediately
after reuniting with individuals from whom they had been separated. The modal
number of minutes to pass between fusion events and the onset of greeting was
one. On average, hyenas greeted within 6.3 i 0.2 minutes of fusion events
(Range: 1 to 137 minutes post-fusion). Interestingly, however, focal hyenas (N =
267) engaged in conciliatory greetings signiﬁcantly later in the post-fusion interval
(7.4 i 0.4 minutes) than they engaged in non-conciliatory greetings (5.9 i 0.1
minutes; Wilcoxon-Signed Ranks Test: Z = 2.866, P = 0.004).

Because our main goal here was to explain the function of non-conciliatory

183

 

greetings, we next removed conciliatory greetings from the data set. Non-
conciliatory greetings (N = 11,759) typically occurred within 5.8 i 0.2 minutes
after fusion (N = 448 hyenas), and became less frequent as time passed after
fusion (Spearman rank correlation: R3 = -0.958, P < 0.00001, Figure 5.5). More
than half of these greetings (57.7%) occurred within the ﬁrst minute, 77.7% within
5 minutes, and 86.6 % within 10 minutes after fusion.

Females were selective when greeting after fusion events

Away from kills, focal adult females (N = 33) were signiﬁcantly more likely to be
responsible for initiating non-conciliatory greetings when they were responsible
for initiating fusion (58.2 i 4.0 %) than when they were joined by another female
(41.8 :t 4.0 %, Wilcoxon-Signed Ranks Test: 2 = 2.16, P = 0.031). In contrast, the
initiation of greetings at kills was less structured; focal females (N = 24) were no
more likely to initiate greetings upon arriving at subgroups (47.9 i 8.3 %) than
when they were joined by other females (52.1 i 8.3 %, Wilcoxon-Signed Ranks
Test: 2 = 0.355, P = 0.723). Arriving females (N = 37) typically joined subgroups
on their own such that, on average, adult females arrived with less than one (0.8
i 0.1) companion. These females joined subgroups containing, on average, 7.9 i '
0.2 hyenas, of which 2.8 i 0.5 were also adult females. Females were therefore
selective with respect to greeting decisions, greeting, on average, only 7.1 i 0.6
% of the adult females available to them after fusion events.

Modeling factors to explain greeting decisions after fusion events

Next we identiﬁed the subset of candidate predictor variables that signiﬁcantly

184

 

% Of greetings

 

 

 

T l T l l T T

‘3 O ‘3 O ‘3 O ‘3 O ‘3 O ‘3 O O
w b b :1 5» :5 :5 k > b ‘3 So 6

_L N w uh 0| G N on
O O O O O O O O O
J l l L l l l l J

 

Number of minutes following fusion event

Figure 5.5. Percentage of all non-conciliatory greetings (N = 11,759) that
occurred among spotted hyenas within the 5 minute-intervals directly after
subgroup reunions (“fusion events”) throughout our longitudinal study.

185

explained: 1) decisions to participate in non-conciliatory greetings with adult
females after fusion and 2) the symmetry of non-conciliatory greetings between
adult females. Whereas both models were consistent with the predictions of the
social bonding hypothesis, neither model strongly supported the submission or
tension reduction hypotheses.

Limited support for the submission or tension rchJction hypotheses

Overall, our statistical models failed to support the main predictions of the
submission or tension reduction hypotheses. In contrast to the submission
hypothesis, neither the relative social rank nor the rank distance of the arriving
female to potential greeting partners explained whether females greeted after
fusion (Table 5.3). Contrary to the predictions of the tension reduction
hypothesis, females were actually least likely to greet per opportunity at kills
(Table 5.3, Figure 5.6). Moreover, the act of greeting in non-conciliatory contexts
also failed to protect hyenas from receiving aggression immediately after fusion
(Table 5.3). Speciﬁcally, females greeting hyenas with whom they had not
previous fought during the same session were just as likely to ﬁght with them
after participating in non-conciliatory greetings as were those females who failed
to greet after fusion. Interesting, however, arriving females greeted high-ranking
females most often per opportunity (Table 5.3). This ﬁnal result suggests that,
irrespective of their relative rank to potential partners, choosy females prefer to
greet high-ranking social allies.

Evidence supporting the social bonding hypothesis
As predicted by the social bonding hypothesis, but in direct contrast to the

186

 

Table 5.3. Independent variables predicting whether or not arriving adult females
greeted particular females present in joined groups at each opportunity to do so.

 

Coefﬁcients: Estimate i S.E. Z-value P-value
(Intercept) -1.747 :I: 0.233 -7.495 < 0.0000001
Subgroup size (at fusion) -0.086 :I: 0.013 6609 < 0.0000001
lntragroup coalition partners 1.038 :I: 0.258 4.023 0.000057
Food present (kill scene) 0556 :I: 0.150 -3.713 0.000205
Clan war participants 2.477 :I: 0.843 2.939 0.003290
Absolute rank of partner -0.030 :I: 0.011 -2.681 0.007344
Association index, 3.605 :I: 1.925 1.873 0.061059
Cooperatively attack lions 0.795 :I: 0.451 1.765 0.077638
Coefficient of relatedness 1.926 :I: 0.558 3.453 0.000555
Prey abundance 0.096 :I: 0.151 0.636 0.524833
Relatedness * Prey abundance -2.502 :I: 0.752 -3.326 0.000880

 

The following additional factors failed to improve the ﬁt of the best model
predicting whether or not females greeted after fusion events: Maternal kinship
(0.341 :t 0.270, 2 = 1.262, P = 0.207), Relative rank (arriving female subordinate
to potential partner, 0.174 :I: 0.142, 2 = 1.231, P = 0.218), rank distance between
potential partners (-0.020 :t 0.017, 2 = -1.172, P = 0.241), absolute social rank of
the arriving female (0.003 d: 0.011, 2 = 0.305, P = 0.761), whether females
cooperatively hunted (-0.467 :I: 0.635, 2 = -0.735, P = 0.462) or fought each other
after fusion (0.035 :I: 0.186, 2 = 0.190, P = 0.850). Including hyena identity as a

random effect improved the ﬁt of our best (Likelihood ratio test: 2’2 = 14.6, d.f. = 1,
P = 0.0001), which was based on a total of 433 possible greeting pairs of adult
females. Adult females (N = 37) only participated in a total of 369 greetings out of
4,448 potential opportunities after fusion. The negative relationship between the
absolute rank of available partners and greetings per Opportunity reﬂects a
preference by adult females to greet with high-ranking females (e.g. the highest
possible social rank was one).

187

 

 

0.40
Non-food context

 

o. 35 « Non-kin 62

Kin

IE

0.30 -

Food context

 

0.25 —
21

0.20 -
44 777

0.15 ~

0.10 — 391 2101

Greetings at fusion per opportunity

0.05 - l

 

 

 

 

Not coalition Coalition Not coalition Coalition
partners partners partners partners

Figure 5.6. Proportion of fusion events in which adult females greeted other adult
females within 10 minutes of subgroup fusion, out of all opportunities to do so, at
kills (food context) and away from kills (non-food context). Coalition partners
formed at least one aggressive coalition directed towards another hyena or
towards a lion during the minutes after a fusion event, non-coalition partners
failed to form a coalition in the minutes after a fusion event. Although bar color
here indicates whether members of each dyad were non-kin (white) and kin
(black). we entered the coefﬁcient of relatedness for each dyad as a continuous
variable in our statistical model (I' able 5.2). Sample sizes over each bar indicate
the number of opportunities available to arriving females to greet potential adult
female partners. Error bars represent 1 1 standard deviation for binomial trials.

188

predictions of the tension reduction hypothesis, adult females preferentially
greeted kin (Table 5.3, Figure 5.6) and closely associating non-kin (Table 5.3,
Figure 5.7) most often per opportunity after each fusion event. After controlling
for the inﬂuence of kinship, close associates greeted signiﬁcantly more often at

each fusion event than did distant associates regardless of local prey abundance
(Association * Prey Abundance: -2.048 :1: 3.138, 2 = -0.737, P = 0.461). However,

the effects of kinship signiﬁcantly interacted with the effects of local prey
abundance (Table 5.3). Whereas non-kin greeted at similarly low rates
throughout the year (Prey abundance: 0.030 :I: 0.154, 2 = 0.197, P = 0.844), kin
were most likely to greet per opportunity during months when prey were scarce
(Prey abundance: -0.562 :I: 0.185, 2 = -3.041, P = 0.004). Nonetheless, kin were
generally more likely than non-kin to greet during months of low prey (0.642 :I:
0.198, 2 = 3.241, P = 0.002) and high prey (0.362 1: 0.219, 2 = 1.656, P = 0.098).
Overall, these results are most consistent with predictions of the social bonding
hypothesis; females greeted kin most often at fusion during those times of year
when prey scarcity most reduces social cohesion among clan mates. .

After controlling for the effects of kinship and patterns of association,
females that joined forces to form aggressive coalitions after fusion were
signiﬁcantly more likely to greet per opportunity than were non-coalition partners
(Table 5.3, Figure 5.6). This was the case when adult females directed intragroup
coalitions towards other clan members and when females cooperatively attacked
alien intruders at clan wars or when mobbing lions during cooperative defense of

food. However, we detected no relationship between the acts of greeting and

189

 

 

Greetings at fusion per opportunity

 

 

 

0.00-0.04 0.05-0.09 0.10-0.14 0.15-0.19 0.20-0.24
Association index

Figure 5.7. Proportion of fusion events in which adult females greeted other adult
females within 10 minutes of subgroup fusion, out of all opportunities to do so, as
a function of how often pairs of adult females associated with each other. For the
purposes of visual representation only, we present patterns of association binned
into discrete categories, but entered values as a continuous variable into our
statistical model (Table 5.2). Sample sizes over each bar indicate the number of
opportunities arriving adult females had to greet potential adult female partners at
fusion events. Error bars represent :t 1 standard deviation for binomial trials.

190

cooperative hunting (Table 5.3). When females formed coalitions with greeting
partners after fusion (Fig. 5.6), they were more likely to do so in the minutes
immediately after greeting (78.5 i 7.8 %) than in the minutes directly before
greeting (21.5 i 7.8 %, Wilcoxon-Signed Ranks Test: Z = 2.520, P = 0.012, N =
21 females). On average, when these females formed coalitions to attack
another hyena, they did so within 2.7 :I: 0.9 minutes of greeting initiation. The
tendency for females to preferentially form coalitions with greeting partners could
not be explained simply by the amount of time females were observed with social
partners. Females remained in subgroups for similar durations after fusion,
regardless of whether or not they formed a coalition (35.7 i 2.7 vs. 31.4 i 0.5
minutes, Wilcoxon-Signed Ranks Test: 2 = 1.547, P = 0.123, N = 21 females).
Modeling factors to explain the symmetry of greetings

Following East et al. (1993), a single univariate analysis of all greetings (e.g.
conciliatory and non-conciliatory combined) indicated that the asymmetry of leg
lifting increased as the rank distance between adult females increased

(Spearman Rank Correlation: N = 22 rank distances; R3 = 0.553, P = 0.007).

Interestingly, conciliatory greetings involving focal females (N = 52) were
signiﬁcantly more asymmetric (63.3 :I: 4.3 %) than were non-conciliatory
greetings (48.3 i 2.7 %, Wilcoxon-Signed Ranks Test: 2 = 4.049, P = 0.00005).
After we accounted association and relatedness, in contrast to the predictions of
submission hypothesis, rank distance was excluded from our best model
explaining the symmetry of non-conciliatory greetings (Table 5.4). Also in

contrast to the tension reduction hypothesis, but as predicted by the social

191

 

bonding hypothesis, closely associating females generally engaged in the most
symmetric greetings. However, after accounting for patterns of association, non-
kin generally participated in more symmetric greetings than kin, suggesting that
socially bonded non-kin might rely more heavily upon reciprocal greetings to
reinforce bonds than do kin. Overall, our ﬁndings are more consistent with
predictions of the social bonding hypothesis than those of the submission and
tension reduction hypotheses.

