Convergent morphological evolution in hyenas and dogs: the relationship between form and function

PDF

Download

Open document

Transcript (if available)

Content CONVERGENT MORPHOLOGICAL EVOLUTION IN HYENAS AND DOGS:
THE RELATIONSHIP BETWEEN FORM AND FUNCTION

by

Jack Tseng

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(INTEGRATIVE AND EVOLUTIONARY BIOLOGY)

December 2011

Copyright 2011 Jack Tseng
ii
Epigraph

Antiquis auctoribus suus constat honos, atque adeo omnibus; quia non ingeniorum aut
facultatum inducitur comparatio, sed viae; nosque non judicis, sed indicis personam
sustinemus.

-Francis Bacon, Novum Organum (1620)

(Rough translation: The honor of the ancient authors, and of all others, are evident; the
comparison I make is not of wits or faculties, but of the methods; I am not a judge, but a
guide.)
iii
Dedication

To my family.
iv
Acknowledgments

For the nurture and intellectual enlightenment so generously provided to me, I
acknowledge my Alma Mater the University of California Berkeley for the guiding light;
my most sincere gratitude to the University of Southern California and Natural History
Museum of Los Angeles County, for helping me take my baby steps; my PhD advisors
Xiaoming Wang and Jill McNitt-Gray for their patience, guidance, and unrelenting
support; my committee members Dave Bottjer, Henryk Flashner, and Blaire Van
Valkenburgh for guidance and discussion; Gary Takeuchi for teaching me to excel in
fieldwork; my wife Juan Liu for her unwavering love and patience; and finally, to my
mother and sister for being the strongest pillars of my life. Fiat Lux.
v
Table of Contents

Epigraph
Dedication
Acknowledgments
List of Tables
List of Figures
Abstract
Preface
Introduction
Method Overview
Introduction References

Chapter One: Evolution of form in hyenas and dogs
Chapter One Abstract
Chapter One Introduction
Chapter One Materials and Methods
Chapter One Results
Chapter One Discussion
Chapter One Conclusion
Chapter One Acknowledgments
Chapter One References

Chapter Two: Finite Element Analysis as a method to study function
Chapter Two Abstract
Chapter Two Introduction
Chapter Two Materials and Methods
Chapter Two Results
Chapter Two Discussion
Chapter Two Conclusion
Chapter Two Acknowledgments
Chapter Two References

ii
iii
iv
ix
xiv
xxvii
xxviii
1
9
16
20
22
24
29
32
35
50
51
53
71
73
75
78
85
88
95
96
98
vi
Chapter Three: Study of function in the hyena Chasmaporthetes lunensis
Chapter Three Abstract
Chapter Three Introduction
Chapter Three Materials and Methods
Chapter Three Results
Chapter Three Discussion
Chapter Three Conclusion
Chapter Three Acknowledgments
Chapter Three References
Chapter Three Appendix Table 1
Chapter Three Appendix Table 2
Chapter Three Appendix Table 3

Chapter Four: Study of function in the hyena Ikelohyaena abronia
Chapter Four Abstract
Chapter Four Introduction
Chapter Four Materials and Methods
Chapter Four Results
Chapter Four Discussion
Chapter Four Conclusion
Chapter Four Acknowledgments
Chapter Four References

Chapter Five: Study of function in fossil dogs
Chapter Five Abstract
Chapter Five Introduction
Chapter Five Materials and Methods
Chapter Five Results
Chapter Five Discussion
Chapter Five Conclusion
Chapter Five Acknowledgments
Chapter Five References

Chapter Six: Study of cranial function in the percrocutid Dinocrocuta
gigantea
Chapter Six Abstract
Chapter Six Introduction
Chapter Six Materials and Methods
Chapter Six Results
Chapter Six Discussion
Chapter Six Conclusion
Chapter Six Acknowledgments
Chapter Six References
Chapter Six Appendix
112
114
116
121
131
133
142
143
144
148
149
150
161
163
165
171
181
187
199
201
202
218
220
222
225
232
234
243
244
246
262
264
265
268
277
280
289
290
291
296
vii

Chapter Seven: Study of mandibular function in the percrocutid
Dinocrocuta gigantea
Chapter Seven Abstract
Chapter Seven Introduction
Chapter Seven Methods
Chapter Seven Results
Chapter Seven Discussion
Chapter Seven Conclusion
Chapter Seven Aknowledgments
Chapter Seven References
Chapter Seven Appendix A

Chapter Eight: An integrative framework for the study of form and function
Chapter Eight Abstract
Chapter Eight Introduction
Chapter Eight Materials and Methods
Chapter Eight Results
Chapter Eight Discussion
Chapter Eight Acknowledgments
Chapter Eight References
Chapter Eight Appendix 1
Chapter Eight Appendix 2
Chapter Eight Appendix 3

Chapter Nine: Enamel microstructure versus macrostructural evolution
Chapter Nine Introduction
Chapter Nine Materials and Methods
Chapter Nine Results
Chapter Nine Discussion
Chapter Nine Acknowledgments
Chapter Nine References
Chapter Nine Appendix

Chapter Ten: Microstructural evolution and paleodiet indicators
Chapter Ten Abstract
Chapter Ten Introduction
Chapter Ten Material and Methods
Chapter Ten Results
Chapter Ten Discussion
Chapter Ten Acknowledgments
Chapter Ten References
Chapter Ten Appendix 1

307
309
310
315
319
323
330
331
333
337
348
350
352
359
369
371
380
382
389
390
392
403
405
408
409
410
416
417
420
425
426
428
430
433
435
441
442
447
viii
Chapter Ten Appendix 2
Chapter Ten Appendix 3

Chapter Eleven: Conclusion and Synthesis
Chapter Eleven Abstract
Chapter Eleven Introduction
Chapter Eleven Acknowledgments
Chapter Eleven References

Comprehensive References

Appendices
Appendix A. Supplementary Information from Chapter One
Appendix B. Supplementary Information from Chapter Two
Appendix C. Supplementary Information from Chapter Four
Appendix D. Supplementary Information from Chapter Five
Appendix E. A compilation of notes taken in creating the FEA
protocol
Appendix F. Notes taken during dissection of Hyaena hyaena at
LACM and Crocuta crocuta at Berkeley
Appendix G. Curriculum Vitae as of November 2011

450
454
465
467
469
490
491
505
532
537
541
546
551
570
577

ix
List of Tables
Table 1.1:

Table 1.2:

Table 1.3:

Table 1.4:

Table 2.1:

Table 2.2:

Table 3.1:

Table 3.2:

Table 3.3:

Carnivoran species used to represent the East African and North
American modern carnivore guilds, respectively. Lists are in
alphabetical order by family then genus.

List of fossil canids, hyaenids, and stem fossil forms used in the
study. List for Canidae and Hyaenidae are in order of geological
and evolutionary appearance.

Anatomical landmarks used in the respective geometric
morphometrics analyses. See Figure 1.2 for illustration.

Tests for allometry: regression statistics for cranial shape
variables versus ln(Centroid Size). All partial warp (PW) and
uniform component scores of fossil hyaenid and borophagine
specimens were regressed against ln(CS). Phylogenetic
independent contrasts (PIC) for average relative warp scores for
each genus as presented in Figures4–6 were regressed (through
the origin) against PICs for ln(CS). Significant correlation
between dorsal shape and ln(CS) in borophagines (p-value in
bold) is associated with narrower/more tubular dorsal skulls in
larger specimens.

Sensitivity tests performed in this study.

Maximum % changes in bite force and strain energy in the
sensitivity tests.

Sensitivity tests altering the values of Young’s modulus (E) and
Poisson’s Ratio of a Crocuta crocuta model (LACM 30655;
Tseng 2009). GPa, gigapascal; MPa, megapascal; μE,
microstrain.

Summary values for analysis 1: 1000 N muscle input force at
third and fourth premolars. For the original Crocuta model, first
value is from the homogeneous model, second from the
heterogeneous model.

Summary values for analysis 2: All models produced 1180.592
N of bite force at LP3. Data for the original C. crocuta model
are from the homogeneous/heterogeneous models, respectively.

67
68
69
70
110
111
155
156
157
x
Table 3.4:

Table 3.5:

Table 3.6:

Table 4.1:

Table 4.2:

Table 4.3:

Table 5.1:

Summary values for analysis 3: All models produced sufficient
bite force at LP3 to break bovine femur with strength of 150
MPa. Data for the original C. crocuta model are from the
homogeneous/heterogeneous models, respectively.

Summary values for analysis 4: bite forces at LP3 used to
calculate stress slope. Data for the original C. crocuta model are
from the homogeneous/heterogeneous models, respectively.

Bone-cracking index calculated for the finite element models.
For explanation of calculations see methods. Abbreviations:
homo., homogeneous model of the original Crocuta skull;
hetero., heterogeneous model of the original skull.

Carnivoran species examined in this study, their body masses,
and principal diet. Body mass estimates for Ikelohyaena were
calculated using skeletal regression models of different
elements as indicated.

Results of finite element analysis of mastication forces. Up, bite
force of upper cusp in the respective scenario (in Newtons);
L.A., anterior lower cusp bite force; L.P., posterior lower cusp
bite force; MA, mechanical advantage; Adj, adjusted total
strain energy (in Joules); Cran/%, cranial strain energy and
percentage of total strain energy; Mand/%, mandibular strain
energy and percentage of total. Percentage values after bite
forces are fractions of total muscle input force.

Results of finite element analysis of prey apprehension forces.
Extrinsic forces were scaled to muscle input area in each model.
En, cranial strain energy; J, Joules; 1, scenario 1, “no TMJ
rotation”; 2, scenario 2, “TMJ rotation with canine movement”;
3, scenario 3, “TMJ rotation with fixed canines”; LC, left
canine; RC, right canine; N, newtons; MA, mechanical
advantage.

Finite element model parameters. For institutional abbreviations
see main text. Input F, applied total muscle force in Newtons;
SA, total surface area; V, total volume.

158
159
160
215
216
217
257
xi
Table 5.2:

Table 5.3:

Table 5.4:

Table 5.5:

Table 6.1:

Table 6.2:

Measures of performance obtained from finite element analysis
of the six skull models. Muscle force input ratio used was 64%
temporalis, 26% masseter, and 10% pterygoid (Turnbull, 1970).
Total forces were scaled to surface area in each model (Dumont
et al. 2009). En, adjusted total strain energy (in Joules); F
o/i
,
ratio of bite force to total muscle input force.

Von Mises stresses at point-sampling positions along the dorsal
mid-sagittal line of the skulls for a unilateral P4 bite using wolf
muscle ratios (64% temporalis-26% masster-10% pterygoid).
All values are in megapascals. Sampling points: 1. rostral point
of the nasal bone; 2. rostrum at infraorbital foramina; 3. rostrum
at anterior border of orbits; 4. frontal regions between
postorbital processes; 5. frontal region at the postorbital
constriction; 6. center of the parietal bone; 7. caudal-most
region of the sagittal crest.

Bite force production efficiency (ratio between bite force and
muscle input force; F
o/i
), energy of deformation (adjusted total
strain energy, En, in Joules), and their percent changes when
models are loaded with a temporalis- or a masseter-driven bite,
respectively.

Average cross-sectional areas of masticatory muscle groups
estimated using Thomason’s (1991) dry skull method.

Descriptive statistics of Von Mises stress in the Canis lupus
finite element model. Data shown are for both the entire
cranium as well as the frontal region. Scenarios tested include
third premolar (P3), fourth premolar (P4), and first molar (M1)
biting. Both right and left side unilateral loading cases were
analyzed. IQR=interquartile range; MAD=median absolute
deviation from the median; MPa=megapascal; SE=standard
error of the median. For definition of the terms see text.

Descriptive statistics of Von Mises stress in the Dinocrocuta
gigantea finite element model. Data shown are for both the
entire cranium as well as the frontal region. Scenarios tested
include third premolar (P3) and fourth premolar (P4) biting.
Both right and left side unilateral loading cases were analyzed.
Abbreviations as in Table 6.1. For definition of the terms see
text.

258
259
260
261
302
303
xii
Table 6.3:

Table 6.4:

Table 6.5:

Table 7.1:

Table 7.2:

Table 7.3:

Table 8.1:

Descriptive statistics of Von Mises stress in the Crocuta
crocuta finite element model. Data shown are for both the entire
cranium as well as the frontal region. Scenarios tested include
third premolar (P3) and fourth premolar (P4) biting. Both right
and left side unilateral loading cases were analyzed.
Abbreviations as in Table 6.1. For definition of the terms see
text.

Comparative statistics of Von Mises stress in the finite element
models of the Crocuta crocuta, Dinocrocuta gigantea, and
Canis lupus skulls. The comparative data are for left P3
unilateral biting. Abbreviations as in Table 6.1. For definitions
of descriptive statistics see text.

Comparative statistics of Von Mises stress in the finite element
models of the Crocuta crocuta, Dinocrocuta gigantea, and
Canis lupus skulls. The comparative data are for left P4
unilateral biting. Abbreviations as in Table 6.1. For definitions
of descriptive statistics see text.

Von Mises stress and strain values and bite forces from
unmodified finite element models of Crocuta crocuta and Canis
lupus with uniform muscle force input of 5891.63 N. Stresses
are in Mpa, strain in με, and bite force is in Newtons. For
abbreviations see text.

Von Mises stress values for the solid finite element models of
C. crocuta, D. gigantea, and C. lupus with an uniform muscle
force input of 5891.63 N. Percent changes of values compared
to the original models are listed for the extant taxa. Median,
MAD, and IQR of the stresses are in MPa; bite force is in
Newtons. For abbreviations see text.

Von Mises stress values for the filled and scaled finite element
models of C. crocuta, D. gigantea, and C. lupus. All models are
analyzed with a bite force of 1000N at the respective bite
positions.

List of actual models and species measurements of hyaenids
and canids used in the study. For list of specimen numbers see
Chapter Eight Appendix 1.
†
extinct taxon.

304
305
306
345
346
347
401
xiii
Table 8.2:

Table 10.1:

Table 11.1:

Table 11.2:

List of modern North American and East Africa carnivoran
species used to construct contour of carnivoran distribution. For
specimen lists see Tseng and Wang (2011).

ANOVA of microwear features according to categories of HSB
specialization. Statistical testing was done on binned category
data (as in Fig. 10.6).

Adaptive characteristics of the modern spotted hyena Crocuta
crocuta, constituting the definition of a specialized bone-
cracking ecomorph.

Modified categories representing degrees of bone-cracking
specialization in hypercarnivorous ecomorphs. Data come from
studies on hyaenid and borophagine canids.
402
464
503
504

xiv
List of Figures
Figure 1.1:

Figure 1.2:

Figure 1.3:

Figure 1.4:

Figure 1.5:

Phylogenies and cranial morphological diversity of Canidae (A)
and Hyaenidae (B). Data for canid phylogeny are from Wang
(1994), Wang et al. (1999), and Tedford et al. (2009) and for
hyaenids from Werdelin and Solounias (1991) and Koepfli et al.
(2006). Genera used in this paper are labeled in black.

Dorsal (A), lateral (B), and ventral (C) anatomical landmarks
used in geometric morphometrics analysis. See Table 1.3 for
explanation of the landmarks.

Relative warp plots of fossil canids and hyaenids grouped with
extant North American (A–C) and East African (D–F)
carnivoran faunas, respectively. The first two relative warp axes
are displayed. Plots show dorsal (A,D), lateral (B,E), and
ventral (C,F) cranial landmark analyses. Shaded polygons
indicate areas of morphospace occupied by borophagine canids
(A–C) and hyaenids (D–F).

Plots of relative warp scores and shape evolution in
borophagine canids and hyaenids, dorsal cranial landmarks. A,
RW1 versus RW2 scores. B, Phylogenies plotted over RW1
versus RW2 scores. C, Deformation grids showing shape
changes represented by RW1 and RW2 axes.

Plots of relative warp scores and shape evolution in
borophagine canids and hyaenids, lateral cranial landmarks. A,
RW1 versus RW2 scores. B, Phylogenies plotted over RW1
versus RW2 scores. C, Deformation grids showing shape
changes represented by RW1 and RW2 axes. D, RW1 versus
RW3 scores. E, Phylogenies plotted over RW1 versus RW3
scores. F, Deformation grids showing shape changes
represented by RW1 and RW3 axes. Symbols and abbreviations
as in Figure 1.3. Note parallel evolutionary changes in
borophagine canids and hyaenids toward more positive RW1
scores.

59
60
61
62
63
xv
Figure 1.6:

Figure 1.7:

Figure 1.8:

Plots of relative warp scores and shape evolution in
borophagine canids and hyaenids, ventral cranial landmarks. A,
RW1 versus RW2 scores. B, Phylogenies plotted over RW1
versus RW2 scores. C, Deformation grids showing shape
changes represented by RW1 and RW2 axes. D, RW1 versus
RW3 scores. E, Phylogenies plotted over RW1 versus RW3
scores. F, Deformation grids showing shape changes
represented by RW1 and RW3 axes. Symbols and abbreviations
as in Figure 1.3. Note lack of overlap between borophagine
canids and hyaenids in morphospace.

Evolutionary changes in cranial shape in borophagine canids
(A–C) and hyaenids (D–F). Plots show dorsal (A,D), lateral
(B,E), and ventral (C,F) cranial landmarks. Arrows indicate
directions of evolutionary change. Illustrations in central
column depict location of landmarks on the skull. Dorsal and
ventral landmarks mirrored about mid-sagittal axis to aid
interpretation.

Expectations of patterns of convergence between two lineages
through a two-dimensional morphospace (A–C), and
observations made from the current study on borophagine
canids and hyaenids (D–F). A, Two lineages starting with
different morphologies evolved forms that converge in
morphospace (arrows indicate directions of morphological
evolution). B, Two lineages evolved toward each other in
morphology, but do not overlap. C, Two lineages exhibit
parallel pathways of morphological change and do not converge
on each other in morphology. Results from this study are
depicted for dorsal (D), lateral (E), and ventral (F) cranial
shapes. The axes represent relative evolutionary differences
between lineages in morphospace, and not absolute time. A–C
modified from Stayton (2006). For D–F, squares represent stem
borophagines, and circles represent stem hyaenids. In dorsal
cranial shape, stem borophagines and hyaenids shared similar
positions in morphospace, but hyaenids evolved a further
reduced rostrum. In lateral cranial shape, all observed
evolutionary changes in borophagines and hyaenids were
parallel. In ventral shape, hyaenids and borophagines exhibited
parallel changes of different extent, and derived borophagines
overlapped with stem hyaenids.

64
65
66
xvi
Figure 2.1:

Figure 2.2:

Figure 2.3:

Figure 2.4:

Figure 2.5:

Figure 2.6:

Mandible model used in the study. Bal., balancing-side joint;
Work., working-side joint; m1, lower first molar (carnassial);
M.p., deep masseter; M.s., superficial masseter; P.i., internal
pterygoid; T.p., deep temporalis; T.s., superficial temporalis;
T.z., zygomatic part of temporalis; Z.m.,
zygomaticomandibularis. Temporalis and masseter muscle
subgroups were used incrementally in the sensitivity test on
number of muscles. All other models used a four-muscle input:
temporalis-masseter-zygomaticomandibularis-pterygoid.

Sensitivity test on tetrahedral element quantity. A. Element
quantity plotted against solution time (in seconds), with
exponential curve in background. B. Element quantity plotted
against reaction force (in Newtons). C. Element quantity plotted
against strain energy (in Joules), with linear regression line. D.
von Mises stress distribution in the working-side dentary in test
models; lateral view (in Megapascals).

Sensitivity test on balancing-working side ratio. A. Ratio
plotted against reaction force, with second-order polynomial
curves fitted onto the working and balancing reaction forces. B.
Ratio plotted against strain energy. C. von Mises stress
distribution in the working-side dentary in test models.

Sensitivity test on musculature ratio. A. Ratio plotted against
reaction force. B. Ratio plotted against strain energy. C. von
Mises stress distribution in the working-side dentary in test
models. Ratios are given by temporalis-masseter-pterygoid
sequences, with zygomaticomandibularis considered part of the
masseter group.

Sensitivity test on number of muscle groups. A. Number of
groups plotted against reaction force, connected by lines to
show trend. B. Number of groups plotted against strain energy.
C. von Mises stress distribution in the working-side dentary in
test models.

Sensitivity test on nodes at the bite point constraint. A. Nodal
constraints plotted against reaction force. B. Nodal constraints
plotted against reaction force, showing components of the bite
force vector. C. Nodal constraints plotted against strain energy.
D. von Mises stress distribution in the working-side dentary in
test models.

102
103
104
105
106
107
xvii
Figure 2.7:

Figure 2.8:

Figure 3.1:

Figure 3.2:

Figure 3.3:

Figure 3.4:

Sensitivity test on temporomandibular joint constraint. A.
Constraint type plotted against reaction force. B. Constraint
type plotted against strain energy. C. Constraint type plotted
against von Mises strain, showing mean and maximum strain
for the working- and balancing-side joints, respectively. D. von
Mises stress distribution in the working-side dentary in test
models.

Sensitivity test on number of material properties. A. Number of
properties plotted against reaction force, connected by lines to
show trend. B. Number of properties plotted against strain
energy. C. von Mises stress distribution in the working-side
dentary in test models.

Illustration of the lateral view of Crocuta crocuta skull, with the
3
rd
and 4
th
premolar positions marked. Dark gray represents the
area marked as temporalis in the models, light gray represents
masseter.

Sensitivity test of a single finite element cranium model
(LACM30655; Tseng 2009) with different Poisson’s ratios.
Circles indicate stress values, squares are strain values. Stress is
measured in megapascals, strain in microstrain.

Ventral (A-D) and dorsal (E-H) view of the Von Mises stresses
in the finite element models loaded with 1000 N of total muscle
input force. A, E, Crocuta crocuta heterogeneous original
model; B, F, C. crocuta homogeneous original model; C, G, C.
crocuta with bone struts removed; D, H, Chasmaporthetes
lunensis. Warmer and brighter colors indicate high stress;
colder and darker indicate low stress.

The median stress values at different bite force levels for the
original homogeneous Crocuta crocuta (diamond),
Chasmaporthetes lunensis (circle), Crocuta crocuta with bony
struts removed (square), and the heterogeneous Crocuta model
(triangle). For data values see Table 3.5.

108
109
151
152
153
154
xviii
Figure 4.1:

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

Figure 5.1:

Specimens used for finite element modelling. Lateral view. (A)
Crocuta crocuta, (B) Ikelohyaena abronia, (C) Lycaon pictus,
(D) Panthera pardus. Abbreviations: bc, braincase; cf,
condyloid fossa; fs, frontal sinus; mc, mandibular condyle; pa,
paracone of upper premolar 4; pad, paraconid of lower molar 1;
prd, protoconid of lower molar 1; I3, incisor 3; C, canine; P,
premolar; M, molar. Circled numbers represent landmarks used
in principal stress analysis. All scale bars equal 20 mm.

Distribution of von Mises stress for a simulated unilateral bite
at the left third premolar in rostro-lateral (A-D) and dorsal (E-
H) views. (A, E) Crocuta crocuta, (B, F) Ikelohyaena abronia,
(C, G) Lycaon pictus, (D, H) Panthera pardus. Hotter colours
indicate higher stress.

Distribution of von Mises stress for simulated prey
apprehension forces “pull back” (A-D) and “lateral shake” (E-
H). (A, E) Crocuta crocuta, (B, F) Ikelohyaena abronia, (C, G)
Lycaon pictus, (D, H) Panthera pardus. Hotter colours indicate
higher stress.

Average maximum and minimum principal stresses (in
megapascals, MPa) along dorsal cranial landmarks in
simulations of mastication (A, B) and prey apprehension (C, D)
forces. (A) Left P3 biting, (B) Left P4 biting, (C) “Pull back”,
(D) “lateral shake” simulations. Maximum principal (tensile)
stresses have positive values; minimum principal (compressive)
stresses have negative values. For explanation of landmarks see
methods and Figure 4.1.

Results of sensitivity tests of symphyseal material properties.
(A) Non-stimulated side canine bite force as a percentage of
stimulated side canine bite force, (B) Symphyseal strain energy
(in Joules), (C) Maximum symphyseal von Mises stress (in
megapascals), (D) Maximum symphyseal displacement (in
mm). For material properties of different tissue types see Table
S3.

Cranial size and shape diversity of representative forms in the
Canidae. Genera studied are highlighted in black font. All
genera have a fossil record, but all extant canids are in Caninae.
Scale bar (next to Epicyon) equals 50 mm. Stratigraphic ranges
taken from Wang (1994), Wang et al. (2008), Wang et al.
(1999).
210
211
212
213
214
251
xix
Figure 5.2:

Figure 5.3:

Figure 5.4:

Figure 5.5:

Figure 5.6:

Comparison of Von Mises (VM) stress distribution in the
heterogeneous (A-C) and homogeneous (D-F) models of Canis
lupus. A,D: dorsal views; B,E: mid-sagittal section views; C,F:
ventral views. All stress distributions are scaled between 0 and
100 Megapascals (MPa). Hotter/lighter colors represent higher
stress, Cooler/darker colors represent lower stress.

Maximum Von Mises stress (in megapascal, MPa) along the
dorsal mid-sagittal line of the skull models from simulated P4
biting with 64%-26%-10% muscle ratios. Analogous
anatomical sampling points are plotted as percentages of total
skull length as measured from dorsal view, showing relative
proportions of the cranial bones. Note consistently low
maximum stress across the dorsal crania of Borophagus and
Epicyon (bold lines). See Methods for list of sampling points.

Percent increases in bite force production efficiency (F
o/i
, dark
bars) and amount of deformation (total strain energy, light bars)
compared to those obtained using modern muscle ratios. All
models were biting with muscle input force generated solely by
the temporalis muscles (highlighted on Canis model). Symbols
as in Figure 3. Note the small amount of change in deformation
required to substantially increase bite force efficiency in
Epicyon haydeni (white circle) and Borophagus secundus
(black circle).

Comparisons of Von Mises stress in the skull models during a
temporalis-dominated unilateral bite with the left carnassial
tooth (P4). A. Hesperocyon gregarius; B. Mesocyon
coryphaeus; C. Microtomarctus conferta; D. Epicyon haydeni;
E. Borophagus secundus; F. Canis lupus. Skulls are scaled to
the same approximate length; for relative sizes see Fig. 5.1.

Mid-sagittal sections of the skull models during a temporalis-
driven unilateral left carnassial (P4) bite. Stress distributions
shown in both Von Mises stress (A-F) and Principal stress 1
(G-L). For the latter, all colors represent tensile stress;
compressive stress is marked in gray. A,G: Canis lupus; B,H:
Borophagus secundus; C,I: Epicyon haydeni; D,J:
Microtomarctus conferta; E,K: Mesocyon coryphaeus; F,L:
Hesperocyon gregarius.

252
253
254
255
256
xx
Figure 6.1:

Figure 6.2:

Figure 6.3:

Figure 6.4:

Finite element model of Dinocrocuta gigantea with muscle
insertion areas of the temporalis and masseter created (dark
gray areas) using the Boneload program (Grosse et al., 2007).
Anterior dorsolateral view. The length of the cranium is
approximately 322 mm. Other models were constructed
similarly by demarcating regions of temporalis and masseter
attachment. Mandibles were included for identification of
resultant muscle force directions, and then removed before
analyses were run.

Dorsal views of Von Mises (VM) stress distribution during left
P3-biting scenario in the cranium of A. Crocuta crocuta, B.
Dinocrocuta gigantea, and C. Canis lupus. All legends are
scaled to have a range of 0~8 MPa for optimized visualization.
The deeper blue areas represent small or no stress and the red
areas represent highly stressed regions. White patches represent
areas where stress exceeds 8 MPa. The crania are scaled to
approximately the same length in the figure. Right P3 biting and
right and left P4 biting scenarios produced similar stress
distributions that are not statistically different.

Ventral views of Von Mises (VM) stress distribution during left
P3-biting scenario in the cranium of A. Crocuta crocuta, B.
Dinocrocuta gigantea, and C. Canis lupus. Legends as in
Figure 6.2.

Von Mises stress gradients from the anterior to posterior
cranium along the mid-sagittal plane in analogous anatomical
sampling points. Stresses from single node samples of A.
Crocuta crocuta, B. Dinocrocuta gigantea, and C. Canis lupus
and mean stresses from node group samples of D. Crocuta
crocuta, E. Dinocrocuta gigantea, and F. Canis lupus are
shown. The data points (from left to right) represent stress
recorded along the mid-sagittal plane in lateral alignment with
(1) anterior border of nasal bones, (2) infraorbital foramina, (3)
anterior boundary of the orbits, (4) the inter-orbital region
between the post-orbital processes, (5) post-orbital restriction of
the frontal-parietal region, (6) anterior-most point of the sagittal
crest, and (7) posterior-most point of the sagittal crest. The
points are plotted as percentages of skull condylobasal length
(CBL) in the anterior-posterior direction. Left side, filled
symbol; right side, open symbol; P3 biting, solid line; P4 biting,
dashed line; M1 biting, dotted line.

297
298
299
300
xxi
Figure 6.5:

Figure 7.1:

Figure 7.2:

Figure 7.3

Figure 7.4:

Figure 7.5:

Figure 7.6:

Computer tomography images of A. Crocuta crocuta, B.
Dinocrocuta gigantea, and C. Canis lupus taken as lateral
views of the mid-sagittal section of the cranium. The frontal
sinus (fs) is indicated in all three crania, and caudal expansion
of the frontal sinus in C. crocuta and D. gigantea is noted by
arrows. The internal cavities of the D. gigantea specimen are
filled with matrix, which is light grey in color in the frontal
sinus area. All crania are scaled in the figure to the same
approximate length.

Photos of specimens used in the study. A, Crocuta crocuta
(LACM[Mamm] 30655), left mandible; B, Dinocrocuta
gigantea (IVPP V15649), right mandible; C, Canis lupus
(LACM[Mamm] 23010), left mandible. Specimens are scaled to
approximately the same length in figure. Scale bar over
carnassial tooth (m1) equals 10 mm.

Muscle attachment sites on the mandible finite element models,
with Crocuta crocuta as an example. The light areas on top of
the ascending ramus and in the mandibular fossa are attachment
sites for the temporalis. The light area on the angular process is
the attachment site of the masseter. The internal pterygoid
attachment (not shown) is on the medial side of the angular
process.

Strain distributions in the mandible of Crocuta crocuta in A,
p3; B, p4, and C, m1 biting scenarios. Color spectrum
represents strain magnitude, with blue as low strain and white
relatively high strain.

Comparison of strain distributions during a p3 biting scenario in
Crocuta crocuta (A, lingual; D, buccal), Dinocrocuta gigantea
(B, lingual; E, buccal), and Canis lupus (C, lingual; F, buccal).

Median strain values at different bite positions for Crocuta
crocuta (open diamond), Canis lupus (filled triangle), and
Dinocrocuta gigantea (open square).

Cross-section strain profiles for (from left to right): p3-p4, p4-
m1, and post-m1 interdental spaces in A, Crocuta crocuta, B,
Dinocrocuta gigantea, and C, Canis lupus during a p3 bite.
View is from rostral towards caudal; buccal is to the right.

301
339
340
341
342
343
344
xxii
Figure 8.1:

Figure 8.2:

Figure 8.3:

Figure 8.4:

Figure 8.5:

Convergent evolution of skull shapes in borophagine canids (A-
B) and hyaenids (C-D) as observed from two-dimensional
geometric morphometrics analyses. A, C, dorsal views; B, D,
lateral views. Illustrations of skull show the measurements of
width to length (W:L) and depth to length (D:L) taken from
theoretical and actual skull shapes. Modified from Tseng and
Wang (In press).

Theoretical models generated by geometric modification of an
Ictitherium skull. The hybrid morphospace occupied by the 36
models spanned D:L ratios from 0.33 to 0.73 and W:L ratios
from 0.42 to 1.11. Theoretical skull shapes are shown in rostral-
lateral view.

Construction of the functional landscape from theoretical
morphologies. W:L and D:L are plotted on the x- and y-axes,
respectively. The functional properties mechanical advantage
(MA) and skull strain energy (SE, in joules) are plotted on the
z-axis. A, D, three-dimensional plots of the data points from
analysis of theoretical models; B, E, the wireframe mesh
overlaid and interpolated using the theoretical models; C, F, the
theoretical models removed, leaving the mesh representing the
functional landscapes for MA and SE, respectively.

Distribution of actual species on the functional landscapes. A,
D, distribution of hyaenids (dark circles) and fossil canids (light
circles) on two-dimensional contour plots of MA and SE,
respectively. Lines are isoclines. B, E, distribution of hyaenid
and canid species on the three-dimensional functional
landscapes for MA and SE, respectively. C, F, the pathways
occupied by the hyaenid (shaded) and canid (outlined) lineages
on the MA and SE landscapes, respectively. Note continuous
climb on the MA landscape and shifting towards shallower
slopes on the SE landscape.

Locations of optimal functional in the hybrid morphospace.
Small shaded squares represent theoretical models matched by
existing hyaenid and canid species (small unshaded square
shows position of insectivorous Proteles cristata). Suboptimal
regions are shown in larger squares. Regions marked by
parenthesized labels represent optimal areas not occupied by
actual species.

393
394
395
396
397
xxiii
Figure 8.6:

Figure 8.7:

Figure 8.8:

Theoretical and actual MA and SE values. A. distribution of
theoretical models overlaid with values from FE models of
actual species, all scaled by total surface area. B, distribution of
theoretical and actual models, the latter scaled by condylobasal
length of the skull. Red triangles indicate the theoretical
pathway traveled by actual species on the functional landscape.
The positions of hyaenid (darker shade) and canid (lighter
shade) groupings are shown as ovals in (B). Species
abbreviations (hyaenids): Ccr, Crocuta crocuta; Clu,
Chasmaporthetes lunensis; Iab, Ikelohyaena abronia; Ict,
Ictitherium sp.; Pbr, Parahyaena brunnea; Pcr, Proteles
cristata. Canids: Bor, Borophagus secundus; Can, Canis lupus;
Epi, Epicyon haydeni; Lpi, Lycaon pictus; Mes, Mesocyon
coryphaeus; Mic, Microtomarctus conferta. Percrocutid: Dgi,
Dinocrocuta gigantea.

Stress distributions on the FE skull models of actual fossil and
extant species. A, Dinocrocuta gigantea; B, Crocuta crocuta;
C, Parahyaena brunnea; D, Ikelohyaena abronia; E,
Chasmaporthetes lunensis; F, Ictitherium sp.; G, Proteles
cristata; H, Epicyon haydeni; I, Borophagus secundus; J,
Microtomarctus conferta; K, Canis lupus; L, Lycaon pictus; M,
Mesocyon coryphaeus. Phylogenetic relationships for hyaenids
(A-G) based on Werdelin and Solounias (1991), and for canids
(H-M) based on Wang (1994), Wang et al. (1999), and Tedford
et al. (2009). Skulls are scaled to the same length for ease of
presentation.

Distributions of modern North American and East African
carnivoran species on the MA (A), SE (B), and MA:SE (C-D)
landscapes. Arrows indicate pathways of evolution for hyaenids
(light arrows) and borophagine canids (dark arrows). Species
distributions of modern carnivoran species are plotted as solid
contours. Peaks on the MA:SE landscape (D) represent
optimized theoretical skull shapes that are either realized (light
shade) or unoccupied (dark shade, with question mark). (D)
corresponds with Figure 8.5.

398
399
400
xxiv
Figure 9.1:

Figure 9.2:

Figure 10.1:

Figure 10.2:

Light microscope examination of enamel Hunter-Schreger
Bands (HSB). A, Regions of tooth crown recorded for HSB
pattern in this study (top, middle, bottom), shown on lower third
premolar of Osbornodon renjiei, F:AM 63316; B, Example of
transitional acute-angle HSB to zig-zag HSB, dotted circle
indicates the sharp-angled HSB folding that characterizes full
zig-zag HSB; C, Example of undulating HSB with semi-
horizontal light and dark enamel prism bands. Scale bar equals
1 mm for part A only.

Intra-dentition variation in HSB patterns in A, fossil
hesperocyonine and borophagine canids and B, hyaenids.
Three-letter codes in legend indicate the HSB pattern
represented by each of the three regions of the tooth crown
examined, with first letter representing tip of crown, and third
letter bottom of crown. Inferred HSB in incisors is based on the
most specialized HSB that does not supersede cheek tooth
specializations, or from known HSB in one or more teeth in the
incisor group. I, incisors; C, canines; P, premolars; M, molars
(lower case indicates lower jaw). Silhouettes on right represent
dorsal skull shapes and relative sizes of some of the genera
examined, in order to demonstrate the disjoint evolution of zig-
zag HSB and large, robust skulls typical of bone-cracking
specialists.

Tooth positions examined in the study. Black square outlines
indicate approximate size of area examined during each trial.
Note the exposed areas of dentine on the shear facet of m1; all
trials were done on the enamel portion of the teeth only.

Examples of microwear features examined. A. Labial (buccal)
wear facet on p4 of Crocuta crocuta, showing a typical
specimen with moderate tooth crown attrition. B. Examples of
small (thin) scratches. C. Examples of large (thick) scratches.
D. Examples of small pits. E. Examples of large pits. F. p4
crown surface of Lycaon pictus, note paucity of microwear
features. Scale bars represent 1 mm.

423
424
457
458
xxv
Figure 10.3:

Figure 10.4:

Figure 10.5:

Figure 10.6:

Figure 10.7:

Figure 11.1:

Figure 11.2:

Method of enamel microstructure analysis. A. Three regions of
the tooth crown were examined for Hunter-Schreger Bands
(HSB), representing top, middle, and bottom thirds of the
crown. One of three types of HSB was recorded for each
region: B. Examples of a region with mostly (> 50%) zig-zag
HSB. C. Region with acute-angled undulating HSB (note that
some zig-zag HSB is also present, e.g. indicated by a dotted
circle). D. Undulating HSB.

Histograms of variance among specimen microwear trials in p4
(A-D) and m1 (E-H). A, E, small scratches; B, F, large
scratches; C, G, small pits; D, H, large pits. Plot ranges for p4
and m1 were adjusted to show maximum spread of data,
respectively.

Box plots of microwear features across HSB categories in p4
(A-D) and m1 (E-H) specimens. Boxes represent inter-quartile
ranges, horizontal lines within boxes are medians; vertical lines
show upper and lower limits, and asterisks represent outliers.

Plots of mean values and 95% confidence intervals for binned
HSB categories in p4 (A-D) and m1 (E-H) specimens. Mean
values are connected in fossil samples to show trend.

Intra-dentition evolution of HSB microstructure in: A. Fossil
Canidae and B. Hyaenidae. Phylogenies for fossil canids based
on Wang (1994) and Wang et al. (1999), and for hyaenids based
on Werdelin and Solounias (1991). Progressively more derived
HSB patterns are indicated by darker shades of grey.

Phylogenetic relationships of A. Canidae and B. Hyaenidae.
Silhouettes show representative skull shapes and sizes in each
lineage. From Tseng and Wang (2011).

Four possible modes of evolution for a given functional
complex of four adaptive traits (represented by colors) present
in an extant species (“taxon 4”). A. modular with adaptation of
all traits within the complex at a single stage in evolution, B.
two modules evolving at different times in the lineage’s
phylogenetic history, C. gradual accumulation of individual
traits over the course of the lineage’s history in an ordered
manner, D. gradual accumulation of traits in an unordered
manner. These scenarios are neither exhaustive nor mutually
exclusive, and are used for demonstration of concept.
459
460
461
462
463
499
500
xxvi

Figure 11.3:

Figure 11.4:

A graphical depiction of the convergent specializing trend in the
hyaenid and borophagine canid lineages. A. increased demand
for durophagy, B. microwear patterns reflect increased
durophagy (from Tseng, in press), C. Enamel microstructure
adapts in response to increased durophagy (from Tseng, 2011),
D. Skull shape gradually evolves towards stereotypical bone-
cracking ecomorph (from Tseng and Wang, 2011), E.
Transitional species show stereotypical biomechanical stress
dissipation patterns, but lack in large bite forces and mechanical
advantage (from Tseng and Stynder, 2011), F. Specialized
bone-cracking ecomorphs adapt through either biomechanics-
or body size-emphasized pathways (from Tseng, in prep.).

Bone-cracking specialization categories mapped onto
stratigraphic occurrence of A. hyaenids and B. borophagine
canids. Data for A are from Werdelin and Solounias (1991),
with ecomorphological categories from Werdelin and Solounias
(1996). B is modified from Tseng and Wang (2010). Colors
indicate specialization categories (Table 11.2): A (red), B
(orange), C (green), D
1
(blue), and D
2
(purple). For hyaenids,
category D
2
is represented by the closely related percrocutid
Dinocrocuta, which are early late Miocene in age (not drawn).
501
502

xxvii
Abstract

The past 65 million years of evolution in carnivoran mammals exhibits numerous cases of
convergence in ecomorphologies, stereotypical morphologies that represent unique ecological
adaptations. Such examples demonstrate the iterative adaptation of distantly related lineages to a
few lucrative hypercarnivore (meat specialist) niches. To explore the mechanistic explanations
underlying convergent evolution of ecomorphologies, this dissertation documents and reviews
recent advances in our understanding of feeding specializations in one particular hypercarnivore
niche, the bone-crackers. Bone-cracking specialists evolved at least three times in Carnivora, in
the hyaenid, percrocutid, and borophagine canid lineages. These studies in the evolutionary
changes of skull shape, enamel microstructure, enamel microwear, and craniodental
biomechanics of hyaenids and borophagine canids show that the suite of adaptive morphological
characters commonly found in the bone-cracking functional complex evolved in a mosaic manner.
Microstructural changes in the enamel were related to increased durophagy as inferred from
microwear analysis, followed by continuous skull shape changes toward increased robustness and
strength. Subsequently, skull stress dissipation patterns became adapted to handle mechanical
demands of durophagy, followed by a split into even more mechanically efficient terminal species
versus body size specialists. An updated definition of bone-cracking specialization is presented,
and implications for more general understanding of feeding specialization are discussed. An
ordered evolutionary sequence of adaptive traits in a functional complex represents a flexible
mode of evolution that can accommodate different degrees of specialization in increasingly
durophagous lineages, and may serve to explain similar adaptations in other carnivorans and non-
carnivoran mammals.
xxviii
Preface

The turning point in my path towards an advanced degree in biology was during
the spring of 2002. My family went on vacation to the San Diego area, with visits to the
San Diego Zoological Parks, Museums, and SeaWorld. I have been to zoos before, but
that particular vacation left a lasting impression in my mind. I wanted to learn about
animals and biology. I transferred from a computer science track into the Department of
Integrative Biology and began taking biology courses. A subsequent internship at the
Oakland Zoo sealed the deal, where I realized that I wanted to study hyenas and their
powerful jaws.
During the six years that I studied as a graduate student at the University of
Southern California, I learned not only a range of techniques to study the form and
function of extant and extinct mammals, but also the very different aspects of field
paleontology. Xiaoming Wang and Gary Takeuchi of the Natural History Museum of Los
Angeles County trained me in the mountain ranges of the Tibetan Plateau, on expeditions
that otherwise would only be realized in dreams. Through seven summers of fieldwork, I
have traveled the grasslands of Inner Mongolia, past the great Qinghai Lake, and into the
heart of Tibet. My graduate education was as diversified as the faculty members in the
Integrative and Evolutionary Biology Program who mentored me.
The reason for outlining the seemingly mundane laundry list of my experiences
above is that the following chapters represent the research I conducted while not fossil-
hunting in the field. These studies constitute part of the university requirement for a
xxix
doctoral degree, but what I have learned as a researcher, as a paleontologist, and as a
person tallies far beyond the pages in a single volume of summaries. In addition to the
mentors, families, and friends who were acknowledged at the beginning of this work,
collaborators on research projects have also helped me grow both intellectually and
personally.
Wendy Binder (Loyola Marymount University) and I collaborated on a study of
the mandibular biomechanics of the modern Spotted Hyena and the extinct Dinocrocuta
gigantea; Mauricio Antón and Manuel Salesa (Museo Nacional de Ciencias Naturales)
and I collaborated on a study of the Spanish skull specimen of Chasmaporthetes lunensis;
Deano Stynder (University of Cape Town) and I collaborated on a study of the South
African fossil hyena Ikelohyaena abronia, and Tao Deng (IVPP, Chinese Academy of
Sciences) and I collaborated on a study of hunting evidence in the extinct percrocutid
Dinocrocuta. These collaborators not only provided stimulating discussion on our topic
of study, I also learned from their research and paper-writing experiences.
It is therefore with great pleasure that I present the summary of the laboratory
research over the past six years, with a range of topics tied together by the central theme
of my doctoral dissertation: convergent evolution and the relationship between form and
function.

Jack Tseng
April 2011
Edmonton, Alberta, Canada
1
Introduction

Organismal evolution, in its essence, is descent with modification. Many
mechanisms have been proposed as processes of evolution, but among them the most
widely known mechanism, by far, is natural selection (Darwin 1859). Organisms, in their
struggle for survival and reproduction, are selected for, or against, based on favorable
combinations of biological characteristics that are beneficial in their current environment
(Futuyma 1997). The favored characteristics are called adaptations (Lauder 1996). This
relatively straightforward logic underlies the foundation of the study of Earth’s biological
history, where the form and function of the species that replaced each other are studied
for their adaptiveness, as well as the subsequent non-adaptiveness of those that met their
own demise (Lauder 1995). This body of works represents explorations of the link
between form and function, using prominent examples in the evolution of carnivoran
mammals. Close associations between form and function are often cited in this group of
animals, based on the repetitive evolution of convergent traits, characteristics that arise
independently in lineages whose ancestors did not possess those traits (Savage 1977).
Craniodental feeding morphologies, in particular, represent exemplars of such pattern,
and therefore are highly amenable to research questions concerning form and function
(Van Valkenburgh 2007). A brief introduction to the reasoning and foundation of the
studies that follow is outlined below.
Many carnivorans are predators—A sustainable ecological community maintains
a dynamic equilibrium of feeding levels (also called trophic levels). The base of the
2
ecological food web composes of primary producers, which most commonly includes
oxygen-producing plants, algae, and Cyanobacteria. The primary producers are eaten by
primary consumers, which include herbivores and omnivores. Those animals are in turn
consumed by top predators, also known as secondary or tertiary consumers. It is within
this last category that many carnivorans belong. Top predators can impose important
stabilizing top-down effects on the rest of the food web either directly (via consumption
of primary consumers and primary producers) or indirectly (via effects on primary
producers from the impact on primary consumers) by modulating the population sizes of
their prey (Finke and Denno 2004). Thus, an understanding of the form and function of
top predators is crucial for reconstructing past ecological communities, which serve as
important predictors to future changes in modern ecological communities under different
environmental conditions (e.g. climate change, land development, etc.). The fossil record
contains the only data for top predator ecological evolution through geologic time.
However, this record contains almost exclusively of teeth and bones for mammalian
species. This restricts the study of evolution of form and function to that of the dental and
osteological tissues. Fortunately, it has been observed that a major source of diversity in
dietary function is associated with craniodental shape (Van Valkenburgh 1988). Thus, the
subsequent chapters focus on the dental and osteological components of craniodental
shape, both of which can be directly compared across fossil and living predators.
Many terrestrial top predators are in the Order Carnivora—Species of the
mammalian order Carnivora have been recorded as fossils since the beginning of the
Cenozoic, 65 million years ago; 271 living species are known (Nowak 1999). More than
3
380 extinct genera containing hundreds if not thousands of additional species have been
described (McKenna and Bell 1997). Carnivorans have the largest body size range and
global geographical distribution of all mammalian orders, even exceeding Cetacea, the
group including whales and dolphins (Nowak 1999). Even though the ordinal name
suggests a main diet of meat, the diversity of foods that carnivorans utilize are much
greater, ranging from strict meat specialists, to omnivores, herbivores, piscivores, and
insectivores (Ewer 1973, Nowak 1999, Van Valkenburgh 1988). Ecologically, large
carnivorans are often the top predators within their ecosystems. For this reason, many
carnivorans are rare, and must compete with other top predators for limited resources.
The resulting pattern is a variety of morphological specializations that allow some species
to take advantage of resources unavailable to others. Among these specializations, the
fossil record documented the evolution of the bone-cracking predators (Werdelin 1996b).
The bone-cracking ecomorphology will be elaborated on below.
The bone-cracking predator is a unique ecomorphology in the carnivore guild—
Before the idea of ecomorphology can be discussed, the conceptual foundation of
ecology must be clarified. “Ecology is the scientific study of the interactions that
determine the distribution and abundance of organisms” (Krebs, 2001, p.2). Whereas
neoecology, the study of the distribution and abundance of living organisms, focuses on
time frames of seasons and years, the study of paleoecology (sensu lato) examines
patterns of distribution and abundance on a macroevolutionary scale (Valentine 2001).
One important concept that underlies much of the study of ecological interactions is the
dietary niche: the dimensions of an organism’s ecological requirements for survival that
4
concerns the acquisition and incorporation of energy from its environment (Hutchinson
1958, Whittaker et al. 1973). Organisms with similar dietary niches are considered
members of the same dietary guild, which is defined as a group of organisms utilizing
similar dietary resources in similar ways (Root 1967, Simberloff and Dayan 1991).
Ecomorphologies, then, can be defined as subgroups within dietary guilds that show
differential morphological specialization in food acquisition and incorporation (Werdelin
1996b). Ecomorphology may be decoupled from phylogenetic history; the same
ecomorphologies have been repeatedly represented in the fossil record by a wide
diversity of taxa in mammalian and reptilian predator guilds (Van Valkenburgh 1995).
The following chapters focus both on the clarification of the evolutionary pattern of the
bone-cracking ecomorphology in the carnivoran family Hyaenidae (Werdelin and
Solounias 1991), as well as phylogeny-independent evolution of the same
ecomorphology in a North American dog subfamily, the Borophaginae (Wang et al.
1999).
Bone-cracking is a well-defined task—The act of predation for the purpose of
energy gain can be divided into two components: food acquisition and food processing.
These two processes are sometimes considered separately but complementary to each
other in discussions of ecomorphology (Van Valkenburgh 1985, 1988). Clarification
needs to be made between morphological traits associated with acquisition of food (e.g.
long distance pursuit by gracile carnivores, as in the wolf, versus ambush of prey using
powerful and more maneuverable limbs, as in tigers), and processing of the acquired food
(e.g. cutting of prey flesh using shearing teeth versus cracking of bone using blunt teeth).
5
Whereas food acquisition involving use of the skull could produce extremely complex
forces (e.g. lion killing prey with a muzzle bite while the prey thrashes), food processing
often involves stereotypical movements of the skull and more predictable forces (e.g.
dismantling a scavenged carcass by cutting strips of soft tissue using teeth). The
following chapters do not discuss acquisition strategies, but will only concentrate on the
food processing mechanism known as bone-cracking.
Bone consumption is a relatively widespread phenomenon in meat-eating
vertebrates; bones are often inadvertently swallowed along with soft tissue of small prey.
In other cases, very large bones are targeted and consumed with a specific behavioral
sequence (Kruuk 1972b, Van Valkenburgh 1996). This study deals with the latter
condition. Bones that cannot be swallowed directly are crushed and/or cracked during
mastication. Bone-cracking was defined by Lars Werdelin (1989) as the point-to-point
cracking of bone by the anterior cheek teeth; bone-crushing, on the other hand, is the
surface-to-surface grinding of bone using more posterior teeth with broad occlusal
surfaces. Of all living carnivorans, only the hyaenids Crocuta crocuta, Parahyaena
brunnea, and Hyaena hyaena are bone-crackers sensu stricto. In the fossil record,
hyaenids and canids, along with oxyaenid and hyaenodontid creodonts, may have
repeatedly (iteratively) evolved bone-cracking capability (Werdelin 1996b).
Non-carnivoran mammals may consume bone for completely different reasons.
Many ungulate species have been observed to gnaw on bones (osteophagia), possibly to
supplement a phosphate deficient diet (Denton et al. 1986). No data are available on the
optimal diet of spotted hyenas, but inferences made from nutrient requirement of felid
6
carnivorans for calcium (8 g/kg food) and phosphorus (6 g/kg food) (Council, 1986), and
the mid-shaft femur bone density of a spotted hyena prey item such as the wildebeest
(Connachaetes taurinus; 1.16 g/cm3) (Lam et al. 1999), suggest spotted hyenas are
typically not calcium-deficient. A typical hyena meal (averaging 1.5-1.8 kg per individual
per day, up to one-third of individual hyena body weight by observation)(Kruuk 1972b),
which incorporates skin, meat, and bone, most likely represents an excess of calcium and
phosphate intake. Thus a nutrient deficiency-induced osteophagia cannot explain hyena
bone-cracking behavior. Besides, studies of bone accumulation by living striped hyenas
suggest that bone breakage is positively correlated to bone fat content (Leakey et al.
1999). Thus, a reasonable explanation for the bone-cracking behavior in living hyaenids
is to gain access to the internal fatty marrow for protein, and not for the minerals in bone.
The gastrointestinal system of hyaenids does not appear to be specialized from superficial
examination; in fact, regions of the large intestine are well differentiated and
unspecialized (Ewer 1973). However, it is possible that physiological adaptations for
bone digestion are not obvious from superficial anatomy. For the purpose of the
following chapters, bone-cracking morphology will be interpreted in the context of its
role in allowing those individuals to access fats and proteins unavailable to other
carnivorans, and it represents an adaptation that might be supplemented by other
currently cryptic modifications in the digestive system.
The bone-cracking ecomorphology evolved iteratively—Iterative evolution, as
defined by Van Valkenburgh (1991), refers to repeated evolution of the “same”
morphological traits (and inferred ecomorphology) independent of phylogenetic ancestry
7
(for a discussion of sameness see Wake, 1999). Among the most prominent examples of
iterative evolution in animals are saber-toothed mammals (nimravid, felid carnivorans,
creodonts, and thylacosmilid marsupials; Van Valkenburgh, 1991) and winged animals
(insects, mammals, pterosaurs, and birds; e.g. Nordberg, 2002). Depending on the level
of analysis, the independently evolved structures can either be convergent (i.e. only
superficially similar) or parallel (i.e. employing a common set of developmental or
structural mechanisms; Futuyma, 1997). Bone-cracking evolved at least once in hyaenids
(Werdelin 1996b), and multiple times in canids (Van Valkenburgh 1991). Comparison
between the typical bone-cracking ecomorphology of hyaenids and borophagine canids
using the methodologies outlined in the next section will address the question of whether
this case of iterative evolution is only superficial or is supported by biomechanical
similarities.
Bone-crackers are not necessarily scavengers—Contrary to common perception
of the hyena as a scavenger, Kruuk (1972b) has shown through a long-term field study of
living spotted hyenas, that active hunting accounts for more than 50% of the prey
consumed for hyenas living on the Serengeti plains in eastern Africa. In contrast to some
studies which have considered the bone-cracking scavenger as an ecological niche, I
consider scavenging by carnivorans as an option and not an obligation. All carnivorans
scavenge to some extent; in the modern day African savannah, only 30% of the large
ungulate mortality result from predation (DeVault et al. 2003). Thus, a large amount of
biomass is available as unhunted carcass. Sociality is an important determinant of
scavenging behavior; large carnivorans probably cannot function as pure scavengers in
8
large groups, as each carcass may not provide enough energy for all individuals in the
group (Kruuk 1972b). However, current findings have not been able to resolve the issue
of social organization of many fossil bone-cracking carnivorans (Andersson 2005, Van
Valkenburgh et al. 2003b). Even so, it is interesting to note that fossil hyaenids and
canids represent the most commonly preserved carnivoran fossils, sometimes ranging
into thousands of individuals at a single fossil locality (Pei 1934, 1940, Zdansky 1924).
Why did bone-cracking capability evolve?—Interspecific competition for food
may play an important role in shaping carnivoran behavior and morphology. When
interference (direct) competition is encountered, extant large carnivorans adapt several
types of responses: 1) temporal avoidance, as in shifting the time of active hunting to
avoid encountering competitors (e.g. cheetahs hunt diurnally instead of nocturnally), 2)
spatial avoidance, as in stashing food in a location inaccessible to other predators (e.g.
leopards storing their food on a tree branch away from the ground; spotted hyenas storing
food underwater), 3) direct (or aggressive) confrontation, the situation where food is
defended by aggressively chasing off competitors (e.g. between African lions and spotted
hyenas), and 4) rapid feeding, or the immediate consumption of a large portion of the
prey before competitors discover the kill (Biknevicius and Van Valkenburgh 1996). The
living spotted hyena utilizes the last strategy, and observations show that some
individuals are able to consume up to one-third of their weight in one sitting (Kruuk
1972b). Intake speed is increased by consuming skin, muscle, and bone indiscriminately
instead of picking specific tissues when feeding. The ability to crack bones would allow
these predators to do exactly that. In addition, during periods of food shortage the bone-
9
cracking species would have access to an exclusive source of nutrient store in bone
marrow. These evidence support the interpretation that bone-cracking could be an
adaptation to increase competitiveness against other large carnivorans, rather than
decreased competitiveness as scavengers. The widespread abundance and success of
some large fossil hyenas supports this view.

Method Overview

Fundamental approach to research questions ―All organisms have in common
with inanimate objects in that they adhere to physical laws (Norberg 2002). The study of
biomechanics aims to describe the energetics and movement of biological structures
using mechanical laws. Whereas different fossil species probably experienced a spectrum
of different environmental conditions and biological interactions that confound
paleoecological studies using modern analogs, the physical laws governing biomechanics
have not changed. Utilizing this concept, interpretations made of extinct species can
avoid some pitfalls of non-uniformitarianism found in purely descriptive paleoecological
reconstructions (Bottjer 1998). The strength of this approach comes with its shortcoming,
in the implicit assumption of a strong correspondence between form and function.
Whereas extensive correlations have been made between morphological characteristics
and their functional implications, the actual relationship between them may still be highly
complex (Lauder 1995, 1996). A multifaceted approach is best in this case; if multiple
10
pairs of form-function correlations corroborate each other, the findings are more robust,
and can be used to make more confident inferences (termed consilience; Skelton, 2001).
The concept of variation plays a central role in evolutionary biology. When
Charles Darwin (1859) outlined the tenants of his theory of evolution by natural selection,
he put much emphasis on populational variation as the raw material of evolutionary
change. The study of the fossil record, a relatively incomplete but essential source of
information about past life, is often based on fragmentary and/or single individuals of a
species. The taxonomic sampling in the subsequent chapters encompasses interspecific
differences that are assumed to surpass intraspecific variation, and thus are not greatly
affected by them. Along with this assumption, the variation between adult and juvenile
individuals of the same species is of great importance (Biknevicius 1996), but will only
be addressed peripherally as they become relevant within the scope of a specific study.
The methods employed in this study fall into two foci: cranial versus dental
evolution.
The craniodental system as a bone-cracking tool ―Mastication is an activity that
places high stresses on the mammalian craniodental system. The basic mineral
composition of mammalian bone is similar throughout the skeleton; however, different
bones may have different material properties by virtue of their structure. The evolution of
bone-cracking required the application of force by the craniodental system onto prey
bone, and cracking the prey bone without damage to themselves. The bone-cracking teeth
serve as the interface between the predator cranium and prey bone. Thus, two aspects of
11
the craniodental system had to be under selection during evolution of the bone-cracking
ecomorphology: the cranial bone complex, and teeth.
Studies of bone and bone suture strain have shown that muscle forces, not biting
reaction forces, are principal in creating strain in the braincase (Herring et al. 2001).
Studies have also shown that surface enamel of teeth used for mastication can display
characteristic superficial damage and wear from regular use (Fortelius and Solounias
2000, Kruuk 1972b, Solounias and Semprebon 2002). The following chapters address
both of these areas. Two-dimensional geometric morphometrics (Chapter 1) and three-
dimensional finite element modeling (Chapters 2-8) are employed to examine cranial
biomechanics; microstructure (Chapter 9) and enamel microwear (Chapter 10) are
employed to examine tooth structure and use.
The osteological component of craniodental shape ―The mammalian dental
masticatory system is composed of corresponding teeth on the cranium and lower jaws.
The lower jaw is composed of right and left dentary bones, on which the teeth are housed.
The upper tooth row, on the other hand, protrudes from the maxillary and premaxillary
bones, which make up only the anterior portion of the cranium. The dorsal and posterior
portions of the cranium are composed of a number of bones sutured to each other, and
those bones are also often fused together entirely in hyenas. As a whole, the cranium is a
highly complex functional system that performs the task of dental mastication as well as
olfaction, auditory sensing, vision, and protection for the central nervous system.
Chapter 1: Studies of craniodental form has benefited greatly by the combination
of multivariate statistics and anatomy in the field of geometric morphometrics. The
12
fundamental principle of geometric morphometrics is to capture the “essence” of
biological shapes using a few informative anatomical landmarks. In the 2D application of
geometric morphometrics, the landmarks are simply positions on the specimen in the
plane of interest, represented by x and y coordinates. A group of landmarks that record
the unique shapes of each individual is compared across different specimens. After
multivariate statistical calculations, the shape differences between each pair of specimens
are measured by the differences in their relative warp scores, which are coefficients that
summarize the variability in the positions of the original landmarks (Zelditch et al. 2004).
With 2D geometric morphometrics, traditional anatomical comparisons between different
hyena species and the interpretation of feeding adaptations are mathematically quantified
and tested for statistical significance.
Chapters 2-8: Whereas living animals can be studied in vivo, fossil species cannot
be compared in the same way for the obvious reason that they are extinct. The lack of a
methodology that is capable of directly and precisely comparing craniodental differences
and their implication for feeding adaptation was a limitation in the field of evolutionary
biology until the adoption of 3D finite element modeling and analysis within the past
decade (Rayfield 2007). First used as an analytical method in the discipline of mechanical
engineering, finite element modeling reproduces an actual object digitally as a discretized
representation, so researchers can conduct mechanical tests that are too time-consuming,
costly, or even impossible to do in reality. This technique has been adapted to studies of
fossil species, whereby the biomechanics of the craniodental system can be directly
compared using 3D digital models of the original specimen. Results from finite element
13
modeling offer new insights into novel craniodental function in fossil species, as well as
tests for previous hypotheses regarding those functions as inferred from living forms
(Rayfield 2007).
The dental component of craniodental shape―Dental form is one of the most
variable features in the mammalian skeleton. From the total loss of dentition in aardvarks
to the dentition of dolphins containing hundreds of teeth, there is tremendous flexibility
for mammals to adapt to their dietary needs. Carnivorans are no exception, and species in
the hyena and dog lineages have evolved large and strong premolars for cracking through
the bones of prey. Fortunately, teeth are also one of the most readily preserved materials
in the mammalian fossil record, thus it is possible to test hypotheses regarding the
evolution of dental structure in the hyena family in response to changing food types and
demands on dietary function. So far, the dental fossil records for approximately 65
different hyena species are known throughout the world; there are at least as many
borophagine canids in the fossil record as well (Wang et al. 1999).
Chapter 9: The carnivore tooth is covered by enamel. Enamel is constructed by
ameloblast cells, which build an organic matrix beginning at the enamel-dentin junction
outward towards the outer enamel surface, and are under strong genetic control (Paine et
al. 2001). The fully developed enamel is a largely inorganic structure made of a mosaic
hydroxyapatite network. Directional variability in the path of ameloblast enamel
production creates three-dimensional enamel patterns that are visible from the enamel
surface and are referred to as enamel microstructure. A specific type of enamel
microstructure, called Hunter-Schreger Bands (HSB), is found in all carnivores (Stefen
14
1994). HSB patterns are further differentiated into sinusoidal “undulating” bands, “acute
undulating ” bands with some angular patterns, and “zig-zag” bands with very sharp
angular patterns. The evolution of enamel microstructure pattern, in this order, has been
observed in carnivore groups that develop hard-food diets. Therefore, the documentation
of microstructure evolution in hyenas and dogs illuminates the adaptive patterns that
culminated in the large bone-cracking predators known today.
Chapter 10: Mammalian tooth enamel, being the hardest material in the entire
skeleton, is often pitted against very hard substances during feeding (e.g. bone, pebbles,
plant silica, etc.). Thus, the abrasion pattern on the enamel surface of mammals represent
the only firsthand evidence for craniodental function in diet by recording the interaction
between the ingested food item, its associated particles, and the tooth (Solounias and
Semprebon 2002). Whereas the correlation between food item and enamel abrasion on
the microscopic scale (known as microwear) is well established for herbivorous
mammals, those for carnivorous mammals are only recently beginning be studied more
comprehensively (Anyonge 1996, Goillot et al. 2009, Schubert et al. 2010, Van
Valkenburgh et al. 1990). Living hyenas incorporate bone in their diet, leaving
characteristic markings on their teeth (Goillot et al. 2009, Schubert et al. 2010, Van
Valkenburgh et al. 1990). Records of fossilized bone that were chewed by fossil hyenas
are well known in the field of fossil hominid research. Such microwear features are one
of the few direct evidence of past diet, and the correlation of such features to
morphological evolution strengthens and clarifies the linkage between structural form and
function.
15
The convergent evolution of similar feeding morphologies among carnivorans
constitutes a major pattern in the evolution of those mammals. Establishing a more robust
link between form and function allows better characterization of the mechanisms for
morphological convergence. As introduced above, the convergent evolution of bone-
cracking hyenas and dogs are studied with four methodological approaches. Craniodental
shape, representing a major source of diversity in carnivoran species, is studied by their
dental and osteological components, respectively. The methods employed in the
following chapter are (1) geometric morphometrics, (2) 3D finite element modeling, (3)
enamel microstructure analysis, and (4) enamel microwear analysis. The following
chapters represent such components of a pluralistic approach to an evolutionary
biological study of convergence, form, and function.
16
Introduction References
Andersson, K. 2005. Were there pack-hunting canids in the Tertiary, and how can we
know? Paleobiology 31(1):56-72.

Anyonge, W. 1996. Microwear on canines and killing behavior in large carnivores: saber
function in Smilodon fatalis. Journal of Mammalogy 77(4):1059-1067.

Biknevicius, A. R. 1996. Functional discrimination in the masticatory apparatus of
juvenile and adult cougars (Puma concolor) and spotted hyenas (Crocuta crocuta).
Canadian Journal of Zoology 74:1934-1942.

Biknevicius, A. R., and B. Van Valkenburgh. 1996. Design for killing: Craniodental
adaptations of predators. Pp. 393-428. Carnivore behavior, ecology, and evolution,
Vol. 2.

Bottjer, D. J. 1998. Phanerozoic non-actualistic paleoecology. Geobios 30(7):885-893.

Darwin, C. 1859. On the origin of species by means of natural selection or the
preservation of favoured races in the struggle for life. John Murray, London.

Denton, D. A., J. R. Blair-West, M. J. McKinley, and J. F. Nelson. 1986. Problems and
paradigms: physiological analysis of bone appetite (Osteophagia). BioEssays
4(1):40-43.

DeVault, T. L., J. Rhodes, Olin E., and J. A. Shivik. 2003. Scavenging by vertebrates:
behavioral, ecological, and evolutionary perspectives on an important energy
transfer pathway in terrestrial ecosystems. Oikos 102:225-234.

Ewer, R. F. 1973. The carnivores. Cornell University Press, Ithaca.

Finke, D. L., and R. F. Denno. 2004. Predator diversity dampens trophic cascades. Nature
429:407-410.