DISCUSSION

Conciliatory versus non-conciliatory greetings

Overall, the meaning and distribution of greetings varied among contexts,
suggesting that greetings represent complex signals. First, as found in early
studies, here the vast majority of greetings served a non-conciliatory function
(East et al., 1993; Hofer and East, 2000). The low frequency of conciliatory
greeting (e.g. occurring immediately after ﬁghts) is presumably because hyenas
rely most heavily upon dispersive conﬂict resolution to settle disputes (Smith et
al., 2008; Wahaj et al., 2001). Second, whereas conciliatory greetings transpire
primarily among non-kin and are initiated mostly by losers of ﬁghts (Wahaj et al.,
2001), we found here that non-conciliatory greetings mainly involved kin, even
after correcting for opportunities to greet. Third, non-conciliatory greetings were
more symmetric and occurred earlier in the interval directly after fusion events
than did conciliatory greetings. Finally, conciliatory greetings lasted longer than
non-conciliatory greetings; this ﬁnding is consistent with the idea that it takes

more time to renegotiate damaged relationships after ﬁghts than it does to

192

Table 5.4. Independent variables predicting the symmetry of non-conciliatory
greetings among adult females.

 

Coefﬁcients: Estimate i SE. 2 Statistic P-value
(Intercept) -0.246 :1: 0.115 -2.142 0.032
Association index (Al) 4.037 1 1.472 2.742 0.006
Coefﬁcient of relatedness -1.108 :I: 0.418 -2.650 0.008

 

The additional predictor variables failed to improve the ﬁt of our best model: Rank
distance: 0016 :1: 0.014, Z = -1.106, P = 0.269, Maternal kinship: -0.255 :t 0.267,
2 = -0.955, P = 0.339, Cooperative hunting: 0.701 :I: 0.463, 2 = 1.514, P = 0.130,
lntragroup coalition partners: -0.137 :I: 0.232, 2 = -0.592, P = 0.554, Intergroup
coalition partners (clan war): 0.878 :I: 0.894, 2 = 0.982, P = 0.326, Prey
abundance: -0.093 :I: 0.105, Z = -0.885, P = 0.376, Food: -0.051 :I: 0.130, 2 = -
0.396, P =j 0.692). The interaction between association and kinship (Al * Kinship:
-7.657 1: 7.153, 2 = -1.070 P = 0.284), or that of any other terms, failed to
improve the ﬁt our model. A Likelihood ratio test conﬁrmed that the inclusion of
the random effect, “hyena identity”, improved the model’s ﬁt (12 = 11.6, d.f. = 1, P
= 0.0007), which included 1,750 non-conciliatory greetings involving 456 different
pairs of adult females from 1988 to 2000 (N = 57 adult females).

193

 

reinforce existing social bonds (Aureli et al., 2002).

Non-conciliatory greetings fail to signal submission

Although spotted hyena greetings are still cited as one of the key examples of
“reliable ritualized expressions of formal rank” (see p. 444, Cafazzo et al., 2010),
our critical test of the submission hypothesis revealed that this does not in fact
appear to be the case. In general, our study produced results that were
consistent with data obtained in earlier studies when we used the same methods.
Speciﬁcally, younger or subordinate spotted hyenas generally initiated greetings
more often than did older or socially dominant hyenas, as found by East et al.
(1993) and Kruuk (1972). However, when we focused only on non-conciliatory
greetings and simultaneously accounted for potential confounding factors using a
multivariate approach not yet widely available when the earlier studies were
conducted (e.g. Carlin et al., 2001; Faraway, 2006; Halekoh et al., 2006; Hardin
and Hilbe, 2003), we obtained results that differed from those of East et al.
(1993). lmportantly, our work revealed that the strength of social bonds is a
better predictor, than rank relationships, in explaining non-conciliatory greetings
among adult females. Interestingly, this ﬁnding is consistent with the hypothesis
originally proposed by Kruuk (1972).

The availability of new matrix permutation tools allowed us to quantify the
linearity of greeting initiation while explicitly correcting for unknown relationships
(e.g. de Vries, 1995). These analyses revealed that ﬁght outcomes clearly
adhered to the expectations of a linear hierarchy (en sensu Drews, 1993), but

that greetings failed to do so. Directional consistency (DC) is the most reliable

194

 

measure for comparing the strength of competition among taxonomic groups in
part because it is informative even in species for which some group members
interact at relatively low rates (Archie et al., 2006; Isbell and Pruetz, 1998; Isbell
and Young, 2002). Here the initiation of greetings was far more balanced within
dyads (DC = 0.65) than were ﬁght outcomes (DC = 0.98). Further, although ﬁght
outcomes were unidirectional across contexts, the extent to which signaling
during greetings was unidirectional varied among contexts. Our results represent
particularly strong evidence against the submission hypothesis because we
found here that the directional consistency (DC) of dominance interactions within
dyads of adult female spotted hyenas was generally similar to or greater than
008 based on ﬁght outcomes among other adult female mammals (Table 5.5). In
contrast, the DC based on hyena greetings is more akin to DC values based on
afﬁliative behaviors in other species (Table 5.5). Whereas these ﬁndings might
seem surprising, use of these new quantitative methods has similarly revealed
misconceptions about signaling in other species (reviewed by Kutsukake, 2009).

Although grooming among primates is often preferentially directed towards
higher and adjacently ranked coalition partners (Schino, 2001; Seyfarth, 1977;
Seyfarth and Cheney, 1984), grooming clearly does not signal submission. In
addition to its hygienic and hedonistic values, grooming also provides important
“political” information to social partners (Cheney et al., 1986; Dunbar, 1991;
Dunbar and Sharman, 1984). The pivotal role of grooming in social bond

maintenance is well documented among non-human primates living in cohesive

195

 

00.089 2:00 9 090500: «0: 0:03 900 0005 930.9: 05 :_ E0_0>0:0 000. 0:03 0:300:95
0>=0=Em :0 00 :0 005000.: 00:000m .0_0EE0E 0.0..:9 «.300 :9 0:30: 0005 9 090500: 0:03 900 060930 E9:

 

 

802. .5020 0:0 =30; ooe 00E09:0 H9“. 000.0500 030530808 0>0x:0E 90Z0>
Coom .._0 “0 00.950: 00.0 . 00E0930 Ear... 00600900: 0050020: 030$ 00>:0080:
$030 0:00.53 3.: 0080050 Eur. 5:020 9:096. 00:92.. 090000
.ﬁoom $0.55: 3.0 00E09:0 E9“. eEtE$0§E0 003.0050: 20$ 000N:00E_:o
.302 .00:> 0B 56 00E09:0 Emi AO0E00 096$ 0:02.905
A88 .0300 20 5.: mad 8588 Eat 09858 0:0: 09:
600m .._0 :0 022$ v0.0 09:09.6 29“. «0:00.50 9000006.: 95300 :0oE<
Amoom .._0 «0 9.000302 vmd 00E0930 29“. «200.5 009$ :005 :00:0E<
0:00.50 0:0000:00$
SOON .__000_ 0:0 0:00:29 cad 00E09:0 EB“. 0>000mc0E 0900:0905
Amoom .._0 :0 3V cad 09:09.8 £9“. 039:0 0305305009 0530. :0E::0I

:83 6933 0:0 5:09 00.0 00800930 Eon.
:09 0:52 can cmsmzoé $6 005830 39: 0.53:0 .n: 2898 02.0
88v .5020 0:0 .000: mud 09:09.0 EB“. «00:0: 0:0000:£b$ 0>0x:0E 0900
€me .m:_0:0>> 0:0 too... :0>v mud m:§:0 _9_:00 0:00.550: 003:3 00203
680 :0 :0 02.0.10: 006 85850 29: 0:39 89 0260
€me .m:_0:0>> 0:0 too: :03 mod 0530210 ._. 0:00:80: 003:3 00203
3030 0:00.53 mad :03025 95020 5:020 00:00.5. 00:02.. 090000
Aooom .._0 90 9.00.393: mod 05:80:00.2 000:: 0:00:08 0:00.00 923603.005
802 .._0 90 9.000202 mod 0080930 Eur. 0:00:20: 00$ 0000:0m
ﬁcow $0.55: 50.0 00500300 220:". 30:80.02200 003.0030: 00$ 000nc0aEEo
880 .._m e 020.: :3 0558522 08:2: £3008 800$ 99.8.: .850
803 605.9 mvd m:_E00:m0=< 0300950 0:008 0:503:00 00000009003
000:0..901 OD 003. 5:00:95 ”00095

00:50.: 0:0:0m05 mcoE0 0:300:95 0:030:00 :0 A09 >0:90_0:00 05:00.5 .m.m 030...

196

groups (e.g. Gomes et al., 2009; Lazaro-Perea et al., 2004; Matheson and
Bernstein, 2000; Silk et al., 2006), especially when the values of partners are
dynamic (Barrett et al., 2002; Henzi and Barrett, 1999; Henzi and Barrett, 2002).
Although adult female hyenas rarely allogroom, our data are consistent with the
idea that greetings might function to quickly update social bonds among hyenas.
Interestingly, spider monkeys (A. geoffroyi) also exchange greetings, but not
allogrooming, at fusion (Schaffner and Aureli, 2005). In fact, Aureli and Schaffner
(2007) theorized that individuals living in societies characterized by high rates of
ﬁssion-fusion dynamics should greet to quickly update relationships. As
predicted, we found here that friendly greetings among preferred companions
were extremely brief and typically occurred immediately after fusion.
Non-conciliatory greetings fail to reduce tension

Our data generally failed to support the hypothesis that non-conciliatory greetings
reduce tensions. In fact, greetings were least likely to occur when meeting up
after fusion events with distant associates, including non-kin, which presumably
have the least secure social relationships (Wahaj et al., 2001). Moreover, most
greetings occurred in neutral contexts in which tensions are low compared to
situations in which resource competition is most likely (Frank, 1986). Finally, the
act of greeting in non-aggressive contexts failed to protect hyenas from
immediately receiving aggression. This last ﬁnding is consistent with earlier
studies. Even though kin are each other’s the best allies, kinship fails to protect
hyenas (Smith et al., 2010; Wahaj et al., 2004) and many primates (reviewed by

Widdig, 2007) from becoming targets of aggression.

197

Like hyena greetings, genital contacts among adult female bonobos (P.
paniscus) serve multiple functions, and are sometimes used to reconcile ﬁghts
(Hohmann and Fruth, 2000). Non-conciliatory genital contacts reduce tensions
over food among bonobos (Hohmann and Fruth, 2000). In contrast, non-
conciliatory greetings here reinforced alliances among hyenas away fromfood.
Although genital contacts facilitate food sharing within the egalitarian societies of
bonobos (de Waal, 1997a), hyena greetings appear to promote coalition
formation among adult females that function to reinforce the status quo in
contexts away from food (Smith et al., 2010).

Non-conciliatory greetings reinforce social bonds

Overall, our ﬁndings are most consistent with the social bonding hypothesis.
Indeed, adult females only exchanged greetings at fusion events with a small
subset of the adult females present, and they selectively directed these gestures
towards their preferred social companions. Speciﬁcally, females greeted coalition
partners, relatives and close associates most often per opportunity. Thus, the
ﬁnding that females often initiate greetings with high-ranking females (East et al.,
1993 and Table 5.3) likely reﬂects social preferences for powerful allies rather
than functioning to signal submission.