Fortelius, M., and N. Solounias. 2000. Functional characterization of ungulate molars
using the abrasion-attrition wear gradient: a new method for reconstructing
paleodiets. American Museum Novitates 3301:1-36.

Futuyma, D. J. 1997. Evolutionary Biology. Sinauer Associates, Inc., Sunderland.

Goillot, C., C. Blondel, and S. Peigne. 2009. Relationships between dental microwear and
diet in Carnivora (Mammalia) - Implications for the reconstruction of the diet of
extinct taxa. Palaeogeography Palaeoclimatology Palaeoecology 271:13-23.

17
Herring, S. W., K. L. Rafferty, Z. J. Liu, and C. D. Marshall. 2001. Jaw muscles and the
skull in mammals: the biomechanics of mastication. Comparative Biochemistry
and Physiology Part A 131:207-219.

Hutchinson, G. E. 1958. Concluding remarks. Cold Spring Harbor Symposium on
Quantitative Biology 22:415-427.

Krebs, C. J. 2001. Ecology. Benjamin Cummings, San Francisco.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior. The
University of Chicago Press, Chicago.

Lam, Y. M., X. Chen, and O. M. Pearson. 1999. Intertaxonomic variability in patterns of
bone density and the differential representation of bovid, cervid, and equid
elements in the archaeological record. American Antiquity 64(2):343-362.

Lauder, G. V. 1995. On the inference of function from structure. Pp. 9-18. In J. J.
Thomason, ed. Functional Morphology in Vertebrate Paleontology. Cambridge
University Press, New York.

Lauder, G. V. 1996. The Argument From Design. Pp. 55-91. In M. R. Rose, and G. V.
Lauder, eds. Adaptation. Academic Press, Inc., San Diego, California.

Leakey, L. N., A. H. Milledge, S. M. Leakey, J. Edung, P. Haynes, D. K. Kiptoo, and A.
McGeorge. 1999. Diet of striped hyaena in northern Kenya. African Journal of
Ecology 37:314-326.

McKenna, M. C., and S. K. Bell. 1997. Classification of mammals above the species
level. Columbia University Press, New York.

Norberg, U. M. L. 2002. Structure, form, and function of flight in engineering and the
living world. Journal of Morphology 252(1):52-81.

Nowak, R. M. 1999. Walker's carnivores of the world. The Johns Hopkins University
Press, Baltimore.

Paine, M. L., S. N. White, W. Luo, H. Fong, M. Sarikaya, and M. L. Snead. 2001.
Regulated gene expression dictates enamel structure and tooth function. Maxtrix
Biology 20:273-292.

Pei, W.-c. 1934. On the Carnivora from Locality 1 of Choukoutien. Palaeontologia Sinica
Series C 8(1):1-216.

18
Pei, W.-c. 1940. The Upper Cave fauna of Choukoutien. Palaeontologia Sinica Series C
10:1-100.

Rayfield, E. J. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35:541-576.

Root, R. B. 1967. The niche exploitation pattern of the blue-gray gnatcatcher. Ecological
Monographs 37:317-350.

Savage, R. J. 1977. Evolution in carnivorous mammals. Palaeontology 20(2):237-271.

Schubert, B. W., P. S. Ungar, and L. R. G. DeSantis. 2010. Carnassial microwear and
dietary behaviour in large carnivorans. Journal of Zoology 280(3):257-263.

Simberloff, D., and T. Dayan. 1991. The guild concept and teh structure of ecological
communities. Annual Review of Ecology and Systematics 22:115-143.

Skelton, P. W. 2001. Bringing fossil organisms to life. Pp. 367-375. In D. E. G. Briggs,
and P. R. Crowther, eds. Palaeobiology II. Blackwell Science, New York.

Solounias, N., and G. Semprebon. 2002. Advances in the reconstruction of ungulate
ecomorphology with application to early fossil equids. American Museum
Novitates 3366:1-49.

Stefen, C. 1994. Zahnschmelzdifferenzierungen bei Raubtieren: [Carnivora im Vergleich
zu Vertretern der Creodonta, Arctocyonidae, Mesonuchidae, Entelodontidae
(Placentalia), Thylacoleonidae, Dasyuridae und Thylaacinidae (Marsupialia)].
Ph.D. Rheinischen Friedrich-Wilhelms-Universtat zu Bonn.

Valentine, J. W. 2001. Chapter 2. Scaling is everything: brief comments on evolutionary
paleoecology. P. 320. In W. D. Allmon, and D. J. Bottjer, eds. Evolutionary
Paleoecology: the ecological context of macroevolutionary change. Columbia
University Press, New York.

Van Valkenburgh, B. 1985. Locomotor diversity within past and present guilds of large
predatory mammals. Paleobiology 11(4):406-428.

Van Valkenburgh, B. 1988. Trophic diversity in past and present guilds of large
predatory mammals. Paleobiology 14(2):155-173.

Van Valkenburgh, B. 1991. Iterative evolution of hypercarnivory in canids (Mammalia:
Carnivora): evolutionary interactions among sympatric predators. Paleobiology
17:340-362.
19

Van Valkenburgh, B. 1995. Tracking ecology over geological time: evolution within
guilds of vertebrates. Trends in Ecology and Evolution 10(2):71-76.

Van Valkenburgh, B. 1996. Feeding behavior in free-ranging, large African carnivores.
Journal of Mammalogy 77(1):240-254.

Van Valkenburgh, B. 2007. Déjà vu: the evolution of feeding morphologies in the
Carnivora. Integrative and Comparative Biology 47(1):147-163.

Van Valkenburgh, B., T. Sacco, and X. Wang, eds. 2003. Pack hunting in Miocene
borophagine dogs: evidence from craniodental morphology and body size.
American Museum of Natural History, New York.

Van Valkenburgh, B., M. F. Teaford, and A. Walker. 1990. Molar microwear and diet in
large carnivores: inferences concerning diet in the sabretooth cat, Smilodon fatalis
Journal of Zoology 222:319-340.

Wake, D. B. 1999. Homoplasy, homology and the problem of 'sameness' in biology. John
Wiley & Sons, New York.

Wang, X., R. H. Tedford, and B. E. Taylor. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243:1-391.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

Werdelin, L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
Pp. 582-624. In J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution.
Cornell University Press, New York.

Werdelin, L., and N. Solounias. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30:1-104.

Whittaker, R. H., S. A. Levin, and R. B. Root. 1973. Niche, habitat, and ecotope. The
American Naturalist 107(955):321-338.

Zdansky, O. 1924. Jungtertiäre carnivoren Chinas. Palaeontologia Sinica Series C 2(1):1-
149.

Zelditch, M., D. Swiderski, D. Sheets, and W. Fink. 2004. Geometric Morphometrics for
Biologists. Elsevier.

20
Chapter One: Evolution of form in hyenas and dogs

This chapter has been published as:

Tseng, Z. J. and X. Wang. 2011. Do convergent ecomorphs evolved through convergent
morphological pathways? Cranial shape evolution in fossil hyaenids and borophagine
canids (Carnivora, Mammalia). Palaeobiology 37(3): 470-489. doi: 10.1666/10007.1.

A copy of the accepted manuscript begins on the next page.
21
Do convergent ecomorphs evolve through convergent morphological
pathways? Cranial shape evolution in fossil hyaenids and borophagine
canids (Carnivora, Mammalia)

Zhijie Jack Tseng and Xiaoming Wang

RRH: CANID AND HYAENID CONVERGENT EVOLUTION
LRH: TSENG AND WANG

22
Chapter One Abstract

Cases of convergent evolution, particularly within ecomorphological contexts, are
instructive in identifying universally adaptive morphological features across clades.
Tracing of evolutionary pathways by which ecomorphological convergence takes place
can further reveal mechanisms of adaptation, which may be strongly influenced by
phylogeny. Ecomorphologies of carnivorous mammals represent some of the most
outstanding cases of convergent evolution in the Cenozoic radiation of mammals. This
study examined patterns of cranial shape change in the dog (Canidae) and hyena
(Hyaenidae) families, in order to compare the evolutionary pathways that led to the
independent specialization of bone-cracking hypercarnivores within each clade.
Geometric morphometrics analyses of cranial shape in fossil hyaenids and borophagine
canids provided evidence for deep-time convergence in morphological pathways toward
the independent evolution of derived bone-crackers. Both clades contained stem members
with plesiomorphic generalist/omnivore cranial shapes, which evolved into doglike
species along parallel pathways of shape change. The evolution of specialized bone-
crackers from these doglike forms, however, continued under the constraint of a full
cheek dentition and restriction on rostrum length reduction in canids, but not hyaenids.
Functionally, phylogenetic constraint may have limited borophagine canids to crack
bones principally with their carnassial instead of the third premolar as in hyaenids, but
other cranial shape changes associated with durophagy nevertheless evolved in parallel in
the two lineages. Size allometry was not a major factor in cranial shape evolution in
23
either lineage, supporting the interpretation of functional demands as drivers for the
observed convergence. The comparison between borophagines and hyaenids showed that
differential effects of alternative functional “solutions” that arise during morphological
evolution may be multiplied with processes of the “macroevolutionary ratchet” already in
place to further limit the evolutionary pathways available to specialized lineages.

Zhijie Jack Tseng. Integrative and Evolutionary Biology Program, Department of
Biological Sciences, 3616 Trousdale Parkway, University of Southern California,
Los Angeles, California 90089, and Department of Vertebrate Paleontology,
Natural History Museum of Los Angeles County, 900 Exposition Boulevard, Los
Angeles, California 90007. E-mail: jtseng@nhm.org

Xiaoming Wang. Department of Vertebrate Paleontology, Natural History Museum of
Los Angeles County, 900 Exposition Boulevard, Los Angeles, California 90007

Accepted: 28 December 2010

24
Chapter One Introduction

Carnivorous mammals include a diverse range of ecomorphologies that have
evolved iteratively (or repetitively, sensu Van Valkenburgh 1991) in different lineages;
among the major categories are doglike and hyena-like forms (Van Valkenburgh 2007).
To better understand patterns of ecomorphological evolution within these two categories,
we examined the evolution of cranial shapes in the Canidae and Hyaenidae. The family
Canidae is the most basal of the extant Caniformia, the former with a fossil record
spanning 40 Myr (Wang et al. 2008), whereas the family Hyaenidae (Feliformia) has its
earliest record ca. 25 Ma (Turner et al. 2008) and an estimated molecular divergence date
of 29 Ma from its feliform sister groups (Koepfli et al. 2006). The last common ancestor
of these two families has an estimated geologic age near the split of the Caniformia-
Feliformia, ca. 43 Ma (Wesley-Hunt and Flynn 2005). Fossil materials from these two
families represent some of the most common and widespread carnivore fossils in their
regions of origin (Old World for hyaenids and New World for canids). The two families
lived their entire evolutionary histories in allopatry prior to the late Miocene (Qiu 2003),
until the arrival of “Canis” cipio ca. 8–7 Ma in Spain (Crusafont-Pairó 1950) and Vulpes
at 7 Ma in Africa (De Bonis et al. 2007). Subsequently, one single hyaenid genus,
Chasmaporthetes, made the reciprocal immigration to the New World during the
Pliocene ca. 4 Ma (Berta 1981, Hay 1921, Kurtén and Werdelin 1988, Qiu 2003). Among
the three canid subfamilies, Hesperocyoninae, Borophaginae, and Caninae, borophagines
overlapped most extensively with hyaenids in both temporal occurrence and
25
morphological similarity, and surpassed Hyaenidae in taxonomic richness and ecologic
breadth, the latter aspect evident in ecomorphs such as the extremely hypocarnivorous
Cynarctus and Cynarctoides (Fig. 1.1) (Wang et al. 1999). Comparison of evolutionary
changes in cranial shape between borophagine canids and hyaenids are therefore
particularly appropriate for more in-depth analysis of previously suggested trends of
convergence between the two families (Munthe 1989, Wang et al. 1999, Werdelin 1989).
Background.—Convergent evolution, a major topic in evolutionary biology, has
been shown to be a prominent and common feature of carnivorous mammals in the
Cenozoic (Van Valkenburgh 2007). It has been demonstrated that there are few adaptive
peaks taken in morphological evolution in response to different dietary needs among
carnivores, resulting in the iterative appearance of similar morphologies that correspond
to similar ecological roles (Werdelin 1996b). These iterative ecomorphologies share
adaptive aspects of the cranial, dental, and postcranial skeletons with modern carnivorans
for which ecological and dietary habits are known (Van Valkenburgh 1985, 1988, 1989a,
1999, 2001, 2007). In addition, each ecomorphology not only evolved iteratively through
time in its respective continent of origin, but often appeared contemporaneously in
geographically distant regions as well (Martin 1989, Werdelin 1989, 1996b). The
existence of similar ecomorphologies through time and space is not unique to mammalian
predators; it has also been shown in guilds of modern avian predators (Van Valkenburgh
1995) and with some variations in theropod dinosaurs (Van Valkenburgh and Molnar
2002), the latter possessing a wider range of ecomorphological adaptations than
previously thought as more discoveries are made (Zanno et al. 2009). These observations
26
indicate that there are universally adaptive aspects to these stereotypical morphologies,
and thus they are ideal cases to test hypotheses of form-function relationships and to
discover mechanistic explanations underlying such extensive morphological convergence.
The most commonly evolved ecomorphology among carnivorous mammals is
hypercarnivory (Crusafont-Pairó and Truyols-Santonja 1956, Van Valkenburgh 2007), or
specialization for eating meat, in which species evolve into catlike, wolf-like, or hyena-
like ecomorphs (Van Valkenburgh 2007, Werdelin 1996b). Quantitative studies in the
shape and function of catlike species have shown that whereas conical-toothed cats
exhibit skull shapes that can be best explained by allometry, the extremely
hypercarnivorous sabertooth ecomorphs evolved cranial shape according to the demands
of the canines as a killing weapon (Slater and Van Valkenburgh 2008). Furthermore, the
evolution of elongate canines may have altered selective pressure toward more efficient
killing mechanisms and secondarily reduced bite forces in those specialized clades
(Christiansen 2008); these data are corroborated by evidence of relatively weak bite
forces in the American sabertooth cat, Smilodon fatalis, demonstrated by finite element
analysis of skull models (McHenry et al. 2007). The iterative evolution of sabertooth
hypercarnivores and the similarity of those forms to each other (rather than their
respective near relatives) strongly demonstrate a common set of selective requirements
during the evolution of a particular ecomorph.
Compared to catlike species, however, wolf-like and hyena-like forms to date
have not received as extensive a treatment in functional studies. The convergent
evolution of these two ecomorphs is best illustrated by their namesake families: Canidae
27
and Hyaenidae, respectively. The diversity of ecomorphs demonstrated by the family
Hyaenidae over their evolutionary history has been amply documented in studies of
systematics and evolution, particularly in the appearance of wolf-like and hyena-like
ecomorphs in the Old World during the Miocene (Turner et al. 2008, Werdelin and
Solounias 1991, Werdelin and Solounias 1996b). Similarly, the evolution of wolf-like
and hyena-like canids has also been documented within the Borophaginae throughout the
Cenozoic of North America (Tedford et al. 2009, Van Valkenburgh 1991, Van
Valkenburgh and Koepfli 1993, Van Valkenburgh et al. 2004, Wang 1994, Wang et al.
1999). The current state of knowledge regarding these two ecomorphs remains at the
interpretation that wolf-like and hyena-like hypercarnivores have evolved repeatedly in
these two families in both craniodental and postcranial characteristics (Berta 1981,
Ferretti 1999, Ferretti 2007b, Kurtén and Werdelin 1988, Stefen 1999, Stefen and
Rensberger 2002, Van Valkenburgh 1985, Werdelin 1989); however, the overall cranial
shape changes that drove the iterative evolution of these ecomorphs and its functional
implications for adaptation have not been investigated in more detail beyond linear
morphometrics and visual comparisons of similar forms (Werdelin 1989, Werdelin and
Solounias 1991). With an extreme ecomorphology such as that of bone-cracking
hypercarnivores, functional demands are expected to be severe, and selective pressures
high (Van Valkenburgh 2007); differences in mechanical performance between modern
bone-cracking and non-bone-cracking carnivoran species provide support for this
interpretation (Tseng 2009). On a macroevolutionary time scale, however, are overall
cranial shape changes evidence of convergent evolution of specialized ecomorphs, as
28
well as convergence among species along the morphological pathways upon which these
specialists evolved? Are wolf-like and hyena-like ecomorphs part of an evolutionary
sequence that occurred in a particular order within the borophagine canids and hyaenids?
What is the evolutionary pattern with regard to ecomorphological convergence over the
entire fossil history of these two carnivoran groups? This study explored these issues
using geometric morphometrics analysis.
Abbreviations.—AMNH, American Museum of Natural History, New York;
CMNH, Carnegie Museum of Natural History, Pittsburgh, Pennsylvania; F:AM, Frick
Collection, American Museum of Natural History, New York; FMNH, Field Museum of
Natural History, Chicago; HMZ, Hezheng Museum of Paleozoology, Gansu, China;
IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of
Sciences, Beijing; JODA, John Day Fossil Beds National Monument, Oregon; LACM,
Natural History Museum of Los Angeles County, Los Angeles; MCZ, Museum of
Comparative Zoology, Harvard University, Cambridge, Massachusetts; MGL, Musée
géologique de Lausanne, Romandy, Switzerland; MNCN, Museo Nacional de Ciencias
Naturales-CSIC, Madrid; MVZ, Museum of Vertebrate Zoology, University of California,
Berkeley; NHMA-MTLB, Mytilinii-1B collection in the Natural History Museum of
Aegean, Mytilinii, Greece; NMB, Naturhistorisches Museum Basel, Switzerland; PPHM,
Panhandle-Plains Historical Museum, Canyon, Texas; TMM, Texas Memorial Museum,
University of Texas, Austin; UAMZ, University of Alberta Museum of Zoology,
Edmonton, Canada; UCMP, University of California Museum of Paleontology, Berkeley,
29
California; USNM/NMNH, National Museum of Natural History (Smithsonian),
Washington, D.C.

Chapter One Materials and Methods

To summarize the variation observed in cranial shapes of borophagine canids and
hyaenids, we used two-dimensional geometric morphometrics analysis. Our analyses
were performed on planar views of the skull specimens, which have been shown to
display meaningful inter- and intraspecific variation in other closely related carnivorans
(Figueirido et al. 2009, Slater and Van Valkenburgh 2008, Tanner et al. 2009). Dorsal,
lateral, and ventral views of each skull specimen were used in three separate analyses.
Because our main interest is in ecomorphological evolution, we placed shape data of
canid and hyaenid crania in the context of two major extant carnivoran guilds: the East
African (including Serengeti) and the North American (“Yellowstone”) carnivore faunas
(Van Valkenburgh 1988). Overlap in the distributions of each species in a morphospace
implies potential overlap in ecomorphology, assuming form, function, and ecology are
closely linked (Lauder 1995, 1996, Raia 2004). The modern species used to represent
each of the two guilds are listed in Table 1.1; the fossil canid and hyaenid species are in
Table 1.2. A total of 413 modern and fossil specimens were used in the analyses
(specimens used are listed in Appendix A, online at http://dx.doi.org/10.1666/10007.s1).
All specimens were photographed with an Olympus μ720sw digital camera
(images with 3072 × 2304 pixels); the skulls were positioned so that in lateral view their
30
mid-sagittal plane is parallel to the lens of the camera, and in dorsal and ventral view the
frontal plane is parallel to the camera. Where possible, a grid was placed behind the
cranium so any distortion at the edge of the photos could be detected. Crania placed
within the central 30% of the viewing area of the camera displayed no distortion around
the edge of the specimens; cranial morphology was therefore faithfully reproduced by the
photos. Several rare species were included in the analyses by digitizing published
photographs of the specimens (Table 1.2); only complete fossil specimens were included,
and no attempt was made to test for distortion in these published photos. Sets of dorsal (n
= 8), lateral (n = 12), and ventral (n = 9) anatomical landmarks were used to represent the
craniodental shapes (Table 1.3, Fig. 1.2). Only Type I and II landmarks were used; these
types of landmarks represent joints, sutures, or features of local maxima that could be
unambiguously identified across the taxonomic sampling chosen in the study (Bookstein
1991). Although many more such landmarks are present than the ones chosen, a balance
had to be struck between capturing the greatest amount of shape information and
including slightly less-complete fossil specimens for larger sample size and wider
taxonomic sampling.
Landmarks were digitized from photographs using TPSDig 2.05 (Rohlf 2006a),
which converted points marked on the photographs into Cartesian x,y coordinates. The
coordinates were then superimposed by a Procrustes fit using CoordGen6 in the
Integrated Morphometrics Package (IMP) (Sheets 2004), where specimens were scaled,
rotated, and aligned on their landmarks, their differences calculated, and absolute size
removed. This operation allowed comparison of the shapes in isolation of their relative
31
sizes, at the same time preserving any allometric effects that might be further analyzed.
Relative warp analysis (RWA) was performed on the entire data set for each view in
TPSRelw (Rohlf 2006b). Morphological variation as summarized by RWA can be
visualized by bivariate plots of relative warp axes, essentially a principal components
analysis weighed by bending energy of shape changes; relative differences in shape were
plotted as thin-plate splines that display deformation grids relative to the mean shape
configuration represented by the origin of the bivariate RW plots (Zelditch et al. 2004).
Because shape differences are commonly summarized by the first few relative warp axes,
as in principal components analysis, we chose to use only the axes that explained more
than 10% of the total variance in shape to visualize evolutionary trends.
Because of differences in the quality and number of preserved skulls among fossil
canids and hyaenids, the fossil specimens were grouped by genera in our analyses;
specimens of congeneric species were combined in calculating mean values in the
relative warp analysis. Although this action has an underlying assumption that cranial
shape changes little within a genus (which may or may not be true), the relatively small
sample sizes of even the best collections of these fossil species necessitated a less
detailed analysis than one possible with modern species. Incorporating more collections
and additional discoveries of fossil canids and hyaenids may increase the sample size and
quality of the current data, and this is probably more so for hyaenids. The fossil canids
used were mainly from the extensive Childs Frick collection of the American Museum of
Natural History. Our fossil hyaenid data come mainly from two collections: the AMNH
32
Frick Collection and HMZ; these data were supplemented with specimens from other
museums when exceptionally complete or rare specimens were available.
To test for allometric effects of cranial shape change, fossil canid and hyaenid
shape data were regressed against the natural logarithm (ln) of centroid size (CS) using
TPSRegr 1.31 (Rohlf 2005). Overall shape as represented by partial warp and uniform
component scores of the specimens were regressed against ln (CS), and multivariate tests
of significance were conducted to evaluate the strength of correlation between size and
shape. In addition, RW axes that explained more than 10% of variance in shape were
examined by regressing genus mean RW scores for those axes against mean ln (CS). To
account for effects of phylogeny on the correlation between skull shape and size, we used
the PDAP module (Midford et al. 2008) of Mesquite (Maddison and Maddison 2009) to
calculate phylogenetic independent contrasts of genus mean RW scores and ln (CS), and
regression analysis was performed on the contrasts. Contrasts were calculated using the
tree topology in Figure 1.1, with equal branch lengths.

Chapter One Results

Dorsal Cranial Shape.—RW1 and RW2 scores were plotted to show fossil canids
and hyaenids with extant faunas of North America and East Africa, respectively (Fig.
1.3A,D). RW1 accounted for 45.42% of the variance in shape, and RW2 accounted for
18.43%. Borophagines and hyaenids occupy similar areas in morphospace, but the former
have more positive RW2 scores, in general. Both display little overlap with extant
33
carnivorans in their respective faunas. Borophagines are bounded in morphospace by
small-bodied generalists (more positive RW1 scores), canines (more negative RW1
scores), hesperocyonines (more positive RW2 scores), and Ursus (more negative RW2
scores) (Fig. 1.3A). Hyaenids are bounded by small-bodied generalists (more positive
RW1 scores), canines (more negative RW1 scores), and Panthera (more negative RW2
scores) (Fig. 1.3D).
The most ancestral genera of borophagines and hyaenids tend to have large,
positive RW1 scores, corresponding to shorter rostrum and relatively long cranial region
(Fig. 1.4A; e.g., Ota, Por). Subsequent borophagine genera and doglike hyaenids tend to
have more negative RW1 scores, corresponding to longer rostrum and relatively short
cranial region (Fig. 1.4A,B; e.g., Mic, Pro, Hwh, Lyc). The derived bone-cracking
hyaenids, however, have more positive RW1 scores as in the ancestral forms (Fig. 1.4A,B;
e.g., Ccr, Iab). The derived borophagines (Bor, Epi, Ael) have negative RW1 scores
overlapping those of stem borophagines as well as the doglike hyaenids, and do not
approach the derived hyaenids in morphospace. Along the RW2 axis, hyaenids tend to
have more negative RW2 scores, corresponding to a wider cranium, more caudally set
postorbital processes, and more elongate temporal fossae than borophagines (Fig. 1.4C).
In this regard, borophagines evolved toward the condition seen in hyaenids with
increasingly negative RW2 scores (Fig. 1.4B).
Lateral Cranial Shape.—Borophagines and hyaenids, again, share adjacent
positions in morphospace, and the former in general have more positive RW2 scores
compared to hyaenids (Fig. 1.3B,E). Borophagines are bounded by canines (more
34
positive RW1 scores), small-bodied generalists (more negative RW1 scores and more
positive RW2 scores), and hesperocyonines (negative RW2 scores) (Fig. 1.3B). Hyaenids
are bounded by canines (more positive RW1 scores), small-bodied generalists (negative
RW1 scores), and felids (positive RW2 scores) (Fig. 1.3E). RW1 accounted for 36.98%
of total variance in shape, with RW2 accounting for 18.09%, and RW3 15.29% (Fig. 1.5).
In lateral shape, ancestral borophagines and hyaenids tend to have more negative
RW1 scores, closer to the extant small-bodied generalists, and corresponding to shorter
rostrum, longer cranial region, and dorso-ventrally compressed skull (Fig. 1.5A–C; Ota,
Arc, Tsp). More derived forms tend to have increasingly positive RW1 and RW2 scores,
corresponding to a deepening cranial region, a more arching dorsal skull, and a shortened
and deepened rostrum (Fig. 1.5A, Ccr, Bor). The two lineages are parallel in this trend.
The derived bone-cracking borophagine Borophagus (Bor) and hyaenid Crocuta (Ccr)
both share the most positive RW1, RW2, and the most negative RW3 scores in their
respective lineages, corresponding with further deepening of the cranial region and a
more caudal position of the cheek dentition (Fig. 1.5C,F).
Ventral Cranial Shape.—Borophagines and hyaenids occupy distinct and non-
overlapping regions in ventral cranial morphospace (Fig. 1.3C,F). Borophagines are
bounded by small-bodied generalists (positive RW1 scores), canines and hesperocyonine
canids (positive RW2 scores), and felids (negative RW2 scores). Hyaenids are bound by
canines (more positive RW1 and RW2 scores) and felids (more positive RW1, more
negative RW2 scores). RW1 accounted for 43.66% of total variance in shape, with RW2
accounting for 19.10%, and RW3 13.91% (Fig. 1.6A,D).
35
In ventral view more derived forms tend to have more negative RW1 scores in
both borophagines and hyaenids (Fig. 1.6A,B; e.g., Ccr, Hhy, Ael, Bor); this
corresponded to widening of the palate and more caudally set cheek dentition (Fig. 1.6C).
The more derived forms also tend to have more negative RW2 values (except for the
stem borophagine Otarocyon), also corresponding to widening of the palate (Fig. 1.6B,C).
Derived borophagines have more negative RW3 scores, corresponding to more elongate
palate and more caudal position of cheek dentition (Fig. 1.6D–F, Bor, Ael). This last
trend is not observed in hyaenids (Fig. 1.6E).
Regression Analysis.—Ln (CS) was not significantly correlated with overall
cranial shape in fossil hyaenid specimens in dorsal, lateral, or ventral analyses (Table 1.4).
For borophagines, only dorsal cranial shape showed significant correlation with size
(Wilks’s λ = 0.37, F
(12,31)
= 4.38, p < 0.0001). Among the RW axes visualized for
evolutionary trends (Figs. 1.4-6), none had significant correlation with size in regression
analyses using independent contrasts (Table 1.4).

Chapter One Discussion

To examine the morphological pathways that led to the independent evolution of
bone-cracking hypercarnivores in borophagine canids and hyaenids, we conducted
geometric morphometrics analyses of cranial shape in the phylogenetic context of the two
lineages. The evolutionary trends observed in dorsal, lateral, and ventral cranial shape
changes provided strong evidence for convergent evolution not only in the specialized
36
bone-cracking hypercarnivores, but also in species along the morphological pathways
upon which they evolved. Morphological pathways through phylogeny were parallel in
direction for ancestral, intermediate, and specialized borophagines and hyaenids. Even
though cranial shape showed largely convergent or parallel changes in the two lineages,
major differences still existed between them in the extent of the morphological
trajectories along those similar pathways. Phylogenetic constraint in canids is a likely
factor contributing to those differences, as demonstrated below.
Hyaenids and canids are two of the longest surviving carnivoran lineages, with
little interchange between their respective places of origins in the Old and New Worlds
during the Miocene, and their morphological similarities have been used as evidence for
ecological similarity (Qiu 2003, Werdelin and Solounias 1991). Major ecomorphological
groups of modern carnivorans (catlike hypercarnivores, canines, small-bodied generalists)
are clearly segregated in our analyses, suggesting that the cranial shape morphospace
captures some ecological information (Fig. 1.3). Within this context, overall similarity
between borophagines and hyaenids (Fig. 1.3A,B and D,E) is consistent with competitive
exclusion as an ecological mechanism to explain the lack of intercontinental dispersal
during the Miocene, despite the fact that many other large, cursorial carnivorans have
managed to immigrate to North America. In other words, if fossil borophagine canids and
hyaenids were to become sympatric in North America, they would have overlapped more
in morphospace with each other than they would with other carnivorans (Fig. 1.3). In
dorsal cranial shape the hyaenids Lycyaena (Lyc) and Chasmaporthetes (Clu), and the
doglike Hyaenictitherium (Hwh) are very similar to borophagines (Fig. 1.4). In lateral
37
cranial shape these and other hyaenid genera are adjacent to borophagines, but also
overlap with hesperocyonines (Fig. 1.3B,E). Ictitherium has been interpreted as jackal-
like, and Hyaenictitherium hunting doglike, on the basis of overall assessment of
craniodental morphology (Werdelin and Solounias 1991). These genera, and in fact all
but one hyaenid genus, were restricted to the Old World. The only hyaenids to disperse to
North America were members of the genus Chasmaporthetes, which have evolved gracile
postcranial skeletons adapted for increased cursoriality (Berta 1981, Ewer 1973, Kurtén
1968, Kurtén and Werdelin 1988).
In the Old World, specialized bone-cracking hyaenids evolved as early as the late
Miocene (excluding middle Miocene Percrocuta and late Miocene Dinocrocuta, which
may be sister to true hyaenids), represented by genera such as Adcrocuta (Werdelin and
Solounias 1991, Werdelin and Solounias 1996b). This niche continued to be occupied in
the Plio-Pleistocene by Pliocrocuta, Pachycrocuta, and Crocuta (Turner and Antón
1996). Borophagine canids were never found in the Old World, and the first canids
known to cross Beringia were members of Caninae, which had more elongate skulls
lacking the morphological adaptations of bone-crackers (Qiu 2003, Wang et al. 2008).
These observations also provide support for the ecological competitive exclusion model.
Members of these two lineages not only overlapped in their derived bone-cracking
morphology, but also paralleled each other in many aspects of cranial shape changes
during their evolution (Figs. 1.4-6). Accordingly, these two lineages remained allopatric
for much of the Miocene and Pliocene, during which time other carnivorans repeatedly
dispersed across Beringia (Qiu 2003, Werdelin and Solounias 1991).
38
Evolutionary Trends.—Because of the comparable diversity and inferred
ecomorphological convergence between hyaenids and borophagine canids (Van
Valkenburgh 2007, Wang et al. 1999, Werdelin 1989), we concentrated our sampling of
canids within borophagines and hesperocyonines (Tables 1.1, 1.2). Nevertheless, the
extant canids examined in this study cluster in their morphospace distribution and are
distinct from most borophagines and hesperocyonines (Fig. 1.3). In dorsal and lateral
cranial shape, the three subfamilies of canids occupied different areas of morphospace:
canines have more negative dorsal RW1 and more positive lateral RW1 scores than
borophagines, which in turn have less positive dorsal RW2 and more positive lateral
RW2 scores than hesperocyonines (Fig. 1.3A,B). These differences corresponded to
widening of the palate, deepening of the cranial region, and doming of the forehead
between hesperocyonines and borophagines, and then elongation of the rostrum and
shortening of the cranial region between the extinct subfamilies and canines (Fig. 1.4C,
1.5C). Even though the three canid subfamilies evolved independently from a common
ancestor very much like Hesperocyon, and hypercarnivory evolved iteratively within
each clade (Van Valkenburgh 1991, Van Valkenburgh et al. 2004), their distributions in
morphospace remained distinct. The evolutionary pathways taken within borophagines
appear to be a continuation of the trends already seen during the evolution of
hesperocyonines and borophagines; the latter replaced the former as the dominant canid
hypercarnivores of the Neogene (Wang et al. 1999). Canines, however, did not evolve
specialized bone-cracking ecomorphs (Tedford et al. 2009). Thus, our results show that
39
morphospace occupied by hyaenids in East Africa is taken up by hesperocyonines and
borophagines in North America, as consistent with previous interpretation (Fig. 1.3).
Even though the different cranial views provided somewhat different
morphospace arrangements of borophagines and hyaenids, the overarching pattern of
convergence is quite clear (Figs. 1.4-6). The differences that remain are aspects of
borophagine evolution we interpret as products of phylogenetic constraint. Both the
extensive overlap between intermediate borophagines and doglike hyaenids in dorsal
cranial shape and the subsequent separation of derived bone-cracking hyaenids from their
counterparts in Borophaginae are indicative of constraints in the dorsal cranial shape of
borophagines (Fig. 1.4B). Bone-cracking hyaenids evolved a shorter rostrum and
relatively longer cranial regions compared to their doglike predecessors (Fig. 1.5). These
changes are likely associated with enlargement of the masticatory musculature, by
accommodating more muscle attachment areas via expansion of the cranial regions (Fig.
1.4C). Furthermore, the relative shortening of the rostrum created a more favorable
arrangement of the palate for masticatory mechanical advantage (Greaves 1985a,
Werdelin 1989). Derived borophagines appear to be restricted to the part of dorsal cranial
morphospace that corresponds to more elongate rostrum, and this could be a constraint
imposed by the retention of a full set of molars in their cheek dentition. This constraint is
also evident in ventral cranial shape (see below). In some cases, however, terminal
Borophagus specimens do show a very short rostrum comparable to the degree seen in
hyaenids, suggesting that intraspecific variation may exemplify more extreme
morphologies than is observed on the genus level.
40
In contrast, lateral cranial shape does not appear to be restricted in morphological
trajectory in borophagines, and a complete parallel can be seen between them and the
hyaenids (Fig. 1.5B,E). More specifically, in contrast to the consistent trend of increasing
RW1 scores through evolution, RW2 and RW3 scores do not show a dramatic change in
either borophagines or hyaenids until the terminal Borophagus and the modern hyaenines
(Crocuta, Hyaena, Parahyaena) appeared, respectively (Fig. 1.5; Bor, Ccr, Hhy, Pbr). In
these derived genera, RW2 values are more positive, and RW3 more negative, than in
other genera in their respective lineages, corresponding to deeper and more dorsally
domed skulls (Fig. 1.5C,F). These morphological features are associated with increased
skull strength and better dissipation of masticatory stress in extant bone-crackers (Tanner
et al. 2008, Tseng 2009). Parallel changes in these aspects of morphology, even though
identical in their direction and timing, still did not create overlap in the borophagine and
hyaenid ecomorphs. The trends are therefore entirely parallel in nature.
Ventral cranial shape changes are also parallel between borophagines and
hyaenids in their direction of evolution. However, the morphospaces traversed by the two
groups are almost entirely offset along RW1 and RW3 axes (Fig. 1.6E). These offsets
indicate that even in the most derived borophagines, the extent of rostral widening and
caudal placement of the cheek dentition do not approach the degree seen in the stem
generalist hyaenid Tungurictis (Fig. 1.6D,E; Tsp). This observation may be explained by
the same phenomenon seen in dorsal cranial shape changes, where rostrum shortening is
restricted in the derived borophagines. The full set of molars present in canids would,
again, constrain the degree of caudal movement of the premolar dentition. The functional
41
implications of this phylogenetic constraint in canids have already been discussed by
Werdelin (1989), who proposed that as a consequence of molar retention, borophagine
canids used their carnassials as the main bone-cracking teeth in the upper dentition,
instead of the third premolars as in hyaenids (Fig. 1.2). This alternative “solution” to
functional demands of bone-cracking would have hindered the ability of bone-cracking
borophagines to use their carnassials for shearing meat (the plesiomorphic function of
those teeth), shortly after full eruption and use.
The carnassial-based bone-cracking specialization that evolved in borophagines
may be different from the morphological arrangement seen in hyaenids, but those derived
ecomorphs do appear to share functional advantage in bite force and skull strength over
those of their non-robust relatives (Tseng 2009, Tseng and Wang 2010). Thus, alternative
“solutions” to similar functional demands may be equally sufficient, at least in a
functional sense. On the macroevolutionary scale, however, repeated specialization for
hypercarnivory in canids has been shown to decrease long-term evolutionary potential
(Van Valkenburgh et al. 2004, Werdelin 1989) as well as morphological disparity in
those lineages (Holliday and Steppan 2004). The interaction between phylogenetic
constraint (which may be partially derived from previous bouts of specialization) and the
process of adaptation resulted in convergent bone-cracking ecomorphs between
borophagines and hyaenids, but the former was accordingly restricted in the distance
traveled in the same morphological pathways through evolution (Figs. 1.4-6). On top of
that, the loss of carnassial shearing function is arguably detrimental to long-term
evolutionary fitness of the lineage: the Borophaginae became extinct at the end of the
42
Blancan North American Land Mammal Age, with the bone-cracking Borophagus as its
last remaining members (Wang et al. 1999). Many of our findings are consistent with this
“macroevolutionary ratchet” model of specialization (Van Valkenburgh et al. 2004), but
our results also show that there is still some flexibility even in lineages with restricted
evolutionary potential, and functionally comparable bone-cracking ecomorphs can
nevertheless arise through parallel processes under different background constraints.
In our preceding discussion of evolutionary trends with regard to hypercarnivory,
we have largely ignored the myrmecophagous aardwolf, Proteles cristatus (Pcr in
figures). Given its peculiar adaptation for a hyaenid, its placement in the evolutionary
changes leading to bone-cracking hypercarnivory is unclear. Furthermore, the ancestry
and phylogenetic position of the aardwolf is still controversial;, morphology and third-
position transversions of the cytochrome b gene seem to indicate a much earlier
divergence from other hyaenids (Jenks and Werdelin 1998, Werdelin and Solounias 1991)
than suggested by molecular analysis with the mitochondrial cytochrome b sequence
(Koepfli et al. 2006). In our analysis, the Hyaenidae seem to follow a more or less
gradual path through morphospace in dorsal, lateral, and ventral cranial shapes; the
position of Proteles, offset from the main course of hyaenid evolution (Figs. 1.4-6),
creates difficulties in resolving its functional or phylogenetic position using only cranial
shape. Therefore, more detailed analyses regarding its evolutionary affinities are required
before it can be interpreted within the current context.
Among other extant hyaenids, only Crocuta hunt large prey and are observed to
routinely consume large amounts of bone (Kruuk 1972b). Parahyaena (Pbr) and Hyaena
43
(Hhy) are generalists, consuming fruit and vegetables as a sizable proportion of their diet,
and do not actively hunt prey larger than themselves (Mills 1990, Mills and Mills 1978).
Dorsal and ventral cranial shape analyses did not differentiate between these dietary
differences, and the derived hyaenines share more positive RW1 scores in dorsal shape
(Fig. 1.4A) and more negative RW1–RW3 scores in ventral shape (Fig. 1.6A,D)
compared to other hyaenids. Differentiation is observed in lateral cranial shape, however.
The derived hyaenines have similar lateral RW2 and RW3 scores, but lateral RW1 scores
in the large generalists are more negative than in most other hyaenids (Fig. 1.5B,E). In
the extant carnivoran faunas, species with the most negative lateral RW1 scores are
small-bodied generalists (Fig. 1.3B,E). The generalist diet of Hyaena and Parahyaena is
thus correlated with lateral RW1 scores that are more negative than in the bone-cracking
hypercarnivores, and is similar to the diet of ancestral hyaenids and extant generalist
carnivorans in this regard. This differentiation, which is visible only in lateral cranial
shape, highlights the importance of examining different views of the skull in interpreting
shape evolution trends.
Among the bone-crackers, whereas Borophagus showed parallel lateral cranial
shape changes with Crocuta relative to their near relatives in lateral RW1–RW3 scores,
the former approached hyaenids in ventral cranial shape only in ventral RW2 and RW3
scores. The borophagine Aelurodon was by far the closest to the hyaenids in ventral RW1
score (Fig. 1.6, Ael). The negative RW1 score corresponds to a widening palate and a
more caudally positioned cheek dentition; this is associated with the different
evolutionary direction toward hypercarnivory taken by Aelurodon compared to
44
Borophagus. Whereas Borophagus evolved bone-cracking capability by adapting the
upper P4 (and corresponding lower p4–m1) as a biomechanically favorable locus for
durophagy, instead of the third premolar used by hyaenids (Werdelin 1989), Aelurodon
and its sister genera showed increasing size in all of the premolars through their evolution
(Wang et al. 1999). The latter trend is similar to that observed in hyaenids; accordingly,
the caudal shift of the cheek dentition proceeded furthest in Aelurodon among
borophagines. This may indicate increased reliance on anterior premolars for durophagy
in Aelurodon relative to Borophagus. Even so, the caudal shift of Aelurodon’s palate is
still far from the degree seen in the most derived hyaenids (Fig. 1.6). Again, the
constraint of having posterior molars would have prevented further evolutionary changes
toward more negative ventral RW1 scores from occurring in borophagines.
The morphological insights obtained from cranial shape analysis allow us to
outline a more detailed history of shape evolution and the associated specialization in the
two clades. Parallel evolutionary pathways in cranial shape were observed in the lateral
and ventral shape analyses: ancestral borophagines and hyaenids had a short rostrum, an
elongate but flat cranial region, more elongate dentition, and narrower palates (Figs. 1.5,
1.6). In dorsal view, the doglike hyaenids overlapped with most borophagine canids in
having a moderately elongate rostrum and short temporal fossae (Fig. 1.3). Lateral and
ventral cranial shape of doglike hyaenids evolved in parallel with their borophagine
counterparts by gradual deepening of the skull, shortening of the rostrum, elongation of
the cranial region, doming of the forehead, and widening of the palate (Figs. 1.5, 1.6).
Evolution of specialized bone-crackers largely continued this trend in lateral and ventral
45
shape, but in dorsal shape the rostrum was further shortened only in derived hyaenids,
and not borophagines (Fig. 1.4). These trends are summarized by vectors indicating
directions of shape change through evolution, superimposed onto deformation grids (Fig.
1.7). The result of these morphological changes over time is that the P4 in borophagine
canids and P3 in hyaenids are in approximately the same positions relative to the
forehead and the temporomandibular joints (TMJ), providing mechanical advantage with
which to efficiently produce large bite forces required in bone-cracking (Greaves 1983,
2000, Werdelin 1989).
Tests for correlation between cranial shape and size indicated that allometry did
not play a major role in the shape changes observed in hyaenids (Table 1.4). This is true
for borophagines as well, except for overall dorsal cranial shape, which was significantly
correlated with size (p < 0.0001). A basic explanation for an allometric trajectory of
cranial shape evolution would be that the physical requirements of movement simply
require different shapes to evolve at different body sizes (Gould 1966). In the case of
mastication, proportionally larger muscle attachment areas with larger body size may be
expected, for example, simply to maintain comparable force production in larger-bodied
forms (e.g., to move heavier jaws, the required muscle force is proportional to muscle
cross-section area, but mass is directly proportional to volume), regardless of whether
those larger species have bone-cracking adaptations. However, dental specialization in
canids toward hypercarnivory evolved in concert with enlarging body size (Van
Valkenburgh et al. 2004), suggesting that larger species were also changing their ecology,
perhaps under energetic constraint of their food choice (Carbone et al. 1999).
46
Furthermore, size allometry is not significantly correlated with borophagine cranial shape
other than in the dorsal shape analysis, suggesting that the majority of evolutionary
changes in the borophagine skull are associated with other factors, namely functional
demands. As discussed above, dorsal cranial shape in borophagines appears restricted in
rostral length reduction as a product of phylogenetic constraint. Therefore, derived
borophagines were not able to parallel hyaenids along dorsal RW1 toward more positive
scores, effectively highlighting size allometry as a means of shape change instead (Fig.
1.4). The allometric trends show increasingly tubular dorsal skulls in larger borophagines
(data not shown), which may be associated with resisting torsional forces that travel
laterally across the skull. Such allometry does not characterize the Hyaenidae, however;
the lack of significant correlations between cranial shape and size in hyaenids indicates
that evolutionary shape changes were associated with selection and functional demands.
This is further evidenced by the fact that hyaenid bone-crackers evolved toward more
extreme RW1 scores (dorsal, lateral, and ventral) than any of the borophagines, a trend
that is associated with shortened rostrum and reduced dentition involving the
evolutionary reduction and loss of the first and second upper molars (Werdelin and
Solounias 1991).
A recent study on the cranial biomechanics of fossil canids demonstrated that the
derived Epicyon and Borophagus possess bone-cracking capability far superior to that of
living canines and fossil hesperocyonine canids (Tseng and Wang 2010). In light of our
current findings, it would be interesting to further compare cranial biomechanics across
fossil hyaenids and borophagines to see if parallel morphological pathways correspond
47
with convergent biomechanics in derived as well as intermediate and ancestral forms.
Bite force studies done on domestic canids with different cranial shapes showed
significant correlation between brachycephalic skulls and higher bite force (Ellis et al.
2009); the patterns observed in this study suggest that a similar relationship might be
expected for an evolutionary sequence of borophagines and hyaenids. In addition, one
should also examine the borophagine Aelurodon to determine whether its skull
biomechanics differ from Borophagus because of its premolar enlargement and more
extreme caudal movement of the cheek dentition, both features closely approaching those
observed in hyaenids. Along another line of inquiry, analysis of mandibular shape across
a broader sample of carnivorans demonstrated that allometry is a main factor in
determining corpus shape across fissiped families; Meloro et al. (2008) predicted that the
mandibles of borophagine canids could adapt to hypercarnivory via allometry. In addition,
the corresponding lower teeth (p4 and m1 in borophagines) that became adapted for
durophagy may have evolved in close association with changes in mandibular shape, as
observed in the evolution of the upper dentition with cranial shape. Whether or not the
cranium and mandible of borophagine canids show similar adaptive pathways and
constraint to specialization remains an interesting topic to be further explored. Beyond
the skull, there is evidence from studies of borophagine postcranial functional
morphology that indicates hunting techniques potentially quite different from those of
modern hyenas (Andersson 2005), suggesting that current ecomorphological categories
may not be defined with sufficient understanding of the organisms as a whole. How
constraints in cranial shape evolution demonstrated here might interact with postcranial
48
evolution is unclear, and future work should integrate these aspects to gain a fuller
understanding of this ecomorphology and to better define such a prominent example of
convergence.
Implications for Understanding Convergence.—Morphological convergence,
being a broad and complex topic, can nevertheless be summarized by several scenarios of
expected outcomes (Fig. 1.8). True morphological convergence occurs where two
lineages, starting with divergent morphologies, evolved to become identical or
overlapping in morphology (Fig. 1.8A). However, a converging trend that does not end in
overlapping morphologies can also exist, in theory (Fig. 1.8B). Finally, two lineages may
be considered convergent if they evolved via parallel pathways of shape change, perhaps
along a similar functional continuum (Fig. 1.8C). Stayton (2006) attempted to quantify
differences between these scenarios by using mathematical means to distinguish between
the first two cases. In reality, the boundaries between any of these scenarios may exist as
continuums in themselves, making categorization difficult. Examination of evolutionary
trends characterized by the RW axes in this study (Figs. 1.4-6) indicates that
morphological evolution of borophagines and hyaenids is best described by largely
parallel trends of change (Fig. 1.8C).
To demonstrate this, the overall evolutionary trends given by the different cranial
shape analyses were simplified as in the theoretical scenarios (Fig. 1.8D–F). Dorsal shape
evolution in borophagines overlapped with that of earlier hyaenids (Fig. 1.8D); however,
borophagines did not evolve any further in shortening the rostrum. Derived, bone-
cracking hyaenids were unique in this regard (Fig. 1.4). In lateral cranial shape,
49
borophagines and hyaenids never converged, but evolved in parallel toward the bone-
cracking ecomorphology (Figs. 1.5, 1.8E). In ventral cranial shape, parallel evolutionary
pathways were also observed for the two lineages; however, borophagines never reached
the extent hyaenids did in caudal movement of the dentition (Figs. 1.6, 1.8F). These
representations of evolutionary trends show additional complexity in actual observations
that may not be easily categorized, especially when pervasive phylogenetic constraint
may be present in lineages undergoing a “macroevolutionary ratchet” model of
specialization. Further studies of similar evolutionary cases may shed light on whether
interaction between constraint and specialization could produce different outcomes than
the ones seen here.
Limitations.—Geometric morphometrics analysis is a powerful technique that can
be used to compare overall morphological shape within a sample of specimens, yielding a
more informative comparison than isolated univariate or bivariate measurements
(Zelditch et al. 2004). However, the necessity of comparing homologous landmarks is a
limitation in the analysis of incomplete specimens. The breadth of the skull across the
zygomatic arches traditionally has been used as a measure of adductor muscle mass
(Binder and Van Valkenburgh 2000, Tanner et al. 2010), and this information would have
been informative in our shape analyses as to the evolutionary changes associated with
masticatory musculature size. However, fossil specimens often have incompletely
preserved zygomatic arches, and in our sample of fossil canids and hyaenids most are not
well preserved (e.g., deformed or missing) in this regard. Therefore, we were constrained
to dorsal and ventral landmarks that extend laterally only to the base of the zygomatic
50
arches (Fig. 1.2). Multivariate calculations that estimate the position of missing
landmarks have been proposed (Adams et al. 2004), but the objectivity of those methods
and their influence on important functional landmarks, which could be used to interpret
macroevolutionary trends, remain to be further evaluated.

Chapter One Conclusion

We utilized geometric morphometrics analysis to study the evolution of cranial
shape changes in fossil borophagine canids and hyaenids, two carnivoran groups that
have converged multiple times in wolf-like and hyena-like ecomorphologies. Results
showed parallel morphological pathways of change throughout their evolutionary
histories, ending in specialized bone-cracking hypercarnivores. This pattern of parallel
change was mediated by phylogenetic constraint in borophagines, limiting the amount of
change possible in the shape and length of the rostrum. These and other findings indicate
that although specialization of the bone-cracking ecomorphology was limited in the
extent of morphological change in borophagines, bone-cracking function was not
necessarily compromised. Interaction between phylogenetic constraint and adaptive
evolution may affect the subsequent evolutionary potential of lineages, but their
mechanisms require more in depth study. On the other hand, convergent ecomorphs may
share similar functional capability evolved via identical pathways, even if their initial
phylogenetic constraints differ. Differential effects of alternative functional “solutions”
that arise during morphological evolution may be multiplied with processes of the
51
“macroevolutionary ratchet” already in place to further limit the evolutionary pathways
available to specialized lineages.

Chapter One Acknowledgments

We thank J. Galkin, J. Kelly, J. Meng, X.J. Ni, and M. Spaulding for help with
and access to the American Museum of Natural History Frick Collection; S.Q. Chen and
W. He for access to specimens in the Hezheng Paleozoology Museum; Z.X. Qiu for
access to specimens in the Institute of Vertebrate Paleontology and Paleoanthropology; R.
Purdy for access to National Museum of Natural History fossil specimens; J. Liu for help
with data collection in the University of Alberta Museum of Zoology; J. Dines for help
with data collection in the Department of Mammalogy, Natural History Museum of Los
Angeles County; E. Lacey and C. Conroy for access to Museum of Vertebrate Zoology
specimens. G. Koufos provided information on Plioviverrops skulls and a photo of the
most complete specimen; T. Rowe and Digimorph.org provided access to a 3-D
reconstruction image of ‘Miacis’ cognitus; L. Werdelin provided discussion and helped
track down references. J. X. Samuels and an anonymous reviewer provided constructive
and detailed comments that greatly improved the structure and content of the manuscript.
N. Atkins performed detailed copyediting on the final version of the manuscript.
Research was supported by a National Science Foundation Graduate Research Fellowship,
a National Science Foundation Doctoral Dissertation Improvement Grant (DEB-
52
0909807), an American Museum Collection Study Grant, and a U.S. Fulbright Grant (to
Z.J.T.).

53
Chapter One References
Adams, D. C., F. J. Rohlf, and D. E. Slice. 2004. Geometric morphometrics: ten years of
progress following the ‘Revolution.’ Italian Journal of Zoology 71:5-16.

Andersson, K. 2005. Were there pack-hunting canids in the Tertiary, and how can we
know? Paleobiology 31:56-72.

Berta, A. 1981. The Plio-Pleistocene hyaena Chasmaporthetes ossifragus from Florida.
Journal of Vertebrate Paleontology 1:341-356.

Binder, W. J., and B. Van Valkenburgh. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252:273-283.

Bookstein, F. L. 1991. Morphometric tools for landmarks data. Cambridge University
Press, London.

Carbone, C., G. M. Mace, S. C. Roberts, and D. W. Macdonald. 1999. Energetic
constraints on the diet of terrestrial carnivores. Nature 402:286-288.

Christiansen, P. 2008. Evolution of skull and mandible shape in cats (Carnivora: Felidae).
PLoS ONE 3(7):e2807.

Crusafont-Pairó, M. 1950. El primer representante del género Canis en el Pontiense
eurasiatico (Canis cipio nova sp.). Boletín de la Real Sociedad Española de
Historia Natural (Geologia) 48:43-51.

Crusafont-Pairó, M., and J. Truyols-Santonja. 1956. A biometric study of the evolution of
fissiped carnivores. Evolution 10:314-332.

de Bonis, L., S. Peigne, A. Likius, H. T. Mackaye, P. Vignaud, and M. Brunet. 2007. The
oldest African fox (Vulpes riffautae n. sp., Canidae, Carnivora) recovered in late
Miocene deposits of the Djurab desert, Chad. Naturwissenschaften 94:575-580.

Ellis, J. L., J. J. Thomason, E. Kebreab, K. Zubair, and J. France. 2009. Cranial
dimensions and forces of biting in the domestic dog. Journal of Anatomy
214:362-373.

Ewer, R. F. 1973. The carnivores. Cornell University Press, Ithaca, N.Y.

Ferretti, M. P. 1999. Tooth enamel structure in the hyaenid Chasmaporthetes lunensis
lunensis from the late Pliocene of Italy, with implications for feeding behavior.
Journal of Vertebrate Paleontology 19:767-770.
54

———. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience 100:41-52.

Figueirido, B., P. Palmqvist, and J. A. Perez-Claros. 2009. Ecomorphological correlates
of craniodental variation in bears and paleobiological implications for extinct taxa:
an approach based on geometric morphometrics. Journal of Zoology 277:70-80.

Gould, S. J. 1966. Allometry and size in ontogeny and phylogeny. Biology Review
41:587-640.

Greaves, W. S. 1983. A functional analysis of carnassial biting. Biological Journal of the
Linnean Society 20:353-364.

———. 1985. The generalized carnivore jaw. Zoological Journal of the Linnean Society
85:267-274.

———. 2000. Location of the vector of jaw muscle force in mammals. Journal of
Morphology 243:293-299.

Hay, O. P. 1921. Descriptions of species of Pleistocene Vertebrata, types or specimens
most of which are preserved in the United States National Museum. Proceedings
of the United States National Museum 59:599-642.

Holliday, J. A., and S. J. Steppan. 2004. Evolution of hypercarnivory: the effect of
specialization on morphological and taxonomic diversity. Paleobiology 30:108-
128.

Jenks, S. M., and L. Werdelin. 1998. Taxonomy and systematics of living hyaenas
(Family Hyaenidae). IUCN/SSC Hyaena Specialist Group, Gland, Switzerland.

Koepfli, K.-P., S. M. Jenks, E. Eizirik, T. Zahirpour, B. Van Valkenburgh, and R. K.
Wayne. 2006. Molecular systematics of the Hyaenidae: relationships of a relictual
lineage resolved by a molecular supermatrix. Molecular Phylogenetics and
Evolution 38:603-620.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior. University
of Chicago Press, Chicago.

Kurtén, B. 1968. Pleistocene Mammals of Europe. Weindenfeld and Nicholson, London.

Kurtén, B., and L. Werdelin. 1988. A review of the genus Chasmaporthetes Hay, 1921
(Carnivora, Hyaenidae). Journal of Vertebrate Paleontology 8:46-66.

55
Lauder, G. V. 1995. On the inference of function from structure. Pp. 9-18 in J. J.
Thomason, ed. Functional morphology in vertebrate paleontology. Cambridge
University Press, New York.

———. 1996. The argument from design. Pp. 55-91 in M. R. Rose and G. V. Lauder, eds.
Adaptation. Academic Press, San Diego.

Maddison, W. P., and D. R. Maddison. 2009. Mesquite: a modular system for
evolutionary analysis, Version 2.72. http://mesquiteproject.org.

Martin, L. D. 1989. Chapter 19. Fossil history of the terrestrial Carnivora. Pp. 536-568 in
J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution. Cornell
University Press, New York.

McHenry, C., S. Wroe, P. D. Clausen, K. Moreno, and E. Cunningham. 2007.
Supermodeled sabercat, predatory behavior in Smilodon fatalis revealed by high-
resolution 3D computer simulation. Proceedings of the National Academy of
Sciences USA 104:16010-16015.

Meloro, C., P. Raia, P. Piras, C. Barbera, and P. O’Higgins. 2008. The shape of the
mandibular corpus in large fissiped carnivores: allometry, function and phylogeny.
Zoological Journal of Linnean Society 154:832-845.

Midford, P. E., T. Garland Jr., and W. P. Maddison. 2008. PDAP package of Mesquite,
Version 1.14. http://mesquiteproject.org/pdap_mesquite/

Mills, M. G. L. 1990. Kalahari hyenas: comparative behavioral ecology of two species.
Blackburn Press, Caldwell, N.J.

Mills, M. G. L., and M. E. J. Mills. 1978. The diet of the brown hyaena Hyaena brunnea
in the southern Kalahari. Koedoe 21:125-149.

Munthe, K. 1989. The skeleton of the Borophaginae (Carnivora, Canidae), morphology
and function. University of California Publications Bulletin of the Department of
Geological Sciences 133:1-115.

Qiu, Z.-x., ed. 2003. Dispersals of Neogene carnivorans between Asia and North America.
American Museum of Natural History, New York.

Raia, P. 2004. Morphological correlates of tough food consumption in large land
carnivores. Italian Journal of Zoology 71:45-50.

Rohlf, F. J. 2005. tpsRegr, Version 1.31. Department of Ecology and Evolution, State
University of New York, Stony Brook.
56

———. 2006a. tpsDig, digitize landmarks and outlines, Version 2.05. Department of
Ecology and Evolution, State University of New York, Stony Brook.

———. 2006b. tpsRelw, Version 1.44. Department of Ecology and Evolution, State
University of New York, Stony Brook.

Sheets, H. D. 2004. Integrated morphometrics package. Department of Physics, Canisius
College, Buffalo, N.Y.

Slater, G. J., and B. Van Valkenburgh. 2008. Long in the tooth: evolution of sabertooth at
cranial shape. Paleobiology 34:403-419.

Stayton, C. T. 2006. Testing hypotheses of convergence with multivariate data:
morphological and functional convergence among herbivorous lizards. Evolution
60:824-841.

Stefen, C. 1999. Enamel microstructure of recent and fossil Canidae (Carnivora:
Mammalia). Journal of Vertebrate Paleontology 19:576-587.

Stefen, C., and J. M. Rensberger. 2002. The specialized enamel structure of hyaenids
(Mammalia, Hyaenidae): description and development within the lineage—
including percrocutids. Zoologische Abhandlungen 52:127-147.

Tanner, J. B., E. R. Dumont, S. T. Sakai, B. L. Lundrigan, and K. E. Holekamp. 2008. Of
arcs and vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta
crocuta). Biological Journal of the Linnean Society 95:246-255.

Tanner, J. B., M. Zelditch, B. L. Lundrigran, and K. E. Holekamp. 2010. Ontogenetic
change in skull morphology and mechanical advantage in the spotted hyena
(Crocuta crocuta). Journal of Morphology 271:353-365.

Tedford, R. H., X. Wang, and B. E. Taylor. 2009. Phylogenetic systematics of the North
American fossil Caninae (Carnivora: Canidae). Bulletin of the American Museum
of Natural History 325:1-218.

Tseng, Z. J. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96:51-67.

Tseng, Z. J., and X. Wang. 2010. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology 271:1386-1398.

57
Turner, A., and M. Antón. 1996. The giant hyaena, Pachycrocuta brevirostris
(Mammalia, Carnivora, Hyaenidae). Geobios 29:455-468.

Turner, A., M. Antón, and L. Werdelin. 2008. Taxonomy and evolutionary patterns in the
fossil Hyaenidae of Europe. Geobios 41:677-687.

Van Valkenburgh, B. 1985. Locomotor diversity within past and present guilds of large
predatory mammals. Paleobiology 11:406-428.

———. 1988. Trophic diversity in past and present guilds of large predatory mammals.
Paleobiology 14:155-173.

———. 1989. Carnivore dental adaptations and diet: a study of trophic diversity within
guilds Pp. 410-436 in J. L. Gittleman, ed. Carnivore behavior, ecology, and
evolution. Cornell University Press, New York.

———. 1991. Iterative evolution of hypercarnivory in canids (Mammalia: Carnivora):
evolutionary interactions among sympatric predators. Paleobiology 17:340-362.

———. 1995. Tracking ecology over geological time: evolution within guilds of
vertebrates. Trends in Ecology and Evolution 10:71-76.

———. 1999. Major patterns in the history of carnivorous mammals. Annual Review of
Earth and Planetary Science 27:463-493.

———. 2001. Chapter 5. The dog-eat-dog world of carnivores: a review of past and
present carnivore community dynamics. Pp. 101-121 in C. B. Stanford and H. T.
Bunn, eds. Meat eating and hominid evolution. Oxford University Press, New
York.

———. 2007. Déjà vu: the evolution of feeding morphologies in the Carnivora.
Integrative and Comparative Biology 47:147-163.

Van Valkenburgh, B., and K.-P. Koepfli. 1993. Cranial and dental adaptations to
predation in canids. Symposium of the Zoological Society of London 65:15-37.

Van Valkenburgh, B., and R. E. Molnar. 2002. Dinosaurian and mammalian predators
compared. Paleobiology 28:527-543.

Van Valkenburgh, B., X. Wang, and J. Damuth. 2004. Cope’s Rule, hypercarnivory, and
extinction in North American canids. Science 306(5693):101-104.

Wang, X. 1994. Phylogenetic systematics of the Hesperocyoninae (Carnivora: Canidae).
Bulletin of the American Museum of Natural History 221:1-207.
58

Wang, X., R. H. Tedford, and B. E. Taylor. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243:1-391.

Wang, X., R. H. Tedford, and M. Antón. 2008. The dog family, Canidae, and their
evolutionary history. Columbia University Press, New York.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

———. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective. Pp.
582-624 in J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution.
Cornell University Press, New York.

Werdelin, L., and N. Solounias. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30:1-104.

———. 1996. The evolutionary history of hyenas in Europe and western Asia during the
Miocene. Pp. 290-306 in R. L. Bernor, V. Fahlbusch, and S. Rietschel, eds. Later
Neogene European biotic evolution and stratigraphic correlation. Columbia
University Press, New York.

Wesley-Hunt, G. D., and J. J. Flynn. 2005. Phylogeny of the Carnivora: basal
relationships among the carnivoramorphans, and assessment of the position of
‘Miacoidea’ relative to Carnivora. Journal of Systematic Palaeontology 3:1-28.

Zanno, L. E., D. D. Gillete, L. B. Albright, III, and A. L. Titus. 2009. A new North
American therizinosaurid and the role of herbivory in ‘predatory’ dinosaur
evolution. Proceedings of the Royal Society of London B 276:3503-3511.

Zelditch, M., D. Swiderski, D. Sheets, and W. Fink. 2004. Geometric morphometrics for
biologists. Elsevier, New York.

59
Figure 1.1. Phylogenies and cranial morphological diversity of Canidae (A) and
Hyaenidae (B). Data for canid phylogeny are from Wang (1994), Wang et al. (1999), and
Tedford et al. (2009) and for hyaenids from Werdelin and Solounias (1991) and Koepfli
et al. (2006). Genera used in this paper are labeled in black.

60
Figure 1.2. Dorsal (A), lateral (B), and ventral (C) anatomical landmarks used in
geometric morphometrics analysis. See Table 1.3 for explanation of the landmarks.

61
Figure 1.3. Relative warp plots of fossil canids and hyaenids grouped with extant North
American (A–C) and East African (D–F) carnivoran faunas, respectively. The first two
relative warp axes are displayed. Plots show dorsal (A,D), lateral (B,E), and ventral (C,F)
cranial landmark analyses. Shaded polygons indicate areas of morphospace occupied by
borophagine canids (A–C) and hyaenids (D–F).

62
Figure 1.4. Plots of relative warp scores and shape evolution in borophagine canids and
hyaenids, dorsal cranial landmarks. A, RW1 versus RW2 scores. B, Phylogenies plotted
over RW1 versus RW2 scores. C, Deformation grids showing shape changes represented
by RW1 and RW2 axes.

63
Figure 1.5. Plots of relative warp scores and shape evolution in borophagine canids and
hyaenids, lateral cranial landmarks. A, RW1 versus RW2 scores. B, Phylogenies plotted
over RW1 versus RW2 scores. C, Deformation grids showing shape changes represented
by RW1 and RW2 axes. D, RW1 versus RW3 scores. E, Phylogenies plotted over RW1
versus RW3 scores. F, Deformation grids showing shape changes represented by RW1
and RW3 axes. Symbols and abbreviations as in Figure 1.3. Note parallel evolutionary
changes in borophagine canids and hyaenids toward more positive RW1 scores.

64
Figure 1.6. Plots of relative warp scores and shape evolution in borophagine canids and
hyaenids, ventral cranial landmarks. A, RW1 versus RW2 scores. B, Phylogenies plotted
over RW1 versus RW2 scores. C, Deformation grids showing shape changes represented
by RW1 and RW2 axes. D, RW1 versus RW3 scores. E, Phylogenies plotted over RW1
versus RW3 scores. F, Deformation grids showing shape changes represented by RW1
and RW3 axes. Symbols and abbreviations as in Figure 1.3. Note lack of overlap between
borophagine canids and hyaenids in morphospace.

65
Figure 1.7. Evolutionary changes in cranial shape in borophagine canids (A–C) and
hyaenids (D–F). Plots show dorsal (A,D), lateral (B,E), and ventral (C,F) cranial
landmarks. Arrows indicate directions of evolutionary change. Illustrations in central
column depict location of landmarks on the skull. Dorsal and ventral landmarks mirrored
about mid-sagittal axis to aid interpretation.

66
Figure 1.8. Expectations of patterns of convergence between two lineages through a two-
dimensional morphospace (A–C), and observations made from the current study on
borophagine canids and hyaenids (D–F). A, Two lineages starting with different
morphologies evolved forms that converge in morphospace (arrows indicate directions of
morphological evolution). B, Two lineages evolved toward each other in morphology, but
do not overlap. C, Two lineages exhibit parallel pathways of morphological change and
do not converge on each other in morphology. Results from this study are depicted for
dorsal (D), lateral (E), and ventral (F) cranial shapes. The axes represent relative
evolutionary differences between lineages in morphospace, and not absolute time. A–C
modified from Stayton (2006). For D–F, squares represent stem borophagines, and circles
represent stem hyaenids. In dorsal cranial shape, stem borophagines and hyaenids shared
similar positions in morphospace, but hyaenids evolved a further reduced rostrum. In
lateral cranial shape, all observed evolutionary changes in borophagines and hyaenids
were parallel. In ventral shape, hyaenids and borophagines exhibited parallel changes of
different extent, and derived borophagines overlapped with stem hyaenids.

67
Table 1.1. Carnivoran species used to represent the East African and North
American modern carnivore guilds, respectively. Lists are in alphabetical order by
family then genus.

East Africa (incl. Serengeti) North America / Yellowstone
Canidae
1. Canis aureus (n = 8)
2. Lycaon pictus (n = 8)
3. Otocyon megalotis (n = 7)
Felidae
4. Acinonyx jubatus (n = 2)
5. Caracal caracal (n = 1)
6. Felis sylvestris (n = 7)
7. Panthera leo (n = 23)
8. Panthera pardus (n = 9)
Herpestidae
9. Atilax paludinosus (n = 5)
10. Bdeogale crassicauda (n = 5)
11. Herpestes sanguineus (n = 8)
12. Ichneumia albicauda (n = 2)
Hyaenidae
13. Crocuta crocuta (n = 45)
14. Hyaena hyaena (n = 2)
15. Parahyaena brunnea (n = 2)
16. Proteles cristatus (n = 6)
Mustelidae
17. Ictonyx striatus (n = 3)
18. Mellivora capensis (n = 1)
Nandiniidae
19. Nandinia binotata (n = 5)
Viverridae
20. Civettictis civetta (n = 4)
21. Genetta maculata (n = 8)
22. Genetta pardina (n = 1)
23. Genetta rubiginosa (n = 5)

Canidae
I. Alopex lagopus (n = 10)
II. Canis latrans (n = 10)
III. Canis lupus (n = 11)
IV. Urocyon cinereoargenetus
(n = 1)
V. Vulpes vulpes (n = 12)
Felidae
VI. Lynx canadensis (n = 5)
VII. Lynx rufus (n = 5)
VIII. Puma concolor (n = 10)
Mephitidae
IX. Mephitis mephitis (n = 4)
Mustelidae
X. Gulo gulo (n = 5)
XI. Martes pennanti (n = 10)
XII. Mustela frenata (n = 10)
XIII. Mustela nivalis (n = 1)
XIV. Neovison vison (n = 2)
XV. Taxidea taxus (n = 6)
Procyonidae
XVI. Procyon lotor (n = 3)
Ursidae
XVII. Ursus americanus (n = 10)
XVIII. Ursus arctos (n = 11)

68
Table 1.2. List of fossil canids, hyaenids, and stem fossil forms used in
the study. List for Canidae and Hyaenidae are in order of geological
and evolutionary appearance.

Canidae Hyaenidae
Hesperocyoninae
A. Hesperocyon gregarius (n = 13)
B. Paraenhydrocyon josephi (n = 3)
C. Mesocyon temnodon (n = 1)
M. coryphaeus (n = 4)
M. brachyops (n = 1)
D. Cynodesmus thooides (n = 3)
E. Enhydrocyon crassidens (n = 1)
E. pahinsintewakpa (n = 1)
E. stenocephalus (n = 1)
F. Osbornodon fricki (n = 1)
O. iamonensis (n = 1)

Borophaginae
G. Archaeocyon pavidus (n = 1)
A. falkenbachi (n = 1)
H. Otarocyon cooki (n = 1)
I. Rhizocyon oregonensis (n = 1)
J. Phlaocyon minor (n = 1)
P. multicuspus (n = 1)
P. leucosteus (n = 1)
K. Desmocyon thomsoni (n = 2)
D. matthewi (n = 1)
L. Paracynarctus sinclairi (n = 1)
M. Microtomarctus conferta (n = 2)
N. Protomarctus optatus (n = 3)
O. Tomarctus brevirostris (n = 1)
P. Aelurodon stirtoni (n = 1)
A. mcgrewi (n = 1)
A. ferox (n = 5)
A. taxoides (n = 1)
Q. Paratomarctus temerarius (n = 3)
P. euthos (n = 2)
R. Epicyon saevus (n = 5)
E. haydeni (n = 4)
S. Borophagus littoralis (n = 1)
B. secundus (n = 5)
Caninae
T. Canis dirus (n = 2)
Ictitherinae
a. Plioviverrops orbignyi (n = 2) †
b. Tungurictis spocki (n = 1)
c. Ictitherium sp. (n = 2)
d. Hyaenictitherium wongi (n = 10)
e. H. hyaenoides (n = 4)
Hyaeninae
f. Lycyaena dubia (n = 1)
g. Chasmaporthetes lunensis (n = 1)
h. Ikelohyaena abronia (n = 1)
i. Adcrocuta eximia (n = 11)

OTHER
Stem carnivoramorphans
‘Miacis’ cognitus (n = 1)
Ooedectes herpestoides (n = 1)
Prohesperocyon wilsoni (n = 1) *

Percrocutidae
Dinocrocuta gigantea (n = 7)

69
Table 1.3. Anatomical landmarks used in the respective geometric morphometrics
analyses. See Figure 1.2 for illustration.

Dorsal Lateral Ventral
1. Suture, premaxilla/maxilla 1.
Premaxilla/maxilla/canine
junction 1.Premaxilla/maxilla/canine junction
2. Dorsal rim, infraorbital foramen 2. Caudal maxilla/canine junction 2. Lateral border, P2/P3 junction
3. Rostral edge of orbit 3. Maxillar/P3/P4 junction 3. Lateral border, P3/P4 junction
4. Lateral tip of postorbital process 4.Caudal maxilla/P4 junction 4. Caudal tip, P4
5. Rostral base of zygomatic arch 5. Caudal suture, maxilla/jugal 5. Medial edge, post-glenoid process
6. Caudal base of zygomatic arch 6.Tip, post-glenoid process 6. Mastoid process (widest point of basicranium)
7. Caudal tip of sagittal crest 7. Inflection point, caudal and
ventral faces of temporal fossa
7. Ventral-midsagittal edge, foramen magnum
8. Suture, nasal/frontal 8.
Inflection point, sagittal/occipital
crests 8. Caudal-midsagittal edge, palatine
9.
Junction, frontal crest/dorsal edge
of cranium 9. Caudal-midsagittal point, incisive foramina
10. Dorsal edge, long axis of orbit
11. Rostral junction, nasal/premaxilla
12.
Mid-point, rim of infraorbital
foramen

70
Table 1.4. Tests for allometry: regression statistics for cranial shape variables versus
ln(Centroid Size). All partial warp (PW) and uniform component scores of fossil hyaenid
and borophagine specimens were regressed against ln(CS). Phylogenetic independent
contrasts (PIC) for average relative warp scores for each genus as presented in Figures
1.4–6 were regressed (through the origin) against PICs for ln(CS). Significant correlation
between dorsal shape and ln(CS) in borophagines is associated with narrower/more
tubular dorsal skulls in larger specimens.

Specimen PW/uniform scores Genera RW (PIC-adjusted)
Wilks’s λ F p F p
Hyaenidae Dorsal 0.64 F
(12,19)
= 0.88 0.58 F
(3,6)
= 1.02 0.45
(fossil only) Lateral 0.32 F
(20,16)
= 1.67 0.15 F
(3,7)
= 0.56 0.66
Ventral 0.43 F
(14,11)
= 1.03 0.49 F
(3,5)
= 0.31 0.82

Borophaginae Dorsal 0.37 F
(12,31)
= 4.38 <0.0001 F
(3,9)
= 0.96 0.45
Lateral 0.45 F
(20,29)
= 1.72 0.09 F
(3,9)
= 0.55 0.66
Ventral 0.67 F
(14,28)
= 0.98 0.50 F
(3,8)
= 1.82 0.22

71
Chapter Two: Finite Element Analysis as a method to study function

This chapter has been published as:

Tseng, Z. J., McNitt-Gray, J. L., Flashner, H., Wang X., and R. Enciso. 2011. Model
sensitivity and use of the comparative finite element method in mammalian jaw
mechanics: mandible performance in the Gray Wolf. PLoS ONE 6(4): e19171.
doi:10.1371/journal.pone.0019171.

A copy of the accepted manuscript begins on the next page.
72
Model sensitivity and use of the comparative finite element method in
mammalian jaw mechanics: mandible performance in the Gray Wolf

ZHIJIE JACK TSENG
1,2,*
, , JILL L. MCNITT-GRAY
1,3
, HENRYK FLASHNER
1,4
,
XIAOMING WANG
1,2
, AND REYES ENCISO
1,5

1
Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, Los Angeles, California, United States of America
2
Department of Vertebrate Paleontology, Natural History Museum of Los Angeles
County, Los Angeles, California, United States of America
3
Departments of Kinesiology, Biological Sciences, and Biomedical Engineering,
University of Southern California, Los Angeles, California, United States of America
4
Department of Aerospace and Mechanical Engineering, University of Southern
California, Los Angeles, California, United States of America

5
Ostrow School of Dentistry, University of Southern California, Los Angeles, California,
United States of America
*Corresponding author: jtseng@nhm.org
73
Chapter Two Abstract

Finite Element Analysis (FEA) is a powerful tool gaining use in studies of
biological form and function. This method is particularly conducive to studies of extinct
and fossilized organisms, as models can be assigned properties that approximate living
tissues. In disciplines where model validation is difficult or impossible, the choice of
model parameters and their effects on the results become increasingly important,
especially in comparing outputs to infer function. To evaluate the extent to which
performance measures are affected by initial model input, we tested the sensitivity of bite
force, strain energy, and stress to changes in seven parameters that are required in testing
craniodental function with FEA. Simulations were performed on FE models of a Gray
Wolf (Canis lupus) mandible. Results showed that unilateral bite force outputs are least
affected by the relative ratios of the balancing and working muscles, but only ratios
above 0.5 provided balancing-working side joint reaction force relationships that are
consistent with experimental data. The constraints modeled at the bite point had the
greatest effect on bite force output, but the most appropriate constraint may depend on
the study question. Strain energy is least affected by variation in bite point constraint, but
larger variations in strain energy values are observed in models with different number of
tetrahedral elements, masticatory muscle ratios and muscle subgroups present, and
number of material properties. These findings indicate that performance measures are
differentially affected by variation in initial model parameters. In the absence of validated
input values, FE models can nevertheless provide robust comparisons if these parameters
74
are standardized within a given study to minimize variation that arise during the model-
building process. Sensitivity tests incorporated into the study design not only aid in the
interpretation of simulation results, but can also provide additional insights on form and
function.
75
Chapter Two Introduction

Finite element analysis (FEA), the discretization of structures and approximation
of their mechanical behavior (the response of structure to load), has traditionally been an
analytical technique in the engineering disciplines as an important component of the
development process to improve design. More recently, however, its use in functional
studies of biological structures has become more common (Fastnacht et al. 2002, Kupczik
2008, Ross 2005). FEA has been applied in vertebrate biomechanics research across
diverse taxonomic groups, including crocodiles , non-avian dinosaurs , birds , lizards ,
fishes , and a variety of mammals . FEA complements in vivo experimental studies by
allowing simulations using user-defined input assumptions regarding the study system,
which could otherwise be impossible to implement. Currently, most studies of this type
address the mechanical behavior of the craniodental system.
Given the diverse functional questions that could be examined using the FE
approach, the current available data from FE publications are largely incomparable across
studies precisely because of the comparative nature of current applications. Even within
narrow clades of closely related genera and species, lack of absolute values from FE
results means published stress and strain values cannot be used to evaluate relative
performance of models across different studies (e.g. models of felid species in McHenry
et al., 2007 versus those in Slater and Van Valkenburgh, 2009). In many studies, different
approaches in how muscles and constraints are modeled also make comparisons difficult.
Furthermore, the diversity of taxonomic groups that can potentially be studied using this
76
technique, accompanied by the different software programs and protocols used by
researchers in FE model construction, further complicates any attempts at the synthesis of
current FE knowledge across vertebrate groups. The current diversity of input
assumptions in FE models used in comparative biology suggests a need to quantify the
sensitivity of performance measures to those parameters, in order to build a general
context for comparing results within and across different studies.
Several previous studies have addressed the choice of model parameters and their
implications for comparing FE analytical results to those obtained from in vivo
experiments for masticatory muscle forces (Ross et al. 2005), bite forces (Davis et al.
2010), and elastic bone properties (Strait et al. 2005). However, few have addressed the
comparison of relative values in the growing literature on vertebrate FE models, which is
becoming more numerous given the flexibility of this approach in allowing tests of form
and function (Dumont et al. 2009). In one attempt, Sellers and Crompton (2004)
conducted a sensitivity study of human bite force prediction with FEA using a large
number of combinations of model parameters and found that masticatory muscle
insertion points, as well as the modeled mobility of the temporomandibular joints (TMJ),
had a large effect on resulting jaw forces. Even if current FEA applications in vertebrate
functional morphology cannot provide accurate mechanical values for comparing to
experimental results, and in most cases FE models do not have corresponding in vivo data
for validation tests, standardized comparisons can nevertheless be highly informative
(Dumont et al. 2009). Furthermore, a comparative approach has the advantage of being
77
able to include extinct forms for which material properties and other parameters cannot
be directly obtained.
In order to provide this context with which to evaluate the effect of different
modeling parameters on the resulting stresses and strains in comparative mammalian
mandible FE models, we conducted sensitivity analyses on a model of the carnivoran
Canis lupus by testing a range of values for seven required parameters that vary among
comparative FE studies (Table 2.1, Fig. 2.1). The effects of variation in those parameters
on performance measures were evaluated by examining bite force output, strain energy,
temporomandibular joint reaction forces, and stress distribution (Dumont et al. 2009).
Bite force output (or other related measures, such as mechanical advantage) is a key
performance variable in evaluating and comparing masticatory function of the
craniodental system, as larger bite forces permit a species to consume harder and tougher
foods, as well as predating on large prey. Both of these adaptations mediate the
ecological interactions within the predator guild and across trophic levels. Strain energy
has been used as a measure of the work-efficiency of the craniodental system under
simulated loads [30]. Selection should favor such work efficiency given the functional
demands and trade-offs of achieving maximum stiffness with a given structural quantity
and weight (i.e. bone). Joint reaction forces have been shown to exhibit consistent
patterns during the mastication cycle, and represent indicators of whether the joint region
is being properly modeled (Hylander 1979). Distributions of von Mises stress is used to
show likely areas of failure when the bone undergoes ductile fracture (Dumont et al. 2005,
Nalla et al. 2003). This study aims to test how input assumptions in FE models affect
78
these performance measures, which are in turn used to test functional hypotheses and in
comparisons of functional capability across species.

Chapter Two Materials and Methods

We used the Gray Wolf Canis lupus mandible model from Tseng and Binder
(2010) for sensitivity tests. The structure of interest included both dentaries of the
specimen. The specimen was CT-scanned with a Siemens Definition 64 scanner (Siemens
Medical Solutions, Forcheim, Germany) with 0.6 mm slice thickness, 0.37 mm pixel
resolution, and a size of 512x512 pixels. 499 images were obtained. We chose the
mandible for modeling and sensitivity analysis because of the simplicity of the lower jaw,
which is composed of two dentary bones with three joints and no sutures (Scapino 1965).
Compared to the cranium, the function of the mandible is not complicated by the
multitude of roles, such as the protection of several sensory organs, played by the former
structure. In addition, cranial sutures render the cranium a composite structure, and may
mediate the location and magnitude of strain during muscle contraction and mastication
(Herring and Teng 2000). Fewer anatomical features need to be accounted for when
modeling the mandible, therefore allowing us to focus on the choices in model resolution,
material properties, and boundary conditions and their effects on analysis results.
Models were constructed following the protocol used in Dumont et al. (Dumont et
al. 2009, Dumont et al. 2005) and Tseng and Binder (Tseng and Binder 2010). The
starting point for the tests was a base model with 383,319 4-noded tetrahedral elements,
79
0.6 balancing-working side ratio, 55%-26(9)%-10% temporalis-
masseter(zygomaticomandibularis)-pterygoid muscle ratio, single-node bite point and
TMJ constraints, and a single material property (E = 20 GPa, Poisson ratio = 0.3). All
models simulated the biological phenomenon of a unilateral bite with the left lower first
molar (the carnassial tooth). We isolated seven main parameters in FE model-building for
our sensitivity tests: number of finite elements used to represent the mandibular
morphology, balancing- versus working-side muscle activation ratios, relative muscle
forces among the masticatory muscle groups, the number of sub-groups within each
masticatory muscle group, the size of the bite point constraint, the constraint used at the
temporomandibular joints, and number of material properties assigned (Appendix B
Table S1). All models had linear elasticity, and static equilibrium equations were solved
in analyses. The variations tested within each category are described in more detail below.
1. Number of finite elements. Craniodental FE models in the current literature are
mainly built from four-noded tetrahedral elements; these constant stress elements have
three degrees of translation freedom per node. Likewise, all models analyzed in this study
used four-noded tetrahedrals only. In contrast, ten-noded tetrahedral elements provide
more detailed information regarding the distribution of stress and strain within each
element, but craniodental FE models built at ~250,000 elements showed variation in
results with 10% between four- versus ten-noded elements (Dumont et al. 2005). This
observation has been cited to justify using four-noded tetrahedral models built with large
numbers of elements (>1,000,000) as being sufficient for the general functional questions
being asked. Given the current widespread use of four-noded elements in craniodental
80
and in fact most other FE models, we ran eight tests with the more commonly used four-
noded tetrahedral elements. Number of triangular elements in each model were adjusted
in Geomagic Studio 10 (Geomagic, Inc.) before they were meshed into 4-noded
tetrahedral elements in Strand7 2.4.1 (G+D Computer Pty Ltd, Sydney, Australia). The
number of tetrahedral elements ranged from ~100,000 to ~1,400,000, typical of the
resolution seen in most published FE studies (Appendix B Table S2).
2. Muscle activation schemes. Many of the currently published craniodental FE
models use symmetrical bilateral muscle forces, even in unilateral biting simulations.
Electromyography studies have shown, however, that at least in Canis, mastication of
bone and meat is achieved without maximum bilateral recruitment of the jaw muscles
(Dessem 1989). Feedback from periodental nerves also plays a role in mediating the use
of muscle forces to produce large bite forces, at the same time maintaining joint stability
(Dessem 1989). Therefore, unilateral bite simulations with maximum bilateral muscle
recruitment may represent theoretical maxima and not realistic voluntary maxima (Ellis
et al. 2008). Among mammals, a range of muscle recruitment ratios is present both within
individuals and across clades; the adaptation of the mandible to particular modes of
muscle loading may be informative in themselves in reflecting typical loading scenarios
in a given species (Dessem 1989, Hylander et al. 2000). We tested unilateral bites at the
carnassial tooth (lower molar 1) with 11 models ranging from no balancing side muscle
contribution to fully bilateral muscle activation. Working- and balancing-side muscle
differences were tested in 10% increments (Appendix B Table S3). This range
81
encompasses the ratios observed in several mammalian groups (Dessem 1989, Hylander
et al. 2000).
3. Muscle proportions. The relative contributions of the three main jaw-closing
muscles (temporalis, masseter, and pterygoid) have been estimated in craniodental FE
models using either physiological cross-sectional area (PCSA), an estimated of muscle
cross-sectional area using dry skulls (Thomason 1991), or by mass of dissected muscles.
PCSA has been shown to be a good predictor of muscle force and bite force in bats
(Davis et al. 2010), but in most cases this information is not available for living
vertebrates, let along fossil species. We tested a wide range of muscle proportions which
encompasses several estimates that have been made for Canis (Schumacher 1961,
Turnbull 1970). Eight models were made, including muscle activation of each of the
three major jaw-closing muscle groups in isolation (represented by numbers in the order
of temporalis-masseter-pterygoid; Appendix B Table S4). Even though the masticatory
muscle groups do not activate in isolation in reality, their contribution to, and effects on,
the resulting biomechanical performance measures can nevertheless reflect potential
adaptations (Tseng and Wang 2010). The lateral pterygoid muscle is proportionally
smaller than the other jaw muscles mentioned, accounting for about 3% of total PCSA or
< 0.6% by wet weight in Canis lupus (Schumacher 1961); thus, this relatively minor
muscle was not included in our analysis.
4. Number of muscles. The main jaw-closing muscles temporalis, masseter, and
pterygoid can be subdivided into subgroups based on their gross anatomy, and
mammalian craniodental models have been made with a range of muscle groups from
82
temporalis and masseter muscles only (Dumont et al. 2005) to all three muscles and their
subgroups (McHenry et al. 2007). We tested for the effect of number of muscle groups on
model outcomes by building seven models ranging from a single jaw closing muscle to
seven muscle groups, including subgroups of the temporalis and masseter muscles (Fig.
2.1, Appendix B Table S5). The total input force remained the same, and the relative
contributions of muscle groups are calculated from the initial 55%-26(9)%-10%
temporalis-masseter(zygomaticomandibularis)-pterygoid muscle ratio used in other test
categories. Forces among additional subgroups within each major muscle group (when
present) are distributed by their respective surface areas.
5. Bite point constraint. The evaluation of bite force in craniodental FE models is
often done by sampling reaction forces of nodal constraints at the bite points; however,
the range of variation in bite force magnitude estimated by a single node constraint versus
constraint distributed over an area is unknown. In carnivorans with self-sharpening
carnassial teeth, the cusps remain pointed through time, and thus the first point of contact
during mastication is situated near the tip of the crown. Therefore, we varied the number
of nodes representing the bite point, starting with a single constraint at the tip of the
carnassial protoconid. We tested a range of nodal constraint quantity from a single node
to ~65 nodes (covering the entire cusp) using six models (Appendix B Table S6).
6. The temporomandibular joint (TMJ). The TMJ has been modeled as rotating
around single nodes (Dumont et al. 2005), a row of nodes along the condyloid
fossa/mandibular condyle (Tseng 2009), or around a beam through the axis of rotation
connected to the joints by rigid links (McHenry et al. 2007). The use of single nodes
83
creates artificially elevated stress values immediately around the constraint, but the
overall distribution of stress in the structure is not affected further away from those
constraints. We tested four ways of constraining the TMJ in order to examine their
differences: (1) single node constraint at each TMJ; (2) a row of nodal constraints along
the mandibular condyle; (3) a single rigid link between the mandibular condyle and a
beam in the axis of the TMJ with no translation or rotation other than rotation in the
dorsoventral plane; (4) rows of rigid links connecting the mandibular fossa to the axial
beam with dorsoventral rotational freedom (Appendix B Table S7). More recently,
Dumont et al. ( 2011) used another method to constrain the TMJ, namely allowing
translation in the axis connecting the left and right TMJ, in addition to rotational freedom
in the sagittal plane. This alternative was not tested in the current study.
7. Number of material properties. Two main methods of material property
assignments in the current literature on craniodental FEA are (1) assigning properties of a
single material to the entire model, or (2) assigning multiple categories of materials based
on Hounsfield Units (HU), the gray values representing densities in CT image data. We
tested six models ranging from homogeneous single-property to heterogeneous 10-
property models (Appendix B Table S8). Bone properties were assigned based on HU
intervals obtained during examination of the CT data using published HU-density and
density-modulus equations (Rho et al. 1995, Schneider et al. 1996), and tooth enamel and
dentine were assigned properties based on published values (Habelitz et al. 2001, Haines
1968, Qin and Swain 2004). No calibration standard was available from the CT data; the
density and modulus equations were applied assuming similar relationships existed for
84
the data (for example, cortical bone properties calculated using unadjusted HU from the
CT data provided a density of 1.77 g/cm
3
and Young’s modulus of 19.39 GPa, well
within measured range of typical mammalian cortical bone). All materials were treated as
isotropic, and all analyses were linear static (Appendix B Tables S8, S9).
Three-dimensional reconstructions were built from CT image data in Mimics 13.1
(Materialise N.V., Leuven, Belgium), reconstructions were cleaned in Geomagic Studio
10 (Geomagic, Inc., Research Triangle Park, North Carolina, USA), and then remeshed in
Mimics. The solid mesh FE models were built in Strand7 2.4.1. The cranium of the
specimen was used as reference to identify the direction of muscle forces on the mandible;
the relative positions of the cranium and mandible were modified from zero load state
(full occlusion, 0º gape) by a 10º rotation of the mandible about the TMJ. This
modification created a 10º gape angle that simulated mastication of a small food item
between the carnassial teeth. Segmentation of the reconstruction from image data was
done using both automated functions in Mimics, as well as manual delineation of bone
boundaries. Meshes represented the overall macrostructure of the mandible, without
differentiation of microstructural architecture in bone or teeth. Masticatory muscle forces
were modeled using the Boneload program written by Grosse et al. (2007). 1000 N of
total muscle force was used for all models tested. The model results used for comparison
of sensitivity tests were the reaction forces (in Newtons) at the carnassial bite position
and the working- and balancing-side joint constraints. Total strain energy (equivalent to
the work done in deforming the mandible) values were also compared (Dumont et al.
2009). In addition, the von Mises stress distributions were visualized on the models. A
85
total of 44 models of the Canis lupus mandible were constructed, each given a unique
identification number (Table 2.1, Appendix B S1; models deposited at Dryad:
doi:10.5061/dryad.8961).

Chapter Two Results

1. Number of finite elements. Reaction forces at the bite point and balancing-side
joint were lower in the low resolution model (101,674 elements), and working-side joint
forces higher, than all of the other models. Higher-resolution models showed no clear
trend in increasing or decreasing reaction forces, although some variation is present (Fig.
2.2). Strain energy values showed small increase with tetrahedral element quantity, but
the slope was on the order of 10
-9
and does not represent a significant trend. Model
solution time increased exponentially between ~100,000 and ~1,200,000 tetrahedral
elements. The low-resolution model showed lower von Mises stress distributions across
the ascending ramus than all other models, which do not show visible differences in
stress distribution (Fig. 2.2D)
2. Muscle activation schemes. Balancing-side reaction forces increased, and
working-side decreased, with increasing ratio of balancing-working side muscle
activation (Fig. 2.3). Bite force remained largely invariant. Joint reaction forces are lower
than the bite force on both working- and balancing-sides between the ratios 0.4 to 0.6.
Balancing-side joint reaction forces are higher than working-side reaction forces, a
pattern consistent with experimental values, at ratios larger than 0.5 (Hylander 1979).
86
Strain energy values are lowest between ratios of 0.3 to 0.5, and are elevated in both
higher and lower ratios. Higher balancing-side muscle activation is correlated with
decreased von Mises stress on the working-side ascending ramus, but increased stress in
the mandibular corpus below the premolars (Fig. 2.3C).
3. Muscle proportions. Using the internal pterygoideus or the masseter muscle in
isolation created elevated working-side TMJ reaction forces (Fig. 2.4). Strain energy
values increased when the pterygoideus and temporalis muscles were used in isolation.
All other muscle ratios exhibited comparable levels of reaction forces and strain energy
values, with the lowest bite force in the 55-30-15 (temporalis-masseter-pterygoideus)
model. The ventral side of the mandibular corpus is more stressed in masseter- and
pterygoideus-only models, and the ascending ramus is more stressed in temporalis-only
models. All other models showed little difference in von Mises stress distribution (Fig.
2.4C).
4. Number of muscles. Reaction forces and strain energy values decreased with
increasing number of muscle subgroups modeled. Reaction forces decreased by 20%
from the one-muscle model to the 2-4 muscle models, and the latter showed little
difference among themselves. A further decrease of ~15% was observed from the 2-4
muscle models to the 5-7 muscle models; again, the latter group showed little difference
among themselves. A larger drop in strain energy (40%) was observed from one-muscle
to 2-4 muscle models, and a ~25% drop from 2-4 muscle models to 5-7 muscle models.
Models with more muscle subgroups showed lower stresses in the ascending ramus and
the corpus ventral of the premolars (Fig. 2.5).
87
5. Bite point constraint. Bite force increased by 60% from a single-node bite
point to a 66-node bite point, whereas joint reaction forces stayed constant. Strain energy
decreased by <10% from a single-node to a 6-node constraint, but stayed constant for
higher numbered constraints. Components of the bite force vector show no significant
increases with node number, indicating that the directions of the vector were instead
becoming more aligned in the dorsoventral direction, increasing the magnitude of the
resultant (Fig. 2.6B). No differences in stress patterns are observed across the models in
areas other than immediately around the bite point, which showed more widespread stress
with higher numbered node constraints (Fig. 2.6).
6. The temporomandibular joint (TMJ). The 10-node model had similar bite
force to the 1-node model, but the former had elevated joint reaction forces that exceeded
the bite force, and higher working-side TMJ forces than balancing side forces. Bite force
decreased <10% in the link models, which had no joint reaction forces at the nodes.
Strain energy values decreased by ~50% from single-node/single-link models to 10-
node/10-link models, respectively. Von Mises strain is higher in the link models than in
the node models. The single-node/single-link models showed higher von Mises stress in
the caudal half of the mandible compared to the other models (Fig. 2.7).
7. Number of material properties. Bite force increased by 30%, and joint reaction
forces decreased by 20%, from 1-3 property models to 6-8 property models. Strain
energy values increased more than 20 fold between those models. The modeling of
enamel and dentin had a significant effect on the stress distribution of the models, with
88
most of the stress being contained at the biting tooth in the 6-10 property models (Fig.
2.8C).

Chapter Two Discussion

We conducted sensitivity tests on performance measures by altering seven input
parameters that are required in FE modeling building, but which vary among comparative
studies in the literature. Results showed that varying the values of initial parameters had a
wide range of effects on bite force (1% to 60% maximum difference) and strain energy
(14.7% to >100% maximum difference). The balancing-working muscle activation ratio
had the smallest effect on bite force output over the range tested (0.0-1.0), and for
estimates of unilateral bite force one might be tempted to discount its influence on the
results. However, plots of changes in joint reaction forces showed that only above a ratio
of 0.5 were working-side reaction forces smaller than balancing side reaction forces, as
predicted by experimental data (Fig. 2.3A)(Hylander 1979). Furthermore, the joint
reaction forces were lowest relative to the bite force, and therefore the simulated bite was
least stressful to the TMJ, in the 0.4-0.6 ratio range. This range overlaps with the 0.6 ratio
obtained experimentally by Dessem (1989), who suggested that balancing-side muscle is
not fully activated during maximum bite force production, partly because of the need to
stabilize the jaw joints. The lowered joint reaction forces observed in the FE models are
consistent with this hypothesis. In addition, strain energy values are also lowest in the
0.4-0.5 range, suggesting that this configuration provides optimal mandible performance
89
on the basis of work-efficiency (Dumont et al. 2009). Even though von Mises stress
distributions on the mandible showed no significant differences across the range of ratios
tested, using an activation ratio of 0.4 to 0.6 between the balancing- and working-side
jaw musculature returned lower joint reaction forces and higher work-efficiency (Fig.
2.3A)(Slater et al. 2010, Tseng 2009).
Bite force output showed most significant changes in models that differed in
number of bite point constraints (Fig. 2.6). Constraints that cover a larger area of contact
produced higher bite forces than single-node constraints, and this difference approached
60%. In estimating bite forces, comparative FEA studies have used both a distributed
area of bite point constraints and single nodes . Results from our analyses showed that,
everything being equal, sum of forces from a multi-node constraint would be larger than
in the single-node estimate. In most cases comparative FEA are consistent in their model
constraints within each study, but care must be taken when one attempts to evaluate bite
forces estimated from different studies with different approaches. This is especially true
for extinct taxa; where possible, taxon-specific validation should be coupled with
modeling different bite constraints to test the range of reasonable bite force estimates that
can be made by FE models (Davis et al. 2010). It remains difficult to use FEA for bite
force estimates of extinct organisms before generalizations are made on how best to
model bite points. Furthermore, the increasing number of constraints placed at the biting
tooth could have over-constrained the models beyond realistic scenarios, and this would
partially explain the large differences in bite force observed.
90
Strain energy values were least affected by the type of bite point constraint (Fig.
2.6C), but were significantly more variable in models that differed in number of material
properties (Fig. 2.8B). This is to be expected because increased number of material
properties also created more elements that have lower density and modulus values in the
model. Interestingly, very high strain energy (i.e. low work efficiency) was observed in
models that had more than six material properties, and von Mises stress in those models
are concentrated in the biting tooth (Fig. 2.8C). The stress distribution indicates that most
of the deformation in models with more material properties was concentrated on and
within the biting tooth, which was modeled with a plate covering of enamel, and a single-
element thick layer of dentine. The large difference that exists in material properties
between the tooth and the surrounding bone may explain stress concentration in the
former. Evolutionary specializations of enamel microstructure in durophagous
carnivorans are consistent with increased selection for stronger teeth, which are required
to withstand large stresses incurred from contact against hard food (Rensberger 1995,
Stefen 1997). However, increased stress concentration in the biting tooth was not
observed until at least four material properties were present (Fig. 2.8C), indicating that
sufficient differentiation in tooth-bone material properties are required to model this
effect. For applications in extinct taxa, fossilized bone often does not provide enough
resolution or faithful reproduction of relative bone densities to enable such tests (Tseng et
al. 2011a). In cases where such differentiation is possible, however, multiple-property
models would tend to increase bite force and also strain energy, and would need to be
standardized before comparisons are made across homogeneous and heterogeneous
91
models. The current study did not explicitly consider variation in the ranges of material
properties represented in multi-property models. Models with increasing number of
material properties also had increasing ranges of densities and modulus values
represented by those properties; it remains to be seen how wider or narrower ranges of
material properties for a given multi-property model can affect results. An additional
factor that has been validated in FE models recently is the localized effect of periodontal
ligament on strain in the alveolus; the effect of excluding this tissue from FE models on
overall results appear to be small, however (Panagiotopoulou et al. 2011).
A summary of maximum changes in bite force and strain energy is shown in
Table 2.2. In all but one case, variation in model parameters had larger effects on strain
energy than on bite force. In addition, increasing the complexity and magnitude of the
values within each parameter can either increase or decrease the performance variables.
Theoretically, using a mosaic combination of values in comparisons of any two species
models can produce differences where there is none (false positive), or a result of no
difference when a difference actually exists (false negative). Functional factors behind a
two-model comparison can, therefore, be confounded with variation in input parameters.
Whereas balancing-working muscle ratios, bite point and joint constraints, and number of
material properties are often standardized across species models, and therefore should not
constitute as large a source of error, the number of elements and musculature ratios are
rarely identical among currently published models. On average, the doubling of
tetrahedral elements in the mandible model led to a ~12% increase in strain energy. One
reason that differences in element numbers change the magnitude of performance
92
measures is the different internal densities of elements as dictated by automated meshing
functions in the FE software program. In the program used for this study (Strand7),
coarse models are calculated by minimizing steps required to transition to the maximum
element size (which is determined by the initial surface mesh), whereas fine models are
built without much constraint on numbers of elements with maximum size. As a result,
finer models contain larger quantity of small elements. Compounded with the fact that the
number of small elements tend to be higher within each model in regions of high
curvature or shape change, stress increases can be observed without changing inputs
other than element quantity. The number of elements required to consistently represent a
model of complex morphology can only be acquired through convergence analyses of
each unique model, and a recent study by Bright and Rayfield (Bright and Rayfield 2011)
provides a specific example of convergence analysis in mammalian cranium models.
Findings also show that musculature ratios that span the available estimates for
Canis can produce a ~20% difference in bite force and 25% difference in strain energy in
otherwise identical models. PCSA has been shown to be a good predictor of bite force in
bats (Davis et al. 2010), but in comparisons where PCSA is not available, the results
indicate that higher estimates of temporalis would tend to return higher bite force and
strain energy values. The pattern of performance changes with changing musculature
ratios is inherently interesting, and may reveal functional traits not apparent with
comparisons of single models (Tseng and Wang 2010). In these cases building multiple
models from the same individual with different musculature ratios would be more
93
informative than choosing among the available means of estimates of masticatory muscle
force to build a single model.
In summary, the variations that arise in FEA results from changing initial
parameters can be confounded with functional differences in model comparisons. More
confidence can be placed in model comparisons where these factors are examined by
sensitivity and convergence analyses, and in some cases standardized. In standardizing
models, it is more important to keep bite point constraints and the number and range of
material properties constant in evaluating bite force outputs, and keeping material
properties, musculature ratios, and muscle subgroups constant for strain energy
comparisons. The relationship between TMJ joint reaction forces on the balancing-
versus working side jaw should be examined along with bite forces to ensure the forces
acting on the models are reasonably comparable to experimental results. Visual stress
distributions are affected more by number of material properties than by any other factor
examined. Comparisons between different modeling protocols, if they are to be made,
should consider these influences.
Other parameters. The cross-sectional shape of mandibles is an important
predictor of feeding performance and bending strength in carnivorans (Biknevicius and
Ruff 1992). However, in studies of fossil species the internal structures of the skull may
not be preserved, and in some cases filled models can provide reasonable estimates of
mechanical behavior in the original morphology (Tseng and Binder 2010). In cases where
internal morphology simply cannot be reconstructed with any confidence, the filled
models may be sufficient for broad comparison purposes. However, researchers may
94
wish to reconstruct the internal cavity by approximating its boundary if the evolution of
corpus cortical thickness is of interest.
The mandibular symphysis, which exhibits variation in composition and gross
anatomy across mammal species, is a key location that affects the distribution of stresses
across the dentaries (Scapino 1965). Tseng and Stynder (In Press) tested a range of
material properties to approximate the mechanical behavior of the mandibular symphysis
in their carnivoran models, and found that in most cases the stress is conducted through
the symphysis, but modeling the joint as cortical bone can increase regional stress. Their
results are superficially similar to those presented by McHenry et al. (2007) and Wroe et
al. (2007), and suggest that at least for the symphysis, those models show elevated stress
compared to ones constructed with material properties closer to ligament or fibrocartilage.
Homogeneous models, which are built using a single set of material properties,
usually representing average cortical bone, are common in comparative studies (Slater et
al. 2009, Tseng 2009, Wroe et al. 2007). A sensitivity test of typical elastic modulus and
Poisson ratio values used in construction of homogeneous models was conduced by
Tseng et al. (2011a), who showed that the middle range of elastic modulus (15-30 GPa)
and Poisson ratio (0.1-0.4) used by many studies gave comparable results in stress and
strain. Thus, stress distributions of homogeneous models built with values within those
ranges are not expected to be significantly influenced by modeling artifacts when used in
comparisons.
The sensitivity tests performed in this study are by no means exhaustive, and the
range of input assumptions represented by the current set of models can be expanded
95
upon to include more extensive or specific tests that pertain to specific research questions.
The models discussed in this study are available in the Dryad Digital Repository
(doi:10.5061/dryad.8961).

Chapter Two Conclusion

We conducted a series of sensitivity tests to evaluate the range of variation among
the modeling parameters required in studies of functional morphology using FEA.
Findings indicate that not all parameters are equally variable, and consideration needs to
be given to particular sets of parameters, based on the research question being asked. In a
purely comparative context, a Gray Wolf mandible model required only ~300,000
elements to produce reaction forces and strain energy values close to those obtained from
higher-resolution models (>1,400,000 elements). Whereas PCSA, mass, or other
estimates of muscle ratios did not greatly affect the results, the adjustment of the
balancing-working side ratio in unilateral biting simulation does have an effect on joint
reaction forces. For comparative purposes, the number of muscle subgroups, the area of
bite point constraints, the TMJ constraint, and the number and range of material
properties should be kept consistent across models within a single study. Across different
studies, the compound effects of variation among those factors may be large, and
differences up to 50% can be observed by extreme values in a single parameter.
Validation of FE models of living species is needed to determine the set of input
parameters that would give the most realistic results in a given study, but comparative
96
studies can nevertheless be highly informative especially if sources of variation can be
identified within the particular set of values used to construct the models. Lastly, the
pattern of variation obtained through tests of a given parameter within each model may
be instructive in itself, thus researchers may wish to consider sensitivity tests as part of a
study design of comparative form and function using FEA.

Chapter Two Acknowledgments

We thank J. Dines (LACM) for providing the specimen used in the study. M.
McNitt-Gray (UCLA) CT-scanned the specimen. J. Liu provided valuable comments on
the content of the manuscript. The editor and anonymous reviewers provided constructive
criticism that improved the content of the paper.
97
Chapter Two References
1. Rayfield EJ (2007) Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35: 541-576.

2. Ross CF (2005) Finite element analysis in vertebrate biomechanics. The Anatomical
Record 283A: 253-258.

3. Kupczik K (2008) Virtual biomechanics: basic concepts and technical aspects of finite
element analysis in vertebrate morphology. Journal of Anthropological Sciences
86: 193-198.

4. Fastnacht M, Hess N, Frey E, Weiser H-P (2002) Finite element analysis in vertebrate
palaeontology. Senckenbergiana lethaea 82: 195-206.

5. McHenry C, Clausen PD, Daniel WJT, Meers MB, Pendharkar A (2006)
Biomechanics of the rostrum in crocodilians, a comparative analysis using finite
element modeling. The Anatomical Record 288A: 827-849.

6. Bell PE, Snively E, Shychoski L (2009) A comparison of the jaw mechanics in
hadrosaurid and ceratopsid dinosaurs using finite element analysis. The
Anatomical Record 292: 1338-1351.

7. Xing L, Ye Y, Shu C, Peng G, You H (2009) Structure, orientation and finite element
analysis of the tail club of Mamenchisaurus hochuanensis. Acta Geologica Sinica
83: 1031-1040.

8. Arbour VM, Snively E (2009) Finite element analyses of ankylosaurid dinosaur tail
club impacts. The Anatomical Record 292: 1412-1426.

9. Rayfield EJ, Norman DB, Jorner CC, Horner JR, Smith PM, et al. (2001) Cranial
design and function in a large theropod dinosaur. Nature 409: 1033-1037.

10. Rayfield EJ (2004) Cranial mechanics and feeding in Tyrannosaurus rex. Proceedings
of the Royal Society of London, Series B 271: 1451-1459.

11. Rayfield EJ (2005) Aspects of comparative cranial mechanics in the theropod
dinosaurs Coelophysis, Allosaurus, and Tyrannosaurus. Zoological Journal of the
Linnean Society 144: 309-316.

98
12. Soons J, Herrel A, Genbrugge A, Aerts P, Podos J, et al. (2010) Mechanical stress,
fracture risk and beak evolution in Darwin's ground finches (Geospiza).
Philosophical Transactions of the Royal Society of London B Biological Sciences
365: 1093-1098.

13. Moazen M, Curtis N, Evans SE, O'Higgins P, Fagan MJ (2008) Combined finite
element and multibody dynamics analysis of biting in a Uromastyx hardwickii
lizard skull. Journal of Anatomy 213: 499-508.

14. Moreno K, Wroe S, Clausen PD, McHenry C, D'Amore DC, et al. (2008) Cranial
performance in the Komodo dragon (Varanus komodoensis) as revealed by high-
resolution 3-D finite element analysis. Journal of Anatomy 212: 736-746.

15. Hulsey CD, Roberts RJ, Lin ASP, Guldberg R, Streelman JT (2008) Convergence in
a mechanically complex phenotype: Detecting structural adaptations for crushing
in cichlid fish. Evolution 62: 1587-1599.

16. Tanner JB, Dumont ER, Sakai ST, Lundrigan BL, Holekamp KE (2008) Of arcs and
vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta).
Biological Journal of the Linnean Society 95: 246-255.

17. Strait DS, Weber GW, Neubauer S, Chalk J, Richmond BG, et al. (2009) The feeding
biomechanics and dietary ecology of Australopithecus africanus. Proceedings of
the National Academy of Sciences 106: 2124-2129.

18. Slater GJ, Van Valkenburgh B (2009) Allometry and performance: the evolution of
skull form and function in felids. Journal of Evolutionary Biology 22: 2278-2287.

19. Dumont ER, Piccirillo J, Grosse IR (2005) Finite-element analysis of biting behavior
and bone stress in the facial skeletons of bats. The Anatomical Record 283A: 319-
330.

20. Wroe S (2008) Cranial mechanics compared in extinct marsupial and extant African
lions using a finite-element approach. Journal of Zoology 274: 332-339.

21. Wroe S, Clausen PD, McHenry C, Moreno K, Cunningham E (2007) Computer
simulation of feeding behaviour in the thylacine and dingo as a novel test for
convergence and niche overlap. Proceedings of the Royal Society B: Biological
Sciences 274: 2819-2828.

22. McHenry C, Wroe S, Clausen PD, Moreno K, Cunningham E (2007) Supermodeled
sabercat, predatory behavior in Smilodon fatalis revealed by high-resolution 3D
computer simulation. Proceedings of the National Academy of Sciences 104:
16010-16015.
99

23. Slater GJ, Dumont ER, Van Valkenburgh B (2009) Implications of predatory
specialization for cranial form and function in canids. Journal of Zoology 278:
181-188.

24. Farke AA (2008) Frontal sinuses and head-butting in goats: a finite element analysis.
The Journal of Experimental Biology 211: 3085-3094.

25. Tseng ZJ (2009) Cranial function in a late Miocene Dinocrocuta gigantea
(Mammalia: Carnivora) revealed by comparative finite element analysis.
Biological Journal of the Linnean Society 96: 51-67.

26. Tseng ZJ, Binder WJ (2010) Mandibular biomechanics of Crocuta crocuta, Canis
lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia).
Zoological Journal of Linnean Society 158: 683-696.

27. Ross CF, Patel BA, Slice DE, Strait DS, Dechow PC, et al. (2005) Modeling
masticatory muscle force in finite element analysis: sensitivity analysis using
principal coordinates analysis. The Anatomical Record 283A: 288-299.

28. Davis JL, Santana SE, Dumont ER, Grosse I (2010) Predicting bite force in mammals:
two-dimensional versus three-dimensional lever models. Journal of Experimental
Biology 213: 1844-1851.

29. Strait DS, Wang Q, Dechow PC, Ross CF, Richmond BG, et al. (2005) Modeling
elastic properties in finite-element analysis: how much precision is needed to
produce an accurate model? The Anatomical Record 283A: 275-287.

30. Dumont ER, Grosse I, Slater GJ (2009) Requirements for comparing the performance
of finite element models of biological structures. Journal of Theoretical Biology
256: 96-103.

31. Sellers WI, Crompton RH (2004) Using sensitivity analysis to validate the predictions
of a biomechanical model of bite forces. Annals of Anatomy 186: 89-95.

32. Hylander WL (1979) An experimental analysis of temporomandibular joint reaction
force in macaques. American Journal of Physical Anthropology 51: 433-456.

33. Nalla RK, Kinney JH, Ritchie RO (2003) Mechanistic failure criteria for the failure of
human cortical bone. Nature Materials 2: 164-168.

34. Scapino RP (1965) The third joint of the canine jaw. Journal of Morphology 116: 23-
50.

100
35. Herring SW, Teng S (2000) Strain in the braincase and its suture during function.
American Journal of Physical Anthropology 112: 575-593.

36. Dessem D (1989) Interactions between jaw-muscle recruitment and jaw-joint forces
in Canis familiaris. Journal of Anatomy 164: 101-121.

37. Ellis JL, Thomason JJ, Kebreab E, France J (2008) Calibration of estimated biting
forces in domestic canids: comparison of post-mortem and in vivo measurements.
Journal of Anatomy 212: 769-780.

38. Hylander WL, Ravosa MJ, Ross CF, Wall CE, Johnson KR (2000) Symphyseal
fusion and jaw-adductor muscle force: An EMG study. American Journal of
Physical Anthropology 112: 469-492.

39. Thomason JJ (1991) Cranial strength in relation to estimate biting forces in some
mammals. Canadian Journal of Zoology 69: 2326-2333.

40. Schumacher GH (1961) Funktionelle Morphologie der Kaumuskulatur. Jena,
Germany: VEB Gustav Fischer Verlag. 262 p.

41. Turnbull WD (1970) Mammalian masticatory apparatus. Fieldiana: Geology 18: 149-
356.

42. Tseng ZJ, Wang X (2010) Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology 271: 1386-1398.

43. Dumont ER, Davis JL, Grosse I, Burrows AM (2011) Finite element analysis of
performance in the skulls of marmosets and tamarins. Journal of Anatomy 218:
151-162.

44. Rho JY, Hobatho MC, Ashman RB (1995) Relations of mechanical properties to
density and CT numbers in human bone. Medical Engineering and Physics 17:
347-355.

45. Schneider U, Pedroni E, Lomax A (1996) The calibration of CT Hounsfield units for
radiotherapy treatment planning. Physics in Medicine and Biology 41: 111-124.

46. Qin Q-H, Swain MV (2004) A micro-mechanics model of dentin mechanical
properties. Biomaterials 25: 5081-5090.

47. Habelitz S, Marshall SJ, Marshall Jr. GW, Balooch M (2001) Mechanical properties
of human dental enamel on the nanometre scale. Archives of Oral Biology 46:
173-183.
101

48. Haines DJ (1968) Physical properties of human tooth enamel and enamel sheath
material under load. Journal of Biomechanics 1: 117-125.

49. Grosse I, Dumont ER, Coletta C, Tolleson A (2007) Techniques for modeling
muscle-induced forces in finite element models of skeletal structures. The
Anatomical Record 290: 1069-1088.

50. Slater GJ, Figueirido B, Louis L, Yang P, Van Valkenburgh B (2010) Biomechanical
consequences of rapid evolution in the polar bear lineage. PLoS ONE 5: e13870.
doi:13810.11371/journal.pone.0013870.

51. Stefen C (1997) Chapter 7. Differentiations in Hunter-Schreger bands of carnivores.
In: Koenigswald Wv, Sander PM, editors. Tooth enamel microstructure.
Rotterdam: A.A. Balkema. pp. 123-136.

52. Rensberger JM (1995) Determination of stresses in mammalian dental enamel and
their relevance to the interpretation of feeding behaviors in extinct taxa. In:
Thomason JJ, editor. Functional morphology in vertebrate paleontology. New
York: Cambridge University Press. pp. 151-172.

53. Tseng ZJ, Antón M, Salesa MJ (2011) The evolution of the bone-cracking model in
carnivorans: Cranial functional morphology of the Plio-Pleistocene cursorial
hyaenid Chasmaporthetes lunensis (Mammalia: Carnivora). Paleobiology 37:
140-156.

54. Panagiotopoulou O, Kupczik K, Cobb SN (2011) The mechanical function of the
periodontal ligament in the macaque mandible: a validation and sensitivity study
using finite element analysis. Journal of Anatomy 218: 75-86.

55. Bright JA, Rayfield EJ (2011) The response of cranial biomechanical finite element
models to variations in mesh density. The Anatomical Record 294: 610-620.

56. Biknevicius AR, Ruff CB (1992) The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology 228:
479-507.

57. Tseng ZJ, Stynder D (2011) Mosaic functionality in a transitional ecomorphology:
skull biomechanics in stem Hyaeninae compared to modern South African
carnivorans. Biological Journal of the Linnean Society 102: 540-559.

102
Figure 2.1. Mandible model used in the study. Bal., balancing-side joint; Work.,
working-side joint; m1, lower first molar (carnassial); M.p., deep masseter; M.s.,
superficial masseter; P.i., internal pterygoid; T.p., deep temporalis; T.s., superficial
temporalis; T.z., zygomatic part of temporalis; Z.m., zygomaticomandibularis.
Temporalis and masseter muscle subgroups were used incrementally in the sensitivity test
on number of muscles. All other models used a four-muscle input: temporalis-masseter-
zygomaticomandibularis-pterygoid.

103
Figure 2.2. Sensitivity test on tetrahedral element quantity. A. Element quantity plotted
against solution time (in seconds), with exponential curve in background. B. Element
quantity plotted against reaction force (in Newtons). C. Element quantity plotted against
strain energy (in Joules), with linear regression line. D. von Mises stress distribution in
the working-side dentary in test models; lateral view (in Megapascals).

104
Figure 2.3. Sensitivity test on balancing-working side ratio. A. Ratio plotted against
reaction force, with second-order polynomial curves fitted onto the working and
balancing reaction forces. B. Ratio plotted against strain energy. C. von Mises stress
distribution in the working-side dentary in test models.

105
Figure 2.4. Sensitivity test on musculature ratio. A. Ratio plotted against reaction force. B.
Ratio plotted against strain energy. C. von Mises stress distribution in the working-side
dentary in test models. Ratios are given by temporalis-masseter-pterygoid sequences,
with zygomaticomandibularis considered part of the masseter group.

106
Figure 2.5. Sensitivity test on number of muscle groups. A. Number of groups plotted
against reaction force, connected by lines to show trend. B. Number of groups plotted
against strain energy. C. von Mises stress distribution in the working-side dentary in test
models.

107
Figure 2.6. Sensitivity test on nodes at the bite point constraint. A. Nodal constraints
plotted against reaction force. B. Nodal constraints plotted against reaction force,
showing components of the bite force vector. C. Nodal constraints plotted against strain
energy. D. von Mises stress distribution in the working-side dentary in test models.

108
Figure 2.7. Sensitivity test on temporomandibular joint constraint. A. Constraint type
plotted against reaction force. B. Constraint type plotted against strain energy. C.
Constraint type plotted against von Mises strain, showing mean and maximum strain for
the working- and balancing-side joints, respectively. D. von Mises stress distribution in
the working-side dentary in test models.

109
Figure 2.8. Sensitivity test on number of material properties. A. Number of properties
plotted against reaction force, connected by lines to show trend. B. Number of properties
plotted against strain energy. C. von Mises stress distribution in the working-side dentary
in test models.

110
Table 2.1. Sensitivity tests performed in this study.
Parameter
#
Models Tests
Number of elements 8
Increasing element quantity from 101,674 to 1,404,279
Balancing-Working
Ratio
11
+0.1 increment of ratio from 0 to 1.0
Muscle ratio 8 PCSA, mass, dry skull estimates plus individual muscle
groups
Muscle number 7 Temporalis only to 6 subgroups of the temporalis and
masseter
Bite point constraint 6
Single node constraint to area with 66 nodes
TMJ constraint 4
single node, single link, row of nodes, row of links
Material properties 6 Homogeneous model to 10-property heterogeneous
model

A total of 44 models of the same mandible were used in the analyses; some models were
used in multiple test categories.
111
Table 2.2. Maximum % changes in bite force and strain energy in the sensitivity tests.

Parameter ΔValue max Δ Bite force
max Δ Strain
Energy
Number of elements 102k-1404k +10.2% +60.0%
Balancing-Working Ratio 0.0-1.0 -1.1% -39.2%
Muscle ratio Ptery.-Temp. +12.0% -63.7%
Muscle number 1-7 -31.6% -63.3%
Bite point constraint 1-66 +60.0% -14.7%
TMJ constraint nodes-links -6.3% -49.6%
Material properties 1-10 +38.2% +>100%

Value ranges given are for the full range of tests conducted. Changes in bite force and
strain energy are maximum differences within the range of each test. Ptery., pterygoid
muscles; temp., temporalis muscles.

112
Chapter Three: Study of cranial function in the hyena Chasmaporthetes lunensis

This chapter has been published as:

Tseng, Z. J., M. Antón, and M. J. Salesa. 2011. The evolution of the bone-cracking model
in carnivorans: Cranial functional morphology of the Plio-Pleistocene cursorial hyaenid
Chasmaporthetes lunensis (Mammalia: Carnivora). Paleobiology 37(1):140-156. doi:
10.1666/09045.1.

A copy of the accepted manuscript begins on the next page.
113
The evolution of the bone-cracking model in carnivorans: Cranial
functional morphology of the Plio-Pleistocene cursorial hyaenid
Chasmaporthetes lunensis (Mammalia: Carnivora)

Zhijie Jack Tseng, Mauricio Antón, and Manuel J. Salesa

RRH: EVOLUTION OF BONE-CRACKING MORPHOLOGY
LRH: TSENG, ANTÓN, AND SALESA
114
Chapter Three Abstract

Fossil species of the family Hyaenidae represent a wide range of ecomorphological
diversity not observed in living representatives of this carnivoran group. Among them,
the cursorial meat and bone specialists are of particular interest not only because they
were the most cursorial of the hyaenids, but also the only members of this family to
spread into the New World. Here we conduct a functional morphological analysis of the
cranium of the cursorial meat and bone specialist Chasmaporthetes lunensis by
comparing it with the living Crocuta crocuta, a well-known bone-cracking carnivoran,
using finite element modeling. As found with previous finite element studies on hyaenid
crania, C. lunensis is not differentially adapted for stress dissipation between the bone-
cracking and meat-shearing teeth. A smaller occlusal surface on the more slender P3 cusp
of C. lunensis allowed this species to use less bite force to crack a comparably-sized bone
relative to C. crocuta, but higher muscle masses in the latter probably allow it to process
larger food items. We use two indices, the stress slope and the bone-cracking index, to
show that C. lunensis has a well-adapted cranium for stress dissipation given its size, but
the main stresses placed on its cranium might have been from subduing prey and less
from cracking bones. Throughout the Cenozoic, other carnivores besides hyaenids
convergently evolved similar morphologies, including domed frontal regions, suggesting
an adaptive value for a repetitive mosaic of features. Our analyses add support to the
hypothesis that bone-cracking adaptations are a complex model that has evolved
convergently several times across different carnivoran families, and these predictable
115
morphologies may evolve along a common gradient of functionality that is likely to be
under strong adaptive control.

Zhijie Jack Tseng. Integrative and Evolutionary Biology Program, Department of
Biological Sciences, 3616 Trousdale Parkway, University of Southern California,
Los Angeles, California 90089, U.S.A. and Department of Vertebrate Paleontology,
Natural History Museum of Los Angeles County, 900 Exposition Boulevard, Los
Angeles, California 90007, U.S.A. E-mail: jack.tseng@usc.edu

Mauricio Antón and Manuel J. Salesa. Departamento de Paleobiología, Museo Nacional
de Ciencias Naturales-CSIC, C/José Gutiérrez Abascal, 2. 28006 Madrid, Spain.

116
Chapter Three Introduction

Carnivorans represent the major order of carnivorous mammals living today, and
many are top predators in their respective ecological communities (Ewer 1973, Nowak
1999). Whereas studies of modern carnivorans can reveal the diverse ecological roles
different species are adapted to, only those that incorporate the fossil record can shed
light on how each carnivoran lineage has diversified and become specialized in those
roles (Werdelin 1996b). The feliform family Hyaenidae, hyenas and aardwolves, has one
of the richest fossil records of all carnivorans (Werdelin and Solounias 1991). Members
of this family represent one of the longest running examples of morphological convergent
evolution (Turner et al. 2008). Six major categories of hyaenid ecomorphologies have
been recognized: (1) civet-like, (2) mongoose-like, (3) jackal-like, (4) cursorial
meat/bone specialists, (5) transitional bone crackers, and (6) full bone crackers (Werdelin
and Solounias 1996a).
Extant hyaenids (with the exception of the aardwolf, Proteles cristatus, which is a
specialized ant eater), are widely recognized as the most extreme example of bone-
cracking adaptation among living carnivorans, and as such are the obligate reference for
any study of the evolution of bone-consuming in fossil mammals. While the phenomenal
efficiency of the modern hyenas as consumers of carcasses is self-evident, the abilities of
fossil taxa must remain the subject of inference based on the extrapolation of function
from structure, a process that has its own set of strengths and weaknesses (Lauder 1995).
Previous studies have provided definitions of the mosaic of morphological features to be
117
expected in a bone-cracking carnivoran (Van Valkenburgh 2007), including the following:
a) Their snouts are somewhat shortened to increase the mechanical advantage of jaw-
closing muscles; b) The post-carnassial molars are reduced in size and number, but the
premolars are relatively massive and conical in shape; c) The skull displays a prominent
sagittal crest that rises behind the orbits giving the skull a dome-like profile, also
reflecting the expansion of attachment area of the temporalis muscle, the primary jaw
adductor muscle; in addition, the dome-shape appears to strengthen the skull by
dissipating compressive forces that occur during bone-cracking (Werdelin 1989); d)
Similarly, in response to high loads placed on the jaws during bone-cracking, the
dentaries are deep dorsoventrally, with relatively thick cortical bone (Biknevicius and
Ruff 1992); and e) The microstructure of the teeth has been modified in ways that resist
fracture, with enamel prisms layered in a highly complex architecture that resists
propagation of cracks (Stefan and Rensberger 1999).
Bone-crackers, as defined by these morphological features, differ from another set
of carnivorans, including many canids, which are best defined as “bone crushers”. Such
carnivorans use their post-carnassial molars to crush small bones, and the size of the
objects to be crushed in that way is limited by gape because of the posterior position of
the molars, which are close to the cranio-mandibular articulation. Hyaenids, and by
inference other extinct bone-crackers, such as the members of the extinct family
Percrocutidae (Tseng 2009), crack bones with their premolars, which, being more
anteriorly positioned, allow them to bite larger objects and thus access the nutrients inside
the long bones of ungulates.
118
While modern hyaenids and their closest relatives in the fossil record obviously fit
in the full bone-cracker ecomorph (category number 6 in the system of Werdelin and
Solounias [1996]), the relative abilities for bone cracking in the extinct ecomorphs are
more difficult to assess due to the lack of extant analogues. The purpose of this study is to
examine the cranial functional morphology of one of the ecomorphologies, the cursorial
meat/bone specialist hyaenids. Within this category, Werdelin and Solounias (1996a)
included the genera Lycyaena, Hyaenictis, and Chasmaporthetes, which form a
monophyletic clade in the analysis of Werdelin and Solounias (1991). Lycyaena is the
earliest member of this clade, found mostly in the late Miocene deposits of Asia
(Werdelin and Solounias 1991). Hyaenictis is relatively poorly known, with occurrences
in the late Miocene deposits of Europe and southern Africa (Werdelin and Solounias
1991). Chasmaporthetes is the best known genus of this ecomorphological group,
becoming widespread during the Pliocene of Eurasia, and was the only hyaenid form
present in North America (Hay 1921).
The postcranial anatomy of Chasmaporthetes indicates that this genus evolved
increasingly cursorial limb morphology and proportions (Berta 1981; Kurtén and
Werdelin 1988). It also possessed premolars that were slender, trenchant and with strong
posterior accessory cusps, a morphology which made them more adequate for cutting
flesh and skin than the robust, conical premolars of many other hyaenids (Berta 1981,
Khomenko 1932, Schaub 1941, Werdelin and Solounias 1991). This set of dental and
postcranial features has been recognized as indication of more actively predatory habits,
and less reliance on scavenging and bone-cracking than in modern hyenas. However, the
119
functional adaptations of the cranium are a more complex issue. Antón et al. (2006)
pointed out several morphological features suggesting a less complete adaptation of the
skull for bone cracking in Chasmaporthetes than in Crocuta and other extreme bone-
crackers, and those observations were in line with previous work by Kurtén and Werdelin
(1988). But such qualitative morphological observations can be tested and augmented
with the use of quantitative techniques.
Here we compare a cranium of a Pliocene Chasmaporthetes lunensis with a
cranium of the modern spotted hyena Crocuta crocuta, in order to assess mechanical
similarities and differences between the cranial structures of these two taxa. We utilize
the engineering technique of finite element modeling, a method useful in examining the
stress distributions in representation of real-world objects and pinpointing likely areas of
material failure. This method has been applied in carnivorans to test form-function
relationships in some of the major ecomorphologies (McHenry et al. 2007; Slater et al.
2009; Slater and Van Valkenburgh 2009; Wroe et al. 2007; Wroe 2008). We test the
hypothesis that, as interpreted from the more slender dental morphology of C. lunensis
which indicates a less osteophagous habit, its cranium is similarly weaker in conducting
stress during a simulated bone-cracking bite when compared to the modern C. crocuta.
Specifically, the structurally weaker C. lunensis cranium would exhibit more
concentrated stress in the frontal-parietal dome region, where mechanical stress is
conducted from the bone-cracking P3 premolar to the caudal cranium (Tanner et al. 2008,
Werdelin 1989). Furthermore, the more slender zygomatic arches of C. lunensis would
120
also experience higher and more concentrated stress during bone cracking compared to C.
crocuta.
Previous interpretations suggest that Chasmaporthetes evolved apart from a more
“classic”, meat-bone hyaenid morphology, to become more of a pure meat specialist. One
feature supporting that scenario is the very posterior position of the carnassials relative to
the jaw joint, an adaptation of earlier hyaenids to improve the leverage of pre-carnassial
premolars by bringing them closer to the glenoid (Kurtén end Werdelin, 1988). If,
contrary to our expectations above, the skull of Chasmaporthetes should prove to be well
adapted to handle the stresses produced during bone-cracking, this could be an additional
argument in favor of the idea of secondary adaptation to an increased proportion of meat
in the diet of this animal.
An additional benefit of this study will be the possibility of further testing the
ability of the FEA methodology to reflect the functional characteristics of structures such
as the skull of living hyaenids, whose performance at tasks like cracking bone is known
from direct observation. While FEA has become a widely used tool for assessing stress
distributions in the skulls and skeletons of fossil taxa, the technique as it is currently used
cannot account for all the material properties of bone, let alone of the soft tissue that is
combined with it in the living, moving animal whose performance we want to
approximate. Soft tissues and spaces such as that filling the frontal sinuses in the
carnivoran skull may play a role in the response of the skull to stress (Tanner et al. 2008).
Other vital aspects of the functional anatomy of the masticatory apparatus that can escape
detection by current FEA simulations are the influence of tooth morphology and
121
microstructure in the bone-cracking capabilities of a given taxon. All these considerations
will be taken into account when trying to draw reasonable conclusions from the FEA
analyses performed in this study, and we will discuss the implications of such
considerations for future studies of this type.
Abbreviations. —MNCN, Museo Nacional de Ciencias Naturales-CSIC, Madrid,
Spain; USNM, National Museum of Natural History, Smithsonian Institution,
Washington, D.C., U.S.A.