Although mutual inspection of the highly vulnerable genitalia might
improve the efﬁcacy of gestures signaling bond strength (Smuts, 2002; Zahavi,
1977b), hyenas generally minimized this risk by selecting greeting partners who
posed the least risk to them, and by greeting these partners most often away

from contexts in which aggression is most common. First, because the

198

 

reproductive careers of maternal kin are closely linked to one another, through
direct and indirect ﬁtness beneﬁts (Hamilton, 1964), females should theoretically
be disinclined to damage the reproductive organs of their maternal kin. Second,
we found here that hyenas generally greeted in neutral contexts in the absence
of direct feeding competition. Similarly, male baboons (Papio cynocephalus
anubis) and domestic dogs (Canis lupus familians) greet most often when there
are no immediate resources at stake (Smuts, 2002; Smuts and Watanabe, 1990).
In peaceful contexts, greetings might permit group members to assess the
intentions of their partners when the risk of injury to the genitalia is greatly
reduced. Within their highly competitive, female-dominated society, greetings
among adult female spotted hyenas appear to offer a behavioral mechanism by
which females can assess the cooperative tendencies of potential allies.
Non-conciliatory greetings promote collective action
Although greetings failed to play an important role in preparing hyenas for
cooperative hunting, African wild dogs (Lycaon pictus) almost always engage in
greetings prior to hunting (Creel and Creel, 2002). Wild dogs live in cohesive
groups and increase their per capita energy intake by hunting in large groups
(Creel, 1997). In contrast, spotted hyenas typically hunt alone or in pairs because
individuals often suffer reduced energy gains when hunting in large parties
(Smith et al., 2008). Thus, hyenas might be subject to opposing selection
pressures contraindicating the use of greetings to coordinate group hunting.
lmportantly, we found strong evidence suggesting that greetings facilitate

coalition formation; this was true even after controlling for effects of kinship,

199

 

association patterns, and immediate ecological context. Although many workers
theorize that greetings promote coalition formation, direct empirical evidence for
this is limited. Smuts (2002) noted that a pair of male olive baboons engaged in
more risky and more symmetric genital touching than did non-coalition partners.
Ritualized embracing and mutual genital inspection appear to mediate social
relationships in many non-human primates (Alfaro, 2008; Colmenares, 1991;
Okamoto et al., 2001; Perry, 1998; Perry et al., 2003; Smuts and Watanabe,
1990; Wang and Milton, 2003; Whitham and Maestripieri, 2003). Our work
importantly extends these ﬁndings by demonstrating for the ﬁrst time a temporal
link between the patterning of greetings and coalition formation.

Cognitive demands of bond maintenance in dispersed societies

Whereas shifting ecological environments are known to favor enhanced cognitive
abilities in animals (e.g. Braithwaite and Salvanes, 2005; Kotrschal and
Taborsky, 2010), dynamic variation in social group composition should
theoretically also impose cognitive demands upon species in which individuals
recognize the relationships among group mates from whom they are often
separated (Amici et al., 2008; Aureli et al., 2008; Barrett et al., 2003; Connor,
2007). Because resource competition often forces members of ﬁssion-fusion
societies to spend much of their time in fragmented subgroups (Aureli et al.,
2008; Schino, 2000; Smith et al., 2008; Wrangham et al., 1993), many social
animals, including spotted hyenas (East and Hofer, 1991; Theis et al., 2007),
have evolved contact calls to “stay in touch” over long distances (e.g. McComb et

al., 2000; McComb et al., 2003; McCowan and Reiss, 2001; Ramos-Fernandez,

200

2005; Smolker et al., 1993; Spillmann et al., 2010). Moreover, hyenas maintain
cohesion within their social network by depositing individually distinct scent
marks (Burgener et al., 2009; Drea et al., 2002; Theis, 2008).

Although vocal and olfactory cues effectively communicate a hyena’s
identity when clan members are spatially separated, and although adult female
spotted hyenas understand third-party relationships among group mates (Engh et
al., 2005), experiments suggest that vocal cues fail to communicate information
about third-party relationships (Holekamp et al., 1999a). Our ﬁndings suggest
that greetings represent reliable signals with which hyenas can quickly conﬁrm
relationship status in a society in which group members spend much of their time
apart. These signals appear to be especially important in maintaining bonds
among kin during those times of year when resource limitation most strongly
constrains sociality (Smith et al., 2008).

Natural selection should theoretically favor efﬁcient signaling that
coordinates collective behaviors when those behaviors confer an evolutionary
advantage (reviewed by Conradt and Roper, 2005; Noé, 2006). Here greetings
signaled a hyena’s immediate commitment to alliances within a continuously
shifting social milieu. Thus, our ﬁndings elucidate the fundamental role of
signaling in coordinating cooperation among social partners within spatially and
temporally dynamic social landscapes. More broadly, our ﬁndings extend a
growing body of literature suggesting that ritualized signals are centrally
important to the maintenance of cooperative partnerships in complex societies

(e.g. Flack and de Waal, 2007; Rossano, 2009).

201

 

LITERATURE CITED

Agrell J, Wolff JO, Ylonen H, 1998. Counter-strategies to infanticide in mammals:
costs and consequences. Oikos 83:507-517.

Agresti A, Coull BA, 1998. Approximate is better than "exact" for the interval
estimation of binomial proportions. American Statistical Association
52:119-126.

Alexander RD, 1974. The evolution of social behavior. Annual Review of Ecology
and Systematics 52325-383.

Alfaro JL, 2008. Scream-embrace displays in wild black-horned capuchin
monkeys. American Journal of Primatology 70:551-559.

Altmann J, 1974. Observational study of behavior: sampling methods. Behaviour
49:227-267.

Amici F, Aureli F, Call J, 2008. Fission-fusion dynamics, behavioral ﬂexibility, and
inhibitory control in primates. Current Biology 18:1415-1419.

Appleby MC, 1983a. Competition in a red deer stag social group: rank, age and
relatedness of opponents. Animal Behaviour 31 :91 3-918.

Appleby MC, 1983b. The probability of linearity in hierarchies. Animal Behaviour
31:600-608.

Archie EA, Morrison TA, Foley CAH, Moss CJ, Alberts SC, 2006. Dominance
rank relationships among wild female African elephants, Loxodonta
afn'cana. Animal Behaviour 71 :1 17-127.

Aureli F, Cords M, Van Schaik CP, 2002. Conﬂict resolution following aggression
in gregarious animals: a predictive framework. Animal Behaviour 64:325-
343.

Aureli F, de Waal FBM, 2000. Natural Conﬂict Resolution. Berkeley: University of
California Press. ‘

Aureli F, Schaffner CM, 2007. Aggression and conflict management at fusion in
spider monkeys. Biology Letters 3:147-149.

202

 

Aureli F, Schaffner CM, Boesch C, Bearder SK, Call J, Chapman CA, Connor R,
Fiore AD, Dunbar RIM, Henzi SP, Holekamp K, Korstjens AH, Layton R,
Lee P, Lehmann J, Manson JH, Ramos-Femandez G, Strier KB, Schaik
CPv, 2008. Fission-fusion dynamics: new research frameworks. Current
Anthropology 49:627-654.

Baird RW, Dill LM, 1996. Ecological and social determinants of group size in
transient killer whales. Behavioral Ecology 7:408-416.

Baldwin JD, Baldwin J, 1972. Ecology and behavior of squirrel monkeys (Saimin'
oerstedr) in a natural forest in western Panama. Folia Primatologica
18:161-184.

Barrett L, Gaynor D, Henzi SP, 2002. A dynamic interaction between aggression
and grooming reciprocity among female chacma baboons. Animal
Behaviour 63:1047-1053.

Barrett L, Henzi P, 2001. The utility of grooming in baboon troops. In: Economics
in Nature: Social Dilemmas, Mate Choice, Biological Markets (Noe R, van
Hooff JARAM, Hammerstein P, eds). Cambridge: Cambridge University
Press; 119-145.

Barrett L, Henzi P, Dunbar R, 2003. Primate cognition: from 'what now?‘ to 'what
if?‘ Trends in Cognitive Sciences 72494-497.

Barrett L, Henzi SP, Weingrill T, Lycett JE, Hill RA, 1999. Market forces predict
grooming reciprocity in female baboons. Proceedings of the Royal Society
of London Series B-Biological Sciences 266:665-670.

Barton RA, Whiten A, 1993. Feeding competition among female olive baboons,
Papio anubis. Animal Behaviour 46:777-789.

Bates D, Maechler M, 2010. Linear mixed-effects models using S4 classes. In: R
package version 0.999375-33; http://lme4.r-forge.r-project.org.

Bercovitch F B, 1988. Coalitions, cooperation and reproductive tactics among
adult male baboons. Animal Behaviour 36:1198-1209.

Bergman TJ, Beehner JC, Cheney DL, Seyfarth RM, 2003. Hierarchical
classiﬁcation by rank and kinship in baboons. Science 302:1234-1236.

Blumenschine RJ, Caro TM, 1986. Unit ﬂesh weights of some East African
bovids. African Journal of Ecology 24:273-286.

Boesch C, Boesch-Achermann H, 2000. The Chimpanzees of the Tai Forest:
Behavioural Ecology and Evolution. Oxford: Oxford University Press.

203

Boinski S, Kauffman L, Ehmke E, Schet S, Vreedzaam A, 2005. Dispersal
patterns among three species of squirrel monkeys (Saimin' oerstedii, S.
boliviensis and S. sciureus): l. Divergent costs and beneﬁts. Behaviour
142:525-632.

Borries C, 1993. Ecology of female social relationships: Hanuman langurs
(Presbytis entellus) and the Van Schaik Model. Folia Primatologica 61 :21-
30.

' Boydston EE, Kapheim KM, Szykman M, Holekamp KE, 2003. Individual
variation in space use by female spotted hyenas. Journal of Mammalogy
84:1006-1018.

Boydston EE, Kapheim KM, Van Horn RC, Smale L, Holekamp KE, 2005.
Sexually dimorphic patterns of space use throughout ontogeny in the
spotted hyena (Crocuta crocuta). Journal of Zoology, London 2672271-
281.

Boydston EE, Morelli TL, Holekamp KE, 2001. Sex differences in territorial
behavior exhibited by the spotted hyena (Hyaenidae, Crocuta crocuta).
Ethology 107:369-385.

Braithwaite VA, Salvanes AGV, 2005. Environmental variability in the early
rearing environment generates behaviourally ﬂexible cod: implications for
rehabilitating wild populations. Proceedings of the Royal Society B-
Biological Sciences 272:1 107-11 13.

Bramblett CA, 1970. Coalitions among gelada baboons. Primates 11:327-333.

Brown JL, 1983. Cooperation: a biologist's dilemma. Advances in the Study of
Behavior 13:1-37.

Burgener N, Dehnhard M, Hofer H, East ML, 2009. Does anal gland scent signal
identity in the spotted hyaena? Animal Behaviour 77:707-715.

Burnham KP, Anderson DR, 1998. Model selection and inference: A practical
information—theoretic approach. New York: Springer-Verlag.

Burnham KP, Anderson DR, 2002. Model selection and inference: A practical
infonnation-theoretic approach, 2nd ed. New York: Springer-Verlag.

Cafazzo S, Valsecchi P, Bonanni R, Natolib E, 2010. Dominance in relation to

age, sex, and competitive contexts in a group of free-ranging domestic
dogs. Behavioral Ecology 21 :443-455.

204

Cairns SJ, Schwager SJ, 1987. A comparison of association indices. Animal
Behaviour 35:1454-1469.

Campbell CJ, 2006. Lethal intragroup aggression by adult male spider monkeys
(Ateles geoﬁroyr). American Journal of Primatology 68:1197-1201.

Carlin JB, Wolfe R, Brown CH, Gelman A, 2001. A case study on the choice,
interpretation and checking of multilevel models for longitudinal binary
outcomes. Biostatistics 2:397—416.

Caro TM, Collins DA, 1987. Male cheetah social organization and territoriality.
Ethology 74:52-64.

Chancellor RL, Isbell LA, 2009. Female grooming markets in a population of
gray-cheeked mangabeys (Lophocebus albigena). Behavioral Ecology
20:79-86.

Chapais B, 1988. Rank maintenance in female Japanese macaques:
experimental evidence for social dependency. Behaviour 104241-59.

Chapais B, 1992. The role of alliances in the social inheritance of rank among
female primates. In: Coalitions and Alliances in Humans and Other
Animals (Harcourt AH, de Waal F BM, eds). Oxford: Oxford Science
Publications; 29-60.

Chapais B, Girard M, Primi G, 1991. Non-kin alliances and the stability of
matrilineal dominance relations in Japanese macaques. Animal Behaviour
41 :481-492.