Chapter Three Materials and Methods

For anatomical data, we used the most complete cranium of Chasmaporthetes
lunensis known, MNCN-67100, from the MN (Mammal Neogene) unit 17 site of La
Puebla del Valverde (Antón et al. 2006). For the extant spotted hyena, we used the skull
of an adult, wild-caught Crocuta crocuta (USNM 164506). Computed tomography (CT)
data of C. crocuta scanned at slice thickness of 0.35 mm were kindly provided by Dr.
Blaire Van Valkenburgh (University of California, Los Angeles) and the Digital
Morphology Project (University of Texas, Austin). Chasmaporthetes lunensis was
scanned at slice thickness of 0.5 mm (Antón et al. 2006). Both datasets were exported as
1024 x 1024 pixel images (pixel size = 0.21 mm for MNCN-67100, ~0.26 mm for USNM
164506).
The reconstruction protocol, for the most part, follows those reported in Tseng
(2009). Briefly, the CT images were imported into the reconstruction program Mimics 13
122
(Materialise Software, Leuven, Belgium). A 3D reconstruction of the bone was created
there, and imported into Geomagic Studio 10 (Geomagic Inc., Research Triangles Park,
North Carolina, USA) for cleaning. A solid mesh model was then created in Strand7 2.3.7
(G + D Computing Pty Ltd, Sydney, Australia), where the finite element model was
given material properties and load conditions. For the Chasmaporthetes specimen, one of
the zygomatic arches is incomplete. Thus, the better preserved side was mirrored about
the mid-sagittal plane to represent a complete, bilaterally symmetrical model.
During the construction of the Chasmaporthetes model, it was discovered that the
overall frontal sinus morphology was similar to that of Crocuta, but the frontal sinus
struts in the fossil cranium were too difficult to differentiate objectively from its enclosed
matrix; therefore they were not included in the model. To account for this morphological
difference in our comparison of the fossil and extant taxa, we constructed three finite
element models from the same Crocuta cranium: one was built with all of the original
morphology preserved and was assigned a single material property; the second was built
with the same original morphology, but with eight different material properties assigned
to the finite elements according to the density values in their corresponding locations in
the CT data (Chapter Three Appendix Table 1). The third model was built with the
frontal sinus struts removed from the same locations as the matrix-filled regions in the
Chasmaporthetes model, and assigned a single material property. In this way, results
from the three Crocuta models can be compared to Chasmaporthetes, and a sense of the
effect of sinus strut removal and bone density differences quantified.
123
All reconstructed models were kept at their original dimensions. In addition, the
models were built to contain roughly one million tetrahedral elements, as to keep the
resolution of the finite element solutions equivalent (Chapter Three Appendix Table 2).
All craniodental structures in the skulls were assumed to be isotropic, and the analyses
conducted were linearly static. Mid-range values of Young’s modulus (20 GPa) and
Poisson’s ratio (0.3) for mammalian cortical bone were used as in Tseng (2009) to build
single material-property (homogeneous) models. The multiple material-property
(heterogeneous) model was built using eight categories of bone densities based on
Hounsfield Units (HU) obtained from the CT image data; we found the general HU
distribution to be similar between our model and the one studied by McHenry et al.
(2007), therefore we adopted the same relative properties for the eight intervals in our
Crocuta model. Behavior of homogeneous models have been shown to differ from results
obtained with heterogeneous models (Wroe et al. 2007); however, localized diagenesis in
the fossil skull prevented us from making any confident decisions in assigning multiple
material properties based on density values alone. Thus, we removed all cancellous bone
in our homogeneous models to more closely simulate skulls containing only one type of
material (i.e. cortical bone), and we included the heterogeneous model in all analyses as a
control. A sensitivity analysis comparing maximum and minimum stress and strain values
under different combinations of Young’s modulus and Poisson’s ratio values was
conducted using the Crocuta crocuta model in Tseng (2009). Two major masticatory
muscle groups, the temporalis and masseter, were modeled (Fig. 3.1). The relative
contributions of the two muscles were estimated using their cross-sectional areas as
124
calculated by the dry skull method (Thomason 1991). The muscle areas were modeled as
tessellated plates on the skull surface, contracting towards the muscle centroid on the
mandible; the BoneLoad program was used for this simulation (Grosse et al., 2007). All
biting scenarios simulated were unilateral, therefore a difference of 40% between the
working and balancing side muscle activation based on experimental Canis data was used
(Dessem 1989). Four sets of analyses were conducted.
Analysis 1 included loading all models with an arbitrarily chosen total muscle
force of 1000N to observe the resulting bite force. Bite force was measured at the third
and fourth premolar positions on both left and right sides of all models in separate runs
(Fig. 3.1). Once the symmetry of the models was confirmed with comparable results from
the left and right sides, all subsequent analyses were conducted on the left third premolar
bite only. This simulated a unilateral bone-cracking bite. Analysis 2 included loading all
models with a bite force equivalent to that expected of a 80-month old adult Crocuta
crocuta, using the regression equation (1) from Binder and Van Valkenburgh (2000),
derived from empirical bite force measurements in a wide age range of captive Crocuta
crocuta:

Bite force = 165.952 + 12.683*age (1)

A bite force of 1180.592 N calculated from this equation represents a bite force near the
cessation of bite strength increase through ontogeny (Binder and Van Valkenburgh 2000),
and is thus a reasonable estimate for maximum bite force in living captive Crocuta
125
crocuta based on empirical data. Analysis 3 simulated bite force required to overcome the
compressive strength of bovine femur in the transverse direction at 150 MPa (Cowin
1989). The bite force was calculated in proportion to the occlusal surface area of the third
premolar of each model. Because bone undergoes dynamic remodeling (Cowin 1989),
each scenario in this analysis examines biomechanical response of the cranial
morphology at the specific stage of wear (and thus age) of the individual specimens used
(Kruuk 1972b). Analysis 4 tested a wide range of bite force output from 400N to 1000N
in 200N intervals. In combination of stress values obtained in the previous three analyses,
this test provides sensitivity data on the steepness of the stress value slopes in these linear
static analyses. This analysis provides a gauge for the ability of the different models to
handle stress. Skulls that are more adapted to high stresses should have a shallower slope;
in other words, the stress should increase more slowly as muscle input and biting output
forces increased.
All stress and strain values reported in this study are “scaled” values (Tseng 2009).
In calculating stress and strain, smaller finite elements will show higher values than
larger elements for the same amount of force applied to them; therefore, the individual
elemental stress and strain values from each analysis were multiplied by their respective
element volumes and then divided by the median element volume of each model. It
should be noted that the scaled values, although useful in permitting direct comparison
between similarly constructed models, exceed the known failure strength values for
actual bone; the values thus serve a comparative purpose only. Subsequently, the
descriptive statistics used to report these values are median, median absolute deviation
126
(MAD) from the median, and maximum of the scaled values; the first two measures are
robust to skew in the distribution of stress and strain values, which is expected when
finite element models include point loads such as the constraints placed at the biting tooth
(Dumont et al. 2005). One important point to make about the interpretation of all stress
and strain values reported here is that besides the fact the scaled values are artificially
high compared to experimental results from actual bone specimens, the homogeneous
modeling method employed here approximates general stress distribution patterns but
does not exactly replicate finer details such as forces across neighboring bones of very
different properties; thus the numerical results should be interpreted as comparative
values on the level of the entire cranium, and in the context of the heterogeneous model.
The muscle input forces and model parameters are summarized in Chapter Three
Appendix Tables 2 and 3.
An index of bone-cracking capability.— In order to estimate the relative bone
sizes that each carnivoran is (or was) capable of cracking, we turn to food fracture
mechanics. For simpler cases of cracking of epiphyseal and/or flat bones in which the
contact area of the bone with teeth is a relatively small part of the entire bone being
consumed, the bone-cracking bite in hyaenids can be reasonably represented by a three-
point bending scenario, whereby the lower p3 and p4 serve as support and the upper P3
the driving point. Cracking of bone shafts with a hollow cortex is more complex and thus
is not considered here, but the basic premise of bending and crack initiation is still valid
(in those cases the p3-p4 distance becomes more of a proxy for gape in calculating this
127
index). If we then think of a food item as a beam, the maximum stress created in the food
item by the three-point biting can be represented by equation (2):

σ
F
= 3Fl / 2Ebt
2
(Lucas et al. 2002) (2)

where F is the bite force being applied by the P3, l is the distance between the lower p3
and p4 bite points, E is the elastic (Young’s) modulus of the food item (cortical bone), b
is the length of the food item, and t the depth or diameter of the bone. σ
F
is also known as
the modulus of rupture. Under this bending scenario the P3 is applying compressive
stress at the bite point, and the maximum stress in the food item is the tensile stress at the
midpoint between p3 and p4; thus, when σ
F
exceeds the tensile strength of the food item,
the food particle fails and a crack is initiated. For the purpose of this estimation, the point
of crack initiation is assumed to occur at the point of maximum tensile stress and not the
P3 bite point (where prey bone is under compressive stress, and thus stronger, and is
more likely to experience localized indentation when the bone-cracking cusp is sharp).
This equation for describing food fracture mechanics is analogous to one for calculating
the modulus of rupture in more general structural engineering contexts (equation 3):

σ
F
= 6M / bt
2
(Bazant and Li 1995) (3)

128
where M is the ultimate bending moment on the beam being stressed; this is analogous to
the application of bite force at P3 with two supporting points at p3 and p4, respectively,
which also creates a bending moment in the bone being cracked.
The lack of accurate material properties for different sized ungulate long bones
prevents us from making absolute estimates of prey size, but relative conclusions can be
made by comparison to the theoretical values calculated for Crocuta crocuta and from
published observations of prey preference in that species. If we rearrange the previous
equation to solve for the depth (or diameter) of the food item t, the equation (eq. 4)
becomes:

t = (3Fl / σ
F
2Eb)
1/2
(4)

If we assume that the elastic modulus (E) for the bone being consumed is constant (or
varies very little) over the range of medium to large sized ungulates (Ashby et al. 1995),
and that the length b of the food item is always much longer than the transverse width of
the jaw of the bone-cracking predator (which holds true when the tooth contact area is
small relative to size of bone consumed), then the thickness of the food item a predator is
capable of cracking becomes proportional to (eq. 5):

t ≈ (3Fl / 2 σ
F
)
1/2
(5)

129
As can be seen from the original equation, the modulus of rupture is inversely
proportional to the squared value of t. This means for every doubling of bone diameter,
the maximum tensile stress required to initiate a crack in the bone using the same biting
configuration will be multiplied by a factor of four. Thus, in order to increase the bone-
cracking capability to process larger diameter bones, the predator needs to increase both
F, the bite force, and l, the p3-p4 interdental distance, to counteract the increase in
ultimate tensile strength of the food item. Both bite force and interdental distance are
increased in absolute terms by enlarging body size, but this equation also predicts the
evolution of relatively strong muscle forces for a given skull size, the associated
buttressing of cranial bones to withstand this stress, as well as a larger gap in the region
of the bone-cracking teeth to increase the bending moment during the bone-cracking bite.
To assess the extent that each carnivoran skull is specialized for bone-cracking
given their skull size, we calculated the relative bite forces of each finite element model
under the scenario where all models exhibit identical median cranial stress (arbitrarily
chosen at 10 MPa). Because the finite element analyses conducted in this study are
linearly elastic, the absolute magnitudes of the median stress values will not alter the
relationships of the relative bite forces. Furthermore, we assume the ultimate strength of
the modeled skull bone is identical within a reasonable range of behavior across species
because they are composed of similar material (i.e. mammalian bone). Under these
assumptions, the relative capabilities calculated for a given median cranial stress provide
the same relative magnitudes as for median cranial stress near maximum strength, thus
results can be extrapolated for maximum capability. We measured the P3-P4 interdental
130
distance of each species as a proxy for the p3-p4 distance because no associated
mandibles are known for the Chasmaporthetes specimen, and interdental distances are
closely correlated between upper and lower cheek teeth for proper occlusion (we
measured associated dentitions in a fossil Chasmaporthetes specimen housed in the
Institute of Vertebrate Paleontology and Paleoanthropology, IVPP V7275, and observed a
close correspondence between P3-P4 and p3-p4 inter-dental distances at the tallest cusps).
We then calculated the product of F and l to represent bone-cracking capability (i.e.
relative maximum prey bone size). Using the orbital-occipital length (OOL) of the
respective skulls as a proxy for body mass (Van Valkenburgh 1990), we then calculated
the ratio between capability and OOL (eq. 6) as a measure of bone-cracking
specialization (i.e. the bone-cracking index, BCI):

(F
s
l) / OOL (6)

where F
s
is the P3 bite force produced at identical cranial median stress values, l is the
interdental distance between P3 and P4, and OOL is orbital-occipital length. While
relatively simple and not taking into account the full range of complex mechanical
behavior that exists during the manipulation by hyenas of different kinds of prey bone for
cracking, this index provides an elementary quantification of bone-cracking capability
which has thus far been descriptive.

131
Chapter Three Results

Sensitivity test.— Maximum and minimum von Mises stress and strain values are
reported in Table 3.1. With Young’s modulus held constant at 20 GPa, Poisson’s ratio
was increased in 0.1 intervals from 0.0 to 0.49, the range observed in most materials.
Poisson’s ratio ( ν) is a measure of compressibility; the increase in this ratio is
accompanied by decrease in maximum stress and strain. Von Mises strain increases
slightly at ν=0.1 and 0.2, then values for ν=0 and ν=0.3 are equal, and strain decreases
for ν=0.4 and 0.49 (Fig. 3.2). Minimum strain stayed at zero throughout the range, but
minimum stress increases dramatically from ν=0.4 to 0.49. For Young’s modulus, a
Poisson’s ratio was held constant at ν=0.36 while modulus values ranged from E=15 GPa
to 30 GPa, the range observed in mammalian femoral cortical bone (Erickson et al. 2002).
Young’s modulus is a measure of stiffness; both maximum and minimum von Mises
stress did not change through this interval (Table 3.1). Maximum strain, however, peaked
at E= 20 GPa and decreased with both large and smaller values of E.
Analysis 1 (Table 3.2, Fig. 3.3). — In the original heterogeneous Crocuta crocuta
model, P3 produced an average bite force of 158.02 N and P4 192.50 N; the
homogeneous model produced only slightly higher forces at P3 (165.34 N) and P4
(243.55 N) (Fig. 3.3A-B, E-F). P3 biting tends to have lower median stress and
dispersion around the median compared to P4 biting. Maximum stress and strain values
are slightly higher for P4 biting than for P3; the heterogeneous model had higher
maximum values than the homogeneous model across all tooth positions analyzed. In the
132
Crocuta crocuta model with frontal bony struts removed, P3 produced an average of
200.97 N and P4 249.67 N, respectively (Fig 3.3C, G). The stress and strain values
overlap with each other, with no clear differences among biting scenarios.
Chasmaporthetes produced an average of 184.54 N at P3 and 227.09 N at P4 (Fig. 3.3D,
H). The stress values are within the same range across all biting scenarios.
Analysis 2 (Table 3.3). — When all models were given enough muscle force input
to produce a bite force of 1180.592 N at LP3, the median scaled stress is lowest in the
heterogeneous Crocuta crocuta model, followed by the homogeneous C. crocuta model
with bony struts removed. The patterns hold true for median strain values, although
Chasmaporthetes shares similar strain magnitudes with the heterogeneous C. crocuta
model. Maximum scaled strain and stress are lowest in Chasmaporthetes.
Analysis 3 (Table 3.4). — When the models were loaded with enough bite force to
crack a bovine femur with strength of 150 MPa, the median stress is highest in the
homogeneous original Crocuta model. Chasmaporthetes produced stress values between
those of the Crocuta models, but had the lowest strain values.
Analysis 4 (Table 3.5, Fig. 3.4). — In addition to scenarios already tested at LP3
from the previous three analyses, muscle forces producing bite forces of 400, 600, 800,
and 1000 N at LP3 were tested for all models (Table 3.5). Because all models were built
as elastic isotropic material being loaded in linear static conditions, these bite force
output values were used to calculate the linear slope of stress response (Fig. 3.4). The
highest slope belonged to the homogeneous original Crocuta model (0.00194). This is
133
followed by lower slope in Chasmaporthetes (0.00166), and then the modified Crocuta
model (0.00132). The heterogeneous model had the lowest slope at 0.00081.
Bone cracking capability (Table 3.6). — Given identical levels of median cranial
stress, the heterogeneous Crocuta model produced the highest bite force, followed by
Chasmaporthetes and the simplified Crocuta model. Comparing bone-cracking capability
(F
s
l), the estimated maximum bone size Chasmaporthetes can theoretically process is less
than that of Crocuta (88% of Crocuta maximum). For the bone-cracking index (BCI),
which takes into account the body size of the species for evaluating their bone-cracking
capability, the patterns are different. Chasmaporthetes is more specialized than the living
Crocuta in bone-cracking capability when considering its size. The BCI does not address
the hunting abilities of these taxa, however (see discussion).

Chapter Three Discussion

The sensitivity test showed that stress and strain values vary gradually between
Poisson’s ratios of 0.0 to 0.4 (Fig. 3.2). Thus, for comparative purposes, the value of 0.3
used in this study should give reasonable midrange results. The range of Young’s
modulus values tested, however, showed a peak in maximum strain at E=20 GPa but no
change in stress (Table 3.1). The modulus of 20 GPa used in this study would therefore
result in more conservative estimates of maximum strain in the skull models.
The summary statistics given for left and right P3, P4 biting demonstrate the
similarity in overall cranial stress in all biting scenarios (Table 3.2), as found by Tseng
134
(2009). This means that for all models studied, the cranial morphology is not optimized
for specifically P3 or P4 biting. Furthermore, the general distribution of stresses as
illustrated by Figure 3.3 shows overall similarity of stress distribution in the two species;
the fronto-parietal dome and the zygomatic arches of both Chasmaporthetes and Crocuta
exhibit similar levels of stress during a bone-cracking bite. This result does not support
our initial hypothesis that the cursorial meat and bone specialist Chasmaporthetes has
weaker fronto-parietal dome and zygomatic arches.
The heterogeneous Crocuta model generated slightly lower bite forces than
models built with a single homogeneous material property, but had almost half the
magnitude of median stresses compared to the homogeneous models (Table 3.2).
However, peak stress and strain are in general higher in the heterogeneous model. The
denser (and thus stiffer) elements in the heterogeneous model might have transmitted
more of the stress to the bite, instead of conducting them to regions of low density bone
which are more easily stressed, thereby lowering the overall median stress levels. The
visualization of cranial stress in Figure 3E shows higher levels of stress across the fronto-
parietal region than in the homogeneous models, whereas stresses in other regions of the
cranium are lower. The domed frontal region has been hypothesized to serve a dissipating
function, and the channeling of more stress into this region seen in the heterogeneous
model could be evidence for this mechanism in hyaenids (Werdelin 1989). Otherwise, the
overall stress distributions are quite similar among the different Crocuta models,
providing support for the results obtained in the homogeneous Chasmaporthetes model as
an approximation of its mechanical behavior.
135
In the simulation of P3 biting with force production of 1180.592 N,
Chasmaporthetes showed overall cranial stress closest to that of the homogeneous
original Crocuta model (Table 3.3). Given the effect of bony struts in increasing overall
cranial stress as shown by the Crocuta models, Chasmaporthetes probably would have
had higher overall cranial stress if those bony struts were included in the model. On the
other hand, the heterogeneous Crocuta model had the lowest overall stress; therefore the
Crocuta results actually bracket those in the Chasmaporthetes model. This overlap in
estimated masticatory stress levels between the various Crocuta models and
Chasmaporthetes support the interpretation of their having similarly adapted crania for
bone-cracking, as also evident from the bone-cracking index analysis highlighted below
(Table 3.6). Moreover, our results are consistent with those reported by Tanner et al.
(2008), who showed that a modified Crocuta model with filled frontal sinus conduct
forces more efficiently than the original morphology by creating a solid homogeneous
pathway. Along the same lines, the removal of bony struts within the frontal sinus in our
Crocuta model probably also had a homogenizing effect on the force conduction pathway.
What our models do not take into account is the presence of soft tissue in the frontal sinus,
which could have an effect on the dissipation of stress. The enlarged frontal sinus has
evolved convergently in hyaenids, percrocutids (Tseng 2009), and borophagine canids
(Tseng and Wang in press; Werdelin 1989). While this morphological feature is not
necessarily adaptive a priori simply because they are associated with bone cracking
dentition in one group of carnivorans, their independent development in all three groups
is strongly suggestive of such adaptive value. Furthermore, comparative finite element
136
analyses have shown that the domed frontal region of the bone crackers Crocuta and
Dinocrocuta receive lower and more evenly distributed stress than the non-domed Canis
lupus (Tseng 2009).
When occlusal area is considered during a P3 bone-cracking bite with enough
force to overcome bovine femur strength, the results show no obvious adaptations in the
skulls of Crocuta or Chasmaporthetes for bone-cracking, as the stresses are higher in all
models (Table 3.4). The stress and strain both increased in the following order:
heterogeneous Crocuta, Chasmaporthetes, modified Crocuta, and the original Crocuta
model (Table 3.4). This is in part correlated to the calculated P3 occlusal areas of each
model (Chapter Three Appendix Table 2). The Crocuta specimens had the larger absolute
P3 surface area, which required a higher force input to produce enough force output to
crack a bone compared to smaller occlusal areas. Contrary to an expectation that cranial
bone remodeling through ontogeny might compensate for the increased bite force
required for bone cracking by changes in cranial shape and bone thickness (Binder and
Van Valkenburgh 2000), the stress and strain response in the Crocuta models were high.
This apparent contradiction between the physical ability of a pointed tooth to more easily
puncture, and thus damage bone, and the greater crushing power in a mature individual
with worn teeth is highlighted by our results. All bone-cracking carnivorans develop
horizontal wear in their premolars relatively early in ontogeny, thus enlarging the
occlusal area and demanding more of their masticatory muscles in order to crack bones.
We suspect that increased muscular power in mature individuals still gives them an
advantage in bone-cracking over young individuals; evidence from FE modeling of
137
borophagine canid crania suggest that a change in muscle activation ratios and increased
reliance on temporalis muscle use may have improved efficiency of bite force production
in bone-cracking dogs (Tseng and Wang in press). The efficiency provided by a small
occlusal surface, while making sense mechanically, is probably balanced through the
evolution of bone-cracking forms by the necessary wear of the tooth surface through
contact with such abrasive material as bone. Given the influence of occlusal surface size
in our results, it is worth considering the convenience of performing specific,
comparative FEA studies of the premolars of carnivorans, to test if the analyses reflect
the advantages of the robust, conical teeth of specialized bone-crackers over the more
fragile, blade-like premolars of meat specialists, despite the apparent advantage of the
smaller occlusal surface of the latter when both are subject to wear. A high frequency of
tooth fracture is a clear cost of bone-cracking in large carnivorans (Van Valkenburgh
1988) and the greater resistance of broad-based teeth to fracture provides enough
advantage to compensate for increasing occlusal surfaces in worn teeth, but a whole-skull
analysis may be too coarse-grained to properly assess such differences. In this connection,
a fruitful future direction would be to use finite element modeling to compare the
performance of the premolars within a specimen, across different taxa, and through
ontogenetic stages in order to reveal any tooth-specific adaptations for stress dissipation
(Rensberger 1995, Rensberger and Stefen 2006).
The stress response slopes obtained by testing a range of bite forces in each model
provided evidence for the differential capabilities of the models in handling cranial stress
(Fig. 3.4, Table 3.5). Chasmaporthetes has a steeper slope compared to the modified and
138
the heterogeneous Crocuta models. The homogeneous original Crocuta model shows
steeper stress than Chasmaporthetes, and it is expected that the original Chasmaporthetes
skull would have had a slope falling somewhere between the high-slope homogeneous
and the low-slope heterogeneous Crocuta models (Fig. 3.4). Even though analysis using a
single material property is potentially more difficult to interpret for biological meaning,
the bracketing of the Chasmaporthetes slope between the homogeneous and
heterogeneous Crocuta models could be seen as a rough confidence interval for the range
of mechanical behavior expected in the original Chasmaporthetes cranium. It might be
interesting to compare available mandibles of these taxa with the same technique, as
canine mandibles have shown to behave in an elastically isotropic manner, akin to a
single material object (Ashman et al. 1985). In addition, mandibles are also simpler in
structure and function, when compared to the cranium which serves multiple sensory
roles. A study of mandibles paralleling the analyses conducted here would be the next
step in further understanding jaw mechanics and extrapolating potential clade-wide
constraints across different carnivore ecomorphologies in a narrower functional context
(McHenry et al. 2007; Tseng and Binder 2010; Wroe et al. 2007; Wroe 2008). As far as
we know, a lower mandible of Chasmaporthetes lunensis comparable in completeness to
the cranium specimen used in this study is not available at the present, thus
reconstructions would have to be made for such an analysis.
The bone-cracking capability estimated from interdental distance and relative bite
force production indicates that the Chasmaporthetes cranium examined in this study had
relatively lower capacity to crack bones when compared to a typical adult spotted hyena
139
(Table 3.6). This should not come as a surprise, because the Chasmaporthetes specimen
examined is both smaller and more slender, especially at the zygomatic arches, than the
modern Crocuta cranium. However, when body size is included into consideration,
Chasmaporthetes holds up very well on its own, and its bone-cracking index (106) is
comparable to the spotted hyena (100). This result would indicate that, at least cranially,
Chasmaporthetes was just as adapted for handling stress incurred during bone-cracking
behavior as the modern Crocuta. The constraint on the prey bone size that
Chasmaporthetes was capable of consuming, then, would be placed on other factors such
as dental morphology and overall gape. Of course, bone-cracking capability does not
offer any direct insight to the maximum prey size of Chasmaporthetes, as large prey
animals with bones that exceed the processing ability of the predator could nevertheless
have been hunted for consumption of soft tissues and skeletal elements other than long
bones. The results do indicate, however, that if larger prey were hunted, Chasmaporthetes
probably could not have utilized the carcass to the degree seen in modern Crocuta. The
premolar toothrow of Chasmaporthetes is relatively elongate and slender, with well-
developed accessory cusps. These teeth are not as well suited for bone-cracking as the
thick conical premolars of the modern Crocuta. Furthermore, mandibles of
Chasmaporthetes we have observed at other fossil localities are also more slender than
those of Crocuta. Considering the totality of these morphological features of the
craniodental apparatus of Chasmaporthetes, the robust cranium might resist high stresses
less during bone-cracking, but perhaps more during the killing stage when the predator
might lock its jaws on a struggling prey. This interpretation, along with the observation
140
that Chasmaporthetes had slender limbs for cursorial motion (Berta 1981, Kurtén and
Werdelin 1988), would point to a killing strategy combining pursuit and a holding kill
bite. This type of behavior is also observed in modern Crocuta, where individuals lock
their jaws onto parts of a running prey in order to bring it down (Kruuk 1972b).
Chasmaporthetes might have utilized this hunting technique to a larger extent than the
modern Crocuta, and consumed less bone because of its more slender dentition and
smaller size (Turner et al. 2008, Werdelin and Solounias 1991, 1996a). Both prey
acquisition and food processing probably imposed selective pressure on the evolution of
such a flexible hunting and bone-cracking predator; mechanical forces resulting from the
former behavior, potentially involving slashing of prey using anterior teeth during pursuit
and then locking the jaws into the prey during subsequent struggle, are probably
significantly more unpredictable and complex compared to a stereotypical bite with the
purpose of cracking bones, the latter being the scenario tested in this study. For future
work, the addition of torsional loads representative of extrinsic forces during active
hunting is required to see if the skull is more or less adapted to such behavior compared
to Crocuta.
Independent of the advantages that the cranial structure of Chasmporthetes could
provide for its inferred, actively hunting lifestyle, it is also feasible that its skull
morphology was largely inherited from ancestors that had a more pronounced bone-
cracking habit, as evidenced for instance by the location of the carnassials closer to the
jaw joint than in other carnivorans, a typical hyaenid feature (Kurtén and Werdelin 1988).
The intermediate state of the enamel microstructure adaptation in Chasmaporthetes also
141
indicates that it was more specialized than its middle Miocene hyaenid ancestors, but the
former never evolved the sharp-angle zig-zag Hunter Schreger Bands of the full bone-
crackers (Ferretti 1999, 2007). Evidence from gross premolar morphology, superficial
and internal enamel structure, and aspects of cranial morphology all point to
Chasmaporthetes as intermediate in the gradient of morphologies evolving towards full
bone-crackers. At any rate, the overall adaptations of Chasmaporthetes allowed this
animal to be a proficient bone cracker, and the full consumption of most carcasses of its
own prey as well as opportunistic scavenging were well within its range of likely feeding
behaviors. This inferred paleobiological model for Chasmaporthetes implies considerable
room for competition with members of the family Canidae, including pack-hunting wolf-
like species throughout the geographical range of the genus, and the specialized bone-
cracking canid Borophagus in North America. The latter belongs to the borophagine
canid lineage that showed the same evolutionary progression of acquiring a mosaic of
features towards bone-cracking as in hyaenids, and some of the species must have
occupied similar intermediate ecological roles as Chasmaporthetes. The direct
competition implied by this repetitive mode of adaptation is likely to have contributed to
the ultimate demise of this hyaenid in the New World, and potentially of several Eurasian
hyaenids during the dispersal of canids across Beringia.

142
Chapter Three Conclusion

We built and compared finite element models of adult crania of the fossil hyaenid
Chasmaporthetes lunensis from the Pliocene site of La Puebla del Valverde (Spain) and
the modern spotted hyena Crocuta crocuta, a carnivoran whose extreme bone-cracking
abilities are well documented. The Chasmaporthetes model experienced higher stresses
when performing a similar bone-cracking task at the P3 as Crocuta, but the overall
distributions of stress are similar between the two taxa. Using an index of bone-cracking
capability, we showed that Chasmaporthetes, even though smaller in size and probably
not able to consume prey bone as large as Crocuta could, was nevertheless very well
adapted in the cranium to withstand high stresses for its size. This result interpreted in
combination with other craniodental morphological features, especially the more
trenchant and less robust shape of the premolars, indicates that the strong cranium of
Chasmaporthetes might be a tool to deal more with struggling prey during hunting, than
with frequent bone cracking. As suggested by previous workers, Chasmaporthetes was
likely a highly cursorial hunter cranially adapted for grabbing moving prey, but not as
extremely adapted for bone-cracking as Crocuta. The robust, stress-resistant morphology
of the skull of Chasmaporthetes may be in part inherited from ancestors adapted to a diet
including more bone and less meat. Besides true hyaenids, other carnivores, such as
percrocutids and borophagine canids, have convergently evolved similar morphologies,
including domed frontal regions. While the shared presence of such features is already
suggestive of their adaptive value, comparative finite element analyses show that domed
143
frontal regions receive less stress than non-domed ones as found in the wolf, Canis lupus.
There is, thus, a growing body of functional evidence suggesting that the bone-cracking
adaptations shape a complex model that has evolved convergently several times across
different carnivoran families, producing similar morphologies composed of a mosaic of
traits likely to be under strong adaptive control.

Chapter Three Acknowledgments

We thank B. Van Valkenburgh at the University of California, Los Angeles for
providing CT data of Crocuta crocuta; X. Wang at the Natural History Museum of Los
Angeles County for stimulating discussions and ideas about studying functional
morphology using FEA; J. Totten of Mimics Software for technical support; The
Dinosaur Institute at the Natural History Museum of Los Angeles County for research
space. M. Carrano and two anonymous reviewers provided insightful suggestions that
greatly improved the content of this paper. This study was funded by a National Science
Foundation Graduate Research Fellowship, University of Southern California Zumberge
Grant, and American Society of Mammalogists grant-in-aid of research (to ZJT), and
CGL2008-00034/BTE (Spanish Ministry of Science and Innovation).
144
Chapter Three References
Antón, M., A. Turner, M. J. Salesa, and J. Morales. 2006. A complete skull of
Chasmaporthetes lunensis (Carnivora, Hyaenidae) from the Spanish Pliocene site
of La Puebla de Valverde (Teruel). Estudios Geológicos 62(1):375-388.

Ashby, M. F., L. J. Gibson, U. Wegst, and R. Olive. 1995. The mechanical properties of
natural materials. I. Material property charts. Proceedings of the Royal Society of
London, Series A 450:123-140.

Ashman, R. B., G. Rosinia, S. C. Cowin, and M. G. Fontenot. 1985. The bone tissue of
the canine mandible is elastically isotropic. Journal of Biomechanics 18(9):717-
721.

Bazant, Z. P., and Z. Li. 1995. Modulus of rupture: size effect due to fracture initiation in
boundary layer. Journal of Structural Engineering 121(4):739-746.

Berta, A. 1981. The Plio-Pleistocene hyaena Chasmaporthetes ossifragus from Florida.
Journal of Vertebrate Paleontology 1(3-4):341-356.

Biknevicius, A. R., and C. B. Ruff. 1992. The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology
228:479-507.

Binder, W. J., and B. Van Valkenburgh. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252:273-283.

Cowin, S. C. 1989. Bone mechanics. CRC Press, Inc., Boba Baton, FL.

Dessem, D. 1989. Interactions between jaw-muscle recruitment and jaw-joint forces in
Canis familiaris. Journal of Anatomy 164:101-121.

Dumont, E.R., Piccirillo, J., and I.R. Grosse. 2005. Finite-element analysis of biting
behavior and bone stress in the facial skeletons of bats. Anatomical Record Part A
283A:319-330.

Erickson, G. M., J. Catanese III, and T. M. Keaveny. 2002. Evolution of the
biomechanical material properties of the femur. The Anatomical Record 268:115-
124.

Ewer, R. F. 1973. The carnivores. Cornell University Press, Ithaca.

145
Ferretti, M.P. 1999. Tooth enamel structure in the hyaenid Chasmaporthetes lunensis
lunensis from the Late Pliocene of Italy, with implications for feeding behavior.
Journal of Vertebrate Paleontology 19:767-770.

___. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia, Carnivora).
Swiss Journal of Geoscience 100:41-52.

Grosse, I.R., Dumont, E.R., Coletta, C., and A. Tolleson. 2007. Techniques for modeling
muscle-induced forces in finite element models of skeletal structures. Anatomical
Record 290:1069-1088.

Hay, O. P. 1921. Descriptions of species of Pleistocene Vertebrata, types or specimens
most of which are preserved in the United States National Museum. Proceedings
of the United States National Museum 59:599-642.

Khomenko, I. P. 1932. Hyaena borissiaki n. sp. iz russil'onskoj fauny Bessarabii.
Travaux de l'Institut Paléozoologique de l'Academie des Sciences de l'U.R.S.S.
1:81-134.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior. The
University of Chicago Press, Chicago.

Kurtén, B., and L. Werdelin. 1988. A review of the genus Chasmaporthetes Hay, 1921
(Carnivora, Hyaenidae). Journal of Vertebrate Paleontology 8(1):46-66.

Lauder, G. V. 1995. On the inference of function from structure. Pp. 1-18 in J. J.
Thomason, ed. Functional morphology in vertebrate paleontology. Cambridge:
University of Cambridge Press.

Lucas, P. W., J. F. Prinz, K. R. Agrawal, and I. C. Bruce. 2002. Food physics and oral
physiology. Food Quality and Preference 13:203-213.

McHenry, C., S. Wroe, P. D. Clausen, K. Moreno, and E. Cunningham. 2007.
Supermodeled sabercat, predatory behavior in Smilodon fatalis revealed by high-
resolution 3D computer simulation. Proceedings of the National Academy of
Sciences 104:16010-16015.

Nowak, R. M. 1999. Walker's carnivores of the world. The Johns Hopkins University
Press, Baltimore.

Rensberger, J. M. 1995. Determination of stresses in mammalian dental enamel and their
relevance to the interpretation of feeding behaviors in extinct taxa. Pp. 151-172 in
J. J. Thomason, ed. Functional morphology in vertebrate paleontology.
Cambridge University Press, New York.
146

Rensberger, J. M., and C. Stefen. 2006. Functional differentiations of the microstructure
in the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A
278(1-6):149-162.

Schaub, S. 1941. Ein neues Hyaenidengenus von der Montagne de Perrier. Eclogae
Geologicae Helvetiae 34:279-286.

Slater, G. J., E. R. Dumont, and B. Van Valkenburgh. 2009. Implications of predatory
specialization for cranial form and function in canids. Journal of Zoology
278:181-188.

Slater, G. J., and B. Van Valkenburgh. 2009. Allometry and performance: the evolution
of skull form and function in felids. Journal of Evolutionary Biology 22:2278-
2287.

Stefan, C., and J. M. Rensberger. 1999. The specialized structure of hyaenid enamel:
description and development within the lineage-including Percrocuta. Scanning
Microscopy 13:363-80.

Tanner, J. B., E. R. Dumont, S. T. Sakai, B. L. Lundrigan, and K. E. Holekamp. 2008. Of
arcs and vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta
crocuta). Biological Journal of the Linnean Society 95:246-255.

Thomason, J. J. 1991. Cranial strength in relation to estimate biting forces in some
mammals. Canadian Journal of Zoology 69:2326-2333.

Tseng, Z. J. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96:51-67.

Tseng, Z. J., and W. J. Binder. 2010. Mandibular biomechanics of Crocuta crocuta,
Canis lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia).
Zoological Journal of Linnean Society 158:683-696.

Tseng, Z.J., and X. Wang. In press. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology.

Turner, A., M. Antón, and L. Werdelin. 2008. Taxonomy and evolutionary patterns in the
fossil Hyaenidae of Europe. Geobios 41(5):677-687.

Van Valkenburgh, B. 1988. Incidence of tooth breakage among large, predatory
mammals American Naturalist, 131:291-302.
147

___. 1990. Skeletal and dental predictors of body mass in carnivores. Pp. 181-205 in J.
Damuth, and B. J. MacFadden, eds. Body Size in Mammalian Paleobiology:
Estimation and Biological Implications. Cambridge University Press, Cambridge.

___. 2007. Déjà vu: the evolution of feeding morphologies in the Carnivora. Integrative
and Comparative Biology 47:147-163.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

___. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective. Pp. 582-
624 in J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution. Cornell
University Press, New York.

Werdelin, L., and N. Solounias. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30:1-104.

___. 1996. The evolutionary history of hyaenas in Europe and western Asia during the
Miocene. Pp. 290-306 in R. L. Bernor, S. Rietschel, and W. Mittmann, eds. The
Evolution of Western Eurasian Miocene Mammal Faunas. Columbia University
Press, New York.

Wroe, S. 2008. Cranial mechanics compared in extinct marsupial and extant African lions
using a finite-element approach. Journal of Zoology 274:332-339.

Wroe, S., P. D. Clausen, C. McHenry, K. Moreno, and E. Cunningham. 2007. Computer
simulation of feeding behaviour in the thylacine and dingo as a novel test for
convergence and niche overlap. Proceedings of the Royal Society B: Biological
Sciences 274:2819-2828.

Wroe, S., Moreno, K., Clausen, P., McHenry, C. and Curnoe, D. 2007. High-resolution
three-dimensional computer simulation of hominid cranial mechanics. Anatomical
Records 290:1248-1255

148
Chapter Three Appendix Table 1. Material properties used in the heterogeneous Crocuta
model. Modulus and density values were taken from McHenry et al. (2007). HU,
Hounsfield Units, representing relative bone densities in the skull; MPa, megapascal.

HU range Young's Modulus (MPa) Density (kg/mm
3
)
-779~-305 1530 2.51x10
-7

-304~170 1870 2.92x10
-7

171~645 2220 3.33x10
-7

646~1120 10790 1.09x10
-6

1121~1595 21730 1.86x10
-6

1596~2070 27080 2.19x10
-6

2071~2545 32700 2.53x10
-6

2546~3015 38580 2.86x10
-6

149
Chapter Three Appendix Table 2. Parameters used to construct the finite element models
analyzed in this study. Centroid values refer to the node number in each model towards
which the respective simulated muscle contracts. CBL, condylobasal length; LP3, left
third premolar; SA, surface area.

C. lunensis C. crocuta (no
struts)
C. crocuta
(original)
CBL (mm) 225.00 238.80 238.80
Left temporalis centroid 253320 366364 344219
Right temporalis centroid 255964 366574 259953
Left masseter centroid 277563 396419 294850
Right masseter centroid 293330 396464 289494
Working temporalis % 35.60 34.88 34.88
Balancing temporalis % 21.40 20.93 20.93
Working masseter % 26.90 27.63 27.63
Balancing masseter % 16.10 16.58 16.58
SA LP3 (mm
2
) 46.86 67.16 67.16
Tetrahedral elements 1121102 1174149 1173850
Model volume (mm
3
) 4.61x10
5
4.56x10
5
4.77x10
5

150
Chapter Three Appendix Table 3. Muscle proportions used in the analyses. The original
Crocuta values are for homogeneous/heterogeneous models, respectively, where two
values are given. All forces are in Newtons.

Analysis Force input Output force at LP3
1 1000 185.88 C. lunensis
2 6397 1180.59
3 38090.93 7029.30
4 2168 400.00
3251 600.00
4335 800.00
5419 1000.00
1 1000 204.59 C. crocuta (no struts)
2 5874 1180.59
3 50129.12 10074.45
4 1990 400.00
2986 600.00
3981 800.00
4976 1000.00
1 1000 133.30/156.02 C. crocuta (original)
2 7140/6609 1180.59
3 60928.99/56400 10074.45
4 2419/2239 400.00
3629/3359 600.00
4839/4479 800.00
6048/5598 1000.00
151
Figure 3.1. Illustration of the lateral view of Crocuta crocuta skull, with the 3
rd
and 4
th

premolar positions marked. Dark gray represents the area marked as temporalis in the
models, light gray represents masseter.

152
Figure 3.2. Sensitivity test of a single finite element cranium model (LACM30655; Tseng
2009) with different Poisson’s ratios. Circles indicate stress values, squares are strain
values. Stress is measured in megapascals, strain in microstrain.

153
Figure 3.3. Ventral (A-D) and dorsal (E-H) view of the Von Mises stresses in the finite
element models loaded with 1000 N of total muscle input force. A, E, Crocuta crocuta
heterogeneous original model; B, F, C. crocuta homogeneous original model; C, G, C.
crocuta with bone struts removed; D, H, Chasmaporthetes lunensis. Warmer and brighter
colors indicate high stress; colder and darker indicate low stress.

154
Figure 3.4. The median stress values at different bite force levels for the original
homogeneous Crocuta crocuta (diamond), Chasmaporthetes lunensis (circle), Crocuta
crocuta with bony struts removed (square), and the heterogeneous Crocuta model
(triangle). For data values see Table 3.5.

155

Table 3.1. Sensitivity tests altering the values of Young's modulus (E) and
Poisson's Ratio of a Crocuta crocuta model (LACM 30655; Tseng, 2009).
GPa, gigapascal; MPa, megapascal; μE, microstrain.

Von Mises stress (MPa) Von Mises strain ( μE)
E (GPa) Poisson Maximum Minimum Maximum Minimum
20.0 0.0 4.10E+02 1.33E-09 2.06E-08 0.00E+00
20.0 0.1 3.87E+02 6.16E-10 2.13E-08 0.00E+00
20.0 0.2 3.57E+02 1.30E-09 2.14E-08 0.00E+00
20.0 0.3 3.17E+02 2.32E-09 2.06E-08 0.00E+00
20.0 0.4 2.55E+02 3.09E-09 1.79E-08 0.00E+00
20.0 0.5 1.00E+00 2.47E-05 0.00E+00 0.00E+00
15.0 3.6 2.84E+02 2.33E-09 1.68E-10 0.00E+00
20.0 3.6 2.84E+02 2.33E-09 1.93E-08 0.00E+00
22.6 3.6 2.84E+02 2.33E-09 1.71E-08 0.00E+00
25.0 3.6 2.84E+02 2.33E-09 1.55E-08 0.00E+00
30.0 3.6 2.84E+02 2.33E-09 1.29E-08 0.00E+00

156
Table 3.2. Summary values for analysis 1: 1000N muscle input force at third and fourth
premolars. For the original Crocuta model, first value is from the homogeneous model,
second from the heterogeneous model.

Chasmaporthetes lunensis
LP3 LP4 RP3 RP4
Bite force (N) 185.88 227.55 183.20 226.62
Median scaled stress (MPa) 0.306 0.317 0.301 0.314
MAD stress (MPa) 0.250 0.256 0.248 0.256
Max scaled stress (MPa) 99.280 83.618 98.236 147.738
Max scaled strain ( μE) 0.00644 0.00542 0.00639 0.00959
Crocuta crocuta (no frontal struts)
Bite force (N) 204.59 257.48 197.35 241.85
Median scaled stress (MPa) 0.266 0.278 0.282 0.285
MAD stress (MPa) 0.215 0.224 0.225 0.228
Max scaled stress (MPa) 219.450 305.405 158.286 169.569
Max scaled strain ( μE) 0.01425 0.01985 0.01028 0.01100
Crocuta crocuta(original)
Bite force (N) 133.30/
156.02
243.86/
193.12
197.37/
160.01
243.24/
191.88
Median scaled stress (MPa) 0.291/
0.144
0.350/
0.164
0.310/
0.145
0.328/
0.157
MAD stress (MPa)
0.208/
0.131
0.248/
0.148
0.227/
0.131
0.236/
0.142
Max scaled stress (MPa) 140.405/
271.921
181.288/
286.143
158.014/
167.804
111.563/
179.613
Max scaled strain ( μE) 0.00912/
0.01875
0.01182/
0.10598
0.01029/
0.00750
0.00725/
0.00752
157

Table 3.3. Summary values for analysis 2: All models produced 1180.592 N of bite force at
LP3. Data for the original C. crocuta model are from the homogeneous/heterogeneous models,
respectively.

C. lunensis
C. crocuta (no
struts)
C. crocuta
(original)
Median scaled stress (MPa) 1.96 1.56 2.30/1.00
MAD stress (MPa) 1.60 1.26 1.63/0.90
Max scaled stress (MPa) 634.73 1288.07 1087.25/1781.65
Median scaled strain ( μE) 0.00005 0.00002 0.00011/0.00005
MAD strain ( μE) 0.00005 0.00002 0.00011/0.00005
Max scaled strain ( μE) 0.04127 0.08349 0.07086/0.12270

158
Table 3.4. Summary values for analysis 3: All models produced sufficient bite force at
LP3 to break bovine femur with strength of 150 MPa. Data for the original C. crocuta
model are from the homogeneous/heterogeneous models, respectively.

C. lunensis
C. crocuta
(no struts)
C. crocuta
(original)
Bite force (N) 7029.30 10074.45 10074.45
Med. scaled stress (MPa) 11.65 13.30 19.58/8.20
MAD stress (MPa) 9.51 10.80 13.91/7.40
Max scaled stress (MPa) 3774.60 10972.50 9249.72/15367.70
Med. scaled strain ( μE) 0.00076 0.00087 0.00127/0.00103
MAD strain ( μE) 0.00063 0.00071 0.0009/0.00074
Max scaled strain ( μE) 0.24517 0.71560 0.60042/1.06034

159
Table 3.5. Summary values for analysis 4: bite forces at LP3 used to calculate stress slope.
Data for the original C. crocuta model are from the homogeneous/heterogeneous models,
respectively.

Chasmaporthetes lunensis
Bite force (N) 400 600 800 1000
Median scaled stress (MPa) 0.663 0.965 1.326 1.657
MAD stress (MPa) 0.541 0.778 1.082 1.353
Max scaled stress (MPa) 214.824 283.805 429.649 537.668
Max scaled strain ( μE) 0.01396 0.01844 0.02792 0.03495
Crocuta crocuta (no frontal struts)
Bite force (N) 400 600 800 1000
Median scaled stress (MPa) 0.529 0.794 1.059 1.32
MAD stress (MPa) 0.428 0.642 0.856 1.07
Max scaled stress (MPa) 436.514 655.963 870.642 1091.28
Max scaled strain ( μE) 0.02839 0.04258 0.05677 0.07096
Crocuta crocuta (original)
Bite force (N) 400 600 800 1000
Median scaled stress (MPa) 0.778/
0.351
1.167/
0.515
1.555/
0.710
1.943/
0.880
MAD stress (MPa) 0.553/
0.317
0.828/
0.466
1.105/
0.640
1.380/
0.790
Max scaled stress (MPa) 367.825/
614.924
551.738/
902.755
735.650/
1218.11
919.563/
1513.88
Max scaled strain ( μE) 0.02391/
0.04244
0.03586/
0.06219
0.04787/
0.08397
0.06004/
0.10430

160
Table 3.6. Bone-cracking index calculated for the finite element models. For explanation
of calculations see methods. Abbreviations: homo., homogeneous model of the original
Crocuta skull; hetero., heterogeneous model of the original skull.

regression Fsl
(N*m)
% Fsl BCI % BCI
Chasmaporthetes y=0.00166x-0.00624 148.95 88 1137.02 106
Crocuta (no struts) y=0.00132x+0.00036 169.16 100 1077.45 100
Crocuta (homo.) y=0.00194x+0.00717 115.02 100 732.61 100
Crocuta (hetero.) y=0.00081x+0.04080 274.55 100 1748.73 100

161
Chapter Four: Study of cranial function in the hyena Ikelohyaena abronia

This chapter has been published as:

Tseng, Z. J., and D. Stynder. 2011. Mosaic functionality in a transitional ecomorphology:
skull biomechanics in stem Hyaeninae compared to modern South African carnivorans.
Biological Journal of the Linnean Society 102:540-559. doi: 10.1111/j.1095-
8312.2010.01602.x.

A copy of the accepted manuscript begins on the next page.
162
Mosaic functionality in a transitional ecomorphology: skull
biomechanics in stem Hyaeninae compared to modern South African
carnivorans

ZHIJIE JACK TSENG
1,2,*
and DEANO STYNDER
3

1
Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, Los Angeles, California 90089, U.S.A.
2
Department of Vertebrate Paleontology, Natural History Museum of Los Angeles
County, 900 Exposition Boulevard, Los Angeles, California 90007, U.S.A.
3
Department of Archaeology, Faculty of Science, University of Cape Town, Private Bag,
Rondebosch, 7701, South Africa

RH: TRANSITIONAL BONE-CRACKING ECOMORPHOLOGY

*Corresponding author: Zhijie Jack Tseng, Department of Vertebrate Paleontology,
Natural History Museum of Los Angeles County, 900 Exposition Boulevard, Los
Angeles, California 90007, U.S.A., jtseng@nhm.org, +01-780-952-8293.
163
Chapter Four Abstract

Ecomorphologies are categories of ecological adaptation and function, but intermediates
are not always available to shed light on functionality at the transitional stages between
them. We examined an intermediate bone-cracking carnivoran ecomorphology, the stem
hyaenine Ikelohyaena abronia, using finite element analysis. Skull models of
Ikelohyaena, crown hyaenine Crocuta crocuta, and two other hypercarnivores were
simulated with mastication and prey apprehension forces. Results show that Ikelohyaena
already possessed derived features in skull stress distribution and levels of strain energy,
characteristic of the extant bone-cracking Crocuta; however, estimated bite forces in
Ikelohyaena were significantly lower. Prey apprehension simulations showed similar
patterns; the low skull strain energy and low bite force of the Ikelohyaena mandible
indicate poor individual ability to take down large prey. The mosaic features of
craniodental function in Ikelohyaena suggest that initial evolution of the hyaenid bone-
cracking ecomorphology involved skull shape changes which increased stress dissipation,
permitting incorporation of more hard food into the diet. Subsequent evolution of larger
bite forces were then required to increase the size limit of bones that can be cracked and
consumed. This mode of evolution would have allowed transitional hyaenid
ecomorphologies to continuously increase carcass processing ability both during
competitive feeding as well as scavenging throughout their evolution.

164
ADDITIONAL KEYWORDS: bone cracking - Canidae - Felidae - finite element
analysis - Hyaenidae - Langebaanweg - mandibular symphysis - South Africa
165
Chapter Four Introduction

Ecomorphologies, categories of ecological niche adaptation based on
characteristic functional morphological features of different organisms, are highly
informative groupings with which to study trophic diversity and convergent ecological
evolution in vertebrates (Measey et al. 2009, Solounias and Semprebon 2002, Van
Valkenburgh 1985, 1988, 1995, Van Valkenburgh and Molnar 2002, Wainwright and
Richard 1995, Zweers et al. 1997). However, only in a few cases is the fossil record
sufficiently complete to allow tracking of ecomorphological evolution within more
restrictive clades (Holliday and Steppan 2004, Van Valkenburgh 1991, Van Valkenburgh
et al. 2004). Among the major hypercarnivorous ecomorphologies of the mammal order
Carnivora, wolf-like, sabre-toothed and cat-like, and bone-cracking hyaena-like forms
have evolved iteratively in multiple lineages (Werdelin 1996b). One ecomorphology,
bone-cracking hypercarnivores, is exemplified by the relatively complete and
exceptionally gradual evolutionary sequence of the Hyaenidae (Werdelin and Solounias
1991). In this study we explore the functional implications of gradational evolution of the
bone-cracking ecomorphology in the hyaenid clade.
The Hyaenidae is represented by four living species, one of the smallest
carnivoran families in number of extant members; this stands in great contrast to their
fossil record (70+ species) throughout Africa and the Holarctic region over the past 25
Myr (Werdelin and Solounias 1991). The living hyaenas, with the exception of the
myrmecophagous aardwolf (Proteles cristatus (Sparrman 1783), are the only modern
166
examples of the bone-cracking ecomorphology (Van Valkenburgh 1999), and all belong
in the crown group Hyaeninae (Koepfli et al., 2006). Among them the spotted hyaena
(Crocuta crocuta (Erxleben 1777) is the most studied living bone cracker, a highly social
predator that is capable of consuming entire skeletons of their hunted prey by using their
robust skull, strong musculature, and hypertrophied premolar teeth to crack bones (Kruuk
1972b)(Fig. 4.1A). Similar craniodental morphology has evolved convergently in at least
two other carnivoran lineages: the caniform borophagine canids (Wang et al. 1999) and
the feliform Percrocutidae (Qiu et al. 1988); both are extinct. The presence of distinct
morphological features in these carnivorans indicate that, on one hand, it is relatively
straightforward to identify morphological characters among fossil forms that may have a
functional linkage to bone-cracking capability (Van Valkenburgh 2007). On the other
hand, however, the evolutionary pathways by which convergent bone-cracking
ecomorphologies appeared, including the question of whether functional capability for
bone-cracking could be a synapomorphy on a given clade level within each lineage, is
unknown.
In this study we used a biomechanical modelling approach to examine the skull of
one of the most ancestral members of the Hyaeninae, and tested to what extent the
capability for bone-cracking and its corresponding morphology is symplesiomorphic to
all hyaenines. Taxonomically, bone-cracking crown Hyaenine includes the genera
Parahyaena, Hyaena, Pliocrocuta, Pachycrocuta, Adcrocuta, and Crocuta (Werdelin and
Solounias 1991). Several stem hyaenine genera (Metahyaena, Palinhyaena, Ikelohyaena,
Belbus, and Leecyaena) are known, and these are categorized as transitional bone
167
crackers by Werdelin and Solounias (1996b) and Turner et al. (2008) (Appendix C Fig.
S1). Hyaeninae is a monophyletic clade sister to the Chasmaporthetes lineage, which is
less robust in morphology and also more cursorial (Berta 1981, Kurtén and Werdelin
1988, Werdelin et al. 1994). Only skull fragments are known for the stem hyaenine
genera except for the genus Ikelohyaena, which contains a single species I. abronia
(Hendey 1974). I. abronia was first identified at the South African Mio-Pliocene fossil
site of Langebaanweg ‘E’ Quarry (Hendey 1974). It has subsequently been identified in
Lothagam (Kenya), Laetoli (Tanzania) and possibly Hadar (Ethiopia). Lothagam
represents its earliest occurrence in the fossil record, while Laetoli and Hadar represent
its most recent (Werdelin and Lewis 2005). The holotype from Langebaanweg ‘E’
Quarry (SAM-PQL 14186), which consists of a partial skull (undistorted) and partial
post-cranial skeleton (Hendey, 1974), is currently the most complete specimen of I.
abronia in existence. Therefore, we have chosen the type skull of I. abronia as
representing the most plesiomorphic morphology in Hyaeninae, to compare to the
modern Crocuta in masticatory and hunting capability.
Morphologically, I. abronia is smaller than the living hyaenines, and does not
have the degree of hypertrophied conical premolars or skull robusticity seen in the living
forms (Hendey 1974, Werdelin and Solounias 1991). Furthermore, Ikelohyaena retains
posterior molars (M2, M3, m2) and the first lower premolar, plesiomorphic conditions of
earlier hyaenids (Fig. 4.1B). Beyond these differences, the overall skull shape of
Ikelohyaena is similar to the modern hyaenines. These mixed features of Ikelohyaena are
examples of the highly gradational mode of evolution long recognised in the Hyaenidae
168
(Werdelin and Solounias 1991); hyaenid species during their ~25 Myr evolutionary
history are often difficult to delineate because of essentially continuous and overlapping
ranges of morphological variation between many sister forms (Werdelin 1988a, b). As
such, whereas the major types of ecomorphologies within the family are well established,
their exact delineation, especially those in the hyaenine clade which contains the living
bone crackers, is not entirely resolved because of intermediate features in stem species
(Turner et al. 2008, Werdelin and Solounias 1996b). Hence, the stem hyaenine group in
which Ikelohyaena belongs has been considered a transitional group, which is critical in
understanding bone-cracking adaptation in the crown group (Turner et al. 2008).
We utilized the finite element (FE) method in our study of the transitional form.
Finite element analysis is an engineering technique that approximates the mechanical
behaviour and failure of an object by discretisation of its shape and structure, and testing
of simulated forces and constraints; this method has been successfully applied to the
study of vertebrate biomechanics, and in particular the study of skulls (Dumont, Piccirillo
& Grosse, 2005; McHenry et al., 2006; Rayfield, 2007; Rayfield et al., 2001; Ross, 2005).
Specifically, FE studies on carnivoran skulls have demonstrated the utility of this
technique in discerning functional differences between ecomorphologies (McHenry et al.,
2007; Slater, Dumont & Van Valkenburgh, 2009) and similarities between convergent
forms (Tseng, 2009; Tseng & Binder, 2010; Wroe, 2008; Wroe et al., 2007).
In addition to the Crocuta and Ikelohyaena skulls, we also included in our
analysis the skulls of an extant African hunting dog (Lycaon pictus (Temminck 1820) and
a leopard (Panthera pardus (Linnaeus 1758) to place our study in a broader comparative
169
context. Both L. pictus and P. pardus are obligate hypercarnivores (Table 4.1), but with
different social structures and hunting strategies. Lycaon pictus is a highly social predator,
hunting large game in packs with slashing and shaking bites (Anyonge 1996, Ewer 1973,
Nowak 1999). The skull of L. pictus is robust for a canid, but not to the degree seen in
hyaenids; the mandible is relatively shallow, and a dentition containing a full set of
premolars and molars is present (Fig. 4.1C). Panthera pardus, on the other hand, is a
solitary hunter using a stereotypical forelimb hold and directed killing bite commonly
seen in solitary felids (Anyonge 1996, Ewer 1973). The skull and particularly the rostrum
of P. pardus are short; the dentition is reduced in tooth number but specialised in having
highly sectorial cheek teeth and strong, conical canines typical of felids (Fig. 4.1D).
Lycaon has been suggested to represent a modern analog of ictithere hyaenids, which are
a grade of dog-like species preceding hyaenines in their evolutionary history during the
later Miocene epoch (Werdelin and Solounias 1991)(Appendix C Fig. S1). The hunting
strategy of Crocuta is close to that of Lycaon, although the former does so in smaller
groups (Kruuk 1972b). In comparison, the skull of Crocuta is more robust than either L.
pictus or P. pardus, and the dorsal cranium has a frontal sinus extended caudally all the
way to the occiput, below the sagittal crest; the mandibular corpus of Crocuta is laterally
thick and dorsoventrally deepened, and the teeth are also bulbous and robust (Fig. 4.1).
Using the carnivorans described above, we tested the idea that the skull of the
bone-cracking Crocuta is capable of producing larger bite forces than the meat-eating
Lycaon or Panthera, because the ability to crack bones should be directly related to the
amount of force generated from a bite. Secondly, we tested the idea that large bite forces
170
required for bone-cracking are associated with higher mechanical advantage (i.e. output
force: input force ratio) of the bone-cracking teeth, as mechanical demands are severe for
such a task, and selection would favour more efficient use of a given amount of
masticatory musculature. Lastly, we compared the bite force and overall skull strength
with prey apprehension simulations for (1) Lycaon pictus, which uses slashing bites
during pursuit hunting, (2) Panthera pardus, which use holding bites during ambush
hunting, and (3) Crocuta, which uses a combination of the previous two types of bites
during hunting (Anyonge 1996, Van Valkenburgh and Ruff 1987). The skull of
Ikelohyaena is compared to the modern predators in all of the tests proposed above. In
addition, to evaluate the biological relevance of our models and analysis, we included a
sensitivity analysis of several mandibular symphysis models to test for effects of
modelling parameters on symphyseal stress and mandibular bite force. The mandibular
symphysis varies among carnivorans in both anatomy and composition, and is potentially
an important factor in stress conduction across the dentaries (Scapino, 1965, 1981). Our
expectations are that the stem hyaenine I. abronia, being less robust in craniodental
morphology and closer to the ancestral ictithere morphology, should experience
mastication and prey apprehension stresses similar to those observed in L. pictus.
Nevertheless, similarities in domed frontal shape between Ikelohyaena and Crocuta
should confer a certain degree of advantage to the former in lowering and dissipating
cranial stresses incurred from mastication when compared to Lycaon. Based on the
overall gradational pattern of morphological evolution in hyaenids, Ikelohyaena would
not possess the large bite forces and mechanical advantage which should characterize
171
Crocuta and other crown hyaenines; instead, we would expect Ikelohyaena and Lycaon to
produce bite forces of similar magnitudes, given the characterization of the former as a
transitional ecomorphology.
The institutional abbreviations used in the present study are: SAM, South African
Museum, Cape Town, South Africa.

Chapter Four Materials and Methods

All four skulls were digitized with computer tomography (CT), and subsequent
steps and analyses used the digital representations only. The methodology for
construction of FE models follows those in Dumont (2005), Grosse et al. (2007), and
Tseng (2009), and is outlined below.

Model parameters
The holotype skull of I. abronia and comparative modern skulls of C. crocuta
(SAM-ZM 19469), L. pictus (SAM-ZM 12245), and P. pardus (unaccessioned) were CT-
scanned at the Groote Schuur Hospital (Cape Town) with a Toshiba Aquillon medical CT
scanner (120 KV, 300 mA for P. pardus which contained soft tissues; 120 KV, 75 mA
for others with bones only), with image resolution 512 x 512 pixels, pixel size
0.289~0.468 mm, interslice distance 0.500 mm, and exported in the DICOM (Digital
Imaging and Communications in Medicine) medical imaging format. All modern skulls
were from wild-caught individuals collected in South Africa. Regions of the cranium and
172
mandible were digitized in Mimics 13.1 (Materialise N.V., Leuven, Belgium), and the
reconstructed surface mesh was cleaned and improved in Geomagic Studio 10 (Geomagic,
Inc., Research Triangle Park, North Carolina, USA). Holes and areas of low-quality mesh
in the reconstruction were repaired in Geomagic Studio and then remeshed in the Mimics
remesh module; several iterations of this process were performed to create continuous
and smooth surfaces suitable for solid-meshing. Solid mesh models were converted from
surface reconstructions in Strand7 2.3.7 FE software (G+D Computer Pty Ltd, Sydney,
Australia). Four-noded tetrahedral elements were used; all skull models contained at least
1x10
6
elements to approximate convergence in model behaviour influenced by element
type and quantity (Appendix C Table S1; Dumont et al., 2005). In the fossil Ikelohyaena
model, missing regions of the cranium immediately caudal to the orbits were
reconstructed with curve-fitting functions in Geomagic Studio and filled in manually with
reference from the other three models. The morphology of this part of the cranium is
relatively conservative, and thus straightforward reconstructions were possible. In
addition, only the left dentary was preserved in Ikelohyaena; the model was mirrored and
the right dentary created from a copy of the left side.
All mandibles were positioned at an angle rostroventral to the cranium to
represent a 30º gape, which is an average of the optimal gape angle range of 25º~35º in
an FE study of the canid carnivoran Canis lupus dingo Linnaeus, 1758 (Bourke et al.,
2008). Even though optimal gape angles may be larger for felid bites because of their
killing strategy, we chose to keep this particular factor constant among the four models in
order to limit the number of variables introduced into the analysis. The cranium and
173
dentaries were tested as an isotropic material with Young’s modulus of 20 GPa and
Poisson’s ratio of 0.3, average values for mammalian cortical bone (Dumont et al. 2005,
Erickson et al. 2002). Single isotropic-property FE models approximate biomechanical
behaviour in actual bone which is heterogeneous and anisotropic, and even though the
stress magnitudes may not be exactly equivalent to experimental values, the methodology
is suitable given the nature of our study, which compares relative stress and bite force
values. We chose not to use Hounsfield Units from the CT data to represent relative
densities of the modern and fossil skulls; experimental evidence demonstrate differential
uptake of trace elements between trabecular and cortical bone can occur prior to
fossilization, and potentially has substantial but still largely unquantified influence on
altering relative mechanical properties of fossil bones (Olesiak et al., 2010; Trueman et
al., 2004). The cranium and mandible were kept as separate objects with their own
constraints (see below), and positioned in their relative anatomical positions during
analysis. The mandibular symphysis was modelled as a thin plate between the dentaries,
congruent with the boundaries of the dentary bones and in itself divided into two
approximately equally-sized regions representing dorsally the fibrocartilage pad and
ventrally the cruciate ligaments, respectively (Scapino 1965, 1981). The former was
assigned a modulus of 1.5 MPa and Poisson’s ratio of 0.4 (Hu et al., 2001), and the latter
a modulus of 220 MPa and Poisson’s ratio of 0.4 (Butler et al., 1992); for the Panthera
model, both regions were modelled as cruciate ligaments based on the anatomically
distinct symphyseal morphology in felids (Scapino 1981). Given previous observations of
morphological variation in the symphysis of carnivorans (Scapino 1981), a series of tests
174
were run to examine the sensitivity of different material properties on symphysis
mechanical behaviour, and to assess potential impact on current models. The details of
this sensitivity analysis are explained below in the “Mandibular Symphysis” section. All
analyses were linear static.

Mastication forces
Unilateral biting using the third (P3) versus fourth premolar (P4) was tested in all
models (Fig. 4.1). The third premolar is the main tooth used in hyaena bone-cracking, and
the fourth premolar is used in bone-cracking as well as cutting meat (Van Valkenburgh
1996). To prevent free-body motion, crania were constrained from all but rotational
movement in the dorsoventral plane by two nodes, one at each of the articular fossa of the
temporomandibular joints (TMJ). Mandibles were constrained to the same movement,
with two nodes at each of the mandibular condyles at the medial and lateral ends,
respectively. The P3, p3, p4 crown tips were fully constrained from movement in the P3
bone-cracking simulation. The P4, m1 paraconid, m1 protoconid tips were fully
constrained in the P4 cutting simulation (Fig. 4.1). Forces on temporalis, masseter, and
medial pterygoid muscles were applied in ratios proportional to their masses reported
from literature or from our own estimates (Appendix C Table S2). Muscle attachment
areas were similarly derived (Appendix C Fig. S2). Data for C. crocuta were from a
dissection made by one of us (ZJT) at the Field Station for the Study of Behaviour,
Ecology and Reproduction of the University of California, Berkeley (on deceased captive
individual “Frog”). Data for P. pardus were estimated using the soft-tissue CT scan of the
175
specimen in our dataset (Appendix C Table S2, Fig. S2). Both left and right side
unilateral biting were simulated, and balancing side muscles were adjusted to 60% of the
force applied on the working side, based on data derived from muscle activation patterns
in domestic dogs (Dessem 1989).
To calculate the muscle forces required to simulate maximum measured premolar
bite force in C. crocuta, 1000N of total input force was arbitrarily chosen to calculate an
output bite force at P3 in a trial analysis. Because the analyses in this study are linear
static, necessary muscle forces were then calculated in proportion for a bite force of
~1180N at the upper P3, a value derived from the regression of bite force and ontogenetic
age for a full adult (80-month old) C. crocuta as measured by Binder and Van
Valkenburgh (2000):

Bite force = 165.952 + 12.683 x age

This muscle force was then taken as a point of comparison for all other models,
including I. abronia. To compare the upper tooth bite force magnitudes of I. abronia to C.
crocuta and the other carnivoran species, muscle forces were calculated by scaling the
forces to a constant input force: input area ratio for all models; skulls were left at their
original sizes, but muscle forces were scaled to the respective area of muscle attachment
(Appendix C Table S1). Scaling in this way allowed the comparison of bite force as a
factor of both skull size and shape, instead of scaling to model surface area and strictly
examining influence of shape on performance (Dumont et al., 2009). As the analyses
176
were aimed to evaluate the bite force and skull strength of Ikelohyaena given its
intermediate skull morphology and smaller absolute size relative to Crocuta, scaling by
muscle force input area (with the assumption of equal muscle force produced per unit
area among different species) preserved both size and shape. As a proxy for amount and
location of stress imposed on the skulls, von Mises stress distributions were compared
(Dumont et al., 2005, 2009). Bite force output at the tooth tips and mechanical advantage
of the bites (output force/input force) were also compared; in addition, skull strain energy
scaled to the respective model volume was used to compare the ability of the skulls in
conducting forces from muscle to bite point, with lower strain energy values representing
more efficient use of a given input force (Dumont et al. 2009). Whereas strain energy is a
measure of how efficient a structure transmits forces, von Mises stress is a measure of
strength that approximates behaviour of material failure in bone (Dumont et al. 2009). In
some cases, however, failure in bone is better characterized using strain (Nalla et al.,
2003; Currey, 2004); for the purposes of this study, the von Mises stress reported from
the homogeneous, linear models provide the same information as von Mises strain in
comparisons of biomechanical performance.

Prey apprehension forces
Bilateral muscle forces were simulated along with “pull back” or “lateral shake”
extrinsic forces (McHenry et al. 2007, Slater et al. 2009, Wroe et al. 2007), to
approximate forces placed on the skull during prey apprehension. Total muscle input
forces in each model remained identical to the analysis of mastication forces, but instead
177
of an adjusted unilateral bite, muscle forces were divided equally between the left and
right muscles to simulate bilateral bites. Muscle ratios among the temporalis, masseter,
and pterygoid remained the same. Cranial constraints were similarly placed at the TMJ to
allow rotation in the dorsoventral plane. Two nodes were fully constrained, one at each of
the occipital condyles. For the “pull back” simulation, a total of 300N of rostrally-
directed force was divided equally between two nodes, one at each of the posterior faces
of the canines in the Crocuta model. This force is close to what Wroe et al. (2007)
calculated for struggling prey of Canis lupus dingo. For the “lateral shake” simulation,
the same forces were placed instead on the left surfaces of the canines, and directed to the
left. To compare the strength of the skull models under proportional extrinsic and
intrinsic forces, the extrinsic forces were scaled in other models so that their magnitudes
are proportional to total muscle input forces in each model. As in the mastication
analyses, von Mises stress distribution, bite force, and skull strain energy values were
compared.
The mandible was more difficult to model in these prey apprehension scenarios
because they do not articulate with bones other than the cranium, which could be
constrained at the occipital condyles. Bilateral muscle forces and the respective extrinsic
forces were modelled as in the cranium. Three scenarios constraining free-body
movement in different locations were tested along with extrinsic forces in order to
compare the effect of different constraints on stress and strain energy: (1) No TMJ
rotation: the mandibular condyles were fixed from all movement by one node at each
medial and lateral end of the condyles, respectively, (2) TMJ rotation with canine
178
movement along extrinsic force axis: one node at each canine tip was fixed from all but
movement in the axis of the respective extrinsic force, and the mandible was allowed
rotational movement at the temporomandibular joint in the dorsoventral plane, (3) TMJ
rotation with no movement along extrinsic force axis: the mandible, with dorsoventral
rotational freedom as in the previous scenario, was fully constrained by nodes at the
canine tips. These scenarios were created to explore the possible equilibrium states in
which no free-body movement is present in the mandible, as the exact effects of the
forces acting on the mandible during prey apprehension may be complex and are
balanced by muscles and ligaments. The linear static method used in this study limits the
ability of the models to portray this complexity, thus these three scenarios, which
individually might not be biomechanically realistic, were created for comparison
purposes. Von Mises stress distribution and total strain energy were recorded, in addition
to canine bite forces for the second and third scenarios.
In addition to evaluating strain energy and stress distributions, simulations of both
mastication and prey apprehension forces were also analysed by sampling seven
anatomical landmarks along the dorsal cranium (Fig. 4.1). The landmarks are along the
mid-sagittal axis, and are located (1) at the rostral-most contact between the nasal bones;
(2) between the infraorbital foramina; (3) between the rostral-most point of the orbits; (4)
between the post-orbital processes; (5) between the post-orbital constriction of the
parietal bones; (6) half-way along the length of the braincase, and (7) at the caudal end of
the sagittal crest. These landmarks were chosen to cover the entire mid-sagittal axis in
order to represent the stress distributions numerically. In each landmark region ten nodes
179
were selected and average maximum and minimum principal stresses were calculated,
respectively. Maximum principal stress represented tensile forces, and minimum
principal stress represented the compressive forces acting on the skull. Landmark stress
distributions were then compared across different models and bite simulations.

Mandibular symphysis
Scapino (1965) described in detail the anatomy of the mandibular symphysis in
the domestic dog Canis lupus familiaris, and subsequently expanded the examination to
samples across Carnivora (Scapino 1981). What emerged from his studies was that the
patent symphysis in most carnivorans, in fact, represents a third joint in the skull, and
potentially played an important role in mediating the transfer of muscle forces across the
dentaries, as well as the alignment of the carnassial teeth. The symphysis of canids and
hyaenids were of Type I, with low interdigitation of bony protuberances and a prominent
fibrocartilage pad in the dorsal region of the symphyseal space; those of the large cats
(such as Panthera) were of Type III, with high interdigitation of bony protuberances and
little remaining of a plate-like fibrocartilage region. Whereas Scapino (1981) suggested
that a more interdigitated symphysis behaved essentially like a fused one with efficient
lateral transfer of muscle forces, Dessem (1985) showed that the patent Type I symphyses
nevertheless was also very efficient in force transfer during bilateral canine bites with
unilateral muscle activation, with balancing side canine producing 82~109% of the
working side canine. We set up the Crocuta specimen as in the experiment of Dessem
180
(1985) to test the mechanical behaviour of the mandibular symphysis built with different
materials:
The mandibular condyles of the dentaries were constrained from all but rotational
movement in the dorsoventral plane by two nodes located on the medial and lateral ends
of each condyle, respectively. Full nodal constraints (i.e. no movement) were placed at
the tips of both canine teeth. An unilateral muscle contraction force was applied to the
left dentary (the “stimulated side”) by pulling the regions of the temporalis and masseter
muscle attachments towards the corresponding centroid regions in the cranium using the
distributed force algorithm in the BoneLoad program of Grosse et al (2007). An
arbitrarily chosen 1000N of total force was distributed to the temporalis (700N) and
masseter (300N) regions, representing relative proportions of those muscles by mass
typical of Canis, Felis, and Panthera (Davis 1955, Turnbull 1970). The resulting bite
forces were measured at the canine constraint, and the percentage of muscle forces from
the working side transferred to the balancing side was calculated.
The test outlined above was applied to the same mandible, but with the symphysis
assigned a variety of material properties ranging from cartilage to cortical bone (Table
S3). The Poisson’s ratios of the different materials were assumed to be close to 0.4, so
only Young’s moduli were altered in the sensitivity tests. In addition to bite force, the
total strain energy, maximum Von Mises stress, and maximum lateral displacement of the
symphysis were also analysed. Displacement was examined as a proxy of the lateral
deformation experienced by the symphysis as it is pulled apart ventrally by shearing and
tensile forces (Scapino 1981).
181
Chapter Four Results

Mastication forces
Crocuta crocuta produced the highest upper premolar bite forces in both
masticatory scenarios, followed by I. abronia, L. pictus, and P. pardus, respectively.
Skull strain energy is highest in P3 biting in Panthera and Lycaon, followed by Crocuta
and Ikelohyaena. In P4 biting Crocuta has the highest strain energy, followed by
Panthera, Lycaon, and Ikelohyaena. In both biting scenarios of all models, strain energy
in the mandible is at least 65% of the total amount, with the cranium having relatively
less strain energy. Shifting from P4 biting (meat-shearing/bone-cracking) to a P3 bone-
cracking bite, Crocuta showed an overall drop in strain energy from 0.77/0.78 J (left/right)
to 0.67/0.62 J, whereas both Lycaon and Panthera showed increases in strain energy.
Ikelohyaena has a small drop in strain energy from 0.56/0.57 J to 0.54/0.53 J. The
mandible of Crocuta exhibited proportionally more strain energy compared to the
cranium when switching from a P4 to a P3 bite; the same is true for Lycaon, but no
difference is apparent in Ikelohyaena or Panthera. The magnitude of strain energy in the
cranium, on the other hand, decreased from P4 to P3 biting in both Crocuta and
Ikelohyaena, whereas increases are observed in Lycaon and Panthera (Table 4.2).
Crocuta has the highest mechanical advantage (MA) among the four models
(0.19/0.20 for P3 and 0.23/0.23 for P4); Ikelohyaena and Lycaon have similar ratios for
both P3 (0.13/0.13) and P4 (0.15/0.15). Panthera has the lowest MA for both mastication
simulations (0.09/0.10 for P3, 0.12/0.13% for P4). The lower cusps showed the same
182
pattern for Crocuta. The pattern is much less consistent for lower cusps of Ikelohyaena
and Lycaon, however; Ikelohyaena has slightly higher MA for P3 and P4 mastication at
the anterior lower cusp compared to Lycaon, but both have overlapping MA values for
the posterior lower cusps. Panthera has comparable MA at the lower anterior cusp to
Lycaon and Ikelohyaena in P3 mastication, and slightly higher values compared to the
same species in P4 mastication (Table 4.2).
Distribution of von Mises stress is more similar between Crocuta and Ikelohyaena
than for any other pair-wise comparisons. In all models the highest concentration of
elevated von Mises stress is in the rostral face of the working side mandible’s ascending
ramus (Fig. 4.2). In Lycaon and Panthera the ventral region of the mandibular fossa also
showed elevated stress, and to a lesser extent this is true for Ikelohyaena. In all models
the overall stress distribution is similar between P3 and P4 biting; the body of the
mandible showed widespread but relatively even stress between the ascending ramus and
the biting teeth. In the cranium the maxillary and frontal parts of the face show increased
stress, along with the working side zygomatic arch (Fig. 4.2). In dorsal view the working
side maxillary, frontal, and parietal regions are all more highly stressed in Crocuta and
Ikelohyaena than the other two species. In Lycaon widespread but lower stress is seen
across the dorsal cranium, but elevated stress is restricted to the maxillary part of the face
immediately rostral of the orbit (Fig. 4.2G). In Panthera the dorsal cranium is more or
less bilaterally stressed, with slightly higher stresses in the working side frontal bone and
orbit (Fig. 4.2H). The rostrum region is more stressed in Lycaon and Panthera than in
Crocuta or Ikelohyaena.
183

Prey apprehension forces
“Pull back” and “lateral shake” forces were scaled to muscle input area and then
applied to the respective models; for the “pull back” simulation cranial strain energy is
highest in Crocuta (0.54 J), followed by Ikelohyaena, Lycaon, and then Panthera (Table
4.3). In the “lateral shake” simulation cranial strain energy is highest in Lycaon (0.55 J),
then in Crocuta and Ikelohyaena which have similar values, and lowest in Panthera. For
the mandible, the first scenario (no TMJ rotation) produced very high strain energy more
than three orders of magnitude higher than those observed for the cranium as well other
mandible simulations; Ikelohyaena has the highest value for both “pull back” and “lateral
shake” simulations (4,173.14 J and 2,774.74 J, respectively), followed by Crocuta.
Mandibular strain energy in this simulation is lowest in Lycaon (Table 4.3). The second
scenario (TMJ rotation with canine movement) produced strain energy values comparable
between Lycaon and Panthera for both “pull back” and “lateral shake”; those in Crocuta
are slightly lower, then followed by Ikelohyaena which has the lowest values in both
simulations (0.46 J and 0.43 J, respectively). Mandibular strain energy values for these
scenarios are universally higher than those for the cranium within each species. Canine
bite forces are roughly equal between the two sides in the “pull back” simulation, but
shifts asymmetrically to higher forces in the right canine in “lateral shake” for all models.
In the third scenario (“TMJ rotation, no canine movement”), mandibular strain energy
values are equivalent among Crocuta, Lycaon, and Panthera for both “pull back” and
“lateral shake”; these values are slightly lower than those obtained in the second scenario.
184
Ikelohyaena has the lowest strain energy values for both simulations (0.38 J for “pull
back” and 0.39 J for “lateral shake”). Canine bite forces in the third scenario do not show
the large shift observed in the second scenario between “pull back” and “lateral shake”
simulations, and the bite forces are of similar magnitudes within each model (Table 4.3).
Cranial stress distributions in the “pull back” simulation are more caudally
restricted in Crocuta and Ikelohyaena, whereas in Lycaon and Panthera they are more
widespread across the rostrum (Fig. 4.3). Von Mises stress is symmetrically distributed in
all models, but in Lycaon and Panthera the nasal bones have elevated stress whereas they
are relatively unstressed in the hyaenids. In the “lateral shake” simulation, cranial stresses
are asymmetrically distributed, with the frontal region on the right side more stressed
than the left (shaking forces were oriented to the left); conversely, the left nasal bones are
more stressed than the right one. This is true for all models, even though stress
magnitudes in those two regions appear much lower in the hyaenids compared to Lycaon
and Panthera (Fig. 4.3). Stresses at the temporalis attachment sites appear comparably
distributed in all models but Ikelohyaena, which has elevated stress levels that connect in
distribution with temporomandibular joint stress (Fig. 4.3B, F).
Mandibular stress distributions are similar between “pull back” and “lateral
shake” simulations (Appendix C Fig. S3). Scenario 3 produced the lowest strain energy
values, and scenario 1 produced the highest. In all four models the ascending ramus and
angular process are two main regions of concentrated stress in the second and third
scenarios, although the area stressed is highest in Panthera (Appendix C Fig. S3D).
Lycaon, Ikelohyaena, and Crocuta show progressively less angular process stress,
185
respectively. Even distribution of stress throughout the dentary bone ventral to the
toothrow is observed for all the models, although stress magnitudes are highest in
Panthera and lowest in Ikelohyaena. In the first scenario, Lycaon has the least-stressed
dentary, and Panthera the highest. Regions of peak stress are at the post-canine diastema
and post-m1 ramus in all models.

Landmark stress analysis
In both P3 and P4 mastication simulations, dorsal cranial compressive stresses are
higher than those of tensile stresses (Fig. 4.4). For the prey apprehension simulations the
tensile and compressive stress ranges are more equivalent. In both hyaenids, peak tensile
stress during mastication occurs near the rostral base of the frontal (landmark 3) and peak
compressive stress near the fronto-parietal boundary (Fig. 4.4A, B). In Lycaon and
Panthera the peak tensile stresses are of lower magnitude than those in hyaenids, but they
occur more caudally at the top of the frontal between the postorbital processes (landmark
4). Peak compressive stresses occur in two regions in Lycaon and Panthera for the P3
mastication simulation; one peak at the nasal bones and the other in the parietal (Fig.
4.4A). The same pattern is present for P4 mastication in Lycaon, but in Panthera peak
compressive stress is restricted to the parietal only.
In the prey apprehension simulations, compressive stresses dominated the rostral
region of the cranium (Fig. 4.4C, D). Tensile stresses for both “pull back” and “lateral
shake” have peaks at the caudal end of the dorsal cranium for Crocuta and Ikelohyaena,
but in Lycaon and Panthera the peaks are in the frontal region. For the hyaenids the
186
compressive stresses also peaked near the caudal end of the cranium, but for Lycaon and
Panthera again there are two peaks present, in the same regions as in the mastication
simulations (Fig. 4.4).

Mandibular symphysis
Stimulated side canine bite forces increased with increasing stiffness (Young’s
modulus) in the material used to represent the mandibular symphysis (Appendix C Table
S3). However, non-stimulated side canine bite forces are higher for cartilage and cortical
bone material than for the others intermediate in stiffness; this resulted in lower
transmission percentages in tendons, collagen, and ligament materials (Fig. 4.5A).
Conversely, symphyseal strain energy is higher in tendon, collagen, and ligament than in
cartilage or cortical bone (Fig. 4.5B). Maximum von Mises stresses are low in all
materials compared to cortical bone, and the value decreased with decreasing stiffness.
Cartilage and cortical bone experience larger lateral displacement than tendon, collagen,
or ligament. In all, the combination of fibrocartilage and cruciate ligament in the
symphysis, as modelled in the mastication and prey apprehension analyses, provided
material behaviour more similar to tendon and collagen than to cartilage or cortical bone
as a representation of the symphysis (Fig. 4.5).

187
Chapter Four Discussion

Craniodental function in Ikelohyaena
Crocuta crocuta produced the largest bite forces and had the highest mechanical
advantage out of the four models analysed, supporting our initial expectation (Table 4.2).
This was true for both the P3 bone-cracking simulation and a more generic P4
cutting/cracking simulation. Mechanical advantage at upper P3 and P4 in Crocuta were
0.20 to 0.23, 7 to 8% higher than in identical bite simulations in Ikelohyaena and Lycaon.
These results also indicate that Ikelohyaena is closer to Lycaon in its relative bite force
production capability than to Crocuta; this is supported by both similar mechanical
advantages and bite force magnitudes (Table 4.2). The skull of Ikelohyaena thus probably
had a more limited capability in bone-cracking relative to crown hyaenines, at least in
terms of maximum bone size that can be cracked open. On the other hand, skull strain
energy is lower in Ikelohyaena than in all other models; decrease in strain energy during
switching from a P4 to a P3 bite is observed. In both Lycaon and Panthera strain energy
increased when switching from a P4 to a P3 bite, and together with decreased bite force
they indicate a decrease in performance. In this regard Ikelohyaena resembles Crocuta in
that the cranium is adapted to a strong P3-driven bite over the P4 carnassial bite,
supporting a specialised P3 function compared to those of Lycaon and Panthera. Even so,
the decrease in strain energy in P3 biting is much more obvious in Crocuta than it is in
Ikelohyaena. Thus the strain energy values of the skull of Ikelohyaena appear to be
intermediate between the bone-cracking Crocuta and the other species examined, and
188
these results agree with intermediate skull morphology in Ikelohyaena (Hendey 1974)
and with our initial expectations.
Providing further support for this interpretation, von Mises stress in the cranium
of Ikelohyaena also resembles that of Crocuta in both magnitude and distribution (Fig.
4.2A-B, E-F). The centre of high stress seems to shift from the working side maxilla in
Lycaon and Panthera to the working side frontal region in Ikelohyaena and Crocuta,
indicating a transmission of forces dorsocaudally onto the dorsal cranium (Werdelin
1989). The parietal bones in Ikelohyaena are nevertheless more stressed than in Crocuta;
a smaller sagittal crest and the smaller, less shell-like posterior cranium of Ikelohyaena
probably do not dissipate stress quite as well as those of the more robust Crocuta
(Joeckel, 1998; Tanner et al., 2008). Overall these stress patterns indicate a relatively
weak but nonetheless already adaptive cranial shape of Ikelohyaena in bone-cracking bite
simulations. Considered in tandem with the bite force results, it appears that even though
skull shape in Ikelohyaena corresponds with similar stress dissipation patterns to Crocuta,
lower bite force and mechanical advantage were the potential limiting factor in the bone-
cracking capability of the stem hyaenine.
The mandible of Ikelohyaena had lower strain energy than the other three models,
but unlike Crocuta and Lycaon in which the mandibles had higher strain energy values in
P3 biting (with p3 and p4) than with P4 biting (with m1) when compared to the cranium,
there was no difference in mandibular strain energy between the two simulations in
Ikelohyaena (Table 4.2). This indicates that the mandible of Ikelohyaena is more
generalized than Crocuta or Lycaon, as the mandible does not appear to be differentially
189
adapted for bite positions as observed in the cranium. This result may reflect a more
generalized diet, or it may suggest a mandible of intermediate performance in
Ikelohyaena. The latter interpretation has some support from the observation that the
increases in mandible strain energy proportion occur by different causes in Crocuta and
Lycaon; in Crocuta the cranium exhibits a larger drop in strain energy switching from P4
to P3 biting, but the mandible shows a smaller drop. In Lycaon cranial strain energy
increases slightly when switching from P4 to P3 biting, but the mandible strain energy
showed a more significant increase (Table 4.2). The more P3-bite adapted cranium of
Crocuta and the less P3-bite adapted mandible of Lycaon bracket the values in
Ikelohyaena, which is intermediate in these aspects (however, see discussion on model
limitations below for other factors).
These intermediate aspects of the Ikelohyaena skull in mastication simulations are
also observed in the prey apprehension simulations which offer a context for its potential
hunting behaviour. First, biomechanical differences among the modern skulls modelled
can be correlated to differences in killing strategy. For an ambush predator, the cranium
of Panthera is stronger in both “pull back” and “lateral shake” simulations than the other
carnivorans, but has the lowest bite forces in those scenarios (Table 4.3). The shorter face
in Panthera (and felids in general) may have facilitated more efficient conduction of
forces than the canids or hyaenids (Buckland-Wright 1978). Level mechanics studies
show that the jaws of felids exhibit clear mechanical advantage over those of longer-
faced carnivorans (Biknevicius and Ruff 1992, Van Valkenburgh and Ruff 1987), and
although the resulting bite forces are smaller (as expected given the smaller absolute size
190
of the P. pardus skull compared to other species in this study), the directed killing bites
are sufficient in suffocating or paralyzing prey (Biknevicius and Van Valkenburgh 1996).
The skull of the pursuit predator Lycaon had higher strain energy during “lateral shake”
simulations, and lower strain energy in “pull back” simulations. The skulls of
Ikelohyaena and Crocuta are both higher in strain energy in these prey apprehension
simulations than either Lycaon or Panthera, supporting the interpretation that trade-offs
are present in their craniodental function, in which robust cat-like canines are used in a
canid manner of hunting (Van Valkenburgh and Ruff 1987). Furthermore, functional
requirements of bone-cracking may have placed overarching selective pressure on
mandibular evolution in a derived hyaenine such as Crocuta, as the more slender
mandible of Ikelohyaena maintained lower strain energy in prey apprehension
simulations compared to Crocuta (Table 4.3). At the same time, this suggests that
durophagy was not yet a significant part of the dietary repertoire in Ikelohyaena. The less
robust (and more trenchant) dentition of Ikelohyaena permits the stem hyaenine to be
more proficient in slashing bites similar to those of canids in magnitude (Table 4.3).
Reliance on hunting was more suitable for Ikelohyaena as its craniodental morphology
has evolved only intermediate capability for bone-cracking, and as discussed below, its
post-cranial skeleton is more cursorial than those of Crocuta (Hendey 1974).
Examination of stress distribution in the cranium of Ikelohyaena in prey
apprehension simulations showed similarly weak parietal regions as in the mastication
simulations (Fig. 4.3). It is possible that even though the dentition of Ikelohyaena does
not have the robusticity of crown hyaenines, its skull biomechanics during bone-cracking
191
might not be constrained by robustness as much as the ability of the posterior cranium to
dissipate larger muscle forces associated with such a task. The more shell-like dorsal
cranium that is associated with the caudally extended frontal sinus in Crocuta may serve
to dissipate muscle forces (Tanner et al., 2008), and the lesser development of the dorsal
cranium in Ikelohyaena could be associated with correspondingly weaker temporalis
musculature for its size, which was not considered in the analysis. This observation could
also be influenced by ontogeny, as the Ikelohyaena specimen has a barely developed
sagittal crest and unworn dentition (Hendey 1974). Overall, the skull stress patterns are
more similar between the hyaenids than with Lycaon or Panthera; the main difference
between Ikelohyaena and Crocuta is the smaller bite forces and lower mechanical
advantage in the former, as observed in the mastication simulations.
Stress distributions are similar between Ikelohyaena and Crocuta in prey
apprehension simulation of the mandible (Appendix C Fig. S3). Although mandibles in
both hyaenids performed poorly in the first scenario, their stresses are in general
widespread in the mandibular corpus for the other two scenarios. However, the very low
bite force output in those simulations for Ikelohyaena when compared to Crocuta as well
as the other models suggests that Ikelohyaena would not have been very effective in
producing bite forces large enough to inflict wounds on large ungulate prey to the degree
possible in Crocuta (Table 4.3). Lycaon produces similar levels of bite force in prey
apprehension simulation as Ikelohyaena, but the cranium of the former has lower stress in
“pull back” biting (Table 4.3). Thus, even though larger in body size (Table 4.1),
Ikelohyaena would have performed comparably in resisting prey apprehension forces and
192
producing killing bite forces to Lycaon. The larger Crocuta hunt in small groups of 2-5
individuals (Kruuk 1972b) whereas Lycaon often hunt in larger groups around 3-20
adults (Creel and Creel 1995); both hunt prey of comparable sizes but the stronger bite in
the former is a potentially factor in allowing them to do so with fewer individuals. We
interpret the functional attributes in Ikelohyaena as estimated by the analyses performed
here to be more suitable for relatively weaker slashing bites as in Lycaon. Although there
is little evidence to show whether pack hunting existed in Ikelohyaena, its similar
functional morphology to Lycaon and the relatively abundant remains of the more
ancestral ictitheres at many late Miocene fossil localities indicate that social groupings
may have been present as a plesiomorphy.
Interestingly, the apparently mosaic function of the Ikelohyaena skull as
suggested by the results of the FE analyses is corroborated by the evolution of dental
enamel microstructure in hyaenids as a whole (Rensberger and Stefen 2006, Stefen and
Rensberger 2002). A derived sharp-angle (<90º) zig-zag pattern of the Hunter-Schreger
Band (HSB) enamel microstructure is present in crown hyaenines, but its earliest
occurrence (or some intermediate morphology thereof) may well be among the
transitional stem species (Ferretti 2007b). Examination of enamel microstructure in
ictithere hyaenids indicate the presence of derived zig-zag HSB, suggesting enamel
adaptation may have preceded skull shape adaptation; furthermore, the spread of derived
HSB from the carnassial throughout the dentition probably occurred at the ictithere stage
of evolution as well (ZJ Tseng, unpublished data). The interpretation that extreme
carnivoran ecomorphologies may have gone through mosaic stages with incremental
193
modification in function has also been made for sabre-toothed carnivoran skull shape
evolution (Slater & Van Valkenburgh, 2008). Thus, a mosaic mode of craniodental
function in Ikelohyaena constitutes one example of such a pathway which may have
occurred multiple times in Carnivora, terminating in specialist species that may have
decreased evolutionary potential (Van Valkenburgh et al., 2004).
An additional aspect that agrees with a mosaic function interpretation for
Ikelohyaena is that post-cranial features in Ikelohyaena point to a more canid-like stance
and cursorial ability, rather than siding with Crocuta. The forelimbs of Ikelohyaena are
more slender than crown hyaenines, and the hindlimbs are proportionally longer (Hendey
1974). This suggests that Ikelohyaena did not have the galloping gait characteristic in
extant hyaenines (Spoor and Belterman 1986), but instead had a more typical
appendicular skeleton like its ictithere ancestors as well as canids. Nonetheless,
Ikelohyaena retains metacarpal I and most likely a rudimentary digit I on the forefeet,
which may imply a less reduced phalangeal formula, less reduced feet, and therefore a
less cursorial locomotive mode. Partial post-cranial similarities shared by Ikelohyaena
and dog-like ictithere hyaenids raise the fascinating question of the degree to which
cranial and post-cranial functional changes were modular and disjoint in the transitional
bone-cracking ecomorphology and more broadly over the course of evolution in the
family. Evidence from studies of cranial morphological integration suggests diet is a
mediating factor at more restrictive clade levels for carnivorans (Goswami 2006b), and
this may potentially influence function of postcranial traits as well. More detailed study
of ictithere and hyaenine post-cranial skeletons are needed, and one of the most
194
promising locality areas for additional research is the highly fossiliferous Neogene
deposits in the Gansu Province of China, which have recently produced numerous
complete fossil hyaenid skeletons (ZJT personal observation).

Function of the frontal dome
The inflated and pneumatised frontal dome of the Crocuta skull has long been
proposed as a structure that aids in dissipation of masticatory stress (Werdelin, 1989).
The dome created by a dorsally and caudally expanded frontal sinus (Fig. 4.1) has been
shown in previous FE analyses to be closely associated with stress distribution (Tanner et
al. 2008), and the stress-dissipating utility of this feature is also seen in percrocutids
(Tseng 2009) and borophagine canids (Tseng and Wang In press-a). The landmark stress
analysis performed here further suggests that the skull shape of the stem hyaenine
Ikelohyaena already exhibits this frontal dome function (Fig. 4.4). During mastication, a
single tensile and a single compressive peak exist along the dorsal cranium of the
hyaenids, not doubled in compressive force peaks as seen in Lycaon and Panthera. In the
hyaenids, the tensile force peak occurs at the rostral end of the frontal, and the
compressive peak at the fronto-parietal region; the location where these two types of
forces meet overlaps the frontal dome (Werdelin 1989). The intersection between large
tensile and compressive forces can create bone displacement and deformation where
stress is pooled; therefore by having a domed dorsal profile the crania of Ikelohyaena and
Crocuta divert these deforming stresses to the frontal region and away from facial bones
that contact sensory organs. Double compressive force peaks are present in Lycaon
195
mastication simulations, creating two regions where tensile and compressive forces meet:
nasal-frontal and fronto-parietal regions. The result is a highly concentrated region of
stress on the biting side face, instead of a smooth band of stress seen in hyaenids (Fig.
4.2). Again, it appears that Ikelohyaena had a skull capable of pooling such undesirable
stresses into the forehead, even though it probably did not have sufficient bite force to
process very large bones. Subsequent increase in bite force in stem and crown hyaenines
would have improved the capability to process increasingly large bones; the evolutionary
mechanism for such adaptation is not entirely clear, but may have involved a combination
of increase in absolute body size, as well as thickening of skull bones to further increase
mechanical strength.
One complication in the explanation of a plesiomorphic frontal dome function in
hyaenines is that the location of the tensile-compressive stress intersection in hyaenids
appears to fall within the frontal dome only during mastication; in prey apprehension
simulations the tensile and compressive peaks occur jointly at the parietal bones (Fig.
4.4C-D). In Lycaon and Panthera the double-peak pattern observed in mastication
simulations is essentially duplicated in the prey apprehension simulations, suggesting that
a differential response between the two behaviours is present in hyaenids but not Lycaon
or Panthera. The parietal stress peaks of hyaenids during prey apprehension bites do not
seem to be dissipated, and is actually quite elevated in Ikelohyaena at the temporalis
muscle attachment sites (Fig. 4.3). These patterns indicate that the cranium-deforming
principal stresses and their intersecting location in the frontal dome make hyaenid skulls
stronger in mastication but not prey apprehension; the higher strain energy of the hyaenid
196
crania in prey apprehension simulations also attest to their relatively low capability for
handling prey hunting stresses (Table 4.3). Previous studies suggest that bone-cracking
adaptations in the dentition and mandibular corpus may have been quite strongly selected,
perhaps even at the expense of efficiency in other tasks (Biknevicius 1996, Biknevicius
and Leigh 1997, Biknevicius and Ruff 1992, Van Valkenburgh and Ruff 1987). Thus, it
is possible that strong selection already in place at the transitional stage of the bone-
cracking ecomorphology involved a trade-off between masticatory and hunting ability;
this would seem counterintuitive except in a situation where abundant carcasses allowed
for more opportunistic feeding. Additional evidence from palaeoecological and
taphonomic studies of Ikelohyaena fossil localities may shed light on this issue.

Mandibular symphysis
Sensitivity tests conducted on the symphyseal material used in the models
highlighted some factors and trade-offs of using them to evaluate relative biomechanical
performance in carnivoran skulls. Building a model symphysis out of cortical bone
provides relatively high efficiency in bite force transmission and low symphyseal strain
energy (Fig. 4.5). However, the large displacement and maximum symphyseal stress are
probably maladaptive especially in bone-cracking mastication where large dorsoventral
muscle and bite forces generate shear at the mandibular symphysis (Scapino 1981).
Mammals that have a fused symphysis tend not to be highly carnivorous in diet (e.g.
primates); benefits of such a symphysis could instead lie in the resistance of lateral forces
during chewing and/or grinding (Scapino 1981). In any case, none of the carnivorans
197
examined here have fused symphysis, and the heavily interdigitated symphysis in
Panthera may be linked with buttressing the jaw and efficiently delivering a killing bite
(Biknevicius and Ruff 1992, Biknevicius and Van Valkenburgh 1996).
Cartilage appears to have the same advantages and shortcomings as cortical bone,
but without the high maximum stress (Fig. 4.5C). According to the results, cartilage
might appear to be a highly suitable material for the mandibular symphysis, but as
Scapino (1981) demonstrated, the cruciate ligaments are more suitable in resisting shear
and tension from unpredictable orientations, given their arrangement. The tendon,
collagens, and ligaments, although not very efficient in transmitting force, are better than
cartilage and bone at resisting lateral displacement. This aspect of maintaining jaw
stability may be principal for carnivorans, as proper occlusion and alignment require
small and controlled displacement of the symphysis (Scapino 1981). In all four models
examined, the modelled symphyses do not appear to represent major sources of stress or
strain energy; the stresses are more or less conducted through the symphysis and end up
in the dentary bones (Fig. 5A-D). Stress distributions in our mandible analyses appear to
be superficially similar to those done for Smilodon, thylacine, and dingo models
(McHenry et al. 2007, Wroe et al. 2007), which did not specifically take differential
symphysis anatomies into account.

Model Limitations
Functional interpretations made using results from FE models are inherently
dependent on the parameters chosen to represent biologically and biomechanically
198
relevant phenomena (Rayfield, 2007). However, representations of forces beyond simple
stereotypical motion become increasingly complex to analyse and interpret. In the context
of this study, mastication and prey apprehension forces were simulated in the cranium in
similar manners to analyses done previously on other carnivoran species (Slater et al.,
2009; McHenry et al., 2007). The mandible, however, has received less attention in
comparable FE studies. Specifically, the simulation of multiple bite points on the lower
dentition during a single bite, and the testing of multiple scenarios of mandibular
constraints during prey apprehension simulation both appear to be novel among similar
FE studies, but there are several potential factors that could influence the accuracy of
their representation of mandibular biomechanics in this study.
First, the respective bite forces of the anterior and posterior lower cusps in all
mastication simulations show a 19% to 272% difference, with the posterior lower cusp
generating larger forces than the anterior (Table 4.2). This range exceeds those expected
based on lever mechanics calculations as well as bite forces calculated separately for each
of the lower bite points (for example, P3 mastication in all models returned a 11% to 21%
difference in bite force for the lower cusps when analysed separately; data not shown). At
least two factors could be responsible for the variation seen in the results presented in
Table 4.2: first, the mandible is not able to produce as much bite force when multiple bite
points are used simultaneously, and two, the variations stem from artefacts of modelling,
which did not portray the equilibrium state required by the static analysis accurately.
Given the analyses performed here, these two factors cannot be teased apart; in vivo
experiments and further models that test the validity of simulated forces during
199
mastication are needed to fully explore these and other potential factors. Nonetheless, the
conclusions made from the results remain valid regardless of whether the bite points are
modelled separately or simultaneously.
Secondly, the three scenarios tested for prey apprehension simulation in the
mandible are not necessarily biologically relevant by themselves. The first “no TMJ
rotation” scenario created very large strain energy values that are at least three times
higher than results from the second and third scenarios (Table 4.3). The dramatically
different (and disproportionally large) values in the first scenario indicate that it is the
most biologically improbable of the three scenarios proposed. The latter two scenarios
(allowing TMJ rotation but having different degrees of freedom for canine bite points)
produced results that led to similar conclusions, and only differed in strain energy
magnitude. Thus, it appears that allowing TMJ rotation in mandibular prey apprehension
simulations is a more reasonable portrayal than fixing the TMJ (Table 4.3). The
remaining difference between the second and third scenarios, the degrees of freedom
given to the canine bite points, did not appear to change the resulting bite forces
dramatically; thus some confidence can be given to the conclusions made from the results
which are relatively insensitive to variation in the canine constraints imposed.

Chapter Four Conclusion

The transitional ecomorphology I. abronia was analysed along with modern
South African carnivorans in order to examine the functionality of a stem species in a
200
gradual evolutionary series. Finite element model simulations of mastication and prey
apprehension forces indicate that even at the evolutionary stage of stem Hyaeninae, even
skull stress distribution and low strain energy characteristic of specialized extant bone-
crackers were already present. However, masticatory and killing bite forces remained low
and canid-like in Ikelohyaena. Furthermore, the combination of a strong mandible that
produced low bite forces suggested Ikelohyaena had poor ability to handle large prey
individually; group hunting is thus a plausible mechanism for doing so. Hunting of
smaller prey is entirely within the capability of Ikelohyaena, and carcasses could be
utilized efficiently by bone-cracking with their stress-resisting skulls. The development
of a domed frontal region in hyaenines is associated with that region being the destination
of peak tensile and compressive forces incurred from mastication, effectively diverting
deforming forces away from sensory organs. Our analyses showed no comparable
adaptation of the skull for prey apprehension in either Ikelohyaena or Crocuta. The
Ikelohyaena cranium, relatively weak in prey apprehension, provides evidence that
functional demands for more complete carcass processing may have taken over the
selection of skull morphology for increased hunting ability at the time of the earliest
hyaenines. Widely recognised morphological features of hyaenid bone-cracking
ecomorphologies thus appear to be functionally linked to bone-cracking mastication
without obvious advantages in resisting prey apprehension forces. The emerging picture
paints a pattern of gradual functional evolution in the hyaenine skull in a mosaic fashion,
with stress-resistance evolving before significant increases in bite force. The overall
gradational morphological evolution of the family Hyaenidae is likely to be an underlying
201
factor preserving mosaic features in stem hyaenine craniodental function. Continued
studies in other well-documented lineages with bone-cracking ecomorphologies, such as
borophagine canids, will shed light on whether and how different tempo and mode of
evolutionary morphological transitions might influence skull mechanics in terminal
ecomorphologies.

Chapter Four Acknowledgments

ZJT thanks X. Wang (LACM), J. McNitt-Gray (USC) and J. Liu (University of
Alberta) for valuable discussion in the early stages of the experimental setup. A. Shabel
(UC Berkeley) provided facilities and collaborative effort in several carnivoran
dissections done in 2008. DDS thanks Denise Hamerton (Iziko Museums of Cape Town)
for access to the modern comparative material and Nazlea Peters (Groote Schuur Hospital)
for CT scanning assistance. The editor and anonymous reviewers provided many
comments that greatly improved the paper. Support for software purchases and research
provided by American Society of Mammalogists, NSF GRFP and DDIG (DEB-0909807),
and an USC Zumberge Grant (ZJT). DDS would also like to acknowledge the financial
support of the National Research Foundation African Origins Platform (AOP/West Coast
Fossil Park) and the Palaeontological Scientific Trust (PAST).
202
Chapter Four References
Anyonge W. 1993. Body mass in large extant and extinct carnivores. Journal of Zoology
231: 339-350.

Anyonge W. 1996. Microwear on canines and killing behavior in large carnivores: saber
function in Smilodon fatalis. Journal of Mammalogy 77: 1059-1067.

Berta A. 1981. The Plio-Pleistocene hyaena Chasmaporthetes ossifragus from Florida.
Journal of Vertebrate Paleontology 1: 341-356.

Biknevicius AR. 1996. Functional discrimination in the masticatory apparatus of juvenile
and adult cougars (Puma concolor) and spotted hyenas (Crocuta crocuta).
Canadian Journal of Zoology 74: 1934-1942.

Biknevicius AR, and Leigh SR. 1997. Patterns of growth of the mandibular corpus in
spotted hyenas (Crocuta crocuta) and cougars (Puma concolor). Zoological
Journal of Linnean Society 120: 139-161.

Biknevicius AR, and Ruff CB. 1992. The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology 228:
479-507.

Biknevicius AR, and Van Valkenburgh B. 1996. Design for killing: Craniodental
adaptations of predators Carnivore behavior, ecology, and evolution, Vol. 2. 393-
428.

Binder WJ, and Van Valkenburgh B. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252: 273-283.

Bourke J, Wroe S, Moreno K, McHenry C, and Clausen PD. 2008. Effects of gape and
tooth position on bite force and skull stress in the dingo (Canis lupus dingo) using
a 3-dimensional finite element approach. PLoS ONE 3: e2200.

Buckland-Wright JC. 1978. Bone structure and the patterns of force transmission in the
cat skull (Felis catus). Journal of Morphology 155: 35-62.

Butler DL, Guan Y, Kay MD, Cummings JF, Feder SM, and Levy MS. 1992. Location-
dependent variations in the material properties of the anterior cruciate ligament.
Journal of Biomechanics 25: 511-518.

Christiansen P. 1999. Scaling of the limb long bones to body mass in terrestrial mammals.
Journal of Morphology 239: 167-190.
203

Creel S, and Creel NM. 1995. Communal hunting and pack size in African wild dogs,
Lycaon pictus. Animal Behaviour 50: 1325-1339.

Currey JD. 2004. Tensile yield in compact bone is determined by strain, post-yield
behaviour by mineral content. Journal of Biomechanics 37: 549-556.

Davis D. 1955. Masticatory apparatus in the spectacled bear Tremarctos ornatus.
Fieldiana: Zoology 37: 25-46.

Dessem D. 1985. The transmission of muscle force across the unfused symphysis in
mammalian carnivores. In: Duncker HR and Fleischer G, eds. Vertebrate
Morphology. New York: Gustav Fischer Verlag. 289-291.

Dessem D. 1989. Interactions between jaw-muscle recruitment and jaw-joint forces in
Canis familiaris. Journal of Anatomy 164: 101-121.

Dumont ER, Grosse I, and Slater GJ. 2009. Requirements for comparing the performance
of finite element models of biological structures. Journal of Theoretical Biology
256: 96-103.

Dumont ER, Piccirillo J, and Grosse IR. 2005. Finite-element analysis of biting behavior
and bone stress in the facial skeletons of bats. The Anatomical Record Part A
283A: 319-330.

Egi N. 2001. Body mass estimates in extinct mammals form limb bone dimensions: the
case of North American hyaenodontids. Palaeontology 44: 497-528.

Erickson GM, Catanese III J, and Keaveny TM. 2002. Evolution of the biomechanical
material properties of the femur. The Anatomical Record 268: 115-124.

Erxleben JCP. 1777. Systema regni animalis, Classis I, Mammalia, Lipsiae.

Ewer RF. 1973. The carnivores. Cornell University Press, Ithaca.

Ferretti MP. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience 100: 41-52.

Goswami A. 2006. Morphological integration in the carnivoran skull. Evolution 60: 169-
183.

Grosse I, Dumont ER, Coletta C, and Tolleson A. 2007. Techniques for modeling
muscle-induced forces in finite element models of skeletal structures. The
Anatomical Record 290: 1069-1088.
204

Hendey QB. 1974. The late Cenozoic Carnivora of the South Western Cape province.
Annals of the South African Museum 63: 1-369.

Holliday JA, and Steppan SJ. 2004. Evolution of hypercarnivory: the effect of
specialization on morphological and taxonomic diversity. Paleobiology 30: 108-
128.

Hu K, Radhakrishnan P, Patel RV, and Mao JJ. 2001. Regional structural and viscoelastic
properties of fibrocartilage upon dynamic nanoindentation of the articular condyle.
Journal of Structural Biology 136: 46-52.

Joeckel RM. 1998. Unique frontal sinuses in fossil and living Hyaenidae (Mammalian,
Carnivora): description and interpretation. Journal of Vertebrate Paleontology 18:
627-639.

Koepfli K-P, Jenks SM, Eizirik E, Zahirpour T, Van Valkenburgh B, and Wayne RK.
2006. Molecular systematics of the Hyaenidae: Relationships of a relictual lineage
resolved by a molecular supermatrix. Molecular Phylogenetics and Evolution 38:
603-620.

Kruuk H. 1972. The spotted hyena: a study of predation and social behavior. The
University of Chicago Press, Chicago.

Kurtén B, and Werdelin L. 1988. A review of the genus Chasmaporthetes Hay, 1921
(Carnivora, Hyaenidae). Journal of Vertebrate Paleontology 8: 46-66.

Linnaeus C. 1758. Systema naturae per regna tria naturae, secundum classes, ordines,
genera, species, cum characteribus, differentiis, synonymis, locis. Vol. 1: Regnum
animale. Editio decima, 1758 (12th edition of Linaeus 1758). Societatis
Zoologicae Germanicae, Stockholm.

Lynch HA, Johannessen W, Wu JP, Jawa A, and Elliot DM. 2003. Effect of fiber
orientation and strain rate on the nonlinear uniaxial tensile material properties of
tendon. Journal of Biomedical Engineering 125: 726-731.

Mansour JM. 2008. Biomechanics of Cartilage. In: Oatis CA, ed. Kinesiology: The
Mechanicsand Pathomechanics of Human Movement. 2nd ed. Philadelphia, PA:
Lippincott Williams and Wilkins. 69-83.

McHenry C, Clausen PD, Daniel WJT, Meers MB, and Pendharkar A. 2006.
Biomechanics of the rostrum in crocodilians, a comparative analysis using finite
element modeling. The Anatomical Record Part A 288A: 827-849.

205
McHenry C, Wroe S, Clausen PD, Moreno K, and Cunningham E. 2007. Supermodeled
sabercat, predatory behavior in Smilodon fatalis revealed by high-resolution 3D
computer simulation. Proceedings of the National Academy of Sciences 104:
16010-16015.

Measey GJ, Hopkins K, and Tolley KA. 2009. Morphology, ornaments and performance
in two chameleon ecomorphs: is the casque bigger than the bite? Zoology (Jena)
112: 217-226.

Nalla RK, Kinney JH, and Ritchie RO. 2003. Mechanistic fracture criteria for the failure
of human cortical bone. Nature Materials 2: 164-167

Nowak RM. 1999. Walker's carnivores of the world. The Johns Hopkins University Press,
Baltimore.

Olesiak SE, Sponheimer M, Eberle JJ, Oyen ML, and Ferguson VL. 2010.
Nanomechanical properties of modern and fossil bone. Palaeogeography
Palaeoclimatology Palaeoecology 289: 25-32.

Qiu Z-x, Xie J-y, and Yan D-f. 1988. Discovery of the skull of Dinocrocuta gigantea.
Vertebrata PalAsiatica 26: 128-138.

Rayfield EJ. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35: 541-576.

Rayfield EJ, Norman DB, Jorner CC, Horner JR, Smith PM, Thomason JJ, and Upchurch
P. 2001. Cranial design and function in a large theropod dinosaur. Nature 409:
1033-1037.

Rensberger JM, and Stefen C. 2006. Functional differentiations of the microstructure in
the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A 278:
149-162.

Ross CF. 2005. Finite element analysis in vertebrate biomechanics. The Anatomical
Record Part A 283A: 253-258.

Scapino RP. 1965. The third joint of the canine jaw. Journal of Morphology 116: 23-50.

Scapino RP. 1981. Morphological investigation into functions of the jaw symphysis in
carnivorans. Journal of Morphology 167: 339-375.

Schumacher GH. 1961. Funktionelle Morphologie der Kaumuskulatur. VEB Gustav
Fischer Verlag, Jena, Germany.
206

Silver FH, Bradica G, and Tria A. 2002. Elastic energy storage in human articular
cartilage: estimation of the elastic modulus for type II collagen and changes
associated with osteoarthritis. Matrix Biology 21: 129-137.

Silver FH, Christiansen DL, Snowhill P, and Chen Y. 2000. Role of storage on changes
in the mechanical properties of tendon and self-assembled collagen fibers.
Connective Tissue Research 41: 155-164.

Slater GJ, Dumont ER, and Van Valkenburgh B. 2009. Implications of predatory
specialization for cranial form and function in canids. Journal of Zoology 278:
181-188.

Slater GJ, and Van Valkenburgh B. 2008. Long in the tooth: evolution of sabertooth cat
cranial shape. Paleobiology 34: 403-419.

Solounias N, and Semprebon G. 2002. Advances in the reconstruction of ungulate
ecomorphology with application to early fossil equids. American Museum
Novitates 3366: 1-49.

Sparrman AP. 1783. A voyage to the Cape of Good Hope, towards the Antarctic Polar
Circle, and round the world; but chiefly to the country of the Hottentots and
Caffres, from the year 1772 to 1776. Robinson, G. G. J. and Robinson, J., London.

Spoor CF, and Belterman TH. 1986. Locomotion in Hyaenidae. Bijdragen tol de
Dierkunde 56: 24-28.

Stefen C, and Rensberger JM. 2002. The specialized enamel structure of hyaenids
(Mammalia, Hyaenidae): description and development within the lineage -
including percrocutids. Zoologische Abhandlungen 52: 127-147.

Tanner JB, Dumont ER, Sakai ST, Lundrigan BL, and Holekamp KE. 2008. Of arcs and
vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta).
Biological Journal of the Linnean Society 95: 246-255.

Temminck. 1820. Annales generales des sciences physiques 3: 54.

Trueman CNG, Behrensmeyer AK, Tuross N, and Weiner S. 2004. Mineralogical and
compositional changes in bones expose on soil surfaces in Amboseli National
Park, Kenya: diagenetic mechanisms and the role of sediment pore fluids. Journal
of Archaeological Science 31: 721-739.

207
Tseng ZJ. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96: 51-67.

Tseng ZJ, and Binder WJ. 2010. Mandibular biomechanics of Crocuta crocuta, Canis
lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia).
Zoological Journal of Linnean Society 158: 683-696.

Tseng ZJ, and Wang X. In press. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology.

Turnbull WD. 1970. Mammalian masticatory apparatus. Fieldiana: Geology 18: 149-356.

Turner A, Antón M, and Werdelin L. 2008. Taxonomy and evolutionary patterns in the
fossil Hyaenidae of Europe. Geobios 41: 677-687.

Van Valkenburgh B. 1985. Locomotor diversity within past and present guilds of large
predatory mammals. Paleobiology 11: 406-428.

Van Valkenburgh B. 1988. Trophic diversity in past and present guilds of large predatory
mammals. Paleobiology 14: 155-173.

Van Valkenburgh B. 1990. Skeletal and dental predictors of body mass in carnivores. In:
Damuth J and MacFadden BJ, eds. Body Size in Mammalian Paleobiology:
Estimation and Biological Implications. Cambridge: Cambridge University Press.
181-205.

Van Valkenburgh B. 1991. Iterative evolution of hypercarnivory in canids (Mammalia:
Carnivora): evolutionary interactions among sympatric predators. Paleobiology
17: 340-362.

Van Valkenburgh B. 1995. Tracking ecology over geological time: evolution within
guilds of vertebrates. Trends in Ecology and Evolution 10: 71-76.

Van Valkenburgh B. 1996. Feeding behavior in free-ranging, large African carnivores.
Journal of Mammalogy 77: 240-254.

Van Valkenburgh B. 1999. Major patterns in the history of carnivorous mammals.
Annual Review of Earth and Planetary Science 27: 463-493.

Van Valkenburgh B. 2007. Déjà vu: the evolution of feeding morphologies in the
Carnivora. Integrative and Comparative Biology 47: 147-163.

208
Van Valkenburgh B, and Molnar RE. 2002. Dinosaurian and mammalian predators
compared. Paleobiology 28: 527-543.

Van Valkenburgh B, and Ruff CB. 1987. Canine tooth strength and killing behaviour in
large carnivores. Journal of Zoology 212: 379-397.

Van Valkenburgh B, Wang X, and Damuth J. 2004. Cope's Rule, hypercarnivory, and
extinction in North American canids. Science 306: 101-104.

Wainwright PC, and Richard BA. 1995. Predicting patterns of prey use from morphology
of fishes. Environmental Biology of Fishes 44: 97-113.

Wang X, Tedford RH, and Taylor BE. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243: 1-391.

Werdelin L. 1988a. Studies of fossil hyaenas: the genera Thalassictis Gervais ex
Nordmann, Palhyaena Gervais, Hyaenictitherium Kretzoi, Lycyaena Hensel and
Palinhyaena Qiu, Huang & Guo. Zoological Journal of Linnean Society 92: 211-
265.

Werdelin L. 1988b. Studies of fossil hyaenids: the genera Ictitherium Roth & Wagner
and Sinictitherium Kretzoi and a new species of Ictitherium. Zoological Journal
of the Linnean Society 93: 93-105.

Werdelin L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15: 387-401.

Werdelin L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
In: Gittleman JL, ed. Carnivore behavior, ecology, and evolution. New York:
Cornell University Press. 582-624.

Werdelin L, and Lewis ME. 2005. Plio-Pleistocene Carnivora of eastern Africa: species
richness and turnover patterns. Zoological Journal of Linnean Society 144: 121-
144.

Werdelin L, and Solounias N. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30: 1-104.

Werdelin L, and Solounias N. 1996. The evolutionary history of hyenas in Europe and
western Asia during the Miocene. In: Bernor RL, Fahlbusch V and Rietschel S,
eds. Later Neogene European biotic evolution and stratigraphic correlation. New
York: Columbia University Press. 290-306.

209
Werdelin L, Turner A, and Solounias N. 1994. Studies of fossil hyaenids: the genera
Hyaenictis Gaudry and Chasmoporthetes Hay, with a reconstruction of the
Hyaenidae of Langebaanweg, South Africa. Zoological Journal of the Linnean
Society 111: 197-217.

Wroe S. 2008. Cranial mechanics compared in extinct marsupial and extant African lions
using a finite-element approach. Journal of Zoology 274: 332-339.

Wroe S, Clausen PD, McHenry C, Moreno K, and Cunningham E. 2007. Computer
simulation of feeding behaviour in the thylacine and dingo as a novel test for
convergence and niche overlap. Proceedings of the Royal Society B: Biological
Sciences 274: 2819-2828.

Zweers GA, Vanden Berge JC, and Berkhoudt H. 1997. Evolutionary patterns of avian
trophic diversification. Zoology (Jena) 100: 25-57.

210
Figure 4.1. Specimens used for finite element modelling. Lateral view. (A) Crocuta
crocuta, (B) Ikelohyaena abronia, (C) Lycaon pictus, (D) Panthera pardus.
Abbreviations: bc, braincase; cf, condyloid fossa; fs, frontal sinus; mc, mandibular
condyle; pa, paracone of upper premolar 4; pad, paraconid of lower molar 1; prd,
protoconid of lower molar 1; I3, incisor 3; C, canine; P, premolar; M, molar. Circled
numbers represent landmarks used in principal stress analysis. All scale bars equal 20
mm.

211
Figure 4.2. Distribution of von Mises stress for a simulated unilateral bite at the left third
premolar in rostro-lateral (A-D) and dorsal (E-H) views. (A, E) Crocuta crocuta, (B, F)
Ikelohyaena abronia, (C, G) Lycaon pictus, (D, H) Panthera pardus. Hotter colours
indicate higher stress.

212
Figure 4.3. Distribution of von Mises stress for simulated prey apprehension forces “pull
back” (A-D) and “lateral shake” (E-H). (A, E) Crocuta crocuta, (B, F) Ikelohyaena
abronia, (C, G) Lycaon pictus, (D, H) Panthera pardus. Hotter colours indicate higher
stress.

213
Figure 4.4. Average maximum and minimum principal stresses (in megapascals, MPa)
along dorsal cranial landmarks in simulations of mastication (A, B) and prey
apprehension (C, D) forces. (A) Left P3 biting, (B) Left P4 biting, (C) “Pull back”, (D)
“lateral shake” simulations. Maximum principal (tensile) stresses have positive values;
minimum principal (compressive) stresses have negative values. For explanation of
landmarks see methods and Figure 4.1.

214
Figure 4.5. Results of sensitivity tests of symphyseal material properties. (A) Non-
stimulated side canine bite force as a percentage of stimulated side canine bite force, (B)
Symphyseal strain energy (in Joules), (C) Maximum symphyseal von Mises stress (in
megapascals), (D) Maximum symphyseal displacement (in mm). For material properties
of different tissue types see Appendix C Table S3.

215
Table 4.1. Carnivoran species examined in this study, their body masses, and principal
diet. Body mass estimates for Ikelohyaena were calculated using skeletal regression
models of different elements as indicated.

Species Body mass (kg) Diet Reference
Crocuta crocuta 45-80 Meat/bone Kruuk 1972; Nowak 1999
Ikelohyaena abronia ~37-44 Unknown Humerus and Femur; after Anyonge 1993
~35-63 Limb elements; after Christiansen 1999
~31-39 Cranium; after Van Valkenburgh 1990
~29-47 Humerus, Femur, Tibia; after Egi 2001
Lycaon pictus 18-36 Meat Ewer 1973; Nowak 1999
Panthera pardus 28-90 Meat Ewer 1973; Nowak 1999
216
Table 4.2. Results of finite element analysis of mastication forces. Up, bite force of upper
cusp in the respective scenario (in Newtons); L.A., anterior lower cusp bite force; L.P.,
posterior lower cusp bite force; MA, mechanical advantage; Adj, adjusted total strain
energy (in Joules); Cran/%, cranial strain energy and percentage of total strain energy;
Mand/%, mandibular strain energy and percentage of total. Percentage values after bite
forces are fractions of total muscle input force.

Up (N) MA
L.A.
(N) MA
L.P.
(N) MA
Adj
(J)
Cran
(J) %
Mand
(J) %
LP3 1180.73 0.19 630.14 0.10 1004.73 0.16 0.67 0.15 22 51.86 78
RP3 1217.39 0.20 620.64 0.10 1108.42 0.18 0.63 0.14 21 49.94 79
LP4 1398.91 0.23 526.19 0.09 1517.32 0.25 0.77 0.22 29 55.05 71
Crocuta
RP4 1413.40 0.23 777.31 0.13 1182.90 0.19 0.78 0.22 28 56.52 72
LP3 459.45 0.13 283.18 0.08 337.61 0.10 0.54 0.18 33 36.46 67
RP3 469.79 0.13 256.18 0.07 403.49 0.11 0.53 0.18 33 35.57 67
LP4 529.41 0.15 295.06 0.08 474.04 0.13 0.56 0.19 35 36.26 65
Ikelohyaena
RP4 543.52 0.15 331.54 0.09 425.75 0.12 0.57 0.19 33 38.21 67
LP3 400.42 0.13 136.22 0.04 506.77 0.16 0.80 0.23 28 57.30 72
RP3 416.52 0.13 218.43 0.07 365.51 0.11 0.75 0.19 25 56.76 75
LP4 483.74 0.15 208.23 0.07 454.54 0.14 0.64 0.19 30 45.13 70
Lycaon
RP4 469.84 0.15 216.85 0.07 529.50 0.17 0.65 0.20 31 44.79 69
LP3 218.68 0.09 206.37 0.09 257.79 0.11 0.77 0.24 32 52.56 68
RP3 232.45 0.10 183.05 0.08 310.24 0.13 0.79 0.24 31 54.57 69
LP4 282.02 0.12 222.65 0.10 409.81 0.18 0.72 0.23 32 48.44 68
Panthera
RP4 292.80 0.13 253.93 0.11 386.79 0.17 0.77 0.22 29 54.32 71

217
Table 4.3. Results of finite element analysis of prey apprehension forces. Extrinsic forces
were scaled to muscle input area in each model. En, cranial strain energy; J, Joules; 1,
scenario 1, “no TMJ rotation”; 2, scenario 2, “TMJ rotation with canine movement”; 3,
scenario 3, “TMJ rotation with fixed canines”; LC, left canine; RC, right canine; N,
newtons; MA, mechanical advantage.

218
Chapter Five: Study of cranial function in fossil dogs

This chapter has been published as:

Tseng, Z. J., and X. Wang. 2010. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia). Journal of
Morphology 271(11):1386-1398. doi: 10.1002/jmor.10881

A copy of the accepted manuscript begins on the next page.
219
Cranial Functional Morphology of Fossil Dogs and Adaptation for
Durophagy in Borophagus and Epicyon (Carnivora, Mammalia)

Zhijie Jack Tseng
1,2
and Xiaoming Wang
2

1
Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, 3616 Trousdale Parkway, Los Angeles, CA90089;
2
Department of Vertebrate Paleontology, Natural History Museum of Los Angeles
County, 900 Exposition Boulevard, Los Angeles, CA90007

Short Title: FOSSIL CANID FUNCTIONAL MORPHOLOGY

Corresponding author: Zhijie Jack Tseng, Department of Vertebrate Paleontology,
Natural History Museum of Los Angeles County, 900 Exposition Boulevard, Los
Angeles, CA 90007; jtseng@nhm.org; (780) 952-8293.
220
Chapter Five Abstract

Morphological specialization is a complex interplay of adaptation and constraint,
as similarly specialized features often evolve convergently in unrelated species,
indicating that there are universally adaptive aspects to these morphologies. The
evolutionary history of carnivores offers outstanding examples of convergent
specialization. Among larger predators, borophagine canids were highly abundant during
the Tertiary of North America, and are regarded as the ecological vicars of Afro-Eurasian
hyenas. Borophaginae is an extinct group of 60+ species, the largest forms evolving
robust skulls with prominently domed foreheads, short snouts, and hypertrophied fourth
premolars. These specializations have been speculated to enhance bone-cracking. To test
the extent that the skulls of derived borophagines were adapted for producing large bite
forces and withstanding the mechanical stresses associated with bone-cracking relative to
their non-robust sister clades, we manipulated muscle forces in models of six canid skulls
and analyzed their mechanical response using 3D finite element analysis. Performance
measures of bite force production efficiency and deformation minimization showed that
skulls of derived borophagines Borophagus secundus and Epicyon haydeni are
particularly strong in the frontal region; maximum stresses are lower and more evenly
distributed over the skull than in other canids. Frontal strength is potentially coupled
with a temporalis-driven bite to minimize cranial stress during biting in the two derived
genera, as tensile stress incurred by contracting temporalis muscles is dissipated rostro-
ventrally across the forehead and face. Comparison of estimated masticatory muscle
221
cross-section areas suggests that the temporalis-masseter ratio is not strongly associated
with morphological adaptations for bone-cracking in Borophagus and Epicyon; larger
body size may explain relatively larger temporalis muscles in the latter. Compared with
previous studies, the overall cranial mechanics of the derived borophagines is more
similar to bone-cracking hyaenids and percrocutids than to their canid relatives,
indicating convergence in both morphological form and functional capability.

Keywords: Borophaginae, Canidae, bite force, bone-cracking, finite element analysis,
functional morphology
222
Chapter Five Introduction

The mammal order Carnivora composes plesiomorphically omnivorous species
which became specialized for a diversity of diets and ecological niches throughout their
evolution during the Cenozoic (65 million years ago to present); carnivorans are found on
nearly every continent, and the crown group shows the largest body size range among
living mammal orders (Nowak 1999). Within Carnivora, the dog family Canidae contains
species amongst the most well known and charismatic of mammals, as well as the most
widely distributed living carnivores (Larivière and Pasitschniak-Arts 1996, Wang et al.
2008). Rivaling the dramatic range of morphological variety seen in domestic breeds,
dogs have an evolutionary history going back 40 million years, with complex patterns
and processes of morphological change and evolution that is of intense interest to
evolutionary biologists and morphologists (Boyko et al. 2009, Finarelli 2007, Finarelli
and Flynn 2009, Tedford et al. 2009, Wang 1994, Wang et al. 2008, Wang et al. 1999).
Among the most interesting phenomena observed in the evolution of canids is the
diversity of their ecological habits as inferred from their morphology (Wang et al. 2008);
canids appear to have repeatedly evolved meat specialist species (hypercarnivores), at the
same time retaining a relatively ancestral dentition (Holliday and Steppan 2004, Van
Valkenburgh 1991, Van Valkenburgh et al. 2004). This study focuses on one such
specialist type, the bone-cracking hypercarnivore, and examines their robust cranial
mechanics relative to their less robust relatives.

223
Borophaginae and Hyaenidae
Of the three subfamilies of Canidae (Hesperocyoninae, Borophaginae, Caninae),
the borophagines digressed the most from a typically dog-like appearance towards hyena-
like ecomorphologies in their terminal clades (Martin 1989, Wang et al. 1999, Werdelin
1996b). Borophaginae as a canid subfamily was originally envisioned by Simpson (1945)
to include mostly large, bone-crushing forms. The namesake (type genus) of the
subfamily, Borophagus, was often referred to as “hyenoid dogs” by early vertebrate
paleontologists in recognition of its extraordinarily strong crushing premolars. The
comprehensive review of borophagine systematics by Wang et al. (1999) established this
monophyletic subfamily of the Canidae to be much more diverse both taxonomically and
morphologically than had been previously known (Fig. 5.1). Borophaginae began with
small, fox-like Archaeocyon during the Orellan North American Land Mammal Age
(NALMA), and terminated with the large and robust Borophagus diversidens in the late
Blancan NALMA (Wang et al. 1999). The Borophaginae is not only notable for its
numerous late Tertiary forms that exhibit cranial morphological convergence with
hyaenids, but also for the largest canid known, Epicyon (the largest of which were about
the size of the living Brown Bear Ursus arctos; Wang et al., 2008; Wang et al., 1999).
Such a convergence has been speculated to represent a striking case of ecological
parallelism by members of two distinct families of carnivorans that were isolated within
their own respective continents of origins (Wang et al. 2008).
Like the Eurasian medium and large hyaenids, large borophagine canids were
among the most commonly preserved carnivoran fossils throughout the temporal and
224
geographic extent of their occurrence (Turner et al. 2008, Wang et al. 1999). Because of
their inferred abundance, some borophagine canids have been interpreted as social
hunters (Van Valkenburgh et al., 2003; but see Andersson, 2005). Particularly interesting
are the late Miocene to Pliocene hyaenids and canids; many were large bone-cracking
ecomorphs with robust cranial construction, strong premolars, and heavy tooth wear
(Werdelin 1989). Modern spotted hyenas Crocuta crocuta and their extinct relatives
Pachycrocuta, Pliocrocuta, and Adcrocuta were all capable bone-crackers with very
robust craniodental morphology (Kruuk 1972b, Turner and Antón 1996, Werdelin and
Solounias 1990). The suite of craniodental features observed in Crocuta are correlated
with large bite forces (Binder and Van Valkenburgh 2000, Wroe et al. 2005) and
incorporation of bone as a regular part of their diet (Kruuk 1972b). The closely related
fossil hyaenids are also thought to be capable of producing such high bite forces; Plio-
Pleistocene Pachycrocuta was even larger in size than the modern Crocuta (Turner and
Antón 1996). On the canid side, the robust borophagines were best represented by species
of the genera Borophagus and Epicyon. Previous studies have shown that Borophagus
appears to have specialized into the bone-cracking ecomorph by extensive convergence
with hyaenids (Werdelin 1989), and that the largest bone-cracking borophagine canids
are cranially aberrant from all other canids known before or since their appearance in the
fossil record (Tseng and Wang, unpublished data). To test the assertion that the crania of
large borophagine canids are capable of high force production and resistance of
mechanical stress relative to both their smaller and less derived sister forms and their
225
hypercarnivorous living relatives, we modeled six canid skulls with simulated bone-
cracking bites and observed their cranial response in terms of strength and deformation.

Finite element analysis
We built our models using computer tomography (CT) image data of fossil skulls
and examined them using finite element analysis (FEA). Finite element analysis is a
modeling technique commonly used to test the strength of engineered objects and to
improve design, but more recently it has been applied to functional scenarios in
organismal biology (Rayfield 2007, Ross 2005). Several FEA studies have discussed
functional implications of cranial morphology in various groups of carnivorans
(McHenry et al. 2007, Slater et al. 2009, Tseng 2009, Wroe 2007), but no one has
examined borophagine canids using the FEA technique. To take advantage of the
comparative nature of FEA application in biology (Dumont et al. 2009), we chose skulls
from all three subfamilies of Canidae in order to examine the evolution of cranial
function.

Chapter Five Materials and Methods

In order to have this deep time evolutionary perspective, we modeled a total of six
canid skulls representing all three subfamilies: Hesperocyon gregarius, Mesocyon
coryphaeus (Hesperocyoninae), Microtomarctus conferta, Epicyon haydeni, Borophagus
secundus (Borophaginae), and Canis lupus (Caninae). Our expectations are that
226
differences in absolute size notwithstanding, the cranial morphologies represented by the
robust Borophagus and Epicyon are stiffest in resisting deformation and most efficient in
bite force production of the canids tested, when loaded with muscle input forces properly
adjusted for size differenecs.
Abbreviations: F:AM, Frick Collection, American Museum of Natural History,
New York, USA; LACM, Mammalogy Collection, Natural History Museum of Los
Angeles County, California, USA; LACM-CIT, California Institute of Technology
vertebrate paleontology collection incorporated in the Natural History Museum of Los
Angeles County, California, USA; NPS-JODA, John Day National Monument collection,
National Park Service, Oregon, USA; PPM, Plains-Panhandle Museum, Texas, USA.
All six specimens were CT-scanned at the University of California, Los Angeles
Medical Center with a Siemens Definition 64 scanner (Siemens Medical Solutions) at 0.6
mm slice thickness and interslice distance. The resulting images had a resolution of
640x640 pixels. The generation of finite element models follows the procedure outlined
in Tseng (2009) which is a combination of methods described in Dumont et al. (2005)
and Grosse et al. (2007). Briefly, CT images were segmented and reconstructed in the
medical imaging software Mimics 13 (Materialise NV, Leuven, Belgium), and the
cranium was extracted as a continuous 3D representation. The representation was then
used to generate a surface reconstruction, which was subsequently cleaned and stitched in
the rapid prototyping software Geomagic Studio 10 (Geomagic, Inc., North Carolina,
USA) to create a “water-tight” model (Dumont et al. 2005). The reconstructions were
then converted into solid mesh with 4-noded tetrahedral elements, and boundary
227
conditions were applied to generate finite element models in the software STRAND7
(G+D Computing Pty Ltd, Sydney, Australia). As we were interested in comparing the
relative behavior of the cranial shapes only, we simplified the analysis by assigning a
single isotropic material property to the skulls (but see below). In addition, all analyses
were linear static. Accordingly, to maximize the equivalence of bone densities
represented in each of the fossil models and relative to the modern Canis lupus model, all
thin and cancellous bones were removed during model-building to more closely
approximate a single-material property skull. As the thin bones are often located in parts
of the cranium containing delicate soft tissue (e.g. turbinates, braincase, auditory bulla),
removal of those bones should not significantly affect the evaluation of stress patterns as
we are interested in the higher load-bearing elements (e.g. dentition, areas of muscle
attachment) of the skull only.

Material properties and boundary conditions
A Young’s modulus of 20 GPa and Poisson’s ratio of 0.3, average of typical
mammalian cortical bone (Erickson et al. 2002), were used. An additional Canis lupus
model was constructed using heterogeneous material properties based on bone density
values obtained through the CT-scan data in Hounsfield Units (McHenry et al. 2007).
This extra model gave us a reference point for comparing differences between the
homogeneous models and one built with more “realistic” variable bone properties
(supplementary online material, Appendix D SI Table 1). The heterogeneous model was
assigned nine different material properties. Eight of the properties were based on the
228
magnitude of Hounsfield units (HU) of the CT scan obtained in Mimics 13. The full
range of observed HU was divided into eight equal intervals. Based on the similarity of
HU intervals between the Canis CT image data and those of the lion Panthera leo
obtained by McHenry et al. (2007), we used the same density and modulus values as in
that model. One additional property was designated to the surface of both fourth
premolars in the Canis model to represent dental enamel (supplementary online material,
Appendix D, SI Table 1).
For the fossil models, several repair operations had to be done before the models
were suitable for analysis. The Epicyon specimen had a complete right side and
damaged/crushed left side, thus the skull model was mirrored so that the left side was a
symmetrical copy of the right. A more extensive operation had to be done for
Hesperocyon, which had an intact left rostrum and right cranium; each had to be mirrored
to create bilateral symmetry. Last, the zygomatic arches of the Hesperocyon,
Microtomarctus, and Borophagus specimens are incompletely preserved. These models
had to be repaired in Geomagic Studio 10 before they were analyzed. Zygomatic arches
in the final Hesperocyon model were generated from a copy of the bones in Mesocyon,
here taken as the most closely related species to the former of the available CT data. The
zygomatic arches for the other two incomplete models were generated from copies of the
Epicyon model. The copied bones were deformed to match the morphology observed in
their target species, and modifications were made with reference to more complete
specimens illustrated in Wang (1994) and Wang et al. (1999). Incomplete rostral nasal
bones were also repaired by mirroring the complete side within each specimen. Images of
229
these digital “operations” are included in supplementary online material, Appendix D SI
Figs. 1-5.
Partial constraints allowing for rotation of the skull around the
temporomandibular joints (TMJ) were applied by fixing two nodes, one at each of the
articular processes; the only allowed movement at these nodes is rotational movement
around the axis perpendicular to the sagittal plane. A single node was fully constrained at
either the left or right P4 paracone cusp apex to simulate unilateral mastication at the
carnassial tooth (both left and right P4 biting were analyzed). Areas of attachment of the
temporalis, masseter, and pterygoid muscles on the cranial surface were created based on
published literature (Schumacher 1961, Turnbull 1970) and dissections by one of us
(ZJT). A muscle activation ratio of 64% temporalis, 26% masseter, and 10% pterygoid
was initially used as this value has been applied in a modern canid FEA study and is
relatively consistent for modern canids (Slater et al. 2009, Turnbull 1970). Balancing side
muscles have been shown to contract sub-maximally even under bone-crushing bites with
high forces; thus a 60% reduction in balancing side muscle force relative to the working
side was applied to all three muscle groups (Dessem 1989). Taking these considerations
into account, the muscle forces were distributed so that the total muscle forces in the
three masticatory muscle groups had a force distribution percentage ratio of 64:26:10,
and within each muscle group the working to balancing muscles had a force distribution
ratio of 5:3, respectively.

230
Wolf-like muscle forces
To test the relative strength and efficiency of the skulls, analyses were run with
muscle forces scaled to total surface area of each model (Dumont et al. 2009). Muscle
forces applied in all models were proportional to those required to produce a P4 bite force
of 3407N in Canis, a maximum bite force derived from in vivo measurements (Ellis et al.
2008). Both left and right P4 biting were analyzed. Strength is evaluated by 1)
visualization of distributions of von Mises stress throughout the skull and 2) maximum
von Mises stress at seven anatomical sampling points (nasal, rostrum at infraorbital
foramina, rostrum at anterior border of orbits, frontal region between postorbital
processes, frontal region at the postorbital constriction, center of the parietal bone, and
posterior-most region of the dorsal cranium). Von Mises stress is a combined measure of
the principal stresses acting on finite elements, and is a good estimator of stress under
ductile failure which is the mode seen in cortical bones (Dumont et al. 2009, Nalla et al.
2003). Efficiency of skulls is evaluated by 1) the adjusted total strain energy of each skull
under identical loading scenarios; strain energy is equal to the work done by an input
force in elastically deforming a structure, and is a measure of how efficiently a structure
conducts force instead of storing it as elastic energy; strain energy can be compared
directly when properly adjusted to the surface area/volume ratios of each model (Dumont
et al. 2009), and 2) the ratio between P4 bite force and muscle input force (“bite force
production efficiency”), which can be directly compared across models of different
volumes given identical input force to surface area ratios (Dumont et al. 2009). A
summary of the models is provided in Table 5.1.
231
Temporalis- and masseter-driven biting forces
As most of the morphological specializations shared by bone-cracking ecomorphs
are cranial features in the frontal and parietal regions where the temporalis muscles attach,
we tested the mechanical response of the skull models to a temporalis-driven versus a
masseter-driven bite, in order to test whether these morphological modifications are
associated with adaptations in one particular mode of muscle use. We ran analyses with
identical muscle input forces as in the previous part, but with the force applied to either
the temporalis or the masseter group; the pterygoid muscles are a relatively small group
that acts mainly in jaw opening and stability (Turnbull 1970), and thus were not tested
separately. All other factors remained identical to the previous set of analyses. These
analyses represented extreme theoretical scenarios as all three muscle groups are active
during actual mastication, and play a role in maintaining joint stability (Dessem 1989).
Nevertheless, we feel that such tests are still informative in testing potential
musculoskeletal adaptations. Furthermore, given our focus on overall cranial morphology
and the numerous assumptions made about the models, we did not attempt to test for all
possible combinations of muscle ratios because of potential difficulties in interpreting
minute differences in model behavior which could be attributed to differences in
preservation, reconstruction, or other unknown variables unrelated to performance. All
data reported for P4 biting are averages of LP4 and RP4 biting analyses, which were run
separately to evaluate model asymmetry.

232
Chapter Five Results

Homogeneous vs. heterogeneous Canis models
The overall distributions of stress are similar between the homogeneous and
heterogeneous models in dorsal view (Fig. 5.2A,D). The heterogeneous model
experienced higher stress at both of the temporomandibular joints (TMJ), the bite point,
and the fronto-parietal region. In mid-sagittal view, the heterogeneous model has more
widespread areas of elevated stress, as well as more regionalized stress in the rostrum
(Fig. 5.2B). In ventral view, the stress spreading from the TMJ reach farther in the
heterogeneous model than the homogeneous one, and there is more even stress in the
palate between the carnassial teeth (i.e. the bite point) in the heterogeneous model (Fig.
5.2C). The heterogeneous model produced only slightly higher bite force than in the
homogeneous model, but the former had higher force production efficiency (Table 5.2).
Total strain energy is also higher in the heterogeneous model.

Wolf-like muscle forces
The non-robust canids had much higher and broader regions of stress compared to
the derived borophagines Epicyon and Borophagus. Canis and Hesperocyon had the
highest stress distribution across the entire dorsal region of the skull (Fig. 5.3). The other
canids had similarly stressed snouts, but the fronto-parietal regions show much lower
stress in Epicyon and Borophagus than in Mesocyon and Microtomarctus. Epicyon and
Borophagus also show the most even distribution of cranial stress, as indicated by the
233
low variance across the sampling points (Table 5.3). Hesperocyon, on the other hand, had
the highest variance of stress, particularly in the frontal region (sampling point 5, Table
5.3).

Temporalis- and masseter-driven biting forces
Biting with a temporalis-driven muscle force created larger bite forces in all
models (142~177% of original bite force), resulting in higher bite force production
efficiency F
o/i
(0.80~0.95) as well (Table 5.4). However, total adjusted strain energy
increased by at least 40% in all of the models except for Borophagus (10%) and Epicyon
(-6%). Strain energy in Microtomarctus increased by a factor of 25 from 1.63 J to 41.25 J
(Table 5.4). Biting with masseter-only muscle force resulted in lower bite forces for all
models (91~58% of original) except for Mesocyon (123%) and Hesperocyon (~99.5%;
close to no change); strain energy increased in all models but Canis (-94%) and
Hesperocyon (-38%). The cost of increasing bite forces with a temporalis-driven bite is
thus smallest in Epicyon and Borophagus (Fig. 5.4).
When the stress distribution is visualized for the temporalis-driven bite,
differences in relative stress magnitude between robust borophagines and non-robust
canids become more obvious (Fig. 5.5). The dorsal region of the skull is highly stressed
in the fronto-parietal region of all non-robust canids (Fig. 5.5A-C, F). The zygomatic
arches of Hesperocyon, Microtomarctus, and Canis are particularly stressed. Mesocyon
shows lower stress in those regions compared to the group, but the TMJ still have
relatively high stress (Fig. 5.5B). Epicyon and Borophagus have lower stresses overall;
234
the TMJ are still the points of highest stress in these models as in Mesocyon. The elevated
regions of stress in the frontal region of Borophagus correspond to pockets of space in
the caudally extend frontal sinus among the trabecular struts within the sinus.

Chapter Five Discussion

Effects of multiple material properties
The main difference between the heterogeneous and homogeneous models of
Canis lupus is the higher overall stress in the former; this was expected as the
heterogeneous model contained more material properties representing regions of lower
bone density (supplementary online material, Appendix D SI Table 1). As a result, the
heterogeneous model is not as stiff as the homogeneous one, and stresses are not as
evenly distributed because of smaller contiguous areas of homogeneous materials. Total
strain energy is a measure of deformation; their relative magnitudes in a comparative
context could thus be interpreted as differences in the ability to transmit muscle force into
bite force without absorbing the force into strain energy by way of deformation.
Following this line of reasoning, the homogeneous model is more efficient at transmitting
bite force and minimizing strain energy than the heterogeneous model, as evidenced by
the lower strain energy in the former. On the other hand, bite force production efficiency
is actually higher in the heterogeneous model (Table 5.2). The stiff and homogeneous
nature of the single material property model may thus act as a double-edged sword: the
skull is artificially made stiffer and thus stronger by having a contiguous volume of
235
identical material, but the path of force conduction is probably also undifferentiated,
meaning that less of the input muscle force ends up at the bite point, and instead may be
spread throughout the cranium. The heterogeneous model, on the other hand, may have a
more restricted pathway of force conduction through the stiffer and denser bones of the
cranium, and thereby transmitting more of the input force into bite force, but strain
energy is nevertheless stored in the low density bones along the way, and the cranium as
a whole is more stressed (Fig. 5.2). Examination of the distribution of elements with low
modulus values (i.e. lower than homogeneous model) in the heterogeneous model
showed that approximately half of the parietal bone surface, where the temporalis
muscles attach, are made of more compliant bone than those in the homogeneous model.
These regions are likely to absorb energy by deformation, and thus increasing overall
strain energy in the cranium. This interpretation is informative, not only because it
highlights differences between heterogeneous and homogeneous models, but the trade-off
between skull strength and bite force efficiency may be an actual factor in the adaptation
of bone-cracking ecomorphs. The evolution of higher bite forces for fracturing prey bone
needs to proceed with adequate protection of the skull, which contains organs of various
sensory functions. This constraint may explain why bite force efficiency is not
empirically higher in bone crackers Borophagus and Epicyon (Tables 5.2, 5.4).
Interestingly, a trade-off between strength and efficiency appears to be in place for
differently-sized felids (Slater and Van Valkenburgh 2009).
Given these differences, one might expect the heterogeneous models of the fossil
skulls to act in a similar way, in having higher bite force production efficiency but also
236
higher strain energy than their homogeneous counterparts (Table 5.2). McHenry et al.
(2007) did not find unambiguous differences in stress and strain values for their
homogeneous and heterogenous cat models, but they also used mean stress values which
are sensitive to outliers (e.g. point stresses) and non-parametric distributions, which is
usually the case for finite element stress values in (at least) cranial analyses (Dumont et al.
2005, Tanner et al. 2008). Given these uncertainties, a more detailed evaluation of
relative bone densities in fossil specimens is needed to determine whether in certain cases
building heterogeneous models along with homogeneous models is warranted. In our case,
the fossil skulls all come from different fossil localities with varying degree of diagenesis;
fossil density changes even within a small region of bone that appear uniform in the
analogous region of a modern specimen. In exceptional cases, fossils are preserved as to
allow a direct translation of the relative bone densities into material properties during
model construction by using Hounsfield Units (McHenry et al. 2007), but our sense is
that this is far from the norm, and should be further studied in future research involving
fossil FEA.

Skull performance using a wolf-like bite
The main difference between the non-robust canids and purported bone-crackers
Borophagus and Epicyon is the evenness of cranial stress during a carnassial (P4) bite
(Fig. 5.3). Hesperocyon and Canis, which represent omnivore and meat-specialist
hypercarnivore ecomorphs, respectively (Van Valkenburgh 1991), show the highest
range of stress values along the dorsal region of the skull. Mesocyon and Microtomarctus,
237
which are not specialized hypercarnivores based on dental morphology (Van
Valkenburgh et al. 2004), are intermediate between this last group and the bone-crackers
(Table 5.3). The more even stress distributions are associated with lower strain energy
(Table 5.2). A potential tradeoff is evidenced by higher bite force production efficiency
in Hesperocyon and Canis, which have uneven stress distributions, and lower efficiency
in the other canids which have more evenly distributed cranial stress. The poorer
performance of the skulls of Hesperocyon and Canis suggests that the biting scenario
modeled may not be part of the behavioral repertoire that evolved in these two groups.
Omnivory often involves the retention of sharp cusps and wide basins to serve puncturing
and crushing functions, and thus may require maintaining a flexible mode of mastication,
rather than adapting the entire skull for durophagy. Along the same lines, Werdelin (1989)
distinguished bone-cracking behavior in specialized carnivorans such as hyaenids and
derived borophagine canids from bone-crushing which is very common among
carnivorans; bone-cracking is done with conical and robust anterior premolars, whereas
bone-crushing is done using the more flattened and basined molars. Crushing is a much
more generalized way of mastication that is plesiomorphic across carnivoran families
including lineages that evolved bone crackers, as shown by analysis of cranial shapes
(Tseng and Wang, unpublished data). Meat specialists, on the other hand, may encounter
soft tissue which does not require extremely high bite forces, although they consume
bone to some extent even though their craniodental morphology is not specialized (Van
Valkenburgh and Hertel 1993). The stronger skulls of the bone-crackers Epicyon and
Borophagus and those of Mesocyon and Microtomarctus may thus be evidence of a diet
238
containing more bone, at the expense of force production efficiency. Mesocyon
coryphaeus is a rudimentary hypercarnivore (Van Valkenburgh 1991), and
Microtomarctus is a mesocarnivore that may have occupied an ecological niche akin to
those of geologically older hesperocyonine canids (Wang 1994). Given their medium-
range body size (which in Microtomarctus is a reversal from larger-bodied ancestors),
they would have most likely hunted small prey (Carbone et al. 1999), and their relatively
strong skulls compared to those of omnivorous Hesperocyon and hypercarnivorous Canis
lupus would have allowed them to crush and consume prey flesh with bones. Again, this
type of bone-crushing should be differentiated from bone-cracking (Werdelin 1989).
Mesocyon and Microtomarctus do not exhibit the highly robust premolars present in
derived borophagines (e.g. Aelurodon), or the extensive wear on the carnassial tooth
often observed in Epicyon and Borophagus. Thus, Mesocyon and Microtomarctus
probably did not crack large bones frequently (if at all). Epicyon and Borophagus, on the
other hand, exhibit morphological and tooth wear evidence that converge on modern
hyaenids (Werdelin 1989).

Skull performance using temporalis- versus masseter-driven bite
All models showed increase in bite force production efficiency with a temporalis-
driven bite, but most showed decreased in efficiency with a masseter-driven bite (Table
5.4). This observation is consistent with the fact that the temporalis muscle attaches to the
cranium over a much larger area than the masseter on the zygoma, and thus has a larger
potential to affect bite force. However, total strain energy values also indicate that
239
increase in bite force efficiency is coupled with large decreases in skull performance and
increase in deformation in all models except for Epicyon and Borophagus. Therefore, a
low cost, in terms of skull strain energy, associated with increased bite force efficiency
during temporalis-driven biting is a shared characteristic of the derived borophagines. On
the other hand, the masseter-driven bite is costly for all models except Hesperocyon
(Table 5.4); this basal hesperocyonine canid maintained bite force efficiency but
decreased skull strain energy by 38% during a masseter-driven bite. These results
indicate that a bite more evenly distributed between temporalis and masseter input is the
most efficient for rudimentary hypercarnivore Mesocyon, mesocarnivore Microtomarctus,
and meat specialist Canis lupus. For Hesperocyon, a masseter-driven bite is the most
efficient, and this is perhaps associated with more grinding action in its masticatory
behavior or with rapid jaw closure using the masseter. For bone-crackers, their strong
frontal regions and deeper faces are associated with an efficient bite driven by the
temporalis muscles (Fig. 5.5).

The frontal dome stress dissipation hypothesis of Werdelin (1989)
Werdelin (1989) in his treatment of the morphological convergence between
hyaenids and the borophagine canid Borophagus, proposed that the dome foreheads of
bone-crackers created an uninterrupted arch of bone along which compressive stresses
could be dissipated from the bite point (Werdelin, 1989, p.394:Fig. 4). Examination of
the mid-sagittal section in our FEA results shows that stress is indeed evenly distributed
across the forehead of bone-crackers (Fig. 5.6). However, our results also indicate that
240
the deeper and thicker foreheads in Epicyon and Borophagus also reduced the amount of
stress between the face and the braincase, which are bent dorsally towards each other
during a temporalis-driven bite (Fig. 5.6A-F). Furthermore, visualization of the principal
stress acting on the mid-sagittal section shows that compressive stress occurs internally
between the face and braincase, but does not occur along the “arch” of bone represented
by the forehead (Fig. 5.6G-L). The main stress occurring in the direction parallel to the
sagittal plane of the skull is tensile (i.e. landmark 4 in Table 5.3). Therefore, our results
suggest that the contraction of the temporalis muscles creates an overall tensile stress
pulling the parietal bones laterally away from the mid-sagittal line, and also ventrally
towards the TMJ. This action creates high tensile stress along the fronto-parietal bones,
and which travels the “arch” of bone rostro-ventrally to be dissipated, in reverse of
Werdelin’s (1989) interpretation. The tensile stress is then merged with compressive
stress at the bite point. Thus, the high stresses present on the forehead regions of non-
robust canids (Fig. 5.5A-C, F) are the result of tensile stress pooling up in the forehead,
unable to be dissipated further down the face towards the bite point. With this
interpretation, the dome forehead becomes a conductor of tensile stress from strong
temporalis action, and not the compressive stress incurred by the bite point; it also serves
a role in reducing bending between the face and braincase. The compressive stress in the
bone-cracking tooth, then, would be mostly conducted and potentially dissipated by a
combination of formation of enamel micro-cracks which are self-healing, absorption of
strain energy through the enamel-dentine junction, and minimization of tensile stress by
specialized enamel microstructure which is present convergently in hyaenids,
241
percrocutids, and borophagine canids (Chai et al. 2009, Ferretti 2007a, Stefen 1999,
Stefen and Rensberger 2002). The interface between the bone being cracked, the
specialized cracking tooth, and the maxillary bone still requires additional detailed
analyses in order to further test this interpretation.

Jaw musculature and muscle use
Given our demonstration of the more efficient production of bite force using a
temporalis-driven bite in derived borophagines, it is possible that masticatory
musculature in these specialized species underwent corresponding evolutionary changes.
To evaluate this idea, we estimated the cross-section areas for the temporalis and the
masseter-pterygoid muscle groups, using the measurement method of Thomason’s (1991)
dry skull method (note: we did not attempt to estimate bite force by carrying out the
entire technique because of inaccuracies that are being studied and corrected; see Ellis et
al., 2008). The ratio between the two muscle groups is highest in Epicyon, and lowest in
Hesperocyon (Table 5.5). The bone cracker Borophagus has a temporalis to masseter-
pterygoid ratio of 1.00, very close to Canis and Microtomarctus. Therefore, there is no
consistent pattern between temporalis-driven bite efficiency and estimated temporalis
size. However, the masseter-driven bite efficiently performed by Hesperocyon is
associated with relatively large masseter-pterygoid size. When the muscle group ratios
are plotted against log skull volume, Hesperocyon, Microtomarctus, and Epicyon appear
to fall along a linear path, suggesting possibility of size allometry in the estimate ratio
(supplementary online material, Appendix D SI Fig. 6). If this is true, then Borophagus
242
and Canis both have smaller temporalis size than expected by allometry; Mesocyon, on
the other hand, has a larger than expected temporalis size. These interesting but
prelimnary results hint that there may be a more complex pattern of musculature
evolution not simply correlated with bone-cracking capability. It also allows the
possibility that bone-cracking specialization may involve changes in muscle activation
patterns (e.g. temporalis- versus masseter-driven) rather than increases in absolute size of
the muscle groups.

Limitations and future directions
Whereas the premise of this particular study involved only a test of a specific
biting behavior, the results should not be taken to infer that any adaptations seen in the
skull models evolved solely for bone-cracking; canids retain a rather primitive dental
formula, with retention of molars as well as anterior premolars. Although there may be
optimal positions within the dentition to perform demanding tasks such as bone-cracking
(Greaves 1982), the availability of other teeth provides flexibility in a wide range of
possible behavior. A next step would be to test other tooth positions to see if the skulls
are sensitive to different bite points, as well as incorporating muscle forces from the neck
and extrinsic forces acting on the teeth to simulate prey-capturing scenarios. A
comparison is underway to analyze in more detail the skulls of purported bone-crackers
from the canid, hyaenid, and percrocutid families to examine whether finer details of
stress distributions differ among them, and whether efficiency in a temporalis-driven bite
applies to the bone-cracking ecomorph in general.
243
The fragmentary nature of fossil specimens prevented mandibles from being
incorporated into this analysis, but they may provide additional complementary
information about bone-cracking adaptations (Biknevicius and Leigh 1997, Biknevicius
and Ruff 1992). Furthermore, incorporation of simulation-derived values for muscle
forces and directions can be incorporated into finite element analysis (Moazen et al.
2008), improving the accuracy of the motions being modeled. Improvement in material
properties and understanding the orthotropic behavior is a broad direction that will
improve all FEA studies of present and extinct organisms. This, and along with validation
experiments by biologists, will help towards building models where resulting stresses and
bite forces could be directly compared to actual values in living animals.

Chapter Five Conclusion

We built and analyzed six skull models of fossil and living canids in order to test
the degree to which morphological traits characterizing bone-cracking ecomorphologies
are associated with their purported function. A sensitivity test comparing homogeneous
and heterogeneous models of Canis lupus showed that the former is made artificially
stronger but less efficient in bite force production by a single invariant material property.
Simulations of all skull models of a bone-cracking bite using the carnassial tooth (P4)
with wolf-like muscle ratios showed that the omnivore Hesperocyon gregarius and the
meat specialist Canis lupus had the most efficient but weakest skulls. The
mesocarnivore/hypercarnivore Mesocyon coryphaeus and Microtomarctus conferta had
244
more evenly distributed stresses during the same biting scenario, and the bone-cracking
Epicyon and Borophagus had the lowest and most even stresses. Testing of the same
skulls using temporalis- and masseter-driven bites showed that the Hesperocyon
gregarius was most efficient during a masseter-driven bite, and the bone-crackers
Epicyon and Borophagus were most efficient during a temporalis-driven bite. Estimated
muscle group ratios among the species examined showed no association of temporalis
size and efficiency in temporalis-driven biting. Thus, adaptation in bone-cracking
borophagines may involve modification of muscle activation ratios, not absolute muscle
size. Detailed examination of the path of stress in the skulls modeled showed that tensile
stress is the main stress being conducted from the parietal to the face, and that domed
foreheads in bone crackers provide a path for the dissipation of this tensile stress, as well
as reducing overall bending of the skull between the face and braincase. These findings
suggest that there is a strong linkage between the morphological features of bone-
cracking ecomorphs and their functional capability for such behavior.

Chapter Five Acknowledgments

We thank Michael McNitt-Gray (UCLA) for scanning all of the specimens used
in this study. Alan Shabel (UC Berkeley) and Jim Dines (LACM) for making carnivore
specimens available for dissection. Betsy Dumont (UMass Amherst) for providing the
BoneLoad program for distributing muscle forces. Korkut Brown, Keith Henkel, and Rob
Taft (USC) for assistance with CT digitization. Lars Werdelin for constructive criticism
245
on an earlier draft of the manuscript. The editor and two anonymous reviewers provided
constructive comments that improved many aspects of this paper. We thank Theodore
Fremd (NPS-JODA), Jin Meng (AMNH), and Jeff Indeck (PPM) for access to specimens
under their care. Research funded by ASM grant-in-aid of research, US Fulbright Student
Grant, NSF DDIG (DEB-0909807) & GRF, USC Zumberge Grant (ZJT).

246
Chapter Five References
Andersson K. 2005. Were there pack-hunting canids in the Tertiary, and how can we
know? Paleobiology 31(1):56-72.

Biknevicius AR, Leigh SR. 1997. Patterns of growth of the mandibular corpus in spotted
hyenas (Crocuta crocuta) and cougars (Puma concolor). Zoological Journal of
Linnean Society 120:139-161.

Biknevicius AR, Ruff CB. 1992. The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology
228:479-507.

Binder WJ, Van Valkenburgh B. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252:273-283.

Boyko AR, Boyko RH, Boyko CM, Parker HG, Castelhano M, Corey L, Degenhardt JD,
Auton A, Hedimbi M, Kityo R, Ostrander EA, Schoenebeck J, Todhunter RJ,
Jones P, Bustamante CD. 2009. Complex poplation structure in African village
dogs and its implications for inferring dog domestication history. Proceedings of
the National Academy of Sciences 106(33):13903-13908.

Carbone C, Mace GM, Roberts SC, Macdonald DW. 1999. Energetic constraints on the
diet of terrestrial carnivores. Nature 402:286-288.

Chai H, Lee JJ-W, Constantino P, Lucas PW, Lawn B. 2009. Remarkable resilience of
teeth. Proceedings of the National Academy of Sciences 106(18):7289-7293.

Dessem D. 1989. Interactions between jaw-muscle recruitment and jaw-joint forces in
Canis familiaris. Journal of Anatomy 164:101-121.

Dumont ER, Grosse I, Slater GJ. 2009. Requirements for comparing the performance of
finite element models of biological structures. Journal of Theoretical Biology
256:96-103.

Dumont ER, Piccirillo J, Grosse IR. 2005. Finite-element analysis of biting behavior and
bone stress in the facial skeletons of bats. The Anatomical Record Part A
283A:319-330.

Ellis JL, Thomason JJ, Kebreab E, France J. 2008. Calibration of estimated biting forces
in domestic canids: comparison of post-mortem and in vivo measurements.
Journal of Anatomy 212:769-780.

247
Erickson GM, Catanese III J, Keaveny TM. 2002. Evolution of the biomechanical
material properties of the femur. The Anatomical Record 268:115-124.

Ferretti MP. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience:1-12.

Finarelli JA. 2007. Mechanisms behind active trends in body size evolution of the
Canidae (Carnivora: Mammalia). American Naturalist 170(6):876-885.

Finarelli JA, Flynn JJ. 2009. Brain-size evolution and sociality in Carnivora. Proceedings
of the National Academy of Sciences 106(23):9345-9349.

Greaves WS. 1982. A mechanical limitation on the position of the jaw muscles of
mammals the one-third rule. Journal of Mammalogy 63(2):261-266.

Grosse I, Dumont ER, Coletta C, Tolleson A. 2007. Techniques for modeling muscle-
induced forces in finite element models of skeletal structures. The Anatomical
Record 290:1069-1088.

Holliday JA, Steppan SJ. 2004. Evolution of hypercarnivory: the effect of specialization
on morphological and taxonomic diversity. Paleobiology 30(1):108-128.

Kruuk H. 1972. The spotted hyena: a study of predation and social behavior. Shaller GB,
editor. Chicago: The University of Chicago Press. 335 p.

Larivière S, Pasitschniak-Arts M. 1996. Vulpes vulpes. Mammalian Species 537:1-11.

Martin LD. 1989. Chapter 19. Fossil history of the terrestrial Carnivora. In: Gittleman JL,
editor. Carnivore behavior, ecology, and evolution. New York: Cornell University
Press. p 536-568.

McHenry C, Wroe S, Clausen PD, Moreno K, Cunningham E. 2007. Supermodeled
sabercat, predatory behavior in Smilodon fatalis revealed by high-resolution 3D
computer simulation. Proceedings of the National Academy of Sciences
104(41):16010-16015.

Moazen M, Curtis N, Evans SE, O'Higgins P, Fagan MJ. 2008. Combined finite element
and multibody dynamics analysis of biting in a Uromastyx hardwickii lizard skull.
Journal of Anatomy 213:499-508.

Nalla RK, Kinney JH, Ritchie RO. 2003. Mechanistic failure criteria for the failure of
human cortical bone. Nature Materials 2:164-168.

248
Nowak RM. 1999. Walker's carnivores of the world. Baltimore: The Johns Hopkins
University Press. 313 p.

Rayfield EJ. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35:541-576.

Ross CF. 2005. Finite element analysis in vertebrate biomechanics. The Anatomical
Record Part A 283A:253-258.

Schumacher GH. 1961. Funktionelle Morphologie der Kaumuskulatur. Jena, Germany:
VEB Gustav Fischer Verlag. 262 p.

Simpson GG. 1945. The principles of classification and a classification of mammals.
Bulletin of the American Museum of Natural History 8:1-350.

Slater GJ, Dumont ER, Van Valkenburgh B. 2009. Implications of predatory
specialization for cranial form and function in canids. Journal of Zoology
278:181-188.

Slater GJ, Van Valkenburgh B. 2009. Allometry and performance: the evolution of skull
form and function in felids. Journal of Evolutionary Biology 22(11):2278-2287.

Stefen C. 1999. Enamel microstructure of recent and fossil Canidae (Carnivora:
Mammalia). Journal of Vertebrate Paleontology 19(3):576-587.

Stefen C, Rensberger JM. 2002. The specialized enamel structure of hyaenids (Mammalia,
Hyaenidae): description and development within the lineage - including
percrocutids. Zoologische Abhandlungen 52:127-147.

Tanner JB, Dumont ER, Sakai ST, Lundrigan BL, Holekamp KE. 2008. Of arcs and
vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta).
Biological Journal of the Linnean Society 95:246-255.

Tedford RH, Wang X, Taylor BE. 2009. Phylogenetic systematics of the North American
fossil Caninae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 325:1-218.

Thomason JJ. 1991. Cranial strength in relation to estimate biting forces in some
mammals. Canadian Journal of Zoology 69:2326-2333.

Tseng ZJ. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96:51-67.
249

Turnbull WD. 1970. Mammalian masticatory apparatus. Fieldiana: Geology 18(2):149-
356.

Turner A, Antón M. 1996. The giant hyaena, Pachycrocuta brevirostris (Mammalia,
Carnivora, Hyaenidae). Geobios 29(4):455-468.

Turner A, Antón M, Werdelin L. 2008. Taxonomy and evolutionary patterns in the fossil
Hyaenidae of Europe. Geobios 41(5):677-687.

Van Valkenburgh B. 1991. Iterative evolution of hypercarnivory in canids (Mammalia:
Carnivora): evolutionary interactions among sympatric predators. Paleobiology
17:340-362.

Van Valkenburgh B, Hertel F. 1993. Tough times as La Brea: tooth breakage in large
carnivores of the late Pleistocene. Science 261:456-459.

Van Valkenburgh B, Wang X, Damuth J. 2004. Cope's Rule, hypercarnivory, and
extinction in North American canids. Science 306(5693):101-104.

Wang X. 1994. Phylogenetic systematics of the Hesperocyoninae (Carnivora: Canidae).
Bulletin of the American Museum of Natural History 221:1-207.

Wang X, Tedford RH, Antón M. 2008. The Dog Family, Canidae, and Their
Evolutionary History. New York: Columbia University Press. 219 p.

Wang X, Tedford RH, Taylor BE. 1999. Phylogenetic systematics of the Borophaginae
(Carnivora: Canidae). Bulletin of the American Museum of Natural History
243:1-391.

Werdelin L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

Werdelin L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
In: Gittleman JL, editor. Carnivore behavior, ecology, and evolution. New York:
Cornell University Press. p 582-624.

Werdelin L, Solounias N. 1990. Studies of fossil hyaenids: the genus Adcrocuta Kretzoi
and the interrelationships of some hyaenid taxa. Zoological Journal of Linnean
Society 98:363-386.

250
Wroe S. 2007. Cranial mechanics compared in extinct marsupial and extant African lions
using a finite-element approach. Journal of Zoology.

Wroe S, McHenry C, Thomason JJ. 2005. Bite club: comparative bite force in big biting
mammals and the prediction of predatory behaviour in fossil taxa. Proceedings of
the Royal Society of London, Series B 272:619-625.

251
Figure 5.1. Cranial size and shape diversity of representative forms in the Canidae.
Genera studied are highlighted in black font. All genera have a fossil record, but all
extant canids are in Caninae. Scale bar (next to Epicyon) equals 50 mm. Stratigraphic
ranges taken from Wang (1994), Wang et al. (2008), Wang et al. (1999).

252
Figure 5.2. Comparison of Von Mises (VM) stress distribution in the heterogeneous (A-C)
and homogeneous (D-F) models of Canis lupus. A,D: dorsal views; B,E: mid-sagittal
section views; C,F: ventral views. All stress distributions are scaled between 0 and 100
Megapascals (MPa). Hotter/lighter colors represent higher stress, Cooler/darker colors
represent lower stress.

253
Figure 5.3. Maximum Von Mises stress (in megapascal, MPa) along the dorsal mid-
sagittal line of the skull models from simulated P4 biting with 64%-26%-10% muscle
ratios. Analogous anatomical sampling points are plotted as percentages of total skull
length as measured from dorsal view, showing relative proportions of the cranial bones.
Note consistently low maximum stress across the dorsal crania of Borophagus and
Epicyon (bold lines). See Methods for list of sampling points.

254
Figure 5.4. Percent increases in bite force production efficiency (F
o/i
, dark bars) and
amount of deformation (total strain energy, light bars) compared to those obtained using
modern muscle ratios. All models were biting with muscle input force generated solely
by the temporalis muscles (highlighted on Canis model). Symbols as in Figure 5.3. Note
the small amount of change in deformation required to substantially increase bite force
efficiency in Epicyon haydeni (white circle) and Borophagus secundus (black circle).

255
Figure 5.5. Comparisons of Von Mises stress in the skull models during a temporalis-
dominated unilateral bite with the left carnassial tooth (P4). A. Hesperocyon gregarius; B.
Mesocyon coryphaeus; C. Microtomarctus conferta; D. Epicyon haydeni; E. Borophagus
secundus; F. Canis lupus. Skulls are scaled to the same approximate length; for relative
sizes see Fig. 5.1.

256
Figure 5.6. Mid-sagittal sections of the skull models during a temporalis-driven unilateral
left carnassial (P4) bite. Stress distributions shown in both Von Mises stress (A-F) and
Principal stress 1 (G-L). For the latter, all colors represent tensile stress; compressive
stress is marked in gray. A,G: Canis lupus; B,H: Borophagus secundus; C,I: Epicyon
haydeni; D,J: Microtomarctus conferta; E,K: Mesocyon coryphaeus; F,L: Hesperocyon
gregarius.

257
Table 5.1. Finite element model parameters. For institutional abbreviations see main text.
Input F, applied total muscle force in Newtons; SA, total surface area; V, total volume.

Taxon Specimen
Tetrahedral
Elements SA (mm
2
) V (mm
3
) Input F (N)
Canis lupus (hetero.) LACM23010 1,498,420 147,812 229,553 4,119
Canis lupus (homo.) LACM23010 1,120,780 127,810 207,902 5,089
Borophagus secundus F:AM61640 920,143 96,281 220,845 3,834
Epicyon haydeni PPM JWT-1100 937,061 203,036 932,188 8,084
Microtomarctus
conferta
LACM-
CIT1229 948,735 32,764 44,347 1,305
Mesocyon coryphaeus NPS Joda-3348 1,115,623 49,341 80,194 1,965
Hesperocyon gregarius LACM-CIT621 769,740 15,139 12,241 603

258
Table 5.2. Measures of performance obtained from finite element analysis of the six skull
models. Muscle force input ratio used was 64% temporalis, 26% masseter, and 10%
pterygoid (Turnbull, 1970). Total forces were scaled to surface area in each model
(Dumont et al. 2009). En, adjusted total strain energy (in Joules); F
o/i
, ratio of bite force
to total muscle input force.

Taxon Bite force (N) F
o/i
En (J)
Canis lupus (heterogeneous) 3550.4 0.86 62.01
Canis lupus (homogeneous) 3410.1 0.67 49.05
Borophagus secundus 1916.3 0.50 2.51
Epicyon haydeni 3801.4 0.47 9.11
Microtomarctus conferta 607.6 0.47 1.63
Mesocyon coryphaeus 938.6 0.48 7.91
Hesperocyon gregarius 307.8 0.51 11.65

259
Table 5.3. Von Mises stresses at point-sampling positions along the dorsal mid-sagittal
line of the skulls for a unilateral P4 bite using wolf muscle ratios (64% temporalis-26%
masster-10% pterygoid). All values are in megapascals. Sampling points: 1. rostral point
of the nasal bone; 2. rostrum at infraorbital foramina; 3. rostrum at anterior border of
orbits; 4. frontal regions between postorbital processes; 5. frontal region at the postorbital
constriction; 6. center of the parietal bone; 7. caudal-most region of the sagittal crest.

1 2 3 4 5 6 7 Variance
Canis lupus 5.21 8.27 11.69 11.84 16.34 15.02 7.36 16.65
Borophagus secundus 3.96 2.96 2.27 3.05 4.13 4.52 1.34 1.26
Epicyon haydeni 1.85 2.04 1.79 4.62 3.60 3.86 1.28 1.64
Microtomarctus conferta 1.52 3.93 2.61 3.88 7.04 6.65 5.74 4.29
Mesocyon coryphaeus 2.25 2.02 2.16 4.44 8.91 6.03 2.30 6.95
Hesperosyon gregarius 11.93 8.12 6.82 16.48 21.73 15.40 10.66 27.12

260
Table 5.4. Bite force production efficiency (ratio between bite force and muscle input
force; F
o/i
), energy of deformation (adjusted total strain energy, En, in Joules), and their
percent changes when models are loaded with a temporalis- or a masseter-driven bite,
respectively.

Temporalis-driven Masseter-driven
F
o/i
%ch. En (J) %ch. F
o/i
%ch. En (J) %ch.
Canis lupus 0.95 +42 76.91 +57 0.39 -42 3.11 -94
Borophagus secundus 0.79 +58 2.77 +10 0.34 -32 5.54 >+100
Epicyon haydeni 0.80 +70 8.59 -6 0.39 -7 42.98 >+100
Microtomarctus
conferta 0.83 +77 41.25 >+100 0.43 -9 6.21 >+100
Mesocyon coryphaeus 0.80 +67 11.32 +43 0.58 +23 142.78 >+100
Hesperocyon gregarius 0.80 +57 16.46 +41 0.51 0 7.23 -38

261
Table 5.5. Average cross-sectional areas of masticatory muscle groups estimated using
Thomason’s (1991) dry skull method.

Taxon
Temporalis
(mm
2
)
Masseter/Pterygoid
(mm
2
) Ratio
Canis lupus2357 2339 1.01
Borophagus secundus1958 1961 1.00
Epicyon haydeni8012 5461 1.47
Microtomarctus conferta625 624 1.00
Mesocyon coryphaeus1467 1063 1.38
Hesperocyon gregarius346 412 0.84

262
Chapter Six: Study of cranial function in the percrocutid Dinocrocuta gigantea

This chapter has been published as:

Tseng, Z. J. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of the
Linnean Society 96:51-67. doi: 10.1111/j.1095-8312.2008.01095.x.

A copy of the accepted manuscript begins on the next page.
263
Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis

Zhijie Jack Tseng

Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, 3616 Trousdale Parkway, Los Angeles, California
90089-0371, USA, jack.tseng@usc.edu

Department of Vertebrate Paleontology, Natural History Museum of Los Angeles County,
900 Exposition Boulevard, Los Angeles, California 90007, USA

Running title: Functional cranial morphology of Dinocrocuta gigantea
264

Chapter Six Abstract

Carnivoran ecomorphologies evolved repeatedly during the Cenozoic. Whereas extreme
forms (e.g. sabretoothed predators) probably represent similarities in ecology, other
morphologies are more subtle as to the extent of their shared niche space. Finite element
models of the skulls of Dinocrocuta gigantea, Canis lupus, and Crocuta crocuta were
constructed to test the interpretation of D. gigantea as a bone cracker, an interpretation
made on the basis of its large, conical premolars and robust cranial morphology. D.
gigantea is also of interest because it represents a lineage that has been placed in its own
family, sister to Hyaenidae. Thus, functional similarity in craniodental performance could
represent rapid convergence. Findings indicate that the crania of D. gigantea and C.
crocuta perform better in stress dissipation and distribution than that of C. lupus,
regardless of P3 or P4 biting. In particular, the domed frontal region of the bone crackers
received lower, as well as more evenly distributed stress than C. lupus. Thus, the
craniodental forms of the two bone-crackers are linked by functional advantage over that
of C. lupus. Further examination of lineages such as borophagine canids could elucidate
the extent of functional convergence of the bone-cracking ecomorph across diverse
groups.

Keywords: Bone cracking – Canidae - Carnivora – Cenozoic - China – Finite Element
Analysis – Functional morphology - Hyaenidae – Mammalia - Miocene
265

Chapter Six Introduction

Our understanding of functional morphology in fossil carnivorans has been
greatly augmented by studies of morphospace and ecomorphology in both extant and
extinct predator guilds (Van Valkenburgh 1988, 1989b, 1999, 2001, Werdelin 1996a).
General categories have been established based on correlations between craniodental
form and diet in living carnivorans, and similarity in craniodental form between living
and fossil taxa. One example of an iteratively evolved mammalian carnivore
ecomorphology is the bone-cracking predator, which has been characterized on the basis
of craniodental morphology (Werdelin 1989) and enamel microstructure specialization
(Ferretti 2007a, Rensberger 1995, Stefen and Rensberger 2002). Bone cracking
carnivorans are best represented today by members of the family Hyaenidae. Although
the generalizations are well established, no study has compared hyaenid craniodental
forms in their response to mechanical stress imposed on the skull during bone cracking
behavior. This study presents a comparative analysis of a purported fossil bone cracking
carnivoran, Dinocrocuta gigantea Schlosser, 1903, the extant bone cracking spotted
hyena Crocuta crocuta Erxleben, 1777, using the extant gray wolf Canis lupus Linnaeus,
1758 as a non-bone-cracking hypercarnivore for comparison. It is hypothesized that the
craniodental morphology of Dinocrocuta is more suited for bone-cracking in terms of its
stress-dissipating architecture because of the highly domed frontal region, than the
spotted hyena. Both would perform better under loading conditions simulating bone-
cracking than the gray wolf, which is a meat specialist with capability for bone-crushing
266
with molars (for a definition of crushing versus cracking see Werdelin, 1989) but with a
very shallow forehead.
To test the functional hypothesis regarding the capability of the cranium for bone
cracking, the engineering technique finite element modeling is utilized (Laitman 2005,
Richmond et al. 2005, Ross 2005). The cranial models are compared solely on
craniodental form, with element volumes standardized. Because the goal of this study is
to examine the distribution and concentration of reaction forces in the cranium, stress
(force per unit area) is used instead of strain (change in length relative to original length),
a measure of deformation. The measure used to evaluate craniodental function in stress
dissipation is the Von Mises stress, which is a scalar function incorporating principal
stress in the three orthogonal planes of a three-dimensional object. Von Mises stress is
also used as a criterion to evaluate how close an object is to failure, i.e. it is directly
comparable to yield strength of the object (Irons & Ahmad, 1980). The expectation is that
both the median Von Mises stress of the entire cranium, as well as the maximum Von
Mises stress in the frontal-parietal region, hypothesized to serve an important function in
stress dissipation (Buckland-Wright 1978, Werdelin 1989), would be lowest in
Dinocrocuta, higher in Crocuta, and highest in Canis.
As a more recently applied methodology in the field of evolutionary biology,
finite element modeling has demonstrated high potential as a tool to understand
kinematics of forms (Dumont et al. 2005, Laitman 2005, Rayfield 2004, Rayfield et al.
2001). The fundamental principle of this technique is to sufficiently recreate a
representation analyzable by computers from an extremely complex natural structure
267
such as the skull. To build a finite element model, the basic steps include (1) morphology
of object of interest (shape reconstruction), (2) characteristics of the material composition
of the object which governs the behavior of that object under mechanical loads (material
properties), (3) scenario in which the object is being loaded with mechanical force, and
how much force is involved (boundary conditions). Modeling of complex objects
simplifies the problem at hand by making assumptions. Bone, a living and dynamic tissue,
would require highly complex models to be portrayed accurately. By starting only with
shape, or craniodental morphology, this study eliminates many details of bone
biomechanics (e.g. different material properties for cancellous versus cortical bone;
different properties in different load directions or anisotropy; bone sutures, contact
between bone and tooth, etc.). However, the simplifying assumptions made in the study
makes fossils comparable to extant specimens by standardizing the methods of modeling.
Furthermore, in any comparative analyses involving fossil material, the quality of the
data is often constrained by fossil quality, as many anatomical features might be
incomplete or missing altogether. Assumptions made under these considerations are
discussed in more detail in the model building protocol below.
Institutional abbreviations: IVPP, Institute of Vertebrate Paleontology and
Paleoanthropology, Beijing, China; LACM: Natural History Museum of Los Angeles
County (mammalogy department), Los Angeles, California, USA; UCLA, University of
California, Los Angeles, California, USA. Other abbreviations: CT, computed
tomography; IQR, Interquartile range; MAD, median absolute deviation from the median;
(M/G)Pa, (mega/giga)pascal; SE, standard error of the median.
268

Chapter Six Materials and Methods

Data acquisition and processing

An undeformed skull of Dinocrocuta gigantea (IVPP V15649) was used.
Although not all fine internal structures of the bone are preserved, the specimen
represents the best preserved skull of the species. The skull retains cranium and
associated mandibles; the specimen comes from the late Miocene beds of Fugu, Shaanxi
Province, in northern China. The associated cranium and mandibles retain a complete
dentition, which is fully erupted but unworn, representing a sub-adult or young adult
equivalent when compared to Crocuta crocuta. From the developmental stage of the
cranial bones in comparison with other specimens of the same species, it was determined
that this individual also does not have a fully grown cranial region. Thus, a skull of the
extant spotted hyena Crocuta crocuta (LACM 30655) in a similar stage of ontogenetic
development was chosen for analysis and comparison. In addition, an adult gray wolf
Canis lupus (LACM 23010) skull was chosen as a hypercarnivore comparison.
All three skulls were scanned by computed tomography (CT) using a Siemens
Definition 64 scanner (Siemens Medical Solutions, Forcheim, Germany) at UCLA
Medical Center. Specimens were scanned with 0.6 mm slice thickness and a 0.6 mm
interslice distance. This produced 616 images for Dinocrocuta gigantea, 464 for Crocuta
crocuta, and 499 for Canis lupus. The data were then imported into the image processing
software programs Amira (Visage Imaging, Inc., Carlsbad, California, USA) and
VGStudio Max (Volume Graphics GmbH, Heidelberg, Germany), in which a
269
combination of automated thresholding operations and manual delineation were used to
identify the craniodental morphology from the image scans. For the Dinocrocuta
gigantea skull, the internal cavities were filled with inorganic matrix during burial, and
the matrix has approximately the same density as the fossilized bone. Thus, manual
editing of bone boundaries, visible as small gaps and fissures between fossil bone and
rock matrix, was done for all image slices. To ensure a faithful reconstruction of the
fossilized bone, the slices were edited thoroughly in all three planes (axial, coronal, and
sagittal). Once the regions of interest were defined, three-dimensional representations of
the crania were reconstructed.
The reconstructions were then imported into the rapid prototyping software
program Geomagic Studio 9.0 (Geomagic, Inc., Research Triangle Park, North Carolina,
USA), which allowed operations that improved the quality and consistency of the
reconstructions. The cranial reconstructions were refined, holes filled, and then
decimated to 300,000 triangles, which formed the basic elements of the 3D surface
reconstruction. The fossil specimen required extensive modification such as removing
sharp artifacts in internal bone boundaries created during manual delineation. The
mandible reconstructions were modified and cleaned separately from the cranium, but a
correct orientation of articulation preserved to allow for modeling of muscle forces in the
final models.

Finite element model building

The refined reconstructions were then imported into the finite element analysis
software Strand7 (G + D Computing Pty Ltd, Sydney, Australia). In this software the
270
reconstruction was again checked for errors such as sharp angles between triangles and
very large aspect ratios. In addition, the basic triangles of the surface reconstruction were
zipped and extraneous nodes removed to create a single continuous surface. This
structure was then transformed into a solid mesh of four-noded tetrahedral elements. The
solid mesh reconstructions of Dinocrocuta and Canis were then standardized to that of
the Crocuta crocuta mesh, which had an elemental volume of 326,625 mm
3
. This was
done in the Strand7 software program by dividing the elemental volume (calculated by
the software using the model summary function) of the Crocuta crocuta model by the
volume of the model to be standardized to get a ratio; then, the model of interest was re-
scaled with the cubic-root value of this ratio in each of the x, y, and z axes to obtain the
standardized volume. This step standardized the amount of craniodental material
represented by the finite elements, and allowed differences in analytic results to be
attributed to shape differences represented by the cranial models. This approach is
appropriate for the hypothesis being tested, because once the amount of craniodental
material present in all three cranial models are standardized, the analyses could address
how remaining morphological differences such as frontal shape and thickness affect the
stress-dissipating function of the structures during unilateral premolar biting.
After a reconstruction of the original morphology was achieved, the solid mesh
needed to be assigned (1) material properties and (2) boundary conditions (constraints
and loads). Although voxel-based techniques are now able to import density differences
(as Hounsfield units) in CT data directly into the final model for assigning multiple
material properties (McHenry et al. 2007), this was simply not practical for more typical
271
fossil specimens. For most fossil skulls, including the Dinocrocuta gigantea studied here,
many minute details of the cranium have been obliterated by diagenesis, and cavities are
often filled by different minerals. These modifications are amplified in the scanning
process and localized diagenesis can create density differences in the CT data even along
small distances of a single bone. Thus, a direct import of the density data from CT scans
does not create a correct representation for such specimens. To enable direct comparison
with extant skulls, the craniodental reconstruction was assumed to represent a
homogeneous, isotropic, elastoplastic material. For static analyses, the only required
material properties under this assumption are Young’s modulus (E) and Poisson’s ratio
( ν).
The range of Young’s modulus values provided in Erickson et al (2002) for birds
and mammals is 15 to 30 GPa, and finite element models used in this study were run with
this range of values in 5 GPa intervals. All returned similar results, but only data using E
= 20 GPa, the mean for birds and mammals in Table 1 of Erickson et al (2002), are
presented here. Given the variation in Poisson’s ratio depending on state of fatigue of the
test specimen (Pidaparti and Vogt 2002) and the large range of values (from 0.1 to over
0.5) published for both cranial and post-cranial bones of mammals (Peterson and Dechow
2003, Peterson et al. 2006, Reilly and Burstein 1975, Shahar et al. 2007, Wang et al.
2006), a range of Poisson’s ratio values were tested. When the same models were run
using ν = 0.1 to 0.5 in 0.1 intervals, the median Von Mises stress values decreased
between 1 and 4% for every 0.1 increase in ν (data not shown). A sensitivity study is
currently underway to further quantify the effect of changing ν values on model
272
outcomes. All data presented here used a mid-range Poisson's ratio of 0.3. As data are
interpreted comparatively, the conclusions made here are not likely to be altered by slight
changes in the exact Poisson’s ratio used. All three models were assigned identical
material properties. Using an estimated density of 2 mg/mm
3
for dog cortical bone
(Cowin, 1989), together with standardized bone volumes, the final models had skull mass
close to 653.25 g (see Appendix). Model skull mass exceeds the actual dry skull mass by
137% in Crocuta crocuta and by 193% in Canis lupus. This difference imparts additional
strength in the skull models by treating all skull bone (and teeth) as cortical bone,
therefore the absolute stress values obtained in the analyses are probably lower than
would be present in actual skulls.
The boundary conditions included constraints at three locations on the cranial
models: (1) left temporomandibular joint, (2) right temporomandibular joint and (3) tooth
of interest. The temporomandibular joints were fixed from any movement by ten arbitrary
fixed nodes on each joint, roughly representing the length of contact between the glenoid
and the mandibular articular processes. In addition, the tip of the tooth of interest (e.g. P3
for bone-cracking) was fully constrained from movement. The main cusp of P3, paracone
of P4, and paracone and metacone of M1 were constrained in the respective models.
These boundary conditions are meant to represent the simulation of an instantaneous
linear static load applied to a food item (i.e. bone) at the moment of peak force applied
through the tooth of interest by actions of the temporalis and masseter muscles.
Lastly, the loads applied to the models were the contracted muscles of the
temporalis and masseter on both sides of the cranium. The pterygoid muscles were not
273
modeled, as empirical data are lacking for extant carnivorans, and thus introducing
estimated forces would only increase uncertainty in these models, in particular for
Dinocrocuta. The muscle forces were simulated by creating thin plates over the area of
bone where the respective muscles originate. The insertion sites of the temporalis and
masseter were identified by bone rugosities where muscles attach on the mandibles and
from muscle dissection of an extant specimen of Hyaena hyaena (LACM freezer
catalogue number 42206) conducted at the Department of Mammalogy, Natural History
Museum of Los Angeles County, California, USA. The software program Boneload
(Grosse et al. 2007) was used to create tangentially oriented forces on the cranium,
simulating the wrapping of the muscles around the cranium (Fig. 6.1). The insertion
directions of those two masticatory muscles on the mandibles were pointed toward the
center of the ascending ramus for temporalis, and at the midpoint of the lateral ridge
ventral of the mandibular fossa that extends posteriorly to the angular process for
masseter, respectively. This method simulates the approximate pulling direction of the
contracting musculature, but does not account for differences in muscle fiber angles that
might exist between groups within each muscle.
As all three models were scaled to approximately the same element volume, that
of Crocuta crocuta, and the specimen of Crocuta used represents a sub-adult with fully
erupted permanent dentition but with still developing cranial bones, a bite force of 318.15
N at the P4 was used in all analyses. This bite force was derived from the regression
equation of empirical bite force data taken by Binder & Van Valkenburgh (2000)
calculated for a 12-month old captive spotted hyena.
274
Bite force = 165.952 + 12.683*age
Each model was loaded with an arbitrary 1000N total muscle force, with the proportions
of muscle activation set as described above. As bite force measurements were taken with
a fork force transducer by Binder & Van Valkenburgh (2000), which converts vertical
displacement of the forks to change in detected current, the resulting values represent
force perpendicular to the plane of occlusion. Thus, the resulting bite force produced at
the cusp tip of P4 paracone, perpendicular to the tooth occlusal plane, was analyzed. The
new resultant total muscle force was calculated as below (Dumont et al. 2005), and then
distributed according to the muscle activation scheme:
(F
t
)
new
= (F
exp
/ F
R
n
) * F
t

Where (F
t
)
new
is the resultant muscle force required to produce F
exp
; F
exp
is the
experimentally measured force, in this case 318.15 N from Binder & Van Valkenburgh
(2000). F
R
n
is the resulting bite force from the initial arbitrarily chosen total muscle force,
F
t
.
The relative proportions of muscle activation between the temporalis and masseter,
and between the working and balancing side muscles in unilateral biting, can affect
model results. In all the models constructed, a 60% difference between the balancing side
and working side muscle activation was used. This value is based on Dessem’s (1989)
empirical data for domestic dog, which showed that during unilateral bone-crushing with
M1 the balancing side muscles acted at 60% of the maximum recorded
electromyographic activity. Although quantitative data have been collected on muscle
recruitment in cats (Gorniak and Gans 1980), the Dessem study included M1 bone-
275
crushing behavior, under which loading condition better approximates bone-cracking
than one feeding on soft tissue only (as in the cat study). Thus, the total required muscle
force to produce a bite force of 318.15 N was distributed with the balancing side muscle
force being 60% of working side muscle force. Next, the division of forces between the
temporalis and masseter muscles on each side was made proportional to the estimated
cross-sectional areas of the respective muscles, using photography protocol outlined in
the dry skull method (Thomason 1991, Wroe et al. 2005). Photos of dorsal and ventral
cranium were taken perpendicular to the plane of muscle cross-section as in Thomason
(1991), and the area of the plane measured using ImageJ (Rasband 1997-2007). A
summary of the muscle force values used in the analysis is included in the appendix.

Data analysis

The biting scenarios examined were (1) unilateral P3 biting, simulating a bone-
cracking bite and (2) unilateral P4 biting, simulating a shearing bite for all three models,
and (3) unilateral M1 biting, simulating a crushing molar bite in the wolf Canis lupus. All
scenarios were analyzed for both left and right unilateral biting, in order to identify any
asymmetric biases. In addition, the muscle forces derived from the P4 bite force
calculation were used to model both P3 and M1 biting. Three types of data were collected
from the models for each biting scenario: (1) scaled median Von Mises stress of
tetrahedral elements in the entire cranium, and their respective deviations, (2) scaled
median and maximum Von Mises stress of tetrahedral elements in the frontal-parietal
region (“frontal dome”) where morphological changes have been hypothesized to
represent functional adaptation, and (3) change in raw Von Mises stress of tetrahedral
276
elements along the sagittal plane of the frontal dome, representing changes in stress as it
is dissipated from the originating tooth to the rest of the cranium.
For the same amount of stress and volume, the models having larger number of
tetrahedral elements will have lower raw stress per element. Therefore, the contribution
of each element to the overall stress should be made proportional before comparisons are
made. Scaled median stresses were calculated after multiplying the tetrahedral element
stress results by their respective volumes and then dividing by the median tetrahedral
volume, i.e. to eliminate the effect of high stress simply from isolated small elements in
the solid mesh, thus leaving high stress correlated with model shape. The scaled values
were roughly linear to tetrahedral volume after transformation, thus differences between
models can be attributed to morphological difference in the models.
Because stress distribution in the models was expected to be highly skewed, with
most of the elements under little stress, and a few elements sustaining high stress, the
descriptive statistics employed include the median, standard error of the median (SE),
median absolute deviation from the median (MAD), and interquartile range (IQR). All of
the above statistics are robust measures, which are insensitive to outliers caused by model
singularities and sharp features; at the same time, these descriptive measures are
sufficient in using the entire dataset. Because of variation in model quality across
different specimens as well as fossil versus extant taxa, another robust measure of central
tendency, the trimmed mean (summing average of dataset by trimming a set percentage
from the ends of the data), was not used. The reason is difficulty in objectively
delineating singularities versus “real” data consistently in all three models. All statistical
277
summaries were calculated in the software program JMP IN (SAS Institute, Cary, North
Carolina, USA).
For mid-sagittal plane point sampling, data were collected in each of seven mid-
sagittal landmarks. Moving posteriorly, stresses from single nodes were sampled from the
mid-sagittal point at the position of (1) the mid-sagittal anterior edge of nasal bones, (2)
the central point of the infraorbital foramen, (3) the most anterior point of the orbits, (4)
the tip of the post-orbital processes, (5) the point of maximal post-orbital restriction of
the frontal-parietal region, (6) the anterior-most point of the sagittal crest, and (7) the
posterior-most point of the sagittal crest. The single nodes recorded stress at these
specific landmarks; each was sampled five times. In addition, mean stresses for ten nodes
within a circular area around each landmark, covering approximately 6 mm in diameter,
were sampled.

Chapter Six Results

Within-model comparisons

For Canis lupus, no consistent stress pattern could be discerned from values
measured on the entire cranium for the different biting scenarios (Table 6.1). Furthermore,
left and right biting with the same tooth position vary only slightly. P3 biting seems to
result in higher scaled median stress compared to P4 biting, but not M1. The dispersion
of stress values is comparable across all scenarios as described by the MAD and IQR.
Results from the frontal dome, however, show that the scaled median stresses for all
biting scenarios are at least doubled from the recorded values for the entire cranium. In
278
addition, the dispersion of stress values also increases in the frontal region. The
maximum stress is comparable for all biting scenarios. IQR increases slightly for scaled
stress in the frontal region. In the Dinocrocuta model, all stress measures are comparable
across biting scenarios for the entire cranium (Table 6.2). Only small differences were
found between left and right sides. The scaled median stress increases slightly for the
frontal dome. IQR decreases from the entire cranium to the frontal region in all cases.
Both the scaled and absolute maximum stresses are comparable across all scenarios. The
median scaled stress of the entire cranium has an overlapping range between P3 and P4
analyses in the Crocuta model (Table 6.3). Compared with this, the frontal dome has a
slightly increased median stress. The maximum stress in the inter-orbital region is
comparable for P3 and P4 biting. The IQR of scaled median stress increases slightly for
the frontal region for all scenarios except left P4 biting. Only slight differences were
found between left and right sides.

Between-model comparisons

For the entire cranium, median raw stress is highest in Canis and comparable in
the other two models for P3 and P4 biting (Figs 6.2 and 6.3). Scaled median stress of P4
biting is similar in Dinocrocuta and Canis, but higher during P3 biting for Canis (Tables
6.4 and 6.5). Crocuta had the lowest scaled median stress values for all biting scenarios.
The dispersion of stress values (MAD, IQR) of raw stress is highest in Canis and lowest
in Crocuta; however, the scaled stress dispersion is comparable between Dinocrocuta and
Canis, and somewhat lower in Crocuta. For the frontal dome, both raw and scale median
stress are lowest overall in Dinocrocuta; in the Canis model, the raw and scaled median
279
stress increased by 50% or more over those of the Dinocrocuta and Crocuta models. The
same trend is observed for both raw and scaled measures of dispersion (MAD, IQR). The
raw maximum stress of Crocuta frontal dome is around three times of that in
Dinocrocuta, and that of Canis is around five times as much as in Dinocrocuta. The
scaled maximum stress is highest in Crocuta, and lowest in Dinocrocuta. The highest
overall scaled maximum stress in Crocuta is during left P3 biting, and it is close to ten
times the scaled stress in Dinocrocuta.

Point sampling along mid-sagittal plane of dorsal cranium

Seven points were chosen to document the change in absolute stress along the
mid-sagittal plane, including the frontal region which is highly domed in Dinocrocuta
(Fig. 6.1). All five samples returned similar stress trends; one representative trend from
each biting scenario is presented in Figure 6.4A-C. In the Crocuta model, P3 biting
creates peak stress in the region between anterior borders of the orbits, whereas P4 biting
peaks between the post-orbital processes (Fig. 6.4A). In Dinocrocuta, the pattern is the
same as in Crocuta; the only difference is that the first peak at the anterior orbit boundary
is not as pronounced relative to the inter-orbital region of the post-orbital processes as in
Crocuta (Fig. 6.4B). In Canis, two peaks are present in P3 biting: one between the
infraorbital foramina, the other at the anterior tip of the sagittal crest (Fig. 6.4C). For P4
biting there are also two peaks, but the first one has shifted from infraorbital foramina to
inter-orbital region between post-orbital processes. M1 biting shows a similar pattern as
the P4 data for Canis. When the mean stresses along the identical landmarks are sampled
across a circular area of ten nodes, the same patterns are observed in Crocuta and
280
Dinocrocuta (Fig. 6.4D, E). However, in Canis the stress peaks previously observed
around the infraorbital foramina and the anterior tip of the sagittal crest became less
obvious in the mean sample, but P3 biting still exhibits more abrupt stress increases in
those regions (Fig. 6.4F).