Chapman CA, 1990. Ecological constraints on group-size in three species of
neotropical primates. Folia Primatologica 55:1 -9.

Chapman CA, Pavelka MSM, 2005. Group size in folivorous primates: ecological
constraints and the possible inﬂuence of social factors. Primates 46:1-9.

Chapman CA, Wrangham RW, Chapman LJ, 1995. Ecological constraints on
group-size: an analysis of spider monkey and chimpanzee subgroups.
Behavioral Ecology and Sociobiology 36:59-70.

Cheney DL, Seyfarth RM, 1982. Recognition of individuals within and between
groups of free-ranging vervet monkeys. American Zoologist 22:519-529.

Cheney DL, Seyfarth RM, Smuts B, 1986. Social relationships and social
cognition in non-human primates. Science 234:1361-1366.

205

Clutton-Brock TH, 1998. Reproductive skew, concessions and limited control.
Trends in Ecology & Evolution 13:288-292.

CIutton-Brock TH, 2009. Cooperation between non-kin in animal societies.
Nature 462:51-57.

Clutton- Brock TH, Guinness FE, Albon SD, 1982. Red Deer. Behavioral Ecology
of Two Sexes. Chicago: University of Chicago.

Clutton-Brock TH, Parker GA, 1995. Punishment in animal societies. Nature
373:209-216.

Colmenares F, 1991. Greeting, Aggression, and Coalitions between Male
Baboons - Demographic Correlates. Primates 32:453-463.

Colmenares F, Hofer H, East ML, 2000. Greeting ceremonies in baboons and
hyenas. In: Natural Conﬂict Resolution (Aureli F, de Waal FBM, eds).
Berkeley: University of California Press; 94—96.

Connor RC, 1995. Altruism among non-relatives: alternatives to the ‘Prisoner's
Dilemma’. Trends in Ecology & Evolution 10:84-86.

Connor RC, 2007. Dolphin social intelligence: complex alliance relationships in
bottlenose dolphins and a consideration of selective environments for
extreme brain size evolution in mammals. Philosophical Transactions of
the Royal Society B-Biological Sciences 362:587-602.

Connor RC, Heithaus MR, Barre LM, 1999. Superalliance of bottlenose dolphins.
Nature 397:571-572.

Connor RC, Heithaus MR, Barre LM, 2001. Complex social structure, alliance
stability and mating access in a bottlenose dolphin 'super-alliance'.
Proceedings of the Royal Society of London Series B-Biological Sciences
268:263-267.

Connor RC, Smolker RA, Richards AF, 1992. Two levels of alliance formation
among male bottlenose dolphins (Tursiops sp.). Proceedings of the
National Academy of Sciences of the United States of America 89:987-
990.

Conradt L, Roper TJ, 2000. Activity synchrony and social cohesion: a ﬁssion-
fusion model. Proceedings of the Royal Society of London Series B-
Biological Sciences 267:2213-2218.

Conradt L, Roper TJ, 2005. Consensus decision making in animals. Trends in
Ecology & Evolution 20:449-456.

206

Cooper MA, Bernstein IS, 2008. Evaluating dominance styles in Assamese and
rhesus macaques. International Journal of Primatology 29:225-243.

Cooper MA, Bernstein IS, Hemelrijk CK, 2005. Reconciliation and relationship
quality in Assamese macaques (Macaca assamensis). American Journal
of Primatology 65:269-282.

Cooper SM, 1991. Optimal hunting group size: the need for lions to defend their
kills against loss to spotted hyaenas. African Journal of Ecology 29:130-
136.

Couzin ID, 2006. Behavioral ecology: social organization in ﬁssion-fusion
societies. Current Biology 162R169-R171.

Creel S, 1997. Cooperative hunting and group size: assumptions and currencies.
Animal Behaviour 54:1319-1324.

Creel S, Creel NM, 1995. Communal hunting and pack size in African wild dogs,
Lycaon pictus. Animal Behaviour 50:1325-1339.

Creel S, Creel NM, 2002. The African wild dog: behavior, ecology, and
conservation. Princeton, New Jersey: Princeton University Press.

Creel S, Macdonald D, 1995. Sociality, group size, and reproductive suppression
among carnivores. Advances in the Study of Behavior 24:203-257.

Cross PC, Lloyd-Smith JO, Getz WM, 2005. Disentangling association patterns
in ﬁssion-fusion societies using African buffalo as an example. Animal
Behaviour 69:499-506.

Dally JM, Emery NJ, Clayton NS, 2006. Food-caching western scrub-jays keep
track of who was watching when. Science 312: 1 662-1665.

Darwin C, 1859. On the Origin of Species by Means of Natural Selection.
London: John Murray.

Dasser V, 1988. A social concept in Java monkeys. Animal Behaviour 36:225-
230.

de Villiers MS, Richardson PRK, van Jaarsveld AS, 2003. Patterns of coalition
formation and spatial association in a social carnivore, the African wild dog
(Lycaon pictus). Journal of Zoology, London 260:377-389.

de Villiers PA, Kok OB, 1997. Home range, association and related aspects of

elephants in eastern Transvaal Lowveld. African Journal of Ecology
35:224-236.

207

de Vries H, 1993. The rowwise correlation between two proximity matrices and
the partial rowwise correlation. Psychometrika 58:53-69.

de Vries H, 1995. An improved test of linearity in dominance hierarchies
containing unknown or tied relationships. Animal Behaviour 50:1375-1389.

de Vries H, Netto WJ, Hanegraaf PLH, 1993. Matman: a program for the analysis
of sociometric matrices and behavioral transition matrices. Behaviour
1252157-175.

de Waal F BM, 1977. The organization of agonistic relations within two captive
groups of java monkeys (Macaca fasciculan’s). Zeitschrift fur
Tierpsychologie 44:225-282.

de Waal F BM, 1984. Sex differences in the formation of coalitions among
chimpanzees. Ethology and Sociobiology 52239-255.

de Waal FBM, 1986. The integration of dominance and social bonding in
primates. Quarterly Review of Biology 61:459-479.

de Waal FBM, 1991. Rank distance as a central feature of rhesus monkey social
organization: a sociometric analysis. Animal Behaviour 41:383-395.

de Waal FBM, 1997a. Bonobo, the forgotten ape. Berkeley: University of
California Press.

de Waal FBM, 1997b. The chimpanzee's service economy: Food for grooming.
Evolution and Human Behavior 18:375-386.

de Waal FBM, Harcourt AH, 1992. Coalitions and alliances: a history of
ethological research. In: Coalitions and alliances in humans and other
animals (Harcourt AH, de Waal FBM, eds). Oxford: Oxford University
Press; 1-19.

de Waal FBM, Luttrell LM, 1985. The formal hierarchy of rhesus macaques: an
investigation of the bared-teeth display. American Journal of Primatology
9:73-85.

de Waal FBM, Luttrell LM, 1986. The similarity principle underlying social
bonding among female rhesus monkeys. Folia Primatologica 46:215-234.

de Waal FBM, Luttrell LM, 1988. Mechanisms of social reciprocity in three

primate species: symmetrical relationship characteristics or cognition.
Ethology and Sociobiology 92101-118.

208

de Waal FBM, Luttrell LM, 1989. Toward a comparative socioecology of the
genus Macaca: different dominance styles in rhesus and stumptail
macaques. American Journal of Primatology 19283-109.

de Waal FBM, van Hooff JARAM, 1981. Side-directed communication and
agonistic interactions in chimpanzees. Behaviour 77:164-195.

Derix R, van Hooff JARAM, Devries H, Wensing J, 1993. Male and female
mating competition in wolves: female suppression versus male
intervention. Behaviour 127:141-174.

Dias PAD, Luna ER, Espinosa DC, 2008. The functions of the "greeting
ceremony" among male mantled howlers (Alouatta palliata) on Agaltepec
Island, Mexico. American Journal of Primatology 70:621-628.

Dloniak SM, French JA, Holekamp KE, 2006. Faecal androgen concentrations in
adult male spotted hyaenas, Crocuta crocuta, reﬂect interactions with
socially dominant females. Animal Behaviour 71:27-37.

Drea CM, Carter AN, 2009. Cooperative problem solving in a social carnivore.
Animal Behaviour 78:967-977.

Drea CM, Frank LG, 2003. The social complexity of spotted hyenas. In: Animal
social complexity: intelligence, culture, and individualized societies (de
Waal FBM, Tyack PL, eds). Cambridge, Massachusetts: Harvard
University Press; 121-148.

Drea CM, Vignieri SN, Kim HS, Weldele ML, Glickman SE, 2002. Responses to
olfactory stimuli in spotted hyenas (Crocuta crocuta): ll. Discrimination of
conspeciﬁc scent. Journal of Comparative Psychology 116:342-349.

Drews C, 1993. The concept and deﬁnition of dominance in animal behavior.
Behaviour 125:283-313. ‘

Dugatkin LA, 2002. Animal cooperation among unrelated individuals.
Naturwissenschaften 89:533-541.

Dugatkin LA, Sih A, 1998. Evolutionary ecology of partner choice. In: Cognitive
Ecology (Dukas R, ed). Chicago: University of Chicago Press; 379-403.

Dunbar RIM, 1980. Determinants and evolutionary consequences of dominance
among female Gelada baboons. Behavioral Ecology and Sociobiology
72253-265.

Dunbar RIM, 1991. Functional signiﬁcance of social grooming in primates. FoIia
Primatologica 57:121-131.

209

Dunbar RIM, Sharman M, 1984. Is social grooming altruistic? Journal of
Comparative Ethology 64:163-173.

East M, Hofer H, Turk A, 1989. Functions of birth dens in spotted hyaenas
(Crocuta crocuta). Journal of Zoology, London 219:690-697.

East ML, Burke T, Wilhelm K, Greig C, Hofer H, 2003. Sexual conﬂicts in spotted
hyenas: male and female mating tactics and their reproductive outcome
with respect to age, social status and tenure. Proceedings of the Royal
Society of London Series B-Biological Sciences 270:1247-1254.

East ML, Hofer H, 1991. Loud-calling in a female—dominated mammalian society:
II. Behavioural contexts and functions of whooping of spotted hyaenas,
Crocuta crocuta. Animal Behaviour 42:651-669.

East ML, Hofer H, 2001. Male spotted hyenas (Crocuta crocuta) queue for status
in social groups dominated by females. Behavioral Ecology 12:558-568.

East ML, Hofer H, 2002. Conﬂict and cooperation in a female-dominated society:
a reassessment of the "hyperaggressive" image of spotted hyenas.
Advances in the Study of Behavior 3121-30.

East ML, Hofer H, Wickler W, 1993. The erect 'penis' as a ﬂag of submission in a
female-dominated society: greetings in Serengeti spotted hyenas.
Behavioral Ecology and Sociobiology 33:355-370.

East ML, Honer OP, Wachter B, Wilhelm K, Burke T, Hofer H, 2009. Maternal
effects on offspring social status in spotted hyenas. Behavioral Ecology
20:478—483.

Emery NJ, Seed AM, von Bayern AMP, Clayton NS, 2007. Cognitive adaptations
of social bonding in birds. Philosophical Transactions of the Royal Society
B-Biological Sciences 362:489-505.

Endler JA, 1993. Some general comments on the evolution and design of animal
communication systems. Philosophical Transactions of the Royal Society
of London Series B-Biological Sciences 340:215-225.

Engh AL, Esch K, Smale L, Holekamp KE, 2000. Mechanisms of maternal rank
'inheritance' in the spotted hyaena, Crocuta crocuta. Animal Behaviour
60:323-332.

Engh AL, Funk SM, Van Horn RC, Scribner KT, Bruford MW, Libants S,
Szykman M, Smale L, Holekamp KE, 2002. Reproductive skew among

males in a female-dominated mammalian society. Behavioral Ecology
13:193-200.

210

Engh AL, Siebert ER, Greenberg DA, Holekamp KE, 2005. Patterns of alliance
formation and postconﬂict aggression indicate spotted hyaenas recognize
third-party relationships. Animal Behaviour 69:209-217.

Faraway JJ, 2006. Extending the linear model with R2 generalized linear, mixed
effects, and nonparametric regression models. New York: Chapman and
Hall.

Fashing PJ, 2001. Male and female strategies during intergroup encounters in
guerezas (Colobus guereza): evidence for resource defense mediated
through males and a comparison with other primates. Behavioral Ecology
and Sociobiology 50:219-230.

Faul F, Erdfelder E, Lang AG, Buchner A, 2007. G*Power 3: A ﬂexible statistical
power analysis program for the social, behavioral, and biomedical
sciences. Behavior Research Methods 392175-191.

Feh C, 1999. Alliances and reproductive success in Camargue stallions. Animal
Behaviour 57:705-713.

Fentress JC, Ryon J, McLeod PJ, Havkin GZ, 1987. A multidemnsional approach
to agonistic behavior in wolves. In: Man and Wolf: Advances, Issues and
Problems in Captive Wolf Research (Frank H, ed). Boston: Dr. W. Junk
Publishers; 253-274.

Flack JC, de Waal F, 2007. Context modulates signal meaning in primate
communication. Proceeding of the National Academy of Sciences
104:1581-1586.

Flack JC, Girvan M, de Waal FBM, Krakauer DC, 2006. Policing stabilizes
construction of social niches in primates. Nature 4392426-429.

Flinn MV, Geary DC, Ward CV, 2005. Ecological dominance, social competition,
and coalitionary arms races: Why humans evolved extraordinary
intelligence. Evolution and Human Behavior 26:10-46.

Frank LG, 1983. Reproduction and infra-sexual dominance in the spotted hyena
(Crocuta crocuta) (Ph.D. dissertation): University of California, Berkeley.

Frank LG, 1986. Social organization of the spotted hyaena (Crocuta crocuta). II.
Dominance and reproduction. Animal Behaviour 34:1510-1527.

Frank LG, Glickman SE, Powch I, 1990. Sexual dimorphism in the spotted
hyaena (Crocuta crocuta). Journal of Zoology, London 221 :308-313.

211

Frank LG, Holekamp KE, Smale L, 1995. Dominance, demography, and
reproductive success of female spotted hyenas. In: Serengeti II:
Dynamics, Management, and Conservation of an Ecosystem (Sinclair
ARE, Arcese P, eds). Chicago, Illinois: University of Chicago Press; 364-
384.

Gasaway WC, Mossestad KT, Stander PE, 1991. Food acquisition by spotted
hyaenas in Etosha National Park, Namibia: predation versus scavenging.
African Journal of Ecology 29:64-75.

Gintis H, Smith EA, BowIes S, 2001. Costly signaling and cooperation. Journal of
Theoretical Biology 213:103-1 19.

Gittleman JL, 1989. Carnivore Behavior, Ecology and Evolution. Ithaca, New
York: Cornell University Press.

Glickman SE, Frank LG, Pavgi S, Licht P, 1992. Hormonal correlates of
'masculinization' in female spotted hyaenas (Crocuta crocuta). 1. Infancy
to sexual maturity. Journal of Reproduction and Fertility 95:451-462.

Glickman SE, Zabel CJ, Yoerg SI, Weldele ML, Drea CM, Frank LG, 1997. Social
facilitation, afﬁliation, and dominance in the social life of spotted hyenas.
Annals of the New York Academy of Sciences 807:175-184.

Gomes CM, Mundry R, Boesch C, 2009. Long-term reciprocation of grooming in
wild West African chimpanzees. Proceedings of the Royal Society B
276:699-706.

Gompper ME, 1996. Sociality and asociality in white-nosed coatis (Nasua
nan’ca): foraging costs and benefits. Behavioral Ecology 7:254-263.

Gompper ME, Gittleman JL, Wayne RK, 1997. Genetic relatedness, coalitions
and social behaviour of white-nosed coatis, Nasua nan’ca. Animal
Behaviour 53:781-797.

Gouzoules S, Gouzoules H, 1987. Kinship. In: Primate Societies (Smuts BB,
Cheney DL, Seyfarth RM, eds). Chicago and London: University of
Chicago Press; 299-305.

Grammer K, 1992. Intervention in conﬂicts among children: contexts and
consequences. In: Coalitions and Alliances in Humans and Other Animals
(Harcourt AH, De Waal FBM, eds). New York: Oxford University Press;
259-283.

Green B, Anderson J, Whateley T, 1984. Water and sodium turnover and
estimated food consumption in free-living lions (Panthera Ieo) and spotted
hyaenas (Crocuta crocuta). Journal of Mammalogy 65:593-599.

212

Halekoh U, Hojsgaard S, Yan J, 2006. The R Package geepack for generalized
estimating equations. Journal of Statistical Software 15:1-11.

Hamilton IM, 2000. Recruiters and joiners: using optimal skew theory to predict
group size and the division of resources within groups of social foragers.
American Naturalist 155:684-695.

Hamilton WD, 1964. The genetical evolution of social behavior, I and II. Journal
of Theoretical Biology 7: 1 -52.

Harcourt AH, 1992. Coalitions and alliances: Are primates more complex than
non-primates? In: Coalitions and Alliances in Humans and Other Animals
(Harcourt AH, de Waal FBM, eds). Oxford: Oxford Science Publications;
445-471.

Hardin JW, Hilbe JM, 2003. Generalized Estimating Equations. New York:
Chapman & Hall/CRC.

Hemelrijk CK, 1990a. A matrix partial correlation test used in investigations of
reciprocity and other social-interaction patterns at group level. Journal of
Theoretical Biology 143:405-420.

Hemelrijk CK, 1990b. Models of, tests for, reciprocity, unidirectionality and other
social interaction patterns at a group level. Animal Behaviour 39:1013-
1029.

Hemelrijk CK, 1994. Support for being groomed in long-tailed macaques, Macaca
fasciculan's. Animal Behaviour 48:479-481.

Hemelrijk CK, Ek A, 1991. Reciprocity and interchange of grooming and support
in captive chimpanzees. Animal Behaviour 41:923-935.

Henschel JR, Skinner JD, 1990. The diet of spotted hyaenas Crocuta crocuta in
Kruger National Park. African Journal of Ecology 28:69-82.

Henschel JR, Skinner JD, 1991. Territorial behaviour by a clan of spotted
hyaenas Crocuta crocuta. Ethology 88:223-235.

Henschel JR, Tilson RL, 1988. How much does a spotted hyena eat?
Perspective from the Namib Desert. African Journal of Ecology 26:247-
_ 255.

Henzi SP, Barrett L, 1999. The value of grooming to female primates. Primates
40247-59.

213

Henzi SP, Barrett L, 2002. Infants as a commodity in a baboon market. Animal
Behaviour 63:915-921.

Henzi SP, Barrett L, Gaynor D, Greeff J, Weingrill T, Hill RA, 2003. Effect of
resource competition on the long-term allocation of grooming by female
baboons: evaluating Seyfarth's model. Animal Behaviour 66:931-938.

Hewes GW, 1973. Primate communication and gestural origin of language.
Current Anthropology 1425-24.

Hirsch ET, 2007. Spoiled brats: Is extreme juvenile agonism in ring-tailed coatis
(Nasua nasua) dominance or tolerated aggression? Ethology 1132446-
456.

Hofer H, East M, 2000. Conﬂict management in female-dominated spotted
hyenas. In: Natural Conﬂict Resolution (Aureli F, de Waal FBM, eds).
Berkeley: University of California Press; 232-234.

Hofer H, East ML, 2003. Behavioral processes and costs of co-existence in
female spotted hyenas: a life history perspective. Evolutionary Ecology
17:31 5-331 .

Hohmann G, Fruth B, 2000. Use and function of genital contacts among female
bonobos. Animal Behaviour 60:107-120.

Holekamp KE, Boydston EE, Smale L, 2000. Group travel in social carnivores. In:
On the Move: How and Why Animals Travel in Groups (Boinski S, Garber
PA, eds). Chicago: University of Chicago Press; 587-627.

Holekamp KE, Boydston EE, Szykman M, Graham I, Nutt KJ, Birch S, Piskiel A,
Singh M, 1999a. Vocal recognition in the spotted hyaena and its possible

implications regarding the evolution of intelligence. Animal Behaviour
58:383-395.

Holekamp KE, Cooper SM, Katona CI, Berry NA, Frank LG, Smale L, 1997a.
Patterns of association among female spotted hyenas (Crocuta crocuta).
Journal of Mammalogy 78:55-74.

Holekamp KE, Ogutu JO, Frank LG, Dublin HT, Smale L, 1993. Fission of a
spotted hyena clan: consequences of female absenteeism and causes of
female emigration. Ethology 93:285-299.

Holekamp KE, Sakai ST, Lundrigan BL, 2007. Social intelligence in the spotted

hyena (Crocuta cmcuta). Philosophical Transactions of the Royal Society
B-Biological Sciences 362:523-538.

214

Holekamp KE, Smale L, 1991. Dominance acquisition during mammalian social
development: the "inheritance" of maternal rank. American Zoologist
31:306-317.

Holekamp KE, Smale L, 1993. Ontogeny of dominance in free-living spotted
hyaenas: juvenile rank relations with other immature individuals. Animal
Behaviour 46:451-466.

Holekamp KE, Smale L, 1995. Rapid change in offspring sex ratios after clan
ﬁssion in the spotted hyena. American Naturalist 145:261-278.

Holekamp KE, Smale L, Berg R, Cooper SM, 1997b. Hunting rates and hunting
success in the spotted hyena (Crocuta crocuta). Journal of Zoology,
London 242:1-15.

Holekamp KE, Smale L, Szykman M, 1996. Rank and reproduction in the female
spotted hyaena. Journal of Reproduction and Fertility 108:229-237.

Holekamp KE, Szykman M, Boydston EE, Smale L, 1999b. Association of
seasonal reproductive patterns with changing food availability in an
equatorial carnivore, the spotted hyaena (Crocuta crocuta). Journal of
Reproduction and Fertility 116287-93.

Honer OP, Wachter B, East ML, Hofer H, 2002. The response of spotted
hyaenas to Iong-tenn changes in prey populations: functional response
and interspeciﬁc kleptoparasitism. Journal of Animal Ecology 71:236-246.

HOner OP, Wachter B, East ML, Runyoro VA, Hofer H, 2005. The effect of prey
abundance and foraging tactics on the population dynamics of a social,
territorial carnivore, the spotted hyena. Oikos 108:544-554.

Hoy 8, Bauer J, 2005. Dominance relationships between sows dependent on the
time interval between separation and reunion. Applied Animal Behaviour
Science 90:21-30.

Hunte W, Horrocks JA, 1987. Kin and non-kin interventions in the aggressive
disputes of vervet monkeys. Behavioral Ecology and Sociobiology 20:257-
263.

Irvine AB, Scott MD, Wells RS, Kaufmann JH, 1981. Movements and activities of

the Atlantic bottlenose dolphin, Tursiops truncatus, near Sarasota, Florida.
Fishery Bulletin 79:671-688.

215

Isbell LA, Pruetz JD, 1998. Differences between vervets (Cercopithecus
aethiops) and patas monkeys (Erythrocebus patas) in agonistic
interactions between adult females. International Journal of Primatology
19:837-855.

Isbell LA, Young TP, 2002. Ecological models of female social relationships in
primates: Similarities, disparities, and some directions for future clarity.
Behaviour 139:177-202.

Jacobi EF, 1975. Forming a social group of the spotted hyena (Crocuta crocuta
E.). Der Zoologische Garten 45250-53.

Jennings DJ, Carlin CM, Gammell MP, 2009. A winner effect supports third-party
intervention behaviour during fallow deer, Dama dama, ﬁghts. Animal
Behaviour 77:343-348.

Jones CB, 1980. The functions of status in the mantled howler monkey, Alouatta
palliata Gray: Intraspeciﬁc competition for group membership in a
folivorous Neotropical primate. Primates 21:389-405.