Chapter Six Discussion

All within-model results demonstrate that, given the same muscle force input, P3
and P4 biting generate similar stress reactions in all three models, in addition to M1
biting in the Canis model. Although the more posteriorly placed teeth have more
mechanical advantage by leverage, the differences in stress distribution that might
represent adaptations to specific biting regimes (e.g. P3 bone cracking) are not obvious
from the analyses. Thus the crania of the three carnivorans studied here cannot be said to
have functional advantages for any particular biting scenario tested. It could be that the
teeth are simply too close in proximity for the analyses to detect differences in
performance, e.g. larger differences in stress magnitude and distribution are to be
expected for P4 versus canine biting in all cases by principle of lever mechanics. Another
likely explanation could be that the cranium, a product of complex selective pressure for
different functions (e.g. protection of brain, tuning of sensory organs, bite force, gape,
variation in dental function across tooth row), would not appear optimized for
specifically P3 bone-cracking, even if it represents a mechanically demanding task.

281
Canis lupus
In the Canis model, the median scaled stress of the interorbital region was more
than double that of the entire cranium. In part, this could be interpreted as the poor ability
of the inter-orbital region to dissipate stress in Canis. What was unexpected, again, is that
this doubling of median stress is observed for all three biting cases. One might expect the
inter-orbital region to respond differently when biting with the slender P3 versus biting
with the carnassial or the bone-crushing M1. By visually checking for artificial sharp
features and highly distorted triangular elements using error-checking functions in the
Strand7 Finite element software it was concluded that no major errors ensued during the
creation of the finite element model; thus, it is unlikely that some critical error in
modeling caused this pattern. Canine bite force estimations based on the dry skull method
suggest that Canis lupus has higher bite strength than Crocuta crocuta (Wroe et al., 2005).
The authors of that study also suggested, however, that bite strength does not necessarily
imply bone-cracking, which might be more closely associated with structural adaptations
in teeth and cranial bones. The findings from this study suggest structural adaptation in
the cranium of the bone cracking Crocuta compared to Canis for the scenarios tested.
The high stress observed in the dorsal cranium of Canis is in accordance with the
suggestion of Wroe et al (2005) that larger theoretical bite forces in Canis are not
necessarily realized because of cranial structure constraints to bone-cracking, in addition
to any structural improvement in teeth.
An important point which needs to be noted is that given the differences in cranial
morphology and masticatory muscle attachment sites, the skull of Canis required the
282
highest muscle input in order to produce the same bite force compared to the other two
models (Appendix). The high input force is probably what created the elevated stresses
observed throughout the Canis model. Furthermore, upon examining the regions of
highest stress in all three models, it was found that the ends of the zygomatic arches
generally exhibit relatively high stress values (Figs 6.2-3). The presence of these high
stresses are also likely to be partially attributed to muscle force action, as the downward
pull of the masseter muscles would tend to bend the arches in the ventro-medial direction.
Furthermore, the models constructed in this study do not contain any sutures, which have
been shown to represent sites of high strain and may be important in stress distribution
across cranial bones (Herring and Teng, 2000).

Crocuta and Dinocrocuta
Discussions of correlation between craniodental form and performance in
hyaenids have often pointed to the caudally extended frontal sinus present in members of
the group. The extent of the development of the posterior frontal sinus in hyaenids is
unique among carnivorans (Joeckel 1998), and functional relevance to stress dissipation
during bone cracking has been speculated (Werdelin 1989). More recently, it was shown
through examination of theoretical morphology, Joeckel’s (1998) hypothesis seems to
hold, at least for Crocuta crocuta (J. Tanner, personal communication). In the Crocuta
model, the thin bones delineating the frontal sinus are preserved and included; however,
the Dinocrocuta model has a single, continuous cavity inside the cranium. A caudally
extended frontal sinus is definitely present in Dinocrocuta, but too poorly preserved to be
283
included in the model (Fig. 6.5). Furthermore, the hypothesis of Joeckel (1998) places
functional significance on the formation of a shell-like forehead by the presence of
caudally elongate frontal sinus, not the presence of the sinus per se. Even though the
bony plate between the frontal sinus and brain cavity is incompletely preserved in
Dinocrocuta, it is not expected to be load-bearing because of its contact with the brain
roof in life. Thus, if the enlargement of the sinus is causal in creating the frontal dome of
the bone-cracking carnivorans, it could explain the similarity in stress distribution
between Dinocrocuta and Crocuta. The presence of frontal sinus structure in the Crocuta
model, however, created regions of concentrated stress in the bony struts surrounding the
sinus. The curvature is acute in some of the bony struts, thus stress does not conduct
smoothly through the area. A contributing factor to this result could be the lack of soft
tissue in the models. The frontal sinus of the domestic dog has been shown to contain a
covering of respiratory epithelium (Craven et al. 2007, Reznik 1990) in the same region
where the Crocuta model has concentrated stress (Fig. 6.5). A dissection of the cranium
of Hyaena hyaena (LACM freezer catalogue number 42206) confirmed the presence of
this epithelium throughout the frontal sinus of that hyaenid. Although the material
properties of the respiratory epithelium are not known, its close association with the inner
surface of the frontal sinus might nevertheless impart a certain degree of structural
continuum across which stress could be distributed. Further testing with incorporation of
soft tissue material into the model is needed to clarify whether the dissipation of stress
occurs in those small regions.

284
Mid-sagittal point-sampling stress
Clearer patterns arise when the mid-sagittal point-sampling data are graphed
across biting scenarios for each model (Fig. 6.4A-C). Both Crocuta and Dinocrocuta
show a much smoother increase in stress just caudal of the nasal opening and extending
to the sagittal crest. The down-sloping stress values caudal of the biting point in Crocuta
supports the hypothesis that the curvature of the cranium matches the path of stress
distribution to function in dissipation (Fig. 6.4A). A similar pattern is observed in
Dinocrocuta; however, it is less clear, with peaks distributed between the anterior orbital
border and the post-orbital constriction (Fig. 6.4B). In general, both the Dinocrocuta and
Crocuta models demonstrate gradual change in stress levels across the dorsal cranium,
roughly matching the shape of the frontal dome in their stress distribution patterns. In
stark contrast, point sampling of the Canis model shows multiple stress peaks across
different scenarios along the mid-sagittal plane. The anterior stress peaks demonstrate
that the Canis cranium is not as suited for P3 biting as Crocuta or Dinocrocuta (Fig. 6.4,
Table 6.4). A peak in stress could be better buffered by the cranium if it can be
transmitted caudally to the active temporalis muscles which are undergoing tension
during biting (Buckland-Wright 1978). The interpretation seems to stand for both the
Dinocrocuta and Crocuta models, and may explain the very low stresses in the region of
temporalis muscle action (Fig. 6.2).
Mid-sagittal sampling of P4 biting in Canis returned much lower stresses using
the right P4 than the left (Figs. 6.4C,F); reexamination of the results indicate that the
heightened stress in left P4 biting is not concentrated in small regions, but instead is a
285
general elevation of stress across the entire cranium. The same models were used for the
P3 and M1 scenarios as well, which did not return such asymmetry. It is unclear why this
difference exists; however, other than the magnitude, the general trend of the stress peaks
remains valid for all analyses, and does not affect the interpretation made here. Single
node stress gradients in the Canis model are congruent with theoretical expectation (Fig.
6.4C). The stress gradient along the mid-sagittal plane of the Canis model matches the
basic pattern of bending stress calculation from a beam model of a dry Canis skull
(Thomason 1991): there is a small peak in the region above the infraorbital foramina on
the mid-sagittal plane. The second, higher peak occurs near the region between the post-
orbital processes. No data were shown for the posterior cranium by Thomason (1991).
Although the absolute magnitude of stress is higher in Thomason’s calculations, it is of a
different biting scheme compared to this study, i.e. canine biting instead of P3-M1 biting.
This lends additional support to the models, and that the analyses demonstrate the
relatively less well “designed” cranium of the Canis when compared to the other two
models for the biting scenarios tested (Tables 6.4 and 6.5). With the added advantage of
zig-zag Hunter-Schreger enamel banding (Rensberger and Stefen 2006), which is not
present in Canis lupus (Stefen 1999), the robust craniodental morphology of Dinocrocuta
is thus likely to be both capable and functional for bone cracking.
An interesting pattern is revealed when single node stress is compared to mean
stress over a 6-mm diameter area in the point-sampling analysis. Whereas the stress
patterns for Crocuta and Dinocrocuta remain unchanged between the two sampling
methods (Figs 6.4A, B, D, E), the same cannot be said for Canis. In Canis, a steep slope
286
is still present leading posteriorly to the anterior border of the sagittal crest, but mean
stresses of nodes around a larger area at each landmark appear much smoother than the
single node stresses (Figs 6.4C, F). This could be explained by the presence of sharper
stress gradients in those regions in the Canis model compared to the other two models,
thereby allowing single node sampling to pick up local extremes. In both Crocuta and
Dinocrocuta, the increase in stress from the lateral sides toward the mid-sagittal plan is
visibly more gradual (Figs. 6.2A-B), whereas in the Canis model small patches of higher
stress appear more abruptly (Fig. 6.2C). Stresses collected from a group of nodes around
a landmark would thus tend to average out small areas of high stress. These findings
further suggest that the Canis skull model experiences not only elevated stress along the
entire mid-sagittal plane relative to the Crocuta and Dinocrocuta models (Tables 6.1-3),
it also shows steep and unevenly distributed stress gradients in the nasal and interorbital
region. These results provide potential avenues for validation and testing with in vivo
strain gauge experiments (Herring et al. 2001), which may help to explain additional
nuances in the differences between Crocuta and Canis observed here.

Broader implications and future directions
The Dinocrocuta cranium represents an individual with unworn permanent
dentition and incompletely developed sagittal crest and frontal dome. As the finite
element model of Dinocrocuta was made from a relatively young individual, the patterns
observed here might be affected by ontogeny. From observations made on specimens of
more mature individuals, the shape of the forehead becomes much more pronounced and
287
“vaulted”, and may have been better aligned with vertically oriented stress, thereby
channeling them dorsoposteriorly. Crocuta, on the other hand, shows relatively less
modification of the frontal shape through ontogeny. The smooth curvature in Crocuta
thus might persist throughout growth, altering the stress distribution curve to a lesser
degree than in Dinocrocuta (Fig. 6.4A, B).
Much previous theoretical work has been done on the mechanics of the
mammalian mandibles (Greaves 1982, 1983, 1985b, 2000), which firmly established the
“one third rule”. The rule provides that given all considerations to maximize bite force,
function, and mechanical stability, the resultant muscle force vector from the action of
masticatory muscles fall 30% of the way along the jaw length away from the jaw joint.
To maintain stability and prevent frequent torsional loading in the temporomandibular
joint, no biting should occur within 30% of the length from the jaw joint. In addition, the
proper occlusion required for the shearing carnassials (upper P4 and lower m1) places
evolutionary constraint on the location of those teeth and in turn the arrangement of other
cheek teeth relative to them (Savage, 1977). These are factors that might explain the
number, position, as well as the use of cheek teeth in Dinocrocuta. From attrition patterns
of tooth cusps observed in a sample of eight Dinocrocuta gigantea skulls from Gansu
Province, China, it can be concluded that all cheek teeth are used and worn, just as in
extant spotted hyenas. The cranium thus must respond to the overall biting function over
evolutionary time, and not just to the tooth doing the maximum amount of work (i.e. P3
bone cracking). This again would explain why there is little difference between biting
scenarios within the premolar toothrow.
288
The finite element approach has great potential in reconstructing past
ecomorphology by testing anatomically inferred form-function relationships (Rayfield,
2007). More specifically, the comparison of living and fossil taxa using this technique
holds promises for testing current functional hypotheses and refining ecomorphological
definitions. For example, subtleties in cranial mechanics of carnivorans (e.g. borophagine
canids, amphicyonids) and creodonts (e.g. hyaenodontids) that are inferred bone-crackers
could be elucidated in this manner by comparison to their closest living relatives (e.g.
caniform carnivorans) and with living bone-crackers (i.e. Crocuta crocuta). On the other
hand, the undifferentiated cranial response to premolar bites tested in this study points to
difficulties of identifying bone-cracking adaptations when the cranium might be more
generally adapted as a result of multiple evolutionary constraints. Thus, the application of
finite element analysis could be further refined to test new evolutionary questions that
stem from each additional analysis. Among the anticipated developments, (1) more
rigorous statistical testing techniques for analyzing finite element stress and strain data,
(2) a more fundamental understanding of ecomorphology by finite element analyses of
wide-ranging theoretical morphologies, and (4) the application of comparative finite
element analysis to an entire clade of closely related species to examine function in a
phylogenetic context will all continue to improve the utility of the finite element method
in our understanding of functional morphology evolution.

289
Chapter Six Conclusion

Analyses using three finite element models showed that the crania of Crocuta
crocuta and Dinocrocuta gigantea experienced lower stress for the same P3 and P4 biting
scenarios using identical bite force than Canis lupus. Differences in biting scenarios were
small within each model and the same holds true for stress dissipation in the frontal
region. Of the bone cracking carnivorans, Dinocrocuta experienced lower overall stress
in the inter-orbital region as well as lower maximum stress. Point sampling of stresses
along the mid-sagittal plane of the models demonstrate the ability of Dinocrocuta and
Crocuta crania in smoothly conducting stress into the inter-orbital region, probably
allowing the stress to be dissipated through the shell-like dorsal cranium and/or tension in
the temporalis muscles. This is in stark contrast with multiple peaks of stress in the Canis
model, which does not spread stress evenly. Through the examination of functional
morphology using finite element analysis, this study showed that morphology of the
frontal region plays an important role in conducting stress, regardless of premolar usage.
The actual capability of a bone-cracking individual may be balanced by a continuous
shift in dental and cranial mechanical advantage and requirements during its ontogeny.
Examination of purported bone-cracking taxa in other mammalian lineages would shed
light on the extent of functional similarity that underlies morphological convergence.

290
Chapter Six Acknowledgments

I thank Xiaoming Wang and Gary Takeuchi of LACM for discussion, guidance
and encouragement; Betsy Dumont, Sean Werle, and Ian Grosse for training and
hospitality during the Finite Element Analysis in Biology workshop in June 2007 at the
University of Massachusetts, Amherst; Betsy Dumont and Ian Grosse for access to their
Boneload program for modeling jaw musculature in the skull models; Michael McNitt-
Gray at UCLA Medical Center for CT scanning the specimens; Graham Slater at UCLA
for discussion and access to software; the editor and referees for their dedicated reading
of the manuscript and stimulating ideas that greatly improved the content of this paper;
Jim Dines at LACM for extant specimen loans; Zhanxiang Qiu at the Institute of
Vertebrate Paleontology and Paleoanthropology in Beijing, China for an extended loan of
the Dinocrocuta gigantea skull; Jill McNitt-Gray, Reyes Enciso, Henryk Flashner, and
Faizal Kamaruddin at the University of Southern California (USC) for comments and
help with software programs; Dinosaur Institute at LACM for research space. This
research was funded by a USC Zumberge grant, American Society of Mammalogists
grant-in-aid of research, and a National Science Foundation Graduate Research
Fellowship.

291
Chapter Six References
Binder WJ, and Van Valkenburgh B. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252: 273-283.

Buckland-Wright JC. 1978. Bone structure and the patterns of force transmission in the
cat skull (Felis catus). Journal of Morphology 155: 35-62.

Cowin SC. 1989. The mechanical properties of cortical bone tissue. In: Cowin SC, ed.
Bone Mechanics. Boca Baton, FL: CRC Press. 97-127.

Craven BA, Neuberger T, Paterson EG, Webb AG, Josephson EM, Morrison EE, and
Settles GS. 2007. Reconstruction and morphometric analysis of the nasal airway
of the dog (Canis familiaris) and implications regarding olfactory airflow. The
Anatomical Record 290: 1325-1340.

Dessem D. 1989. Interactions between jaw-muscle recruitment and jaw-joint forces in
Canis familiaris. Journal of Anatomy 164: 101-121.

Dumont ER, Piccirillo J, and Grosse IR. 2005. Finite-element analysis of biting behavior
and bone stress in the facial skeletons of bats. The Anatomical Record Part A
283A: 319-330.

Erickson GM, Catanese III J, and Keaveny TM. 2002. Evolution of the biomechanical
material properties of the femur. The Anatomical Record 268: 115-124.

Erxleben JCP. 1777. Systema regni animalis, Classis I, Mammalia, Lipsiae.

Ferretti MP. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience: 1-12.

Gorniak GC, and Gans C. 1980. Quantitative assay of electromyograms during
mastication in domestic cats (Felis catus). Journal of Morphology 163: 253-281.

Greaves WS. 1982. A mechanical limitation on the position of the jaw muscles of
mammals the one-third rule. Journal of Mammalogy 63: 261-266.

Greaves WS. 1983. A functional analysis of carnassial biting. Biological Journal of the
Linnean Society 20: 353-364.

Greaves WS. 1985. The generalized carnivore jaw. Zoological Journal of the Linnean
Society 85: 267-274.

292
Greaves WS. 2000. Location of the vector of jaw muscle force in mammals. Journal of
Morphology 243: 293-299.

Grosse I, Dumont ER, Coletta C, and Tolleson A. 2007. Techniques for modeling
muscle-induced forces in finite element models of skeletal structures. The
Anatomical Record 290: 1069-1088.

Herring SW, and Teng S. 2000. Strain in the braincase and its suture during function.
American Journal of Physical Anthropology 112: 575-593.

Herring SW, Rafferty KL, Liu ZJ, and Marshall CD. 2001. Jaw muscles and the skull in
mammals: the biomechanics of mastication. Comparative Biochemistry and
Physiology Part A 131: 207-219.

Irons B, and Ahmad S. 1980. Techniques of Finite Elements. John Wily & Sons, New
York.

Joeckel RM. 1998. Unique frontal sinuses in fossil and living Hyaenidae (Mammalian,
Carnivora): description and interpretation. Journal of Vertebrate Paleontology 18:
627-639.

Laitman JT. 2005. A microscope on vertebrate form and function: the power of finite
element analysis. The Anatomical Record Part A 283A: 251-252.

Linnaeus C. 1758. Systema naturae per regna tria naturae, secundum classes, ordines,
genera, species, cum characteribus, differentiis, synonymis, locis. Vol. 1: Regnum
animale. Editio decima, 1758 (12th edition of Linnaeus 1758). Societatis
Zoologicae Germanicae, Stockholm.

McHenry C, Wroe S, Clausen PD, Moreno K, and Cunningham E. 2007. Supermodeled
sabercat, predatory behavior in Smilodon fatalis revealed by high-resolution 3D
computer simulation. Proceedings of the National Academy of Sciences 104:
16010-16015.

Nowak RM. 1999. Walker's carnivores of the world. The Johns Hopkins University Press,
Baltimore.

Peterson J, and Dechow PC. 2003. Material properties of the human cranial vault and
zygoma. The Anatomical Record Part A 274A: 785-797.

Peterson J, Wang Q, and Dechow PC. 2006. Material properties of the dentate maxilla.
The Anatomical Record Part A 288A: 962-972.

293
Pidaparti RM, and Vogt A. 2002. Experimental investigation of Poisson's ratio as a
damage parameter for bone fatigue. Journal of Biomedical Materials Research 59:
282-287.

Rasband WS. 1997-2007. ImageJ. Bethesda, Maryland: National Institutes of Health.

Rayfield EJ. 2004. Cranial mechanics and feeding in Tyrannosaurus rex. Proceedings of
the Royal Society of London, Series B 271: 1451-1459.

Rayfield EJ. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35: 541-576.

Rayfield EJ, Norman DB, Jorner CC, Horner JR, Smith PM, Thomason JJ, and Upchurch
P. 2001. Cranial design and function in a large theropod dinosaur. Nature 409:
1033-1037.

Reilly DT, and Burstein AH. 1975. The elastic and ultimate properties of compact bone
tissue. Journal of Biomechanics 8: 393-405.

Rensberger JM. 1995. Determination of stresses in mammalian dental enamel and their
relevance to the interpretation of feeding behaviors in extinct taxa. In: Thomason
JJ, ed. Functional morphology in vertebrate paleontology. New York: Cambridge
University Press. 151-172.

Rensberger JM, and Stefen C. 2006. Functional differentiations of the microstructure in
the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A 278:
149-162.

Reznik GK. 1990. Comparative anatomy, physiology, and function of the upper
respiratory tract. Environmental Health Perspectives 85: 171-176.

Richmond BG, Wright BW, Grosse I, Dechow PC, and Ross CF, Spencer, M.A., Strait,
D.S. 2005. Finite element analysis in functional morphology. The Anatomical
Record Part A 283A: 259-274.

Ross CF. 2005. Finite element analysis in vertebrate biomechanics. The Anatomical
Record Part A 283A: 253-258.

Savage RJ. 1977. Evolution in carnivorous mammals. Palaeontology 20: 237-271.

Schlosser M. 1903. Die fossilen Säugethiere Chinas nebst einer Odontographie der
recenten Antilopen. Abhandlungen königlich bayerischen Akademie
Wissenschaften 22: 1-221.
294

Shahar R, Zaslansky P, Barak M, Friesem AA, Currey JD, and Weiner S. 2007.
Anisotropic Poisson's ratio and compression modulus of cortical bone determined
by speckle interferometry. Journal of Biomechanics 40: 252-264.

Stefen C. 1999. Enamel microstructure of recent and fossil Canidae (Carnivora:
Mammalia). Journal of Vertebrate Paleontology 19: 576-587.

Stefen C, and Rensberger JM. 2002. The specialized enamel structure of hyaenids
(Mammalia, Hyaenidae): description and development within the lineage -
including percrocutids. Zoologische Abhandlungen 52: 127-147.

Thomason JJ. 1991. Cranial strength in relation to estimate biting forces in some
mammals. Canadian Journal of Zoology 69: 2326-2333.

Van Valkenburgh B. 1988. Trophic diversity in past and present guilds of large predatory
mammals. Paleobiology 14: 155-173.

Van Valkenburgh B. 1989. Carnivore dental adaptations and diet: a study of trophic
diversity within guilds. In: Gittleman JL, ed. Carnivore behavior, ecology, and
evolution. New York: Cornell University Press. 410-436.

Van Valkenburgh B. 1990. Skeletal and dental predictors of body mass in carnivores. In:
Damuth J and MacFadden BJ, eds. Body Size in Mammalian Paleobiology:
Estimation and Biological Implications. Cambridge: Cambridge University Press.
181-205.

Van Valkenburgh B. 1999. Major patterns in the history of carnivorous mammals.
Annual Review of Earth and Planetary Science 27: 463-493.

Van Valkenburgh B. 2001. Chapter 5. The dog-eat-dog world of carnivores: a review of
past and present carnivore community dynamics. In: Stanford CB and Bunn HT,
eds. Meat eating and hominid evolution: Oxford Press. 101-121.

Wang Q, Strait DS, and Dechow PC. 2006. A comparison of cortical elastic properties in
the craniofacial skeletons of three primate species and its relevance to the study of
human evolution. Journal of Human Evolution 51: 375-382.

Werdelin L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15: 387-401.
295

Werdelin L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
In: Gittleman JL, ed. Carnivore behavior, ecology, and evolution. New York:
Cornell University Press. 582-624.

Wroe S, McHenry C, and Thomason JJ. 2005. Bite club: comparative bite force in big
biting mammals and the prediction of predatory behaviour in fossil taxa.
Proceedings of the Royal Society of London, Series B 272: 619-625.

296
Chapter Six Appendix. Summary of the three finite element models analyzed in the study.
The muscle cross-sectional areas were measured from posterodorsal and ventral view
photos as in Thomason (Thomason 1991). Balancing muscles were given 60% of the
magnitude of working muscle forces. The frontal dome region was identified by a section
of similar volume between the post-orbital processes of the three models.

Dinocrocuta Crocuta Canis
Body mass (kg)
‡
199.53 36.31 46.24
(Species body mass range; kg)
†
- 40~86 18~80

Model condylobasal length (mm) 198.15 209.91 275.66
(Actual condylobasal length; mm) 322.36 213.76 244.90
Model maximum width (mm) 147.40 137.45 163.58
(Actual max. width; mm) 251.72 139.56 142.66

4-noded tetrahedral elements 1,532,146 973,734 1,120,780
Total element volume (mm
3
) 324,610.7 326,625.0 324,513.3
Model skull mass (g)* 649.22 653.25 649.03
(Actual dry skull mass; g) - 477.85 335.50

Temporalis cross-section (mm
2
) 6.28 x10
3
2.42 x10
3
3.81 x10
3

Masseter cross-section (mm
2
) 5.67 x10
3
1.90 x10
3
2.93 x10
3

Temporalis contribution (%) 52.54 55.92 56.56
Masseter contribution (%) 47.46 44.08 43.44
Modeled bite force (N) 318.15 318.15 318.15
Required muscle force (N) 1,095.55 1,376.36 1,537.92
Total muscle plate area (mm
2
) 1.13x10
4
1.36x10
4
5.51x10
4

Balancing temporalis (N) 215.46 288.62 326.19
Balancing masseter (N) 194.62 227.51 250.53
Working temporalis (N) 359.10 481.04 543.65
Working masseter (N) 324.37 379.19 417.54
Frontal dome elements 14,432 22,043 16,391
Frontal dome volume (mm
3
) 4.40x10
3
4.43x10
3
4.59x10
3

‡
Estimated from condylobasal length using regression equations for >100 kg (Dinocrocuta)
and 10-100 kg (Crocuta and Canis) categories in Van Valkenburgh (1990)
†
Nowak (1999)
*Estimated using dog femur cortical bone density of 2 mg/mm
3
(Cowin, 1989)
297
Figure 6.1. Finite element model of Dinocrocuta gigantea with muscle insertion areas of
the temporalis and masseter created (dark gray areas) using the Boneload program
(Grosse et al. 2007). Anterior dorsolateral view. The length of the cranium is
approximately 322 mm. Other models were constructed similarly by demarcating regions
of temporalis and masseter attachment. Mandibles were included for identification of
resultant muscle force directions, and then removed before analyses were run.

298
Figure 6.2. Dorsal views of Von Mises (VM) stress distribution during left P3-biting
scenario in the cranium of A. Crocuta crocuta, B. Dinocrocuta gigantea, and C. Canis
lupus. All legends are scaled to have a range of 0~8 MPa for optimized visualization. The
deeper blue areas represent small or no stress and the red areas represent highly stressed
regions. White patches represent areas where stress exceeds 8 MPa. The crania are scaled
to approximately the same length in the figure. Right P3 biting and right and left P4
biting scenarios produced similar stress distributions that are not statistically different.

299
Figure 6.3. Ventral views of Von Mises (VM) stress distribution during left P3-biting
scenario in the cranium of A. Crocuta crocuta, B. Dinocrocuta gigantea, and C. Canis
lupus. Legends as in Figure 6.2.

300
Figure 6.4. Von Mises stress gradients from the anterior to posterior cranium along the
mid-sagittal plane in analogous anatomical sampling points. Stresses from single node
samples of A. Crocuta crocuta, B. Dinocrocuta gigantea, and C. Canis lupus and mean
stresses from node group samples of D. Crocuta crocuta, E. Dinocrocuta gigantea, and F.
Canis lupus are shown. The data points (from left to right) represent stress recorded along
the mid-sagittal plane in lateral alignment with (1) anterior border of nasal bones, (2)
infraorbital foramina, (3) anterior boundary of the orbits, (4) the inter-orbital region
between the post-orbital processes, (5) post-orbital restriction of the frontal-parietal
region, (6) anterior-most point of the sagittal crest, and (7) posterior-most point of the
sagittal crest. The points are plotted as percentages of skull condylobasal length (CBL) in
the anterior-posterior direction. Left side, filled symbol; right side, open symbol; P3
biting, solid line; P4 biting, dashed line; M1 biting, dotted line.

301
Figure 6.5. Computer tomography images of A. Crocuta crocuta, B. Dinocrocuta
gigantea, and C. Canis lupus taken as lateral views of the mid-sagittal section of the
cranium. The frontal sinus (fs) is indicated in all three crania, and caudal expansion of the
frontal sinus in C. crocuta and D. gigantea is noted by arrows. The internal cavities of the
D. gigantea specimen are filled with matrix, which is light grey in color in the frontal
sinus area. All crania are scaled in the figure to the same approximate length.

302
Table 6.1. Descriptive statistics of Von Mises stress in the Canis lupus finite element
model. Data shown are for both the entire cranium as well as the frontal region. Scenarios
tested include third premolar (P3), fourth premolar (P4), and first molar (M1) biting. Both
right and left side unilateral loading cases were analyzed. IQR=interquartile range;
MAD=median absolute deviation from the median; MPa=megapascal; SE=standard error
of the median. For definition of the terms see text.

Entire model
lP3 rP3 lP4 rP4 lM1 rM1
Median scaled ± SE (MPa) 0.5440±0
.0019
0.5061±0
.0018
0.4872±0
.0017
0.4257±0
.0015
0.5203±0
.0019
0.4447±0
.0017
Scaled MAD (MPa) 0.45 0.42 0.41 0.36 0.44 0.38
Scaled IQR (MPa) 1.28 1.20 1.17 1.01 1.26 1.12
Frontal dome
Median scaled ± SE (MPa) 1.2075±0
.0220
1.0956±0
.0194
1.1824±0
.0228
0.8423±0
0155
1.2435±0
.0245
1.1508±0
.0222
Scaled MAD (MPa) 0.78 0.68 0.77 0.53 0.82 0.74
Scaled IQR (MPa) 1.79 1.58 1.86 1.26 2.00 1.81
Maximum scaled (MPa) 18.33 19.41 20.45 16.07 21.49 23.02

303
Table 6.2. Descriptive statistics of Von Mises stress in the Dinocrocuta gigantea finite
element model. Data shown are for both the entire cranium as well as the frontal region.
Scenarios tested include third premolar (P3) and fourth premolar (P4) biting. Both right
and left side unilateral loading cases were analyzed. Abbreviations as in Table 6.1. For
definition of the terms see text.

Entire model
lP3 rP3 lP4 rP4
Median scaled ± SE (MPa) 0.4715± 0.4813± 0.4519± 0.5108±
Scaled MAD (MPa) 0.41 0.41 0.39 0.44
Scaled IQR (MPa) 1.24 1.20 1.17 1.28
Frontal dome
Median scaled ± SE (MPa) 0.7033± 0.6576± 0.6141± 0.5904±
Scaled MAD (MPa) 0.50 0.46 0.43 0.44
Scaled IQR (MPa) 1.18 1.07 1.02 1.04
Maximum scaled (MPa) 6.27 5.63 6.74 6.23

304
Table 6.3. Descriptive statistics of Von Mises stress in the Crocuta crocuta finite element
model. Data shown are for both the entire cranium as well as the frontal region. Scenarios
tested include third premolar (P3) and fourth premolar (P4) biting. Both right and left
side unilateral loading cases were analyzed. Abbreviations as in Table 6.1. For definition
of the terms see text.

Entire model
lP3 rP3 lP4 rP4
Median scaled ± SE (MPa) 0.4074±
0.0019
0.3333±
0.0017
0.3951±
0.0018
0.3704±
0.0017
Scaled MAD (MPa) 0.35 0.29 0.33 0.31
Scaled IQR (MPa) 1.21 1.06 1.11 1.09
Frontal dome
Median scaled ± SE (MPa) 0.6950±
0.0132
0.5862±
0.0118
0.6083±
0.0113
0.5769±
0.0116
Scaled MAD (MPa) 0.48 0.42 0.41 0.40
Scaled IQR (MPa) 1.25 1.12 1.07 1.10
Maximum scaled (MPa) 57.77 40.44 52.76 48.68

305
Table 6.4. Comparative statistics of Von Mises stress in the finite element models of the
Crocuta crocuta, Dinocrocuta gigantea, and Canis lupus skulls. The comparative data
are for left P3 unilateral biting. Abbreviations as in Table 6.1. For definitions of
descriptive statistics see text.

Entire Model
C. crocuta D. gigantea C. lupus
Median scaled ± SE (MPa) 0.4074±0.0019 0.4715±0.0016 0.5440±0.0019
Scaled MAD (MPa) 0.35 0.41 0.45
Scaled IQR (MPa) 1.21 1.24 1.28

Frontal dome
Median scaled ± SE (MPa) 0.6950±0.0132 0.7033±0.0154 1.2075±0.0220
Scaled MAD (MPa) 0.48 0.5 0.78
Scaled IQR (MPa) 1.25 1.18 1.79
Maximum scaled (MPa) 57.77 6.27 18.33

306
Table 6.5. Comparative statistics of Von Mises stress in the finite element models of the
Crocuta crocuta, Dinocrocuta gigantea, and Canis lupus skulls. The comparative data
are for left P4 unilateral biting. Abbreviations as in Table 6.1. For definitions of
descriptive statistics see text.

Entire Model
C. crocuta D. gigantea C. lupus
Median scaled ± SE (MPa) 0.3951±0.0018 0.4519±0.0015 0.4872±0.0017
Scaled MAD (MPa) 0.33 0.39 0.41
Scaled IQR (MPa) 1.11 1.17 1.17
Frontal dome
Median scaled ± SE (MPa) 0.6083±0.0113 0.6141±0.0133 1.1824±0.0228
Scaled MAD (MPa) 0.41 0.43 0.77
Scaled IQR (MPa) 1.07 1.02 1.86
Maximum scaled (MPa) 52.76 6.74 20.45

307
Chapter Seven: Study of mandibular function in the percrocutid Dinocrocuta gigantea

This chapter has been published as:

Tseng, Z. J., and W. J. Binder. 2010. Mandibular biomechanics of Crocuta crocuta,
Canis lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia).
Zoological Journal of Linnean Society 158:683-696. doi: 10.1111/j.1096-
3642.2009.00555.x.

A copy of the accepted manuscript begins on the next page.
308
Mandibular biomechanics of Crocuta crocuta, Canis lupus, and the late
Miocene Dinocrocuta gigantea (Carnivora, Mammalia)

Zhijie Jack Tseng
*,1,3
and Wendy J. Binder
2,3

1
Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, 3616 Trousdale Parkway, Los Angeles, California
90087, jack.tseng@usc.edu;
2
Biology Department, Loyola Marymount University, 1 LMU Drive, Los Angeles,
California 90045, wbinder@lmu.edu;
3
Natural History Museum of Los Angeles County, 900 Exposition Boulevard, Los
Angeles, California 90007

*
Corresponding author
309
Chapter Seven Abstract

The relative simplicity of the mandible and its functional integration with the
upper dentition in carnivorans makes it an ideal subject for functional morphological
studies. To compare the mandibular biomechanics of two convergently evolved bone-
cracking ecomorphologies, we use finite element modeling to analyze mandibular corpus
stress. The bone-cracking spotted hyena Crocuta crocuta is used as a living analog to the
late Miocene percrocutid Dinocrocuta gigantea, using the gray wolf Canis lupus as a
molar bone-crushing outgroup. Mandibular stress values during p3, p4, and m1 biting are
found to be lowest in C. crocuta, and elevated in both C. lupus and D. gigantea. However,
the stress-dissipation patterns of the pre-m1 corpus are similar between C. crocuta and D.
gigantea. Lastly, D. gigantea has a relatively weaker corpus at the post-m1 position than
either C. crocuta or C. lupus. These findings suggest that even though stress patterns are
similar between the bone-cracking ecomorphs, the extinct D. gigantea has a weaker
mandibular structure when performing a comparable bone-cracking task as in C. crocuta
because of its slender post-m1 corpus. Ontogeny could potentially play an important role
in strengthening the post-m1 corpus by growth in the dorsoventral axis, and continuous
increase in biting performance through adulthood in living C. crocuta suggests the
possibility of a relatively more delayed development to full bone-cracking capability in D.
gigantea.
310
Chapter Seven Introduction

Families and species of the mammal order Carnivora show diverse ecological
adaptations from herbivory to hypercarnivory (Werdelin 1996b), and this dietary
diversity is reflected in the specialization of craniodental morphology (Van Valkenburgh
1988, 1989a). Large carnivores have been categorized into guilds by their
ecomorphologies, with a range of killing strategies, and these include groups such bone
crackers, meat specialists, etc. (Biknevicius and Van Valkenburgh 1996). Perhaps the
most well-known and studied extant bone-cracking specialist is the spotted hyena
Crocuta crocuta, which is a dominant social predator in sub-Saharan Africa, and an adept
hunter which also has the ability to crack and consume bones of small and large prey
(Kruuk 1972a).
Common to many bone-cracking specialists, C. crocuta is characterized by a
robust cranium and hypertrophied dentition, in addition to high dorsoventral bending
strength in its mandibular corpus (Biknevicius and Ruff 1992). For a given mandibular
length, the mandibular corpus of C. crocuta is especially robust ventral and caudal to the
precarnassial bone-cracking premolars, which may serve to resist forces produced in
processing particularly hard food items such as bone. These traits appear to be functional
aspects of the bone-cracking ecomorphology.
In the fossil record, the major ecomorphologies within the carnivoran guild have
been continuously occupied throughout the Cenozoic (meat specialists, bone specialists,
omnivores, etc.), albeit by taxa with disparate phylogenetic affinity (Van Valkenburgh
311
1991, Werdelin 1989). The bone-cracking ecomorphology is found in different
phylogenetic groups, and this is particularly amenable to study as durophagous carnivores
are easily identified through their craniodental morphology which includes a robust skull
and highly domed frontal regions in the cranium, along with hypertrophied dentition
adapted for processing hard food (Stefen and Rensberger 2002, Werdelin 1989). Two
large lineages that have evolved these bone-cracking ecomorphologies include the
borophagine canids (Wang et al. 1999) and the Hyaenidae (Werdelin and Solounias
1991). A third group, more closely related to the hyaenid lineage, is a family of feliform
carnivorans of very large size with exceptional craniodental robustness (Qiu et al. 1988).
The largest species in this group, Dinocrocuta gigantea, remains enigmatic for the lack of
a complete post-cranial skeleton to reveal its locomotory mode and thus its ecological
niche.
As an example of specialization, the biomechanics of the skull of durophagous
carnivores are of particular interest as they demonstrate the resistance of the skull to
extreme food types, and can reveal constraints of what extant animals can consume as
well as indicators of the limits and adaptations of extinct carnivores. Studies of feeding
adaptation using mandibles have several advantages over those using crania. The
mandible is critical in producing and also resisting forces incurred during feeding, and
unlike the cranium is a single bone that is not under simultaneous selective constraints for
sensory functions and protection of the brain, etc. Furthermore, mandibles are a relatively
strong skeletal element that is better preserved and represented in the carnivoran fossil
record than complete crania, allowing a more complete understanding of
312
macroevolutionary patterns of morphological and functional change. Previous studies of
jaw geometry, mechanical advantage of jaw musculature, and bone stress and strain using
mandibular models have provided both experimental and theoretical foundations for
further inquiry into the masticatory biomechanics of mandibles (Dechow and Hylander
2000, Greaves 1982, 1985a, Hylander 1985, 1986, Ross et al. 2007).
For these reasons, we use the mandible as the element of comparison between the
living C. crocuta and the extinct D. gigantea bone-cracking ecomorphs. The purpose of
this study is to better understand the bone-cracking carnivoran ecomorphology
exemplified by the living spotted hyena, and compare it with an extinct durophagous
carnivore which evolved in a different lineage during the late Miocene epoch of Asia
(~11-10 million years ago). We evaluate the biomechanical similarities between
morphologically identified bone-crackers by examining the ability of the mandibular
corpus at the toothrow to resist and distribute stress incurred during several biting
scenarios.
To this purpose, we utilize a comparative finite element modeling approach to
analyze the relative stress and strain magnitudes and distributions in the mandibular
corpus of the two carnivorans. As demonstrated by Binder and Van Valkenburgh (2000),
the bite force of C. crocuta continues to increase through adult life, and is an indication
of persistent musculoskeletal growth and/or remodeling. The only suitable specimen of
the fossil percrocutid D. gigantea we could locate for this study represents that of a sub-
adult individual, judging from its partially erupted upper canines, lack of tooth wear on
the cheek teeth, unfused cranial sutures, and relatively small size compared to an
313
ontogenetic series of fossil skulls from contemporaneous sedimentary deposits in Gansu
Province, China. For the bone-cracking comparison, we decided to use a sub-adult C.
crocuta skull which also showed the above features of an immature individual.
The extant gray wolf Canis lupus was used as a control representing a non-bone-
cracking meat specialist. While C. lupus does consume bones, as defined by Werdelin
(1989), the gray wolf is a bone crusher, as it crushes bones with the first and/or second
set of molars, rather than the premolars. In fact, as C. lupus is hypothesized to be an
ecological equivalent of some basal bone-cracking hyaenids (Turner et al. 2008,
Werdelin and Solounias 1996a), this species is a particularly apt outgroup. We chose to
compare an adult C. lupus with subadult C. crocuta and D. gigantea because it makes for
a particularly conservative comparison as we know that it will not underestimate the jaw
strength and stresses of C. lupus in comparison with the larger species.
Finite element modeling is an engineering method used to approximate solutions
of real world complex physical problems, and has been developed and used in
mechanical engineering for decades. More recently, this method has begun to be used
effectively in understanding the functional adaptations and limitations of vertebrates. The
application of this technique to analyze 3D structures in vertebrates was recently
reviewed by Rayfield (2007). For analysis of vertebrate skulls, in particular, earlier work
on theropod dinosaurs using two-dimensional finite element modeling (Rayfield 2005)
has provided an initial example of the utility of this method in comparative functional
morphology. Subsequent work by McHenry et al (2006) on the theoretical rostrum
morphology of various crocodilian forms has also broadened the horizon of the
314
application to discover potential evolutionary pathways and more generalized adaptive
patterns. More recently, work on the mammalian frontal sinus by Farke (2008) and
Tanner et al (2008) have demonstrated uses of computer-modified morphologies in
detecting adaptive signals in current morphological structures. Others examining modern
and recently fossilized skulls have used multiple (heterogeneous) material properties and
in vivo muscle data to approximate actual stresses and strains (McHenry et al. 2007,
Moreno et al. 2008). Using a comparative approach of original morphologies with
technique developed in Dumont et al (2005) and Grosse et al (2007), we test a specific
hypothesis regarding the adaptiveness of the robust mandibular morphology of three
living and extinct carnivoran taxa: the bone-cracking ecomorphs C. crocuta and D.
gigantea have robust mandibular corpuses that resist dorsoventral bending stress incurred
during bone-cracking bites with the premolars, and thus would have lower stress than
molar bone-crushing C. lupus which does not consume large bones with the anterior
premolars.
Institutional Abbreviations: IVPP, Institute of Vertebrate Paleontology and
Paleoanthropology, Beijing, China; LACM[Mamm], mammalogy collection of the
Natural History Museum of Los Angeles County, Los Angeles, California, USA. Other
abbreviations: IQR, Interquartile range; MAD, median absolute deviation from median.

315
Chapter Seven Methods

The methodology utilized in this study is similar to that outlined in detail by
Tseng (in press-b). A shorter summary of the modeling-building protocol is included here
in the main text, but interested readers are referred to Chapter Seven Appendix for a more
technically detailed description. One mandible from each species was used, so that
interpretation of the analytical results could be restricted to the comparison of sagittal
bending in the dentate mandibular corpus of the comparative taxa.
Skulls of a C. crocuta (LACM[Mamm] 30655), C. lupus (LACM[Mamm] 23010),
and D. gigantea (IVPP V15649) were scanned with computer tomography (CT) at the
University of California, Los Angeles Medical Center (Fig. 7.1). The D. gigantea
specimen is the most complete known for this species; however, the unworn dentition and
unfused cranial frontal and parietal sutures indicate that it is a sub-adult individual.
Therefore, a sub-adult individual of C. crocuta was chosen to make the two bone-
cracking ecomorphs more comparable in terms of ontogenetic age. A fully adult C. lupus
skull was chosen to represent a large individual with full capability for hypercarnivory as
a non-bone-cracking control (for definition of bone-cracking see introduction). CT data
were imported into the Mimics medical imaging software (Materialise N.V., Leuven,
Belgium) and a digital reconstruction was created. The reconstruction was then smoothed
and checked for errors and artifacts of digitization in the rapid prototyping software
Geomagic Studio 10 (Geomagic, Inc., Research Triangle Park, North Carolina, USA).
The reconstruction was then built into a 3D element mesh in the finite element modeling
316
and analysis software Strand7 (G + D Computer Pty Ltd, Sydney, Australia), where
model parameters were assigned. The most important anatomical parameters were the
attachment sites of the temporalis, masseter, and internal pterygoid muscles, which were
given their relative forces with lines of action directed at sites of origination in the
cranium.
Muscle insertion sites were identified on the three models with comparison to a
dissection done on a Hyaena hyaena skull (LACM[Mamm] 97200). The masticatory
muscles were removed from the skull by carefully dissecting the fibers along the base of
the cranial origination sites and mandibular insertion sites. The attachment regions of the
temporalis, masseter, and internal pterygoid muscles were reproduced using modeling
clay on a plastic cast of the D. gigantea mandible to visually aid the delineation of those
sites on the computer models (Fig. 7.2). In addition, the relative force contributions of the
modeled muscles were assigned based on the relative mass of those muscles extracted
from the dissection specimen. The estimated mass of the temporalis muscle is 190 g,
extrapolated by the remaining temporalis muscle on the dissection specimen (76 g, ~40%
of total) after a veterinary necropsy removed the parietal region of the cranium to extract
the brain. The masseter muscle complex (superficial and deep) had a mass of 65 g; the
internal pterygoid had a mass of 18 g. A muscle contribution ratio of 69.6% temporalis to
23.8% masseter to 6.6% internal pterygoid was used in all three models. These ratios are
close to those reported by Turnbull (1970) from a C. lupus dissection (68.2% temporalis
to 24.2% masseter to 7.6% pterygoid). Relative muscle masses were used as a proxy for
the relative force production capability of the masticatory muscles; other studies utilize
317
physiological cross-sectional areas (PCSA) of the jaw muscles to estimate input force
(Ross et al. 2005), but as such data are lacking for carnivorans, we chose not to use PCSA
to estimate muscle forces. The implications of this assumption in our models are
discussed below in the context of the results.
Validation studies combining finite element models and in vivo strain data for
macaques have shown that the bones of the cranium are best modeled as elastically
orthotropic elastic material when using homogeneous material properties (Strait et al.
2005). As we restrict our study to the cortical region of a single mandibular bone, we
used material properties of an elastic isotropic material to represent the mandible models;
there is some previous evidence as to the validity of this approach (Ashman et al. 1985).
A Young’s modulus of 20 GPa and a Poisson’s ratio of 0.3 were used in all the models;
these values are typical of mammalian cortical bone (Erickson et al. 2002) and represent
a compromise between the more pliable bone and more brittle enamel which were
modeled as the same material. Because we are interested in the relative mechanical
behavior of the mandibles and not the exact response at boundaries of different materials,
multiple material properties were not utilized. Thus structures such as the connective
tissues between teeth and bone, and the cancellous bones were not included. One added
advantage of this simplified approach is to allow direct comparison between extant and
fossil jaws, since diagenesis prohibits direct observation of density differences in the
teeth and jaws which are proxies for assigning material properties in our methodology.
Three sets of analyses were conducted. In all analyses the models of C. lupus and
D. gigantea were standardized to the length of C. crocuta (Fig. 7.1). By standardizing
318
length, the analytic results can be interpreted in the context of functional differences by
partially removing body size influence (Biknevicius and Ruff 1992). The models were
constrained from movement at the mandibular symphysis, bite position, and the
temporomandibular joint to simulate a bone-cracking bite causing dorsoventral bending.
The analysis simulates an instantaneous event at static equilibrium, and for the purposes
of examining overall patterns of stress distribution in the corpus we did not include
mobility at the temporomandibular joint or mandibular symphysis. Under these
assumptions, we exclude the potential effects of torsion during unilateral biting when
balancing and working side muscles are differentially activated, and instead examined
mainly dorsoventral bending forces. In the first analysis, the extant taxa C. crocuta and C.
lupus, with models of their original morphology, were loaded with the same muscle force
of 5891.63 N. This input force is required to produce the maximum bite force at the lower
carnassial tooth (m1) of a 12-month (the estimated age of the skull used) old C. crocuta
calculated from the regression equation given by Binder and Van Valkenburgh (2000).
Biting at the p3, p4, and m1 were tested, respectively. The results of the analyses are
summarized using Von Mises stress and strain values, which are combined measures of
how close a material is to failure. The stress and strain values were standardized by first
multiplying the element values by their element volumes, then dividing by the median
element volume of each model. Median is used because it is a robust measure of central
tendency in non-parametric distributions such as the distribution of stress and strain
values (Zar 1999). Typical finite element models produce analytic results with a large
number of low-stress elements and a small number of high-stress elements, and thus are
319
poorly characterized by parametric summary statistics. Results for the second and third
analyses are presented in this way as well.
In the second analysis, models of all three taxa were modified so that the
medullary cavities of the corpuses are removed. This was done to directly compare the
fossil D. gigantea mandible, for which the mandibular medullary cavity could not be
distinguished, with the extant specimens. This set of analyses also allows the comparison
between the original and solid models of C. crocuta and C. lupus, providing an estimate
of how much the results of the solid D. gigantea might deviate from an analysis of the
original morphology, if it was preserved. All other parameters and input forces were held
constant as in the first analysis. In the third analysis, the three solid models were
modified so that a uniform bite force of 1000 N was produced at each of the three biting
positions (p3, p4, m1, respectively). This allowed the examination of the relative
efficiencies of the mandibles at distributing and resisting stress when an identical bite
force is produced. Based on these experimental parameters, a total of 18 test cases were
analyzed.

Chapter Seven Results

Analysis 1
For C. crocuta, the difference in median stress is much smaller between any two
biting scenarios than in the same comparisons in C. lupus, and this is reflected in the
stress distribution as well (Fig. 7.3). Median stress ranged from 5.58 MPa for m1 biting
320
to 7.01 MPa for p3 biting (Table 7.1). Median strain followed a similar increase with
0.00036 for m1 biting to 0.00045 for p3 biting (Table 7.1). For C. lupus p3 and p4 biting
produced similar magnitudes of stress and strain throughout the mandible, and m1 biting
produced lower overall values (Table 7.1). The same patterns hold true for values of
median absolute deviation from median (MAD) for both stress and strain of the two
models. The dispersion of stress values around the median (IQR, interquartile range) of
the C. crocuta model is roughly 50% of the corresponding values for C. lupus (Table 7.1).
For overall strain, the IQR for C. crocuta are less than half the value of C. lupus in all
biting scenarios. For the same amount of muscle input force, C. lupus produced higher
bite forces than C. crocuta at all three biting positions. Bite force increased sequentially
from p3 to m1 in C. lupus; in C. crocuta bite force is highest in m1 but lowest at p4
(Table 7.1).

Analysis 2
Stress magnitudes increased slightly for all biting scenarios when an identical set
of analyses were conducted on solid C. crocuta and C. lupus models (Table 7.2). Bite
forces at p3 and p4 increased significantly in the solid Crocuta model (145% and 245%,
respectively), but m1 bite force only increased slightly (8.5%). For C. lupus, elevation in
bite force in the solid model increased in magnitude across the board from p3 to m1 (16%,
25%, and 64%, respectively). Bite forces at p3 and p4 in Dinocrocuta are lower than
either Crocuta or Canis; the m1 bite force of Dinocrocuta is intermediate between
Crocuta and Canis. The percentage changes in stress from the original models to the
321
filled models are of increasing magnitude from the p3 to the m1 position for Canis (5%,
9%, and 13% for the three bite scenarios, respectively). In Crocuta, the increase in stress
is smallest at the p4 position (7%), which also exhibited the largest increase in bite force;
p3 experienced a 17% increase and m1 14% increase in median stress.

Analysis 3
At 1000N of p3 bite force, the mandible of C. crocuta had the lowest stress, and
D. gigantea the highest. The same pattern is true for 1000N of p4 bite force, but for the
m1 bite C. lupus had slightly lower stress than C. crocuta. For both D. gigantea and C.
lupus mandibular stress decreases from a p3 bite to a m1 bite, but for C. crocuta a p4 bite
produced lower stress than in both p3 and m1 bites. MAD is substantially higher (by a
factor of >4) in D. gigantea and C. lupus than in C. crocuta for p3 and p4 biting, but for
m1 C. lupus has the lowest MAD. D. gigantea and C. lupus share similarly elevated
levels of IQR compared to C. crocuta, but for m1 biting C. crocuta has a higher IQR than
the other two taxa.
The visualization of strain in the mandible models shows that high stress in D.
gigantea and C. lupus is most concentrated in the anterior-facing slope of the coronoid
process (Fig. 7.4E, F). The most anterior part of the mandible receives relatively little
strain; visible increase in mandibular corpus strain is seen posterior of the p3 bite point in
C. crocuta, immediately under the bite point in D. gigantea, and both anterior and
posterior of the bite point in C. lupus (Fig. 7.4D-F). In C. crocuta most of the strain is
concentrated ventral of the articular process on the buccal side; the lingual has more
322
widespread but also more even strain around the angular process (Fig. 7.4A, D). In D.
gigantea the entire mandibular corpus ventral and posterior of the p3 bite point has
elevated strain; the highest strained regions are in the dorsal and ventral regions of the
articular process, and the anterior coronoid process (Fig. 7.4E). On the lingual side, high
strain continues around both the anterior and posterior slope of the coronoid process, and
connects on the lingual face of the ascending ramus (Fig. 7.4B). Regions of lower strain
tapers off anterior of the m1, but the corpus below m1 still experiences elevated strain.
Regions of elevated strain in C. lupus extend both anterior and posterior of the corpus
below the p3 bite point (Fig. 7.4F); the corpus immediate below m1 has low strain.
Posterior to this point, the strains are high on the anterior face of the coronoid process,
and ventral of the articular process; the ascending ramus also shows higher strain than in
the other two models (Fig. 7.4D-F). On the lingual side, strain is elevated near the
mandibular symphysis, and gradually increases posterior of m1; the strain is even and
widespread posterior to this point, except for high strain in the center of the coronoid
process and near the angular process (Fig. 7.4C). In the mandibular corpus beneath the
cheek teeth C. crocuta has low and confined stress, whereas C. lupus and D. gigantea
have more widespread stress distribution in the region. The most stressed regions in all
three models are around the base of the ascending ramus.
For all three models, the median stress and strain experienced in the entire
mandible was comparable between p3 and p4 biting, and lower for m1 biting (Fig. 7.5).
This decrease in m1 biting is slight in C. crocuta, but more dramatic in C. lupus and D.
gigantea.
323
All interdental regions examined for C. crocuta (p3-p4, p4-m1, and post m1)
showed relatively little strain compared to the other two models (Fig. 7.6). A small strand
of elevated strain is visible in the post-m1 cross-section on the dorsal-buccal part of the
jaw (Fig. 7.6A). In D. gigantea, the entire buccal side of the p3-p4 and p4-m1 cross-
sections shows elevated strain; the post-m1 cross-section has the highest strain at the
dorsal-buccal edge, with the strain lower and spreading around both the dorsal and
ventral edge onto the lingual side of the mandible (Fig. 7.6B). In C. lupus, the entire
buccal side and the ventral-lingual side of the p3-p4 cross-section show elevated strain; at
the p4-m1 cross-section the strain is only slightly elevated, and is concentrated in the
ventral-buccal region (Fig. 7.6C). In the m1-m2 cross-section, the entire ventral side and
both ventral-lingual and ventral-buccal sides of the mandibular corpus has higher strain.
Overall, the stress and strain values decrease gradually from p3, p4, to m1 biting
for C. crocuta; they are comparable between p3 and p4 biting in C. lupus and D. gigantea,
and lower for m1 biting. The distribution of stress and strain values is less variable in C.
crocuta than in D. gigantea; both are less variable than C. lupus. In cross section, the
mandibular corpus stress is minimal in C. crocuta; in D. gigantea it is concentrated on
the buccal wall, and in C. lupus it is spread around the buccal, ventral, and lingual walls.

Chapter Seven Discussion

Whereas the mammalian cranium is under simultaneous selective pressure on its
sensory and mechanical functions, the mandibles are simpler in that their main function is
324
mastication. This makes the mandible a good system for studying evolutionary questions
regarding dietary adaptations because of its “single-purpose” function and morphological
simplicity in being a single unit of bone. Consistent with previous findings of strong
correlation between mandibular corpus strength and dietary adaptation (Biknevicius and
Ruff 1992, Therrien 2005), the profile of median strain across different biting scenarios
were more consistent in C. crocuta than in C. lupus (Fig. 7.5). Crocuta crocuta, a
habitual bone-cracking predator, demonstrates adaptations for using the p3, p4, or m1
positions equally well in its mandibular corpus structure. Canis lupus, on the other hand,
experienced visibly lower stress and strain in m1 biting than for p3 or p4 biting (Fig. 7.5).
This is likely associated with the use of premolar tooth use to crack bone in C. crocuta,
and the concentration on m1 function for shearing meat in C. lupus. The position of the
lower carnassial in C. crocuta is more posteriorly situated in the dentary than the same
tooth in C. lupus (Biknevicius 1996, Biknevicius and Ruff 1992); this is further
indication of functional differences in the tooth position for greatest force production in
the two species, respectively. Complementing previous data on inferred functional
differences in the biting behavior of the two species from dental morphology and
mandibular structure, our results are consistent with the known differences in C. Crocuta
and C. lupus, the former using premolars to crack bone, while the later uses post-
carnassial molars (including the talonid basin of m1) in processing hard foods.
The cross-section images of the C. crocuta and C. lupus mandible models further
demonstrate the correspondence between corpus strength and predator ecology. The
interdental cross-sections at p3-p4, p4-m1, and post-m1 of the C. crocuta jaw all show
325
relative little strain during p3 biting (Fig. 7.6A). Along with only minor changes in
overall strain when changing bite positions, the mechanical behavior of the mandible (i.e.
high strength throughout the length of the corpus) is consistent with how this species use
both the anterior premolars for bone-cracking, and the carnassial (m1) for meat
consumption. C. lupus, on the other hand, has the lowest strain during m1 biting (Fig.
7.5). This shows the emphasis of this hypercarnivore on the shearing component of the
dentition, as can be seen from the low stress in the entire region ventral of the carnassial
even during p3 biting (Fig. 7.4C). Surprisingly, in this context, the mandible of D.
gigantea is similar to C. lupus in that there is lowered stress and strain incurred during
m1 biting, but it is similar to C. crocuta in terms of similar strain distributions in the p3-
p4 and p4-m1 regions during p3 biting, as would be expected for a bone-cracking species
(Fig. 7.6B). The weakest region of the D. gigantea mandible is in the post-m1 regions,
where the corpus is most likely not yet fully developed in this individual. While it has
been noted in the past that D. gigantea may be a bone-cracker, this analysis functionally
demonstrates a similarity between the biomechanics of Crocuta and Dinocrocuta, giving
further evidence that jaw morphology does indeed demonstrate functionality. The lack of
post-carnassial molars in each may demonstrate a common result of the evolution of the
premolar as the bone cracking teeth in these groups. Another interesting line of
investigation in functional convergence would be to examine the cranial and mandibular
biomechanics of the North American borophagine canids, which would have used a
combination of the upper carnassial and the lower p4 and carnassial for bone-cracking
(Wang et al. 1999, Werdelin 1989).
326
Overall, the mandible of C. crocuta experiences much lower stress and strain
values than in both D. gigantea and C. lupus (Fig. 7.4; Tables 7.1 and 7.2). Given the
same length, C. crocuta is much better at resisting bending than the other two species.
This is unexpected because the cranium of D. gigantea has previously been shown to be
just as “well-designed” for a simulated bone-cracking bite as the spotted hyena (Tseng in
press-b). Even though Tseng (in press-b) used bone volume to standardize the cranium
models whereas this study uses standardized length, a series of checks run with volume-
standardized mandible models return similar results as to the ones shown for length-
standardization (not presented here). A recent study comparing the implications between
different scaling methods suggests that model surface area standardization allows stress
and strain energy to be directly compared given identical loading forces (G. Slater,
personal communication); we obtained similar results using surface area standardization
compared to those presented here, the major difference being smaller differences in
median stress and strain among the three models.
Our interpretation of the results does not change when comparing the original
morphology models to the fossil D. gigantea, or with models where the medullary cavity
is filled (Tables 7.1 and 7.2). This provides a degree of robustness to our results. Given
the same mandible, however, bite force increases in the filled models from the original
models. This is taken to represent the more efficient conduction of forces through a more
homogeneous bone medium in the filled mandibles. Given these results, it is interesting
to note that the medullary cavities of the C. crocuta and D. gigantea (where visible)
appear to be relatively smaller compared to C. lupus during our examination of the CT
327
images. The reduced medullary cavity in the bone-cracking carnivores might thus be a
modification to increase cortical volume within the mandibular corpus and more efficient
force transmission during biting.
As mentioned in the introduction, given the nearly fully erupted permanent
dentition and presence of sutures in the frontal and parietal bones, the D. gigantea
specimen used in this study was most likely a sub-adult, and the C. crocuta specimen was
chosen to roughly match the developmental stage of the D. gigantea specimen studied.
Nevertheless, the manner in which development occurs may not be the same in these
species. The difference in stress and strain in the mandibles of these two specimens could
represent differential ontogenetic paths of bone growth. Whereas the mandibular corpus
appears relatively deep and robust in C. crocuta, the D. gigantea mandibular corpus has a
relatively underdeveloped post-m1 region when compared to several adult specimens in a
late Miocene fossil fauna in northern China (Tseng, personal observation). This gives
support to a relative developmental delay in dorsoventral deepening of the posterior
mandibular corpus in the larger D. gigantea when compared with C. crocuta.
To summarize, our results do not unambiguously support our hypothesis that the
bone-cracking taxa share similar stress distributions and low stress levels in the
mandibular corpus during a simulated bone-cracking bite. Even though C. crocuta and D.
gigantea exhibit similar stress distribution patterns, the stress levels in D. gigantea are
more similar to C. lupus in all biting scenarios tested. This result may be explained by
ontogenetic differences in tooth use and mandibular shape in D. gigantea, and/or a
different masticatory strategy compared to C. crocuta.
328

Implications
Even though the results presented in this study show clear differences between
taxa representing different ecomorphologies, ontogeny appears to be an important factor
that requires further examination. For extant carnivorans, this would require analysis of
an ontogenetic series of mandibles using finite element modeling to reveal differences in
stress and strain distribution in the mandibular corpus through development. A better
understanding of the biomechanical changes in the masticatory apparatus of these
carnivorans around the weaning period would be especially useful for estimating the life
history of juvenile carnivorans in the fossil record. For example, if D. gigantea juvenile
individuals indeed had a relatively delayed development of the posterior mandibular
corpus compared to C. crocuta, it could imply either a prolonged weaning period (similar
to and perhaps extending beyond that of Crocuta), and/or a switch in prey preference
from smaller to larger species or individuals, if D. gigantea was similar to C. crocuta in
its bone-cracking behavior. The critical transitional period between a mixed deciduous-
permanent dentition to fully permanent dentition in C. crocuta occurs around the time of
weaning (Biknevicius 1996); it leaves room for speculation as to whether D. gigantea had
a similar constraint through ontogeny with the development of its mandibular corpus. In
addition, the social structure of each carnivore species could affect the growth phases of
the mandibular corpus which might be involved in agonistic interactions among
conspecifics (Biknevicius and Leigh 1997). This information becomes particularly useful
if the timing and duration of deciduous and permanent dental eruption in fossil taxa can
329
be incorporated (e.g. Feranec, 2004) to estimate absolute durations of juvenile dental
development. There is evidence that there are further morphological changes and
performance changes such as increasing bite force, which can occur even after the
complete eruption of adult teeth and body size (Benoit 2006, Binder and Van
Valkenburgh 2000). While these are difficult to detect using standard morphological
measurements, FEA may be sensitive enough to detect patterns that are associated with
these types of functional changes. Thus better understanding C. crocuta jaw
development may give greater insight into the understanding of fossil species such as D.
gigantea.
The analyses conducted in this study are linearly static, with full constraint of the
temporomandibular joints and the mandibular symphysis. A fruitful future direction
would be to incorporate some degree of mobility at the joints, and using a dynamic
analysis to examine additional dissipation or concentration of stress across these joints
(Greaves 1988). Furthermore, the stress distribution at different gapes can also be studied
in a dynamic analysis to reveal potential optima in gape angle and inferred prey bone size.
In conjunction with the addition of these modified constraints, the left and right
mandibles should be analyzed together to include the asymmetric effects of a unilateral
bite on the stress and strain of balancing versus working side corpus (Greaves 1983).
Lastly, the addition of heterogeneous material properties at least to the enamel of the
dentition would provide a more realistic model.
The comparison of our dissection muscle mass proportions to published estimates
for C. crocuta has some implications for bite force estimates in extinct taxa. The
330
assignment of muscle proportions from the H. hyaena dissection to the model of C. lupus
was a good approximation of the muscle mass proportions for C. lupus in Turnbull
(1970), which listed ratios of temporalis 68.2% to masseter 24.2% to pterygoid 7.6%
(5:3.2:1). As a check, all three models were also analyzed with these proportions, and the
resulting stress and strain were indistinguishable from those obtained using dissection
proportions. Both empirical values are dramatically different from the muscle cross-
sectional areas estimated by the dry skull method (Thomason 1991, Wroe et al. 2005),
which gave proportions of temporalis 55.92% to masseter 44.08 (1.3:1) for C. crocuta.
Given potential differences in muscle densities which might affect assignment of relative
muscle forces as estimated by muscle mass and cross-section area, one might expect
these percentages to not match. However, the published bite force estimate for C. crocuta
using the dry skull method (Wroe et al. 2005) is lower than forces measured in vivo
(Binder and Van Valkenburgh 2000) by up to 100%. Specifically, the contribution of the
temporalis muscle is higher from a dry skull estimate, and the masseter smaller. Since the
temporalis has a longer in-lever arm than the masseter, its underestimation tends to have
a larger effect on underestimating bite force, and thus may partially explain this
difference in muscle force estimates.

Chapter Seven Conclusion

Finite element models of Crocuta crocuta, Canis lupus, and Dinocrocuta
gigantea were constructed and used in a comparative biomechanical analysis of stress
331
and strain in the mandibular corpus. Results suggest that for a given mandibular length, C.
crocuta experiences much lower stress and strain than C. lupus or D. gigantea. The lower
stress and strain in C. crocuta is also more consistent across different biting scenarios,
indicating a mandibular corpus more generally adapted for unpredictable loads. C. lupus
and D. gigantea, on the other hand, showed lowest stress and strain when biting with the
m1 carnassial. However, the stress and strain distributions in D. gigantea during p3 and
p4 biting are more equivalent to C. crocuta than in C. lupus, suggesting that mandibular
corpus of D. gigantea is in part similarly adapted as in C. crocuta. The slight increase in
mandibular stress in filled corpus models from models with medullary cavities, coupled
with more significant increases in bite force, indicates that the thick cortical bone mass of
the bone-cracking taxa provides more efficient force transmission. Further examination
of ontogenetic series of C. lupus and C. crocuta could reveal life history differences that
reflect the divergent ecomorphologies represented in adult individuals, and serve as a
comparative basis for life history studies in extinct carnivoran taxa.