Kays RW, Gittleman JL, 2001. The social organization of the kinkajou Pofos
ﬂavus (Procyonidae). Journal of Zoology, London 253:491-504.

Kingdon J, 1997. Kingdon Field Guide to African Mammals. Princeton: Princeton
University Press.

Kolowski JM, Katan D, Theis KR, Holekamp KE, 2007. Daily patterns of activity
in the spotted hyena. Journal of Mammalogy 88:1017—1028.

Kotrschal A, Taborsky B, 2010. Environmental change enhances cognitive
abilities in ﬁsh. PLoS Biology 8:1-7.

Kr‘L‘Itzen M, Shenrvin WB, Connor RC, Barré LM, Van de Casteele T, Mann J,
Brooks R, 2003. Contrasting relatedness patterns in bottlenose dolphins
(Tursiops sp.) with different alliance strategies. Proceedings of the Royal
Society of London Series B 2702497-502.

Kruuk H, 1972. The Spotted Hyena: A Study of Predation and Social Behavior.
Chicago, Illinois: University of Chicago Press.

Kruuk H, 1976. Feeding and social behaviour of the striped hyaena (Hyaena
vulgan's Desmarest). East African Wildlife Journal 14291-111.

Kruuk H, Parish T, 1982. Factors affecting population density, group size and

territory size of the European badger, Meles meles. Journal of Zoology
196231 -39.

216

Kummer H, 1967. Tripartite relations in hamadryas baboons. In: Social
communication among primates (Altman SA, ed). Chicago: University of
Chicago Press; 63-71.

Kummer H, 1971. Primate societies: group techniques of ecological adaptation.
Chicago: Aldine Publishing Company.

Kummer H, 1978. Value of social relationships to nonhuman-primates: a heuristic
scheme. Social Science Information 17:687-705.

Kun A, Scheuring I, 2009. Evolution of cooperation on dynamical graphs.
Biosystems 96:65-68.

Kutsukake N, 2009. Complexity, dynamics and diversity of sociality in group-
living mammals. Ecological Research 24:521-531.

Kutsukake N, Suetsugu N, Hasegawa T, 2006. Pattern, distribution, and function
of greeting behavior among black-and-white colobus. International Journal
of Primatology 27:1271-1291.

Landau HG, 1951. On dominance relations and the structure of animal societies:
I. Effects of inherent characteristics. Bulletin of Mathematical Biophysics
13:1-9.

Lazaro-Perea C, De Fatima M, Snowdon CT, 2004. Grooming as a reward?
Social function of grooming between females in cooperatively breeding
marmosets. Animal Behaviour 67:627-636.

Lee PC, 1987. Allomothering among African elephants. Animal Behaviour
35:278-291.

Lefebvre L, Henderson D, 1986. Resource defense and priority of access to food
by the mate in pigeons. Canadian Journal of Zoology 64:1889-1892.

Lehmann J, Boesch C, 2004. To ﬁssion or to fusion: effects of community size on
wild chimpanzee (Pan troglodytes verus) social organization. Behavioral
Ecology and Sociobiology 56:207-216.

Lehmann J, Korstjens AH, Dunbar RIM, 2007a. Fission—fusion social systems as
a strategy for coping with ecological constraints: a primate case.
Evolutionary Ecology 21:613-634.

Lehmann L, Keller L, Sumpter DJT, 2007b. The evolution of helping and harming

on graphs: the return of the inclusive ﬁtness effect. Journal of Evolutionary
Biology 20:2284-2295.

217

Leinfelder I, de Vries H, Deleu R, Nelissen M, 2001. Rank and grooming
reciprocity among females in a mixed-sex group of captive hamadryas
baboons. American Journal of Primatology 55:25-42.

Leiva D, Solanas A, Salafranca L, 2008. Testing reciprocity in social interactions:
A comparison between the directional consistency and skew-symmetry
statistics. Behavior Research Methods 40:626-634.

Lewis ME, Werdelin L, 2000. The evolution of spotted hyenas (Crocuta). I.U.C.N.
Hyaena Specialist Group Newsletter 7:34-36.

Liang KY, Zeger SL, 1986. Longitudinal data analysis using generalized linear
models. Biometrika 73:13-22.

Lu A, Koenig A, Borries C, 2008. Formal submission, tolerance and
socioecological models: a test with female Hanuman langurs. Animal
Behaviour 762415-428.

Lusseau D, Schneider K, Boisseau OJ, Haase P, Slooten E, Dawson SM, 2003.
The bottlenose dolphin community of Doubtful Sound features a large
proportion of long-lasting associations - Can geographic isolation explain
this unique trait? Behavioral Ecology and Sociobiology 54:396-405.

Lusseau D, Williams R, Wilson B, Grellier K, Barton TR, Hammond PS,
Thompson PM, 2004. Parallel inﬂuence of climate on the behaviour of

Paciﬁc killer whales and Atlantic bottlenose dolphins. Ecology Letters
7:1068-1076.

Manson JH, Navarrete CD, Silk JB, Perry S, 2004. Time-matched grooming in
female primates? New analyses from two species. Animal Behaviour
67:493-500.

Massey A, 1977. Agonistic aids and kinship in a group of pigtail macaques.
Behavioral Ecology and Sociobiology 2:31-40.

Matheson MD, Bernstein IS, 2000. Grooming, social bonding, and agonistic
aiding in rhesus monkeys. American Journal of Primatology 512177-186.

Matthews HL, 1939. Reproduction in the spotted hyaena, Crocuta crocuta
(Erxleben). Philosophical Transactions of the Royal Society of London B
23021-78.

Maynard Smith J, Price GR, 1973. The logic of animal conﬂict. Nature 246:15-18.

McComb K, Moss C, Sayialel S, Baker L, 2000. Unusually extensive networks of
vocal recognition in African elephants. Animal Behaviour 59:1103-1109.

218

McComb K, Reby D, Baker L, Moss C, Sayialel S, 2003. Long-distance
communication of acoustic cues to social identity in African elephants.
Animal Behaviour 65:317—329.

McCowan B, Reiss D, 2001. The fallacy of 'signature whistles‘ in bottlenose
dolphins: a comparative perspective of 'signature inforrnation' in animal
vocalizations. Animal Behaviour 62:1151-1162.

McMahan CA, Morris MD, 1984. Application of maximum-likelihood paired
comparison ranking to estimation of a linear dominance hierarchy in
animal societies. Animal Behaviour 32:374-378.

Mesterton-Gibbons M, Sherratt TN, 2007. Coalition formation: a game-theoretic
analysis. Behavioral Ecology 18:277-286.

Mills MGL, 1985. Related spotted hyaenas forage together but do not cooperate
in rearing young. Nature 316261-62.

Mills MGL, 1989. The comparative behavioral ecology of hyenas: the importance
of diet and food dispersion. In: Carnivore behavior, ecology, and evolution
(Gittleman JL, ed). Ithaca, New York: Cornell University Press; 125-142.

Mills MGL, 1990. Kalahari Hyaenas: Comparative Behavioral Ecology of Two
Species. London: Unwin Hyman.

Mills MGL, Biggs HC, 1993. Prey apportionment and related ecological
relationships between large carnivores in Kruger National Park. Symposia
of the Zoological Society of London 65:253-268.

Mitani JC, Merriwether DA, Zhang C, 2000. Male afﬁliation, cooperation and
kinship in wild chimpanzees. Animal Behaviour 59:885-893.

Mitani JC, Watts DP, Pepper JW, Merriwether DA, 2002. Demographic and
social constraints on male chimpanzee behaviour. Animal Behaviour
64:727-737.

Mundry R, Fischer J, 1998. Use of statistical programs for nonparametric tests of
small samples often leads to incorrect P values: examples from Animal
Behaviour. Animal Behaviour 56:256-259.

Murray CM, 2007. Method for assigning categorical rank in female Pan

troglodytes schweinfurthii via the frequency of approaches. lntemational
Journal of Primatology 28:853-864.

219

Neaves WB, Grifﬁn JE, Wilson JD, 1980. Sexual dimorphism of the phallus in
spotted hyaena (Crocuta crocuta). Journal of Reproduction and Fertility
59:509-513.

Nishida T, Hosaka K, 1996. Coalition strategies among adult male chimpanzees
of the Mahale Mountains, Tanzania. In: The Great Ape Societies (McGrew
W, Nishida T, Marchandt L, eds). Cambridge: Cambridge University
Press; 114-134.

Noé R, 1984. Lasting alliances among adult male savannah baboons. In: Primate
ontogeny, cognition and social behavior (Else JG, Lee PC, eds).
Cambridge: Cambridge University Press; 381-392.

Noe R, 1990. A veto game played by baboons: a challenge to the use of the
prisoners dilemma as a paradigm for reciprocity and cooperation. Animal
Behaviour 39:78-90.

Noe R, 1992. Alliance formation among male baboons: shopping for proﬁtable
partners. In: Coalitions and alliances in humans and other animals
(Harcourt AH, de Waal F BM, eds). Oxford: Oxford University Press; 284-
321.

Noe R, 2001. Biological markets: partner choice as the driving force behind the
evolution of mutualisms. In: Economics in Nature: Social Dilemmas, Mate
Choice, Biological Markets. (Noe R, van Hooff J, Hammerstein P, eds).
Cambridge: Cambridge University Press; 93-118.

Noe R, 2006. Cooperation experiments: coordination through communication
versus acting apart together. Animal Behaviour 71 :1-18.

Noe R, de Waal FBM, Vanhooff J, 1980. Types of dominance in a chimpanzee
colony. Folia Primatologica 34190-110.

Noé R, Hammerstein P, 1994. Biological markets: supply and demand determine
the effect of partner choice in cooperation, mutualism and mating.
Behavioral Ecology and Sociobiology 3521-11.

Noe R, Hammerstein P, 1995. Biological Markets. Trends in Ecology & Evolution
10:336-339.

Noé R, van l-Iooff JARAM, Hammerstein P, 2001. Economics in Nature: Social
Dilemmas, Mate Choice, Biological Markets. Cambridge: Cambridge
University Press.

Noe R, Vanschaik CP, Vanhooff J, 1991. The market effect: an explanation for
pay-off asymmetries among collaborating animals. Ethology 87297-118.

220

Nowak MA, 2006. Five rules for the evolution of cooperation. Science 314:1560-
1563.

Nowak MA, Sigmund K, 1998. Evolution of indirect reciprocity by image scoring.
Nature 393:573-577.

Nowak MA, Sigmund K, 2005. Evolution of indirect reciprocity. Nature 437:1291-
1298.

O’Brien TG, 1993. Allogrooming behavior among adult female wedge-capped
capuchin monkeys. Animal Behaviour 46:499-510.

Oindo B0, 2002. Body size and measurement of species diversity in large
grazing mammals. African Journal of Ecology 40:267-275.

Okamoto K, Agetsuma N, Kojima S, 2001. Greeting behavior during party
encounters in captive chimpanzees. Primates 42:161-165.

Olson LE, Blumstein DT, 2009. A trait-based approach to the evolution of
complex coalitions in male mammals. Behavioral Ecology 20:624-632.

OtaIi E, Gilchrist JS, 2006. Why chimpanzee (Pan troglodytes schwer'nfun‘hir)
mothers are less gregarious than nonmothers and males: the infant safety
hypothesis. Behavioral Ecology and Sociobiology 59:561-570.

Packer C, 1977. Reciprocal altruism in Papio anubis. Nature 265:441-443.

Packer C, Pusey AE, 1982. Cooperation and competition within coalitions of
male lions: kin selection or game theory? Nature 296:740-742.

Packer C, Scheel D, Pusey AE, 1990. Why lions form groups: food is not
enough. American Naturalist 13621-19.

Pan W, 2001. Akaike's information criterion in generalized estimating equations.
Biometrics 57:120-125.

Parsons KM, Durban JW, Claridge DE, Balcomb KC, Noble LR, Thompson PM,
2003. Kinship as a basis for alliance formation between male bottlenose

dolphins, Tursiops truncatus, in the Bahamas. Animal Behaviour 66:185-
194.

Pawlowski B, Dunbar RIM, 1999. Impact of market value on human mate choice
decisions. Proceedings of the Royal Society of London Series B-Biological
Sciences 266:281-285.

221

Pelchat GO, 2008. Buddies for life: male associations and coalitions in bighorn
sheep, Ovis canadensr’s (Master's of Science): University of Calgary.

Pennisi E, 2005. How did cooperative behavior evolve? Science 309:93-93.

Pereira ME, Kappeler PM, 1997. Divergent systems of agonistic behaviour in
lemurid primates. Behaviour 134:225-274.

Pereira ME, McGlynn CA, 1997. Special relationships instead of female
dominance for redfronted lemurs, Eulemur fulvus rufus. American Journal
of Primatology 43:239-258.

Perry S, 1996. Female-female social relationships in wild white-faced capuchin
monkeys, Cebus capucinus. American Journal of Primatology 40:167-182.

Perry 8, 1998. Male-male social relationships in wild white-faced capuchins,
Cebus capucinus. Behaviour 135: 1 39-172.

Perry 8, Baker M, Fedigan L, Gros-Louis J, Jack K, MacKinnon KC, Manson JH,
Panger M, Pyle K, Rose L, 2003. Social conventions in wild white-faced
capuchin monkeys - Evidence for traditions in a neotropical primate.
Current Anthropology 44:241-268.

Petit O, Abegg C, Thierry B, 1997. A comparative study of aggression and
conciliation in three cercopithecine monkeys (Macaca fuscata, Macaca
nigra, Papio papio). Behaviour 134:415-432.

Petit O, Thierry B, 1994. Aggressive and peaceful interventions in conﬂicts in
Tonkean macaques. Animal Behaviour 48:1427-1436.

Pinheiro JC, Bates DM, 2000. Mixed-effects Models in S and S-PLUS. New York:
Springer-Verlag.

Pollick AS, de Waal FBM, 2007. Ape gestures and language evolution.
Proceedings of the National Academy of Sciences of the United States of
America 104:8184-8189.

Pollick AS, Gouzoules H, De Waal FBM, 2005. Audience effects on food calls in
captive brown capuchin monkeys, Cebus ape/Ia. Animal Behaviour
70:1273-1281.

Preuschoft S, 1999. Are primates behaviorists? Formal dominance, cognition,
and free-ﬂoating rationales. Journal of Comparative Psychology 113:91-
95.

222

Preuschoft S, van Schaik CP, 2000. Dominance and communication: conﬂict
management in various social settings. In: Natural Conﬂict Resolution
(Aureli F, de Waal FBM, eds). Berkeley: University of California Press; 77-
105.

Prudhomme J, Chapais B, 1993. Aggressive interventions and matrilineal
dominance relations in semifree-ranging barbary macaques (Macaca
sylvanus). Primates 34:271-283.

Queller DC, 1985. Kinship, reciprocity and synergism in the evolution of social
behavior. Nature 318:366-367.

Ramos-Fernandez G, 2005. Vocal communication in a ﬁssion-fusion society: Do
spider monkeys stay in touch with close associates? International Journal
of Primatology 26: 1 077-1 092.

Ramos-Fernandez G, Boyer D, Gomez VP, 2006. A complex social structure with
ﬁssion-fusion properties can emerge from a simple foraging model.
Behavioral Ecology and Sociobiology 60:536-549.

Range F, 2006. Social behavior of free-ranging juvenile sooty mangabeys
(Cercocebus torquatus atys). Behavioral Ecology and Sociobiology
59:51 1-520.

Range F, Noe R, 2002. Familiarity and dominance relations among female sooty
mangabeys in the Tai National Park. American Journal of Primatology
56:137-153.

Reeve HK, 1998. Game theory, reproductive skew, and nepotism. In: Game
theory and animal behavior (Dugatkin LA, Reeve HK, eds). Oxford: Oxford
University Press; 118-145.

Reeve HK, Emlen ST, 2000. Reproductive skew and group size: an N-person
staying incentive model. Behavioral Ecology 11:640-647.

Reeve HK, Emlen ST, Keller L, 1998. Reproductive sharing in animal societies:
reproductive incentives or incomplete control by dominant breeders?
Behavioral Ecology 92267-278.

Rice WR, 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

Richter C, Mevis L, Malaivijitnond S, Schulke O, Ostner J, 2009. Social

relationships in free-ranging male Macaca arctoides. International Journal
of Primatology 30:625-642.

223

Ridley J, Yu DW, Sutherland WJ, 2005. Why long-lived species are more likely to
be social: the role of local dominance. Behavioral Ecology 16:358-363.

Rodseth L, Wrangham RW, Harrigan AM, Smuts BB, 1991. The human
community as a primate society. Current Anthropology 32:221-254.

Romero T, Aureli F, 2008. Reciprocity of support in coatis (Nasua nasua).
Journal of Comparative Psychology 122:19-25.

Rossano MJ, 2009. Ritual behaviour and the origins of modern cognition.
Cambridge Archaeological Journal 19:243-255.

Rowell TE, 1966. Hierarchy in organization of a captive baboon group. Animal
Behaviour 142430-&.

Rowell TE, Rowell CA, 1993. The social organization of feral Ovis an'es ram
groups in the pre-rut period. Ethology 95:213-232.

Sachs JL, Mueller UG, Wilcox TP, Bull JJ, 2004. The evolution of cooperation.
Quarterly Review of Biology 79:135-160.

Sachs R, 1967. Liveweights and body measurements of Serengeti game animals
of Serengeti game animals. East African Wildlife Journal 5:24-36.

Schaffner CM, Aureli F, 2005. Embraces and grooming in captive spider
monkeys. International Journal of Primatology 26:1093-1106.

Schaller GB, 1972. The Serengeti Lion. Chicago: University of Chicago Press.

Scheiber IBR, Kotrschal K, Weiss BM, 2009. Serial agonistic attacks by greylag
goose families (Anser anser) against the same opponent. Animal
Behaviour 77:1211-1216.

Scheiber IBR, Weiss BM, Frigerio D, Kotrschal K, 2005. Active and passive
social support in families of greylag geese (Anser anser’). Behaviour
142:1535-1557.

Schilder MBH, 1990. Interventions in a herd of semi-captive plains zebras.
Behaviour 112:53-83.

Schino G, 2000. Beyond the primates: expanding the reconciliation horizon. In:
Natural Conﬂict Resolution (Aureli F, de Waal F BM, eds). Berkeley:
University of California Press; 225-242.

Schino G, 2001. Grooming, competition and social rank among female primates:
a meta-analysis. Animal Behaviour 62:265-271.

224

Schino G, 2007. GrOoming and agonistic support: a meta-analysis of primate
reciprocal altruism. Behavioral Ecology 18:1 15-120.

Schino G, Aureli F, 2009. Reciprocal altruism in primates: partner choice,
cognition, and emotions. In: Advances in the Study of Behavior, Vol 39.
San Diego: Elsevier Academic Press Inc; 45-69.

Schino G, Di Giuseppe F, Visalberghi E, 2009. Grooming, rank, and agonistic
support in tufted capuchin monkeys. American Journal of Primatology
702101-105.

Schino G, Tiddi B, Polizzi di Sorrentino E, 2007. Agonistic support in juvenile
Japanese macaques: cognitive and functional implications. Ethology
113:1151-1157.

Scott DK, 1980. Functional aspects of prolonged parental care in Bewick's
swans. Animal Behaviour 28:938-952.

Semple S, Gerald MS, Suggs DN, 2009. Bystanders affect the outcome of
mother—infant interactions in rhesus macaques. Proceedings of the Royal
Society B 276:2257-2262.

Seyfarth RM, 1976. Social relationships among adult female baboons. Animal
Behaviour 24:917-938.

Seyfarth RM, 1977. Model of social grooming among adult female monkeys.
Journal of Theoretical Biology 65:671-698.

Seyfarth RM, 1980. The distribution of grooming and related behaviors among
adult female vervet monkeys. Animal Behaviour 28:798-813.

Seyfarth RM, Cheney DL, 1984. Grooming, alliances, and reciprocal altruism in
vervet monkeys. Nature 308:541-543.

Silk JB, 1982. Altruism among female Macaca radiate: Explanations and analysis
of patterns of grooming and coalition formation. Behaviour 79:162-188.

Silk JB, 1987. Social behavior in evolutionary perspective. In: Primate Societies
(Smuts BB, Cheney DL, Seyfarth RM, eds). Chicago: University of
Chicago Press; 318-329.

Silk JB, 1992. Patterns of intervention in agonsitic contests among male bonnet
macaques. In: Coaltions and Alliances in Humans and Other Animals
(Harcourt AH, De Waal FBM, eds). New York: Oxford University Press;
215-231.

225

Silk JB, 1993. Does participation in coalitions inﬂuence dominance relationships
among male bonnet macaques. Behaviour 126:171-189.

Silk JB, 2002. Kin selection in primate groups. International Journal of
Primatology 23:849-875.

Silk JB, 2007a. The adaptive value of sociality in mammalian groups.
Philosophical Transactions of the Royal Society B-Biological Sciences
362:539-559.

Silk JB, 2007b. Social components of ﬁtness in primate groups. Science
317:1347-1351.

Silk JB, Alberts SC, Altmann J, 2003. Social bonds of female baboons enhance
infant survival. Science 302:1231-1234.

Silk JB, Alberts SC, Altmann J, 2004. Patterns of coalition formation by adult
female baboons in Amboseli, Kenya. Animal Behaviour 67:573-582.

Silk JB, Altmann J, Alberts SC, 2006. Social relationships among adult female
baboons (Papio cynocephalus) I. Variation in the strength of social bonds.
Behavioral Ecology and Sociobiology 61 :183-195.

Slocombe KE, Zuberbuehler K, 2007. Chimpanzees modify recruitment screams
as a function of audience composition. Proceedings of the National
Academy of Sciences of the United States of America 104:17228-17233.

Smale L, Frank LG, Holekamp KE, 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, Holekamp KE, Weldele M, Frank LG, Glickman SE, 1995. Competition
and cooperation between litter-mates in the spotted hyaena, Crocuta
crocuta. Animal Behaviour 50:671-682.

Smale L, Nunes S, Holekamp KE, 1997. Sexually dimorphic dispersal in
mammals: patterns, causes, and consequences. Advances in the Study of
Behavior 26:181-250.

Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE, 2008. Social and
ecological determinants of ﬁssion-fusion dynamics in the spotted hyaena.
Animal Behaviour 76:619-636.

Smith JE, Memenis SK, Holekamp KE, 2007. Rank-related partner choice in the

ﬁssion-fusion society of spotted hyenas (Crocuta crocuta). Behavioral
Ecology and Sociobiology 61:753-765.

226

Smith JE, Van Horn RC, Powning KS, Cole AR, Graham KE, Memenis SK,
Holekamp KE, 2010. Evolutionary forces favoring intragroup coalitions in
the spotted hyena (Crocuta crocuta). Behavioral Ecology 21:284-303.

Smith K, Alberts SC, Altmann J, 2003. Wild female baboons bias their social
behaviour towards paternal half-sisters. Proceedings of the Royal Society
of London Series B 270:503-510.

Smolker RA, Mann J, Smuts BB, 1993. Use of signature whistles during
separations and reunions by wild bottle-nosed-dolphin mothers and
infants. Behavioral Ecology and Sociobiology 33:393-402.

Smuts B, 2002. Gestural communication in olive baboons and domestic dogs. In:
The Cognitive Animal: Empirical and Theoretical Perspectives on Animal
Cognition (Beckoff M, Allen C, Burghardt GM, eds). Cambridge: MIT
Press; 301-306.

Smuts BB, Watanabe JM, 1990. Social relationships and ritualized greetings in
adult male baboons (Papio cynocephalus anubis). International Journal of
Primatology 11:147-172.

Sommer V, 1988. Male competition and coalitions in langurs (Presbytis entellus)
at Jodhpur, Rajasthan, India. Human Evolution 3:261-278.