Chapter Seven Aknowledgments

We thank X. Wang (LACM) for helpful discussions on the structure of the study;
Z. Qiu (IVPP) for a loan of Dinocrocuta gigantea; J. Dines (LACM) for access to Canis
lupus and Crocuta crocuta; M. McNitt-Gray (UCLA) for scanning the specimen; B.
Dumont (UMass Amherst) provided the Boneload program for modeling muscle forces;
software purchase was supported by a USC Zumberge interdisciplinary grant and a grant-
332
in-aid from the American Society of Mammalogists; research was supported by a NSF
graduate research fellowship (ZJT).

333
Chapter Seven References
Ashman RB, Rosinia G, Cowin SC, Fontenot MG. 1985. The bone tissue of the canine
mandible is elastically isotropic. Journal of Biomechanics 18: 717-721.

Benoit M. 2006. Multiphasic allometric analysis in lions (Panthera leo): Life history
expressed through morphometrics. Journal of Vertebrate Paleontology 26: S41A.

Biknevicius AR. 1996. Functional discrimination in the masticatory apparatus of juvenile
and adult cougars (Puma concolor) and spotted hyenas (Crocuta crocuta).
Canadian Journal of Zoology 74: 1934-1942.

Biknevicius AR, Leigh SR. 1997. Patterns of growth of the mandibular corpus in spotted
hyenas (Crocuta crocuta) and cougars (Puma concolor). Zoological Journal of
Linnean Society 120: 139-161.

Biknevicius AR, Ruff CB. 1992. The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology 228:
479-507.

Biknevicius AR, Van Valkenburgh B. 1996. Design for killing: Craniodental adaptations
of predators Carnivore behavior, ecology, and evolution, Vol. 2. 393-428.

Binder WJ, Van Valkenburgh B. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252: 273-283.

Dechow PC, Hylander WL. 2000. Elastic properties and masticatory bone stress in the
macaque mandible. American Journal of Physical Anthropology 112: 553-574.

Dumont ER, Piccirillo J, Grosse IR. 2005. Finite-element analysis of biting behavior and
bone stress in the facial skeletons of bats. The Anatomical Record Part A 283A:
319-330.

Erickson GM, Catanese III J, Keaveny TM. 2002. Evolution of the biomechanical
material properties of the femur. The Anatomical Record 268: 115-124.

Farke AA. 2008. Frontal sinuses and head-butting in goats: a finite element analysis. The
Journal of Experimental Biology 211: 3085-3094.

Feranec RS. 2004. Isotopic evidence of saber-tooth development, growth rate, and diet
from the adult canine of Smilodon fatalis from Rancho La Brea. Palaeogeography
Palaeoclimatology Palaeoecology 206: 303-310.

334
Greaves WS. 1982. A mechanical limitation on the position of the jaw muscles of
mammals the one-third rule. Journal of Mammalogy 63: 261-266.

Greaves WS. 1983. A functional analysis of carnassial biting. Biological Journal of the
Linnean Society 20: 353-364.

Greaves WS. 1985. The generalized carnivore jaw. Zoological Journal of the Linnean
Society 85: 267-274.

Greaves WS. 1988. A Functional Consequence Of An Ossified Mandibular Symphysis.
American Journal of Physical Anthropology 77: 53-56.

Grosse I, Dumont ER, Coletta C, Tolleson A. 2007. Techniques for modeling muscle-
induced forces in finite element models of skeletal structures. The Anatomical
Record 290: 1069-1088.

Hylander WL. 1985. Mandibular function and biomechanical stress and scaling.
American Zoologist 25: 315-330.

Hylander WL. 1986. In-vivo bone strain as an indicator of masticatory bite force in
Macaca fascicularis. Archives of Oral Biology 31: 149-157.

Kruuk H. 1972. The spotted hyena: a study of predation and social behavior. The
University of Chicago Press: Chicago.

McHenry C, Clausen PD, Daniel WJT, Meers MB, Pendharkar A. 2006. Biomechanics of
the rostrum in crocodilians, a comparative analysis using finite element modeling.
The Anatomical Record Part A 288A: 827-849.

McHenry C, Wroe S, Clausen PD, Moreno K, Cunningham E. 2007. Supermodeled
sabercat, predatory behavior in Smilodon fatalis revealed by high-resolution 3D
computer simulation. Proceedings of the National Academy of Sciences 104:
16010-16015.

Moreno K, Wroe S, Clausen PD, McHenry C, D'Amore DC, Rayfield EJ, Cunningham E.
2008. Cranial performance in the Komodo dragon (Varanus komodoensis) as
revealed by high-resolution 3-D finite element analysis. Journal of Anatomy 212:
736-746.

Qiu Z-x, Xie J-y, Yan D-f. 1988. Discovery of the skull of Dinocrocuta gigantea.
Vertebrata PalAsiatica 26: 128-138.

335
Rayfield EJ. 2005. Aspects of comparative cranial mechanics in the theropod dinosaurs
Coelophysis, Allosaurus, and Tyrannosaurus. Zoological Journal of the Linnean
Society 144: 309-316.

Rayfield EJ. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35: 541-576.

Ross CF, Dharia R, Herring SW, Hylander WL, Liu Z-J, Rafferty KL, Ravosa MJ,
Williams SH. 2007. Modulation of mandibular loading and bite force in mammals
during mastication. Journal of Experimental Biology 210: 1046-1063.

Ross CF, Patel BA, Slice dE, Strait DS, Dechow PC, Richmond BG, Spencer MA. 2005.
Modeling masticatory muscle force in finite element analysis: sensitivity analysis
using principal coordinates analysis. The Anatomical Record Part A 283A: 288-
299.

Stefen C, Rensberger JM. 2002. The specialized enamel structure of hyaenids (Mammalia,
Hyaenidae): description and development within the lineage - including
percrocutids. Zoologische Abhandlungen 52: 127-147.

Strait DS, Wang Q, Dechow PC, Ross CF, Richmond BG, Spencer MA, Patel BA. 2005.
Modeling elastic properties in finite-element analysis: how much precision is
needed to produce an accurate model? The Anatomical Record Part A 283A: 275-
287.

Tanner JB, Dumont ER, Sakai ST, Lundrigan BL, Holekamp KE. 2008. Of arcs and
vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta).
Biological Journal of the Linnean Society 95: 246-255.

Therrien F. 2005. Mandibular force profiles of extant carnivorans and implications for the
feeding behaviour of extinct predators. Journal of the Zoological Society of
London 267: 249-270.

Thomason JJ. 1991. Cranial strength in relation to estimate biting forces in some
mammals. Canadian Journal of Zoology 69: 2326-2333.

Tseng ZJ. in press. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society.

Turnbull WD. 1970. Mammalian masticatory apparatus. Fieldiana: Geology 18: 149-356.

336
Turner A, Antón M, Werdelin L. 2008. Taxonomy and evolutionary patterns in the fossil
Hyaenidae of Europe. Geobios 41: 677-687.

Van Valkenburgh B. 1988. Trophic diversity in past and present guilds of large predatory
mammals. Paleobiology 14: 155-173.

Van Valkenburgh B. 1989. Carnivore dental adaptations and diet: a study of trophic
diversity within guilds In: Gittleman JL, ed. Carnivore behavior, ecology, and
evolution. New York: Cornell University Press. 410-436.

Van Valkenburgh B. 1991. Iterative evolution of hypercarnivory in canids (Mammalia:
Carnivora): evolutionary interactions among sympatric predators. Paleobiology
17: 340-362.

Wang X, Tedford RH, Taylor BE. 1999. Phylogenetic systematics of the Borophaginae
(Carnivora: Canidae). Bulletin of the American Museum of Natural History 243:
1-391.

Werdelin L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15: 387-401.

Werdelin L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
In: Gittleman JL, ed. Carnivore behavior, ecology, and evolution. New York:
Cornell University Press. 582-624.

Werdelin L, Solounias N. 1991. The Hyaenidae: taxonomy, systematics and evolution.
Fossils and Strata 30: 1-104.

Werdelin L, Solounias N. 1996. The evolutionary history of hyaenas in Europe and
western Asia during the Miocene. In: Bernor RL, Rietschel S and Mittmann W,
eds. The Evolution of Western Eurasian Miocene Mammal Faunas. New York:
Columbia University Press. 290-306.

Wroe S, McHenry C, Thomason JJ. 2005. Bite club: comparative bite force in big biting
mammals and the prediction of predatory behaviour in fossil taxa. Proceedings of
the Royal Society of London, Series B 272: 619-625.

Zar JH. 1999. Biostatistical Analysis. Prentice-Hall: Upper Saddle River, New Jersey.

337
Chapter Seven Appendix. Finite element modeling protocol

1. Reconstruction from CT data: The skulls were scanned with a Siemens Definition
64 scanner (Siemens Medical Solutions, Forcheim, Germany) at 0.6 mm slice thickness
and interslice distance. This produced 464 images for Crocuta crocuta, 499 for Canis
lupus, and 616 for Dinocrocuta gigantea. The images were exported in the DICOM file
format into Mimics. Within Mimics, the cranium and mandibles were segmented
separately, but a correct articulation position was maintained. For D. gigantea, the
mandibular cavity was filled with sediment in which the fossil was preserved. No
attempts were made to remove the sediments because of difficulty in distinguishing bone
from sediment. The segmented masks were simplified using Mimics’ remeshing function,
and the resulting 3D reconstruction exported as stereolithography (*.stl) files.
2. Reconstruction improvement and smoothing: the stereolithography files were then
imported into Geomagic Studio. The mandibular surface reconstructions were first
decimated in triangle count and then refined to represent 50,000 total triangles in C.
crocuta and C. lupus, and 100,000 triangles in D. gigantea. The “sandpaper” and “fill
holes” functions were used to remove sharp surface features that are artifacts from
segmentation and surface mesh generation. The reconstructions were checked for
intersecting triangles, and when no errors were found, the reconstructions were exported
as separate cranium and mandibles stereolithography files for each specimen.
3. Finite element model: the stereolithography files were then imported into Strand7,
and first cleaned by removing duplicate nodes that represent junctions of triangle corners
created in the ASCII stl format (as opposed to binary). The surface reconstructions were
338
then checked for errors in free triangle edges and t-junctions. If an error is found, the
reconstruction is modified in Geomagic Studio and then re-imported into Strand7. Once
the surface reconstructions were error-free, they were solid-meshed with 4-noded
tetrahedral elements. The Crocuta crocuta model had 483033 (original morphology) and
125892 (filled corpus) elements, respectively. The Dinocrocuta gigantea mandible model
had 1188858 elements, and the Canis lupus model had 429482 (original morphology) and
130693 (filled corpus elements, respectively. The cranium associated with each set of
mandibles was then imported into Strand7. The forces calculated for each muscle were
saved as a text file, which is imported into the Boneload program (Grosse et al. 2007),
where force vectors simulating muscles wrapping around the bone surface were created
over the entire surface of each muscle attachment sites. The material properties of E = 20
GPa and ρ = 0.3 were assigned, and the mandibular symphysis, bite position, and the
temporomandibular joint were fixed from any movement. A linear static analysis was
then run with each scenario. Calculation time per scenario is approximately 20 minutes
for C. crocuta and C. lupus, and 40 minutes for D. gigantea because of the higher degrees
of freedom created by the larger number of tetrahedral elements.

339
Figure 7.1. Photos of specimens used in the study. A, Crocuta crocuta (LACM[Mamm]
30655), left mandible; B, Dinocrocuta gigantea (IVPP V15649), right mandible; C,
Canis lupus (LACM[Mamm] 23010), left mandible. Specimens are scaled to
approximately the same length in figure. Scale bar over carnassial tooth (m1) equals 10
mm.

340
Figure 7.2. Muscle attachment sites on the mandible finite element models, with Crocuta
crocuta as an example. The light areas on top of the ascending ramus and in the
mandibular fossa are attachment sites for the temporalis. The light area on the angular
process is the attachment site of the masseter. The internal pterygoid attachment (not
shown) is on the medial side of the angular process.

341
Figure 7.3. Strain distributions in the mandible of Crocuta crocuta in A, p3; B, p4, and C,
m1 biting scenarios. Color spectrum represents strain magnitude, with blue as low strain
and white relatively high strain.

342
Figure 7.4. Comparison of strain distributions during a p3 biting scenario in Crocuta
crocuta (A, lingual; D, buccal), Dinocrocuta gigantea (B, lingual; E, buccal), and Canis
lupus (C, lingual; F, buccal).

343
Figure 7.5. Median strain values at different bite positions for Crocuta crocuta (open
diamond), Canis lupus (filled triangle), and Dinocrocuta gigantea (open square).

344
Figure 7.6. Cross-section strain profiles for (from left to right): p3-p4, p4-m1, and post-
m1 interdental spaces in A, Crocuta crocuta, B, Dinocrocuta gigantea, and C, Canis
lupus during a p3 bite. View is from rostral towards caudal; buccal is to the right.

345
Table 7.1. Von Mises stress and strain values and bite forces from unmodified finite
element models of Crocuta crocuta and Canis lupus with uniform muscle force input of
5891.63 N. Stresses are in Mpa, strain in μ ε, and bite force is in Newtons. For
abbreviations see text.

Stress Strain Bite
Median MAD IQR Median MAD IQR Force
p3 7.01 6.04 30.36 0.00045 0.00039 0.00197 334.6
C.
crocuta p4 6.69 5.77 29.6 0.00043 0.00038 0.00192 287.47
m1 5.58 4.87 25.56 0.00036 0.00032 0.00166 468.12

p3 23.68 20.29 70.23 0.00154 0.00133 0.00457 563.31
C. lupus p4 22.41 19.32 66.55 0.00146 0.00127 0.00433 725.83
m1 16.38 14.98 53.85 0.00105 0.00099 0.00351 906.63
346
Table 7.2. Von Mises stress values for the solid finite element models of C. crocuta, D.
gigantea, and C. lupus with an uniform muscle force input of 5891.63 N. Percent changes
of values compared to the original models are listed for the extant taxa. Median, MAD,
and IQR of the stresses are in MPa; bite force is in Newtons. For abbreviations see text.

Median MAD IQR
Bite
force %+Med %+MAD %+IQR %+Bite
p3 8.49 7.55 30.18 822.26 0.17 0.20 -0.01 0.59
C. crocuta p4 7.18 6.44 27.20 991.87 0.07 0.10 -0.09 0.71
m1 6.51 5.96 27.32 472.10 0.14 0.18 0.06 0.01

p3 22.96 20.49 61.21 528.06 - - - -
D.
gigantea p4 22.35 19.84 57.69 663.16 - - - -
m1 17.11 15.54 44.69 975.45 - - - -

p3 25.04 23.61 85.50 653.70 0.05 0.14 0.18 0.14
C. lupus p4 24.50 22.98 80.71 910.69 0.09 0.16 0.18 0.20
m1 18.84 18.24 70.79 1484.09 0.13 0.18 0.24 0.39

347
Table 7.3. Von Mises stress values for the filled and scaled finite element models of C.
crocuta, D. gigantea, and C. lupus. All models are analyzed with a bite force of 1000N at
the respective bite positions.

Median MAD IQR
p3 10.32 9.19 36.71
Crocuta crocuta p4 7.24 6.49 27.42
m1 13.80 12.63 57.87

p3 46.70 41.75 123.85
Dinocrocuta gigantea p4 35.60 31.75 92.20
m1 19.04 17.26 49.54

p3 38.82 36.61 132.00
Canis lupus p4 25.19 23.63 82.71
m1 12.86 12.45 48.45

348
Chapter Eight: An integrative framework for the study of form and function

This chapter is currently under revision as:

Tseng, Z. J. in revision. Testing adaptive hypotheses of convergence with functional
landscapes: A case study of bone-cracking hypercarnivores. Evolution

A copy of the submitted manuscript begins on the next page.
349
Testing adaptive hypotheses of convergence with functional landscapes:
a case study of bone-cracking hypercarnivores

Zhijie Jack Tseng
1,2,
*

1
Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, Los Angeles, California 90089, U.S.A.
2
Department of Vertebrate Paleontology, Natural History Museum of Los Angeles
County, 900 Exposition Boulevard, Los Angeles, California 90007, U.S.A.
*jtseng@nhm.org

Running Title: Theoretical morphology and functional landscapes

350
Chapter Eight Abstract

Morphological convergence is a well documented phenomenon in mammals, and
adaptive explanations are commonly employed to infer similar functions for convergent
characteristics. I present a conceptual framework that adopts aspects of theoretical
morphology and engineering optimization to test hypotheses about adaptive convergent
evolution. Bone-cracking ecomorphologies in Carnivora were used as a case study.
Previous research has shown that skull deepening and widening are major evolutionary
patterns in convergent bone-cracking canids and hyaenids. A two-dimensional design
space, with skull width-to-length and depth-to-length ratios as variables, was used to
examine optimized shapes for two functional properties: mechanical advantage (MA) and
strain energy (SE). Functionality of theoretical skull shapes was studied using finite
element analysis (FEA) and visualized as functional landscapes. The distribution of
actual skull shapes in the landscape followed a convergent trend of plesiomorphically
low-MA and moderate-SE skulls evolving towards higher-MA and moderate-SE skulls;
predictions were corroborated by FEA of 13 actual specimens. Nevertheless, regions
exist in the landscape where optimal high-MA and low-SE shapes are not represented by
existing species; their vacancy is consistent with phylogenetic constraint on skull shape at
higher taxonomic levels. Results highlight the interaction of biomechanical and non-
biomechanical factors in constraining skull shape to localized functional optima through
evolution.

351
Key Words: Finite Element Analysis, Borophagine, Canidae, Hyaenidae, biomechanics,
fossils
352
Chapter Eight Introduction

Convergent evolution is a prominent feature of mammalian evolution in the
Cenozoic, and many cases (e.g. convergently fossorial, arboreal, herbivorous, or
carnivorous forms) have become textbook examples demonstrating the concept in
evolutionary biology (Futuyma 1997). Morphological convergence is often interpreted as
adaptive based on the assumption of a close correspondence between form and function,
and for the precise reason that the traits appear in unrelated clades of species (Lauder
1995, 1996). This study addresses two questions about morphological convergence in
mammalian skull morphology: (1) Do morphologically convergent species actually share
similar functionality? (2) If so, do those morphologies occupy optimal adaptive peaks in
the adaptive landscape? These questions are explored with theoretical morphology and
finite element modeling in a case study of bone-cracking carnivorous mammals.
Adaptation, like convergent evolution, is a central concept in evolutionary biology
research. Studies of patterns and processes of adaptation on the macroevolutionary scale
often rely on morphological characters, essentially those that are preserved in the fossil
record. The concept of the fitness (or adaptive) landscape, as originally proposed to
visualize possible evolutionary pathways of genetic interactions, has been adopted as a
framework to examine morphology in evolutionary and ecological contexts (Arnold 2003,
McGhee 1980, Simpson 1944, Wright 1932). In a demonstration of the concept at its
extremes, Kauffman (1995) used simulations of hypothetical genetic interactions to
create two fitness landscapes, one (“Fujiyama” landscape) with a single adaptive peak,
353
and the other with a random distribution of equally adaptive peaks. Evolution (or more
specifically, the process of adaptation) is thought to proceed on intermediate landscapes
between those extremes, with differentially elevated adaptive peaks, some of which act as
“topological attractors” where examples of convergence can be sought (McGhee 1999,
Thomas and Reif 1993).
In conventional morphometric studies, examples of convergent morphological
evolution can be identified by evolutionary pathways that move toward each other in
empirical morphospace, a morphospace built using existing, observed morphological
diversity (McGhee 1999, Stayton 2006). However, convergent morphologies can also
evolve via parallel evolutionary pathways that do not exhibit obvious trends of such
movement in empirical morphospace (Tseng and Wang In press-b). The complex
craniodental system of vertebrates, particularly those of mammals, is subject to multiple
functional demands not only of mastication and food acquisition, but also a range of
sensory functions (Greaves 1985a, Savage 1977, Vaughan et al. 2000). Understanding
key evolutionary drivers of functional changes in such complex systems can be daunting,
although there is some evidence of modularity to indicate that certain complex features
evolved as integrated units (Goswami 2006a, b). To put the issue at hand as an analogy in
engineering optimization theory: the number of possible designs of an engineered tool is
proportional to the multiplicity of functions it is intended to serve; selective pressures on
multi-tasking biological structures may similarly have resulted in equally fit
morphologies on distinct (but comparable) adaptive peaks in a fitness landscape (Niklas
354
1997). This phenomenon of “many-to-one” form-function relationship has been
recognized as a major feature of adaptive evolution (Wainwright 2007).
With the aid of computer-based simulation tools, such questions can now be
addressed with the creation of fitness landscapes based on hypothetical morphospace
(McGhee 1999). The bulk of previous work on theoretical morphospace has been done in
studies of plants and invertebrate animals (McGhee, 1999 and references therein).
Complex mathematical models have been constructed to simulate growth patterns and
possible (but sometimes non-existent) morphotypes in a variety of organismal groups.
However, few studies have focused on constructing hypothetical morphospaces of
vertebrates, particularly mammals. One factor in the paucity of such studies may lie in the
large number of skeletal elements that exist in vertebrates, and the highly integrated
functionality of many larger animals. Such emergent properties make parameterization of
key morphological traits difficult. Nonetheless, the exploration of form and function
using empirical morphospace and simulation of morphotypes that occur in different
regions of such morphospace has already been proposed and explored in vertebrates
(O'Higgins et al. 2011). A subset of functional simulations currently rely on finite
element analysis (FEA), a technique which has gained wide use in the study of vertebrate
biomechanics, particularly on the craniodental system (Rayfield 2007, Ross 2005).
However, FEA has mostly been applied to studies of existing or fossil morphology, and
has not been used with emphasis on theoretical morphology (but see Preuschoft and
Witzel 2005). This study centers on exploring the union of functional simulations of
craniodental function using FEA with the study of theoretical morphology using adaptive
355
landscapes and hybrid morphospaces. A prominent example of convergent evolution in
the Cenozoic record of mammals, that of bone-cracking hyaenids and borophagine canids,
was used to demonstrate the utility of combining functional and theoretical approaches to
study evolutionary (and potentially adaptive) changes in morphology.

Finite element analysis
FEA was originally devised as a technique in the engineering profession, used in
the design process to conduct mechanical testing on discretized representations of real-
world objects. The term was coined by Clough (1960) for his specific applications in the
civil engineering field. In the past two decades, the application of FEA to studies of
vertebrate functional morphology has seen a notable increase, particularly in the study of
the craniodental system (Dumont et al. 2005, McHenry et al. 2007, Rayfield et al. 2001,
Ross 2005). Application to mammalian craniodental biomechanics has been applied in a
diverse range of research questions, from convergent evolution (Wroe et al. 2007),
ecological niche (Slater et al. 2009), bite force (Davis et al. 2010), to bone strain and
model validation (Ross et al. 2010), among others.
The initial input to biological FEA is the morphology of interest, either derived
from computer-generated models, photos of specimens, or more commonly, computer
tomography images (Rayfield 2007). Representations of the morphology in question are
modified and converted into element meshes, which are mathematical geometric
constructs of the original morphology. Material properties and boundary conditions are
assigned to the mesh model with values derived from experiments, or in the case of
356
extinct organisms, experimental values taken from closely related living taxa (Rayfield
2007). FE analysis software programs can then perform simulations of forces on the FE
model, returning results in the form of stresses, strains, and bite force. The process of
improving models of actual species, usually by digitally repairing incomplete areas of the
structure of interest, is amenable to manipulation and creation of non-existing, theoretical
shapes that can then be tested in the same way as a model of an actual species (Wroe et al.
2010).

Bone-cracking ecomorphology
Ecomorphologies are categories of ecological specialization, based on
characteristic morphological features inferred to be associated with specific functions.
The repetitive evolution of major ecomorphologies in carnivorous mammals is a key
feature of this mammalian group throughout their Cenozoic evolution (Van Valkenburgh
1988, 1999, 2007, Werdelin 1996b). As in stereotypical cat-like and dog-like carnivorans,
the hyena-like forms are hypercarnivores specialized in consumption of vertebrate flesh
(Van Valkenburgh 2007). These hyena-like forms also have robust craniodental
morphological features interpreted as adaptations for durophagy. Strong and bulbous
cheek teeth, deep and rounded foreheads, large, rugose parietal areas for jaw muscle
attachment, and robust zygomatic arches are the main features of bone-cracking
ecomorphologies (Van Valkenburgh 2007, Werdelin 1989). These morphological
features are associated with excellent bone-cracking capability in the extant spotted
hyenas (Binder and Van Valkenburgh 2000, Kruuk 1972b). The generally large-bodied
357
carnivorans that possess these morphological features have been identified in the fossil
record in Hyaenidae (Werdelin and Solounias 1991), borophagine canids (Wang et al.
1999), and Percrocutidae (Qiu et al. 1988, Tseng 2009, Tseng and Binder 2010).
Hyaenids and percrocutids are feliform carnivorans, with the majority of their
evolutionary record in the Old World (Werdelin and Solounias 1991). The earliest
records of both groups are found in middle Miocene deposits of Eurasia; percrocutids did
not survive beyond the Miocene, whereas hyaenids are known today by four species,
composing the smallest living carnivoran family (Nowak 1999). Distinct morphological
differences between percrocutids and hyaenids, which have been proposed to be sister
groups, are established at their earliest occurrences (Chen and Schmidt-Kittler 1983). The
evolution of true hyaenids was quite gradual, with sequential appearance of six
ecomorphological categories through their ~25 m.y. fossil record (Turner et al. 2008,
Werdelin and Solounias 1991). In contrast, the fragmentary fossil record of percrocutids
is currently lacking a comprehensive phylogenetic framework. Nevertheless, it is clear
that the most robust forms in either lineage, the hyaenines and the percrocutid
Dinocrocuta, respectively, possessed capability for bone-cracking comparable to, or
exceeding, the modern spotted hyena (Crocuta crocuta)(Tseng 2009, Tseng et al. 2011a).
Canidae are North American natives, evolving into three subfamilies that
represent some of the most common fossil carnivorans to be found in the Tertiary:
Hesperocyoninae, Borophaginae, and Caninae (Tedford et al. 2009, Wang 1994, Wang et
al. 1999). All modern canids belong in Caninae, with no surviving species from the other
two subfamilies (Wang et al. 2008). Borophaginae contain the most hyena-like canids,
358
some of which have long been considered ecological vicars of true Old World hyaenids
(Van Valkenburgh et al. 2003a, Wang et al. 1999, Werdelin 1989). Craniodental function
in the most specialized borophagine canids has also been shown to allow improved bone-
cracking capability (Tseng and Wang 2010). Furthermore, the bone-cracking
ecomorphologies in the borophagine canids evolved derived craniodental morphology via
parallel evolutionary pathways alongside the macroevlutionary patterns observed in
hyaenids (Fig. 8.1; Tseng and Wang, in press). Such extensive convergence in
craniodental morphology and inferred functional capability proceeded under a complex
interplay of adaptation and constraint (Holliday and Steppan 2004, Tseng and Wang In
press-b, Werdelin 1989).
Considering the evolutionary patterns observed previously for borophagine canids,
hyaenids, and percrocutids, I test the hypothesis that bone-cracking ecomorphologies
were specialized forms that converged on identical or equivalent adaptive peaks on an
adaptive landscape. Form and function are closely linked, and the parallel evolution in
skull shape changes shared by bone-cracking ecomorphologies is reflected in their
functional similarities. Secondly, convergently evolved specialist species in both lineages
represent optimal skull shapes within a theoretically possible range of variable
morphologies, and therefore occupy adaptive peaks in the theoretical morphospace. A
functional landscape constructed using principles of functional morphology and
theoretical morphology is presented as a framework to test these hypotheses. The general
utility of such approach is then demonstrated by tracking evolution of craniodental
359
function, as inferred from FEA simulations, of actual fossil and extant species in the
convergent lineages.

Chapter Eight Materials and Methods

Hybrid morphospace
Strictly speaking, a theoretical morphospace, as defined by McGhee (1999), is
constructed without any morphometric input from actual specimens. The geometric
shapes of organismal morphology are created using mathematical models, spanning a
range that encompasses morphological representations which do not exist in the known
organismal record (McGhee 1999). In contrast, the hybrid morphospace used this
analysis was constructed with an actual ecomorphology in the Hyaenidae: the jackal-like
Ictitherium (Turner et al. 2008, Werdelin and Solounias 1991). A two-dimensional
morphospace was used in conjunction with two functional properties (sensu Wainwright
2007) described below to create functional landscapes. The morphological parameters
were chosen to represent the main axes of evolutionary skull shape changes observed in
both the Hyaenidae and the borophagine canids (Fig. 8.1), which exhibited parallel
evolutionary pathways of change through time towards bone-cracking ecomorphologies
(Tseng and Wang In press-b). These axes are relative skull width (width-to-length ratio,
W:L) and relative skull depth (depth-to-length ratio, D:L).
During the evolution of the hyaenid and borophagine canid lineages, species
evolved from relatively long-snouted, shallow- and narrow-skulled forms to short-
360
snouted, deep- and wide-skulled, highly robust forms (Tseng and Wang In press-b, Wang
et al. 1999, Werdelin and Solounias 1991). These general skull shape changes are
associated with the increased biomechanical capability of the larger and more robust
species to consume hard foods (Tseng 2009, Tseng et al. 2011a, Tseng and Wang 2010).
The exact causal links between incremental morphological changes and functional
improvements are not known (and therefore the morphospaces created here are
hypothetical in nature), but mechanical functions of specific craniodental features in
bone-cracking ecomorphologies have been proposed (Joeckel 1998, Rensberger and
Stefen 2006, Werdelin 1989). Among these features are the development of a dome
forehead and enlarged masticatory muscles, which are manifested in relatively deeper and
wider skulls, respectively (Tanner et al. 2010). Accordingly, hybrid morphospaces were
created to encompass and extend this range of observed evolutionary trends. Both
morphological parameters were altered from the plesiomorphic state seen in the skull of
Ictitherium by (1) increasing dorsoventral skull depth relative to skull length, and (2)
increasing lateral skull width in increments of 25% up to 200% deviation from the
Ictitherium specimen. To examine skull functionality that fall below this area (i.e. in a
more ancestral morphology), relative width and depth ratios of 75% from Ictitherium
were also examined.
The two main axes of cranial shape change identified above formed a two-
dimensional morphospace analogous to Raup’s (1966) classic “cube” of geometric
parameters of shell coiling (Fig. 8.2). The cranial parameters used here, however, do not
constitute theoretical morphospace in the strict sense; the directions, and to a certain
361
extent the ranges, of each morphological axis were chosen for analysis based on previous
work using empirical morphospace (Tseng and Wang In press-b).

Measures of Function
Conventional adaptive landscapes rely conceptually on direct measures of
performance, survival and reproduction; such measures are dependent on environmental
and ecological conditions at the specific temporal and spatial scale being examined
(Arnold 2003, Wainwright 2007). The creation of a functional landscape, as defined here,
aimed to measure more universal features of craniodental systems based on biomechanics.
Bite force, regardless of the means for its estimation in living and extinct organisms, is
one parameter that is crucial for vertebrates in both prey apprehension and mastication
(Meers 2002, Wroe et al. 2005). It is particularly important for bone-cracking
ecomorphologies, as bite force is one direct determinant of the size of prey bone that can
be consumed (Binder et al. 2002, Binder and Van Valkenburgh 2000). Thus, the bite
force performance of fossil and living carnivorans is expected to be of major importance
to the fitness of bone-cracking lineages, as a major parameter of fitness.
Similarly, it has been argued that skull strain energy, a measure of the work done
in deformation of an object in FEA simulations, is a suitable measure of functional
efficiency (Dumont et al. 2009). This argument is based on the suggestion that biological
objects (e.g. skulls) with maximum stiffness for a given volume of material (i.e. low
strain energy during deformation) should be favored by selective processes that maximize
functionality (Dumont et al. 2009). Skull strain energy is used as a second axis of
362
function in this study in addition to bite force. The skulls of species in bone-cracking
lineages are expected to be selected for increased stiffness per amount of skull bone, in
order to perform the intensive bone-cracking behavior which places large amounts of
stress and strain on the skull when compared to soft-food mastication (Dessem 1989).
These two functional properties are examined in hypothetical morphotypes that vary
along the morphological parameters described above, forming a landscape of
functionality with which functional evolution in actual lineages can be compared (Fig.
8.3).

Functional landscape
Analogous to an adaptive landscape, where the third-dimension is a fitness axis
used to document adaptive peaks and valleys over a bivariate plot of morphological
parameters (McGhee 1999), a functional landscape charts functional properties measured
by a biomechanical axis over the bivariate plot of morphological parameters.
The incremental changes in morphological parameters of the hypothetical
morphotypes that show variation along the same directions as observed empirically in
hyaenids and canids (Fig. 8.2). However, the resulting morphologies also embody general
variation in skull shape among other carnivorous mammals. The association between
parameters of skull shape and the ecological habits of extant carnivorans has been
demonstrated in empirical morphospaces created by geometric morphometrics analyses
(Meloro et al. 2008, Wroe and Milne 2007). Conceptually, the morphological parameters
used in this particular study can be supplemented with other functionally relevant
363
parameters which are particular to the research question being addressed. Similarly, there
may be other functional properties in addition to bite force and skull strain energy that are
relevant to the specific type of functional morphology being examined. In its basic
concept, the functional landscape is a functional manifestation of an adaptive landscape,
its fitness axis (commonly the z-axis) having been modified to measure aspects of
biomechanical function, which underlies organismal performance in relevant tasks.

Generation of theoretical models
Theoretical morphotypes representing incremental deviation of the two
morphological parameters from Ictitherium were generated by modification of an
Ictitherium digital skull model. A complete and intact skull of the late Miocene hyaenid
Ictitherium sp. (HMV0163, Hezheng Paleozoology Museum, Gansu Province, China)
was scanned using computer tomography (CT) at Lanzhou University Hospital No. 1
(Gansu Province, China) with a Siemens Somatom Sensation 64 scanner (120 KV,
304.00 mAs); images had a pixel size of 0.2578 mm, resolution 512x512 pixels, and 0.36
mm interslice distance. Data were exported in the DICOM (Digital Imaging and
Communications in Medicine) format. The cranium and mandible of the specimen were
separated and digitized using the software program Mimics 13 (Materialise NV). Digital
reconstructions, including internal morphology, were exported as stereolithography
format (*.stl). The files were then read into Geomagic Studio 10 (Geomagic, Inc.) where
generation of theoretical morphotypes took place.
364
Skull depth and width in theoretical morphotypes were changed by scaling the
original digital model of Ictitherium in the respective axes by a set percentage (75%-
200% of original). The axes of shape change were aligned so that depth increased along
the line connecting the carnassial tooth and the top of the frontal dome (Figs. 8.1-2);
width increased along the axis perpendicular to the long axis of the skull. The modified
theoretical morphotypes were then exported into Strand7 2.3.7 finite element analysis
software program (G + D Computer Pty Ltd), where finite element meshes were
generated.
The finite element meshes representing different morphotypes were modeled with
identical forces, material properties, and boundary conditions. As the fourth premolar
(carnassial tooth) represents a synapomorphy of Carnivora for shearing and masticating
meat, all models simulated unilateral bites with the upper carnassial. Only the cranium
was analyzed; the dentaries were used for reference only (see below). Three jaw-closing
muscle groups were modeled: temporalis, masseter, and pterygoid. The relative
contributions of the muscle groups to total input force were set at 67% (Temporalis), 22%
(Masseter), and 11% (Pterygoid); these values were based on wet weight of the relative
muscles in modern Crocuta crocuta (Tseng and Stynder 2011). Proportions of 64%, 22%,
and 11% have been reported for canids (Davis 1955, Slater et al. 2009, Turnbull 1970);
the small differences between hyaenids and canids were assumed to be negligible for the
model results studied, and the construction of models from actual specimens (including
canids) used the first set of percentages for consistency. Muscle activation on the
balancing (non-biting) side cranium was adjusted to 60% of the total input force on the
365
working (biting) side cranium; ratios across the muscle groups remained the same
(Dessem 1989). Force vectors within each muscle attachment area were divided evenly
over the entire area, with adjustment for wrapping of musculature around the cranial
muscle attachment sites using the Boneload program (Grosse et al. 2007). Muscle force
vectors in the respective muscle groups were oriented toward centroids of each muscle
group at the attachment sites on the corresponding dentaries which were identified using
anatomical texts and dissections. A gape of 30 degrees was simulated for all models,
close to the optimal angle present in Canis lupus dingo (Bourke et al. 2008b). A total of
39,820 N of input muscle force was simulated in all models, and the output bite force was
calculated as mechanical advantage (output force / input force) with a maximum range of
0.0 (no output force) to 1.0 (output force = input force). All models were also adjusted to
have identical total surface areas (1 x 10
6
mm
2
), to allow comparison of performance
variables among theoretical morphotypes as a function of shape changes, but not size
(Dumont et al. 2009). This particular ratio of input force (39,820 N) to surface area ratio
(1 x 10
6
mm
2
) matched the ratio used by Tseng and Wang (2010), which was derived
from the force-surface area ratio in their Canis lupus model that simulated maximal
measured bite force in Canis familiaris (Ellis et al. 2008). Therefore, the models
portrayed an empirically measured skull function.
Three nodal constraints were placed on the cranium models: the left and right
temporomandibular joints (TMJ), and the unilateral bite point. The bite point was
modeled as a nodal constraint fixed from all translational and rotational movements,
located at the tip of the paracone cusp on the carnassial. The TMJ was modeled as a
366
single nodal constraint in the middle of each glenoid fossa, fixed from all but rotational
movement in the sagittal plane. All models were given a single set of material properties
representing typical values for mammalian cortical bone. All analyses were linear and
static, therefore only two of three material parameters were required: Young’s (Elastic)
modulus = 20 GPa, and Poisson’s ratio = 0.3. Heterogeneous models that contain
multiple material properties have been shown to have higher stresses and bite forces
compared to identical models made with a single set of material properties; such
differences in results are acknowledged, but they were assumed to have no great effect on
the comparative context being pursued in this study (McHenry et al. 2007, Tseng et al.
2011a, Tseng and Wang 2010).
Bite force output is measured as mechanical advantage, or output force divided by
input force (MA). Skull strain energy (SE) values, in Joules, were extracted along with
MA from the analyses, and plotted against bivariate plots of the two morphological
parameters (D:L and W:L ratios). Wireframe plots were constructed to represent the
functional landscape, upon which simulation results from models of actual hyaenid and
borophagine species were plotted (Fig. 8.3).

Skull dimensions of actual species
To use the functional landscape in predictions of functional evolution in actual
lineages, the W:L and D:L ratios of actual hyaenids and canid species were measured
from specimen photos. The dataset of fossil and extant hyaenids and canids from Tseng
and Wang (In press-b) was used. Nine hyaenid species and 15 canid species were
367
complete enough to be measured. Means were used where sample size > 1, and
specimens that were used in FEA (see below) were plotted individually (Table 8.1).

Models of actual species
FE models of actual fossil and extant species of Hyaenidae and borophagine
canids were constructed as described above for the theoretical models. Most of the
models were existing ones taken from previous studies (Tseng 2009, Tseng et al. 2011a,
Tseng and Stynder 2011, Tseng and Wang 2010). Bite force and skull strain energy
values were obtained from analyses after all models were standardized so that the ratios
of total muscle input force to total model surface area were kept constant across all
models (Dumont et al. 2009). Such standardization allowed the absolute size of models to
be removed, and comparisons of skull shape and function measured. This type of
comparisons are desired in this case because the functional landscape is constructed from
morphological parameters that approximate evolutionary shape changes, which have no
significant size allometry in hyaenids and canids (Tseng and Wang In press-b). Also,
body size increased dramatically over the course of evolution in the two carnivoran
groups examined, so that bite force would show increases even in absence of
biomechanical adaptations. Therefore, comparisons solely based on skull shape are the
most appropriate.
To examine evolutionary trends predicted by the functional landscape, a series of
FE models that represent different degrees of specialization for bone-cracking in the
hyaenid and borophagine lineages, respectively, were used. The hyaenids Proteles
368
cristatus (J050607T02, ZJT comparative collection), Ictitherium sp. (HMV0163),
Chasmaporthetes lunensis (Antón et al. 2006, Tseng et al. 2011a), Ikelohyaena abronia
(Tseng and Stynder 2011), Parahyaena brunnea (MVZ117842, Museum of Vertebrate
Zoology, University of California, Berkeley; CT scans provided by B. Van Valkenburgh),
and Crocuta crocuta (Tseng et al. 2011a) were analyzed. The fossil and modern canids
analyzed included Mesocyon coryphaeus, Microtomarctus conferta, Epicyon haydeni,
Borophagus secundus, and Canis lupus from Tseng and Wang (2010), and Lycaon pictus
from Tseng and Stynder (2011). In addition, the percrocutid Dinocrocuta gigantea, a
feliform carnivoran that convergently evolved bone-cracking morphology independent of
hyaenids or canids, was included in the analysis using the model from Tseng (2009). A
total of 13 models of actual fossil and extant species were used.
In addition to MA and SE values, the stress distributions on the skulls of actual
species were also visualized. Values of von Mises stress, which approximate materials
that fail under a ductile mode of fracture, were used (Dumont et al. 2005, Nalla et al.
2003). As FEA conducted on models of actual species were scaled in a similar manner to
the theoretical models, the distribution of von Mises stress on the skull represents relative
levels of stress that can be directly compared across species. High levels of stress under
such comparisons can therefore be interpreted as likely areas of material failure.

369
Chapter Eight Results

The hybrid morphospace composed 36 theoretical models (Chapter Eight
Appendix 2), onto which a wire mesh was interpolated to create the functional landscapes
(Figs. 8.2-3). Two separate landscapes were created, one for mechanical advantage (MA;
Fig. 8.3A-C) and the other for skull strain energy (SE; Fig. 8.3D-F). The landscapes
showed predictable trends of variation. Increasing skull depth, regardless of the starting
skull width, translated into higher MA and higher SE (Fig. 8.3). Increasing skull width
generated progressively lower MA and SE at shallower skull depths, but the patterns
became more complex at higher skull depths (Fig. 8.3). Peaks in MA are found at skull
depth-to-length (D:L) ratio of > 0.7 and width-to-length (W:L) ratios of 0.4-0.7 (Fig.
8.3A-C). Lowest MA values are found at D:L < 0.4 and W:L > 0.7 (Fig. 8.3A-C).
D:L and W:L ratios of actual hyaenid and borophagine canid species overlapped
extensively in their distribution on the functional landscape (Fig. 8.4). The species
followed a joint evolutionary pathway from D:L 0.3-0.4 and W:L 0.5-0.6 to D:L ~0.5 and
W:L ~0.7 (Fig. 8.4A, D). This pathway showed a continuous climb towards higher
elevation on the MA landscape (Fig. 8.4B), and a path into an adaptive valley on the SE
landscape (Figs. 8.4E). The pathways occupied by actual hyaenids and canids are
bordered at the bottom right with a large region of low MA and low SE theoretical shapes
(Fig. 8.5). In the upper regions are high MA and high SE shapes; both of these regions
represent suboptimal areas (Fig. 8.5).
370
The regions of the MA vs. SE plot occupied by actual species represent
movements from low MA (~0.18) towards higher MA (~0.25) at or below the fitted curve
(SE = 174.74*MA
2
– 47.04*MA + 5.8609, r
2
= 0.8363) for all theoretical models (Fig.
8.6). The MA values of models of actual species covered a slightly larger range than
predicted by theoretical models, from ~0.16 to ~0.27 (Fig. 8.6A). The exception is the
myrmecophagous hyaenid Proteles cristata, which has an MA of ~0.12, lower than all
actual species and theoretical models. SE values of actual species followed the overall
trend predicted by the theoretical models, but do not follow the theoretical pathways
exactly. To test for potential differences created by scaling factor, models of actual
species were re-analyzed with scaling by total volume, total muscle attachment surface
area, or total skull length (condylobasal length). Volume- and muscle-scaled models
returned essentially identical results as the total surface area method (Fig. 8.6A, only
surface area results are shown). Scaling by skull length returned similar results, except
that MA values for the derived hyaenids Crocuta crocuta and Parahyaena brunnea were
lower, and SE values for Ictitherium and Chasmaporthetes lunensis were also lower (Fig.
8.6B). Such differences did not change the overall trends, however.
Von Mises stress distributions on the actual models showed a trend of
increasingly stressed fronto-parietal regions in hyaenids (Fig. 8.7A-G). The canid models
showed no such trend, and in general had moderate levels of von Mises stress that spread
over the dorsal cranium, except for elevated stress levels in Canis lupus and decreased
levels in Epicyon haydeni (Fig. 8.7H-M). The fronto-parietal region in Ikelohyaena
abronia and Canis lupus showed the highest peak stress, and in all models the
371
temporomandibular joints tend to have elevated stress levels (Fig. 8.7). Model data and
results are included in Chapter Eight Appendix 3.

Chapter Eight Discussion

The concept of functional landscapes was used to test the hypothesis that
convergent morphological evolution in canids and hyaenids can be explained in terms of
functional evolution towards optimal bone-cracking capability. Theoretical skull shapes
showed a general increase in mechanical advantage (MA) with higher D:L ratios,
although the narrower (lower W:L) skulls had largest levels of strain energy (Fig. 8.3).
Actual hyaenid and canid species showed a steady climb up the MA landscape, at the
same time moving along topological isoclines in SE. Such a pattern of evolution is
consistent with optimization theory; in this case two functions of the skull, maximizing
MA and minimizing SE, are optimized by traveling upslope on the MA landscape and
moving along topological isoclines on the SE landscape (Fig. 8.4). Therefore, the
hypothesis that convergent morphologies shared convergent functional capability is
supported.
Hyaenid and canid models are adjacent to each other on the MA vs. SE plots, with
more derived species having higher MA (Fig. 8.6). As predicted by the landscape model,
the ratio MA:SE was maintained or increased through evolution; the path from less
derived species to more specialized species tend to occur downward (i.e. smaller SE for a
given MA) or rightward (i.e. larger MA for a given SE) on the plot (Fig. 8.6). Such
372
distribution is also expected under an optimization model. The functional correlation of
the MA and SE distributions is further supported by the fact that Proteles cristata, a
specialized insectivorous hyaenid, does not crack bones, and accordingly has very low
MA towards the bottom left corner of the MA vs. SE plot (Fig. 8.6). In other words,
function can be lost over time with disuse.
Overall, among the theoretical models used to construct the functional landscape,
only a small number overlapped with actual morphologies (Fig. 8.5). In the upper left and
upper right regions high MA is coupled with high SE, making those morphologies sub-
optimal; those areas are accordingly not occupied by actual species (Fig. 8.5). The bottom
right corner is marked by both low MA and low SE, and is similarly not optimal. The
actual path taken by canid and hyaenids constitutes a route of increasing MA at relatively
small cost in SE increase (Fig. 8.5). Therefore, the functional landscape distributions of
actual species predict skull MA to be maximized relative to increase in skull SE through
evolution. The distribution of MA versus SE values for the 13 actual skull models show
an intermediate position within the range of theoretical morphologies, overlapping the
regions predicted by the functional landscape, supporting the hypothesis that
morphological convergence is coupled with increased optimization of skull function (Fig.
8.6).
The remaining unoccupied regions in the functional landscape, however, indicate
that the hypothesis predicting the derived morphotypes occupying peak functional optima
on the landscape was not supported. Contrary to the expectation, the most optimized
theoretical shapes in the functional landscape are not occupied by actual species (Fig.
373
8.5). Skull shapes with D:L = 0.7, W:L = 0.7 and D:L = 0.5, W:L = 0.8-1.1 tend to have
high MA and relatively low SE, making them more suitable for generating large bite
forces than shapes toward the central and bottom left regions of the landscape, where
canids and hyaenids are actually located (Fig. 8.5). There appears to be no visible barriers
or adaptive valleys on the MA landscape, or prohibitively high SE peaks on the SE
landscape to explain the lack of actual species in those regions (Fig. 8.4). To check
whether this bias in distribution is a function of similarly restricted skull shape changes
specific to hyaenids and borophagine canids, the modern carnivoran dataset composed of
37 North American and East African carnivoran species from Tseng and Wang (In press-
b) was plotted onto the functional landscapes (Fig. 8.8, Table 8.2). The pathways taken
by canid and hyaenids species overlapped with the skull shapes observed among
representatives of modern carnivoran families (Fig. 8.8A-C). Only the cheetah, Acinonyx
jubatus, was distinct from all other carnivorans by much higher W:L ratios which placed
the species on an SE peak (Fig. 8.8).
The extensive overlap of modern carnivorans with the evolution sequence of
canids and hyaenids indicates a general constraint in carnivoran skull shape disparity
within the hybrid morphospace of theoretical possible shapes (Fig. 8.2). Presence of a
higher-level constraint in skull shape has previously been identified in both eutherian and
marsupial carnivores (Wroe and Milne 2007), and accordingly created a limitation on the
evolution of functionally more optimized skull shapes in bone-cracking carnivorans (Fig.
8.8). Of course, the complex suite of functions the mammalian skull plays in mastication,
food acquisition, and sensory reception meant that constraints on the realized skull shapes
374
are more complex than just biomechanical ones. Theoretically speaking, however, one
optimal path to maximize MA and minimize SE would be to travel along topological
lines at D:L between 0.3 and 0.4 towards higher W:L ratios (Fig. 8.8C-D), in absence of
non-biomechanical constraints. At higher W:L ratios SE increases more slowly, and
therefore those theoretical shapes have relatively higher MA. In reality, W:L ratio is
constrained across the modern carnivorans analyzed, and skull shape in bone-cracking
hypercarnivores evolved towards higher D:L with an upper limit of W:L ~0.7. Therefore,
the carnivoran trend observed was to instead increase MA towards local (but not global)
optima (Fig. 8.8).
The high-MA high-SE skull of Dinocrocuta gigantea also demonstrates the
presence of additional factors in determining performance, in addition to the two
functional properties examined. The largest bone-cracking carnivorans examined in this
study, Epicyon haydeni and Dinocrocuta gigantea (both with skull length over 300 mm),
share similarities in skull shape but not in biomechanical properties (Fig. 8.7). Epicyon
has a low-MA and low-SE skull, in contrast to the high-MA high-SE skull of
Dinocrocuta. This seemingly contradicting result can be explained by the one-to-one
form to function property of mechanical advantage (Wainwright 2007). Mechanical
advantage by itself is a scale-free measure of force generation, but in fact a system with
high MA and low absolute muscle force can generate the same resulting bite force as a
system with low MA and large absolute bite force. Therefore, the disparate distributions
of Dinocrocuta and Epicyon can represent similar performing morphologies that
converge along another axis of evolutionary change, namely body size. A large body size
375
would allow the less efficient Epicyon to generate bite forces required to crack bones to a
comparable magnitude as smaller, more shape-adapted skulls of Crocuta and Borophagus.
On the other hand, the large body size of Dinocrocuta would allow a smaller muscle
input to generate sufficient bone-cracking bite forces, therefore not producing the high-
SE predicted at its maximum capability (Fig. 8.6). Such alternatives to evolutionary
changes in skull shape can be further coupled with behavior, in which bones of relatively
smaller prey are cracked and consumed, and bones of relatively larger prey intentionally
left alone. With this interpretation, body size increase in bone-cracking carnivorans as a
masticatory adaptation would be analogous to larger body size in ungulates as a defense
mechanism, in that both increases in body size alone constitutes an adaptation. Whether
“body-size” specialists should constitute a distinct sub-category of bone-cracking
ecomorphology is a fascinating issue that remains to be explored. Archaic mammals such
as creodonts and condylarths, for example, evolved dental morphology and body size
approaching the larger carnivoran bone-crackers, even though the skulls of those
mammals do not share the suite of morphological features seen in carnivorans (Gunnell
1998, Werdelin 1996b).
A concept intimately associated with adaptive landscapes is the
macroevolutionary ratchet, which has been studied in carnivorans (Holliday and Steppan
2004, Van Valkenburgh et al. 2004). The limited number of alternative means of
morphological specialization is associated with decrease in morphological disparity in
repeatedly specialized lineages, which affected the long-term fitness of those lineages
(Holliday and Steppan 2004, Van Valkenburgh 2007). In this context, generalist species
376
are located at lower elevations of the adaptive landscape, and specialists are higher up
adaptive peaks; the macroevolutionary ratchet can be visualized as the evolutionary
process of moving up in elevation on the landscape (Strathman 1978). Catastrophic,
sometimes even localized, events may shift the position of those adaptive peaks, causing
the demise of specialists by their very inability to move or survive in other regions of the
fitness landscape (Strathman 1978). Others argue for the mobility and dynamic nature of
adaptive peaks through time, which may imply a different mode of adaptation and
specialization of organisms which involves more evolutionary “adjustment” to current
peaks (McGhee 1999). The fact that convergent canids and hyaenids evolved via
pathways within the overall distribution of modern carnivorans indicates that a higher
taxonomic level constraint on skull depth and width ratios is present, and such constraints
are more general than can be specified in a macroevolutionary ratchet model for bone-
cracking specialists (Fig. 8.8). Nevertheless, it would be interesting to further explore
whether the pathways on the functional landscape are “one-way streets”, and the distance
already traveled by a particular lineage may indeed represent the macroevolutionary
ratchet in action.
The proxy for functionality used in this study, namely measures of bite force and
skull strain energy, are biomechanical function indicators, arguably not a very complete
measure of fitness (using a definition of the organism’s ability to both survive and
reproduce). However, the fact that terminal members of the lineages studied represent the
best examples of Crocuta-equivalent bone-cracking ecomorphologies in the Cenozoic,
and their evolutionary processes show overwhelming trend towards robust craniodental
377
features, suggest that in this case the functional properties likely would have been quite
important in their evolution. Furthermore, carnassial mechanical advantage is a common
selective parameter for all carnivorans, and measures of its biomechanical function are
directly linked to mastication and food intake. One can also argue that plotting of
evolutionary trends onto the static functional landscape is not greatly affected by the
possibility of shifting adaptive peaks in other types of landscapes which are contingent
upon environmental variations (McGhee 1999); biomechanical function underlies the
capability of different species to utilize harder food, which existed in the form of prey
skeletal remains regardless of their taxonomic identity or the surrounding environment. In
other words, the same selective pressures for masticatory capability would exist
independently of environmental changes, as long as larger vertebrate prey are present.
Thus, performance measures based on physical principles such as mechanical advantage
are suitable rulers to test specific form-function hypotheses in ecomorphological contexts.
Regardless of the simplicity of a two-dimensional framework, the resulting
distribution of actual species on the functional landscape shows a remarkable consistency
of maintaining MA:SE ratios throughout the region occupied by bone-cracking canids,
hyaenids, and the corresponding modern faunas in their respective areas of origin (Fig.
8.8). Such pattern indicates an overarching selection for the maintenance of strong skulls
and efficient bites across Carnivora, attributes which are principal in both active hunting
and passive scavenging behaviors. Despite the outstanding morphological features of the
skull and teeth in specialized bone-cracking ecomorphologies, the functional properties
of those derived ecomorphs still operated within the bounds of carnivoran distribution.
378
Again, the notable exception in the modern east African fauna is the cheetah, Acinonyx
jubatus. Skull shape in the cheetah has fallen off the tall ridges on the functional
landscape, and is located in a valley with low MA. The strict requirements for speed may
have overridden the base functional demands of mastication, demonstrating that such
deviation from the major trend is nevertheless feasible (Fig. 8.8).
Among metazoan animals, redundancy in body segments has been proposed to
enhance evolutionary potential for differentiation in functions (Wainwright 2007). An
analogous explanation can be applied to the plesiomorphically homodont dentition of
vertebrates, which evolved into highly heterodont dentition in mammals. Carnivores
exhibit fine examples of diversified function of heterodont teeth (Van Valkenburgh
1989a). The shallow, slashing bites of pursuit predators are made using the anterior
incisors and canines, and the crushing bites of omnivores are made with the posterior
bunodont molars (Van Valkenburgh 1996). Such differentiation in dental function is
shared by all carnivorans on a more general level, indicating the presence of multiple
axes of functional properties exist for different tooth positions. Carnassial function, the
focus of the current study, for example, should be supplemented with functional
properties in other teeth in order to more fully characterize the potential selective forces
that shape craniodental morphology. Such integration requires more complex
mathematical formulations of a multi-dimensional problem (Arnold 2003), of which the
current study represents a two-dimensional first step that is easily visualized.
The fact that the functional landscapes predicted movement of canid and hyaenid
species through only a small range on a shallow slope in the D:L and W:L morphospace
379
of MA:SE ratios, but the actual evolutionary differences in morphology being quite
striking (Turner et al. 2008, Wang et al. 1999), suggest there are other important factors
besides general skull dimensions in improving mechanical advantage and minimizing
skull strain energy. Movement on the landscape towards deeper and wider skulls also
allows more masticatory musculature to be present in the parietal region, which was not
adjusted in the theoretical shapes analyzed here. In addition, the relative proportions of
the rostrum and the braincase, and also the positions of the dentition relative to the
masticatory muscles both affect mechanical advantage. Such changes require more
sophisticated theoretical models, and more fine-tuned variations in FE skull models,
which might be generated using algorithms derived from geometric morphometrics
analysis (Stayton 2009). Nevertheless, the usage of simple morphological parameters to
create theoretical skull shapes was shown to be informative in discovering potential
biomechanical and non-biomechanical constraints on overall skull shape in convergent
evolution of adaptive morphologies in carnivorous mammals. Many more studies are
needed to explore the begging questions and to improve the completeness of such
theoretical frameworks.
In sum, a functional landscape framework constructed from theoretical
morphologies showed the presence of adaptive peaks that are not attained by actual
species. The pathways that actual species traversed, however, were nevertheless local
optima of relatively high mechanical advantage and moderate skull strain energy.
Predictions from the functional landscape are supported by results obtained using models
of actual species, showing a clear link between form and function in the evolution of
380
bone-cracking ecomorphologies. The restricted region occupied by a wider sampling of
modern carnivorans on the functional landscape indicates higher-level phylogenetic
constraint as an explanation for the unoccupied optimal peaks. The combination of
theoretical morphology and functional modeling with FEA has been shown to be an
informative approach to test adaptive hypotheses regarding morphological convergence,
and has implications for applications in broader taxonomic contexts.

Chapter Eight Acknowledgments

I thank my PhD advisor X. Wang and my committee members for guidance. J.
Liu provided much emotional and intellectual support without which this study could not
have been completed. G. Xie helped with CT scanning of Ictitherium. B. Van
Valkenburgh and the Digimorph project (UT Austin) provided CT images of Parahyaena
and Crocuta. M. Antón, W. Binder, M. Salesa, and D. Stynder contributed collaborative
efforts on previously published computer models used in this paper. The editor and
reviewers provided constructive comments on the paper that improved its content. The
curators and managers of the collections who assisted during the course of my study: S.
Chen, W. He (Hezheng Paleozoology Museum); Z. Qiu (Institute of Vertebrate
Paleontology and Paleoanthropology, Chinese Academy of Sciences); J. Indeck (Plains-
Panhandle Museum); J. Meng (American Museum of Natural History). This research was
supported by a National Science Foundation Graduate Research Fellowship and a
Doctoral Dissertation Improvement Grant (DEB-0909807), American Society of
381
Mammalogists Grant in Aid of Research, United States Fulbright Program, and a
University of Southern California Zumberge Grant.
382
Chapter Eight References

Antón, M., A. Turner, M. J. Salesa, and J. Morales. 2006. A complete skull of
Chasmaporthetes lunensis (Carnivora, Hyaenidae) from the Spanish Pliocene site
of La Puebla de Valverde (Teruel). Estudios Geológicos 62:375-388.

Arnold, S. J. 2003. Performance surfaces and adaptive landscapes. Integrative and
Comparative Biology 43:367-375.

Binder, W. J., E. N. Thompson, and B. Van Valkenburgh. 2002. Temporal variation in
tooth fracture among Rancho La Brea dire wolves. Journal of Vertebrate
Paleontology 22:423-428.

Binder, W. J., and B. Van Valkenburgh. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252:273-283.

Bourke, J., S. Wroe, K. Moreno, C. McHenry, and P. D. Clausen. 2008. Effects of gape
and tooth position on bite force and skull stress in the dingo (Canis lupus dingo)
using a 3-dimensional finite element approach. PLoS ONE 3:e2200.

Chen, G. F., and N. Schmidt-Kittler. 1983. The deciduous dentition of Percrocuta
Kretzoi and the diphyletic origin of the hyaenas (Carnivora, Mammalia).
Palaontologische Zeitschrift 57:159-169.

Clough, R. W. 1960. The Finite Element Method in Plane Stress Analysis. Proceedings of
2nd ASCE Conference on Electronic Computation, Pittsburgh, PA.

Davis, D. 1955. Masticatory apparatus in the spectacled bear Tremarctos ornatus.
Fieldiana: Zoology 37:25-46.

Davis, J. L., S. E. Santana, E. R. Dumont, and I. Grosse. 2010. Predicting bite force in
mammals: two-dimensional versus three-dimensional lever models. Journal of
Experiemental Biology 213:1844-1851.

Dessem, D. 1989. Interactions between jaw-muscle recruitment and jaw-joint forces in
Canis familiaris. Journal of Anatomy 164:101-121.

Dumont, E. R., I. Grosse, and G. J. Slater. 2009. Requirements for comparing the
performance of finite element models of biological structures. Journal of
Theoretical Biology 256:96-103.

383
Dumont, E. R., J. Piccirillo, and I. R. Grosse. 2005. Finite-element analysis of biting
behavior and bone stress in the facial skeletons of bats. The Anatomical Record
Part A 283A:319-330.

Ellis, J. L., J. J. Thomason, E. Kebreab, and J. France. 2008. Calibration of estimated
biting forces in domestic canids: comparison of post-mortem and in vivo
measurements. Journal of Anatomy 212:769-780.

Futuyma, D. J. 1997. Evolutionary Biology. Sinauer Associates, Inc., Sunderland.

Goswami, A. 2006a. Cranial modularity shifts during mammalian evolution. The
American Naturalist 168:270-280.

Goswami, A. 2006b. Morphological integration in the carnivoran skull. Evolution
60:169-183.

Greaves, W. S. 1985. The generalized carnivore jaw. Zoological Journal of the Linnean
Society 85:267-274.

Grosse, I., E. R. Dumont, C. Coletta, and A. Tolleson. 2007. Techniques for modeling
muscle-induced forces in finite element models of skeletal structures. The
Anatomical Record 290:1069-1088.

Gunnell, G. F. 1998. Chapter 5. Creodonta. Pp. 91-109 in C. M. Janis, K. M. Scott, and L.
L. Jacobs, eds. Evolution of Tertiary Mammals of North America. Volume 1:
Terrestrial carnivores, ungulates, and ungulatelike mammals. Cambridge
University Press, Cambridge.

Holliday, J. A., and S. J. Steppan. 2004. Evolution of hypercarnivory: the effect of
specialization on morphological and taxonomic diversity. Paleobiology 30:108-
128.

Joeckel, R. M. 1998. Unique frontal sinuses in fossil and living Hyaenidae (Mammalian,
Carnivora): description and interpretation. Journal of Vertebrate Paleontology
18:627-639.

Kauffman, S. A. 1995. At Home in the Universe. Oxford University Press, Oxford.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior. The
University of Chicago Press, Chicago.

Lauder, G. V. 1995. On the inference of function from structure. Pp. 9-18 in J. J.
Thomason, ed. Functional Morphology in Vertebrate Paleontology. Cambridge
University Press, New York.
384

Lauder, G. V. 1996. The Argument From Design. Pp. 55-91 in M. R. Rose, and G. V.
Lauder, eds. Adaptation. Academic Press, Inc., San Diego, California.

McGhee, G. R., Jr. 1980. Shell form in the biconvex articulate Brachiopoda: a geometric
analysis. Paleobiology 6:57-76.

McGhee, G. R., Jr. 1999. Theoretical Morphology: The concept and its applications.
Columbia University Press, New York.

McHenry, C., S. Wroe, P. D. Clausen, K. Moreno, and E. Cunningham. 2007.
Supermodeled sabercat, predatory behavior in Smilodon fatalis revealed by high-
resolution 3D computer simulation. Proceedings of the National Academy of
Sciences 104:16010-16015.

Meers, M. B. 2002. Maximum bite force and prey size of Tyrannosaurus rex and their
relationships to the inference of feeding behavior. Historical Biology 16:12.

Meloro, C., P. Raia, P. Piras, C. Barbera, and P. O'Higgins. 2008. The shape of the
mandibular corpus in large fissiped carnivores: allometry, function and phylogeny.
Zoological Journal of Linnean Society 154:832-845.

Nalla, R. K., J. H. Kinney, and R. O. Ritchie. 2003. Mechanistic failure criteria for the
failure of human cortical bone. Nature Materials 2:164-168.

Niklas, K. J. 1997. The Evolutionary Biology of Plants. University of Chicago Press,
Chicago.

Nowak, R. M. 1999. Walker's carnivores of the world. The Johns Hopkins University
Press, Baltimore.

O'Higgins, P., S. N. Cobb, L. C. Fitton, F. Groning, R. Phillips, J. Liu, and M. J. Fagan.
2011. Combining geometric morphometrics and functional simulation: an
emerging toolkit for virtual functional analyses. Journal of Anatomy 218:3-15.

Preuschoft, H., and U. Witzel. 2005. Functional shape of the skull in vertebrates: Which
forces determine skull morphology in lower primates and ancestral synapsids?
The Anatomical Record 283A:402-413.

Qiu, Z.-x., J.-y. Xie, and D.-f. Yan. 1988. Discovery of the skull of Dinocrocuta gigantea.
Vertebrata PalAsiatica 26:128-138.

Raup, D. M. 1966. Geometric analysis of shell coiling: General problems. Journal of
Palaeontology 40:1178-1190.
385

Rayfield, E. J. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35:541-576.

Rayfield, E. J., D. B. Norman, C. C. Jorner, J. R. Horner, P. M. Smith, J. J. Thomason,
and P. Upchurch. 2001. Cranial design and function in a large theropod dinosaur.
Nature 409:1033-1037.

Rensberger, J. M., and C. Stefen. 2006. Functional differentiations of the microstructure
in the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A
278:149-162.

Ross, C. F. 2005. Finite element analysis in vertebrate biomechanics. The Anatomical
Record Part A 283A:253-258.

Ross, C. F., M. A. Berthaume, P. C. Dechow, J. Iriarte-Diaz, L. B. Porro, B. G.
Richmond, M. Spencer, and D. S. Strait. 2010. In vivo bone strain and finite-
element modeling of the craniofacial haft in catarrhine primates. Journal of
Anatomy 218:112-141.

Savage, R. J. 1977. Evolution in carnivorous mammals. Palaeontology 20:237-271.

Simpson, G. G. 1944. Tempo and mode in evolution. Columbia University Press, New
York.

Slater, G. J., E. R. Dumont, and B. Van Valkenburgh. 2009. Implications of predatory
specialization for cranial form and function in canids. Journal of Zoology
278:181-188.

Stayton, C. T. 2006. Testing hypotheses of convergence with multivariate data:
morphological and functional convergence among herbivorous lizards. Evolution
60:824-841.

Stayton, C. T