Spillmann B, Dunkel LP, van Noordwijk MA, N. A. Amda R, Lameira AR, Wich
SA, van Schaik CP, 2010. Acoustic properties of long calls given by
ﬂanged male orangutans (Pongo pygmaeus wurmbir) reﬂect both
Individual identity and context. Ethology 116:385-395.

Springer MS, Murphy WJ, Eizirik E, O'Brien SJ, 2003. Placental mammal
diversiﬁcation and the Cretaceous-Tertiary boundary. Proceedings of the
National Academy of Sciences of the United States of America 10021056-
1061.

Springer MS, Murphy WJ, Eizirik E, O'Brien SJ, 2005. Molecular evidence for
major placental clades. In: The Rise of Placental Mammals (Rose KD,
Archibald JD, eds). Baltimore: The John Hopkins University Press; 37-49.

Stammbach E, 1978. Social differentiation in groups of captive female
hamadryas baboons. Behaviour 67:322-338.

Sterck EHM, Watts DP, van Schaik CP, 1997a. The evolution of female social

relationships in nonhuman primates. Behavioral Ecology and Sociobiology
41:291-309.

227

Sterck EHM, Watts DP, vanSchaik CP, 1997b. The evolution of female social
relationships in nonhuman primates. Behavioral Ecology and Sociobiology
41:291-309.

Stevens JMG, Vervaecke H, de Vries H, Van Elsacker L, 2005. The inﬂuence of
the steepness of dominance hierarchies on reciprocity and interchange in
captive groups of bonobos (Pan paniscus). Behaviour 142:941-960.

Symington MM, 1990. Fission-fusion social organization in Ate/es and Pan.
International Journal of Primatology 11:47-61.

Szykman M, Engh AL, Van Horn RC, Funk SM, Scribner KT, Holekamp KE,
2001. Association patterns among male and female spotted hyenas
(Crocuta crocuta) reﬂect male mate choice. Behavioral Ecology and
Sociobiology 50:231-238.

Szykman M, Van Horn RC, Engh AL, Boydston EE, Holekamp KE, 2007.
Courtship and mating in free-living spotted hyenas. Behaviour 144:815-
846.

Theis KR, 2008. Scent marking in a highly social mammalian species, the
spotted hyena, Crocuta crocuta (Dissertation). East Lansing: Michigan
State University.

Theis KR, Greene KM, Benson-Amram SR, Holekamp KE, 2007. Sources of
variation in the long-distance vocalizations of spotted hyenas. Behaviour
144:557-584.

Tilson RT, Hamilton WJI, 1984. Social dominance and feeding patterns of
spotted hyaenas. Animal Behaviour 32:715-724.

Trinkel M, Fleischmann PH, Steindorfer AF, Kastberger G, 2004. Spotted hyenas
(Crocuta crocuta) follow migratory prey. Seasonal expansion of a clan
territory in Etosha, Namibia. J Zool, Lond 264:125-133.

Trivers RL, 1971. The evolution of reciprocal altruism. Quarterly Review of
Biology 46:35-57.

Uehara S, Hiraiwahasegawa M, Hosaka K, Hamai M, 1994. The fate of defeated
alpha male chimpanzees in relation to their social networks. Primates
35:49-55.

Valero A, Schaffner CM, Vick LG, Aureli F, Ramos-Femandez G, 2006.

lntragroup lethal aggression in wild spider monkeys. American Journal of
Primatology 68:732-737.

228

 

Val-Laillet D, Guesdon V, von Keyserlingk MAG, de Passille AM, Rushen J,
2009. Allogrooming in cattle: Relationships between social preferences,
feeding displacements and social dominance. Applied Animal Behaviour
Science 116:141-149.

van Hooff JARAM, Wensing JAB, 1987. Dominance and its behavioral measures
in a captive wolf pack. In: Man and Wolf: Advances, issues and problems
in captive wolf research (Frank H, ed). Dordrecht, The Netherlands: Dr. W.
Junk Publishers; 219-252.

Van Horn RC, Engh AL, Scribner KT, Funk SM, Holekamp KE, 2004a.
Behavioural structuring of relatedness in the spotted hyena (Crocuta
crocuta) suggests direct ﬁtness beneﬁts of clan-level cooperation.
Molecular Ecology 13:449-458.

Van Horn RC, McElhinny TL, Holekamp KE, 2003. Age estimation and dispersal
in the spotted hyena (Crocuta crocuta). Journal of Mammalogy 84:1019-
1030.

Van Horn RC, Wahaj SA, Holekamp KE, 2004b. Role-reversed nepotism among
cubs and sires in the spotted hyena (Crocuta crocuta). Ethology 1102413-
426.

Van Meter PE, 2009. Hormones, stress and aggression in the spotted hyena,
Crocuta crocuta (Ph.D. Dissertation). East Lansing: Michigan State
University.

Van Meter PE, French JA, Dloniak SM, Watts HE, Kolowski JM, Holekamp KE,
2009. Fecal glucocorticoids reﬂect socio-ecological and anthropogenic

stressors in the lives of wild spotted hyenas. Hormones and Behavior
55:329-337.

van Schaik CP, 1999. The socioecology of ﬁssion-fusion sociality in orangutans.
Primates 40:69-86.

Vehrencamp SL, 1983. A model for the evolution of despotic versus egalitarian
societies. Animal Behaviour 31:667-682.

Vervaecke H, De Vries H, Van Elsacker L, 1999. An experimental evaluation of
the consistency of competitive ability and agonistic dominance in different
social contexts in captive bonobos. Behaviour 136:423-442.

Vervaecke H, de Vries H, Van Elsacker L, 2000. Function and distribution of
coalitions in captive bonobos (Pan paniscus). Primates 41:249-265.

229

Vervaecke H, Roden C, De Vries H, 2005. Dominance, fatness and ﬁtness in
female American bison, Bison bison. Animal Behaviour 70:763-770.

Vogel ER, Munch SB, Janson CH, 2007. Understanding escalated aggression
over food resources in white-faced capuchin monkeys. Animal Behaviour
74:71-80.

Wagner AP, 2006. Behavioral ecology of the striped hyaena (Hyaena hyaena).
(Ph.D.). Bozeman, Montana: Montana State University.

Wahaj SA, Guse KR, Holekamp KE, 2001. Reconciliation in the spotted hyena
(Crocuta crocuta). Ethology 10721057-1074.

Wahaj SA, Van Horn RC, Van Horn TL, Dreyer R, Hilgris R, Schwarz J,
Holekamp KE, 2004. Kin discrimination in the spotted hyena (Crocuta
crocuta): nepotism among siblings. Behavioral Ecology and Sociobiology
56:237-247.

Walters J, 1980. Interventions and the development of dominance relationships
in female baboons. Folia Primatologica 34:61-89.

Wang E, Milton K, 2003. lntragroup social relationships of male Alouatta palliata
on Barro Colorado Island, Republic of Panama. lntemational Journal of
Primatology 24:1227-1243.

Wasser SK, Barash DP, 1983. Reproductive suppression among female
mammals: implications for biomedicine and sexual selection theory.
Quarterly Review of Biology 58:513-538.

Wasserman S, Faust K, 1994. Social Network Analysis. Cambridge: Cambridge
University Press.

Watts DP, 1997. Agonsitic interventions in wild mountain gorilla groups.
Behaviour 134:23-57. .

Watts DP, 2002. Reciprocity and interchange in the social relationships of wild
male chimpanzees. Behaviour 139:343-370.

Watts HE, 2007. Social and ecological inﬂuences on survival and reproduction in
spotted hyenas, Crocuta crocuta (Ph.D.). East Lansing: Michigan State
University.

Wechsler B, 1988. Dominance relationships in jackdaws (Corvus monedula).
Behaviour 106:252-264.

230

West SA, Grifﬁn AS, Gardner A, 2007a. Evolutionary explanations for
cooperation. Current Biology 17:R661-R672.

West SA, Grifﬁn AS, Gardner A, 2007b. Social semantics: altruism, cooperation,
mutualism, strong reciprocity and group selection. Journal of Evolutionary
Biology 20:415-432.

West-Eberhard MJ, 1975. The evolution of social behavior by kin selection.
Quarterly Review of Biology 5021-33.

Wey T, Blumstein DT, Weiwei S, Jordan F, 2008. Social network analysis of
animal behaviour: a promising tool for the study of sociality. Animal
Behaviour 75:333-344.

Whateley A, 1980. Comparative body measurements of male and female spotted
hyaenas from Natal. Lammergeyer 28:40-43.

White PA, 2005. Maternal rank is not correlated with cub survival in the spotted
hyena, Crocuta crocuta. Behavioral Ecology 16:606-613.

White PA, 2007. Costs and strategies of communal den use vary by rank for
spotted hyaenas, Crocuta crocuta. Animal Behaviour 73:149-156.

Whitehead H, 2008. Analyzing Animal Societies. Chicago: University of Chicago
Press.

Whitham JC, Maestripieri D, 2003. Primate rituals: The function of greetings
between male Guinea baboons. Ethology 109:847-859.

Widdig A, 2007. Paternal kin discrimination: the evidence and likely mechanisms.
Biological Reviews 82:319-334.

Widdig A, Niirnberg P, Krawczak M, Streich WJ, Bercovitch F, 2002. Afﬁliation
and aggression among adult female rhesus macaques: a genetic analysis
of paternal cohorts. Behaviour 139:371-391.

Widdig A, Streich WJ, Nuernberg P, Croucher PJP, Bercovitch FB, Krawczak M,
2006. Paternal kin bias in the agonistic interventions of adult female

rhesus macaques (Macaca mulatta). Behavioral Ecology and Sociobiology
61:205-214.

Widdig A, Streich WJ, Tembrock G, 2000. Coalition formation among male

Barbary macaques (Macaca sylvanus). American Journal of Primatology
50:37-51.

231

Williams RL, 2000. A note on robust variance estimation for cluster-correlated
data. Biometrics 56:645-646.

Wilson DS, 1975a. Theory of group selection. Proceedings of the National
Academy of Sciences of the United States of America 72:143-146.

Wilson DS, Wilson EO, 2007. Rethinking the theoretical foundation of
sociobiology. Quarterly Review of Biology 82:327-348.

Wilson EO, 1975b. Sociobiology: The New Synthesis. Cambridge: Belknap Press
of Harvard University Press.

Wittemyer G, Douglas-Hamilton l, Getz WM, 2005. The socioecology of
elephants: analysis of the processes creating multitiered social structures.
Animal Behaviour 69:1357-1371.

Woodroffe R, Macdonald DW, 1993. Badger sociality - models of spatial
grouping. Symposia of the Zoological Society of London 652145-169.

Wrangham RW, 1980. An ecological model of female-bonded primate groups.
Behaviour 75:262-300.

Wrangham RW, Conklin-Brittain NL, Hunt KD, 1998. Dietary response of
chimpanzees and cercopithecines to seasonal variation in fruit
abundance. l. Antifeedants. International Journal of Primatology 19:949-
970.

Wrangham RW, Gittleman JL, Chapman CA, 1993. Constraints on group size in
primates and carnivores: population density and day-range as assays of
exploitation competition. Behavioral Ecology and Sociobiology 32:199-
209.

Zabel CJ, Glickman SE, Frank LG, Woodmansee KB, Keppel G, 1992. Coalition
formation in a colony of prepubertal spotted hyenas. In: Coalitions and
alliances in humans and other animals (Harcourt AH, de Waal FBM, eds).
Oxford: Oxford University Press; 112-135.

Zahavi A, 1977a. The cost of honesty (further remarks on the handicap principle).
Journal of Theoretical Biology 67:603-605.

Zahavi A, 1977b. Testing of a bond. Animal Behaviour 25:246-247.

Zahavi A, 1980. Ritualization and the evolution of movement signals. Behaviour
72:77-81.

232

Zimen E, 1976. On the regulation of pack size in wolves. Zeitschrift fi'Ir
Tierpsychologie 40:300-341.

Zuberbtihler K, 2008. Audience effects. Current Biology 18:R189-R190.

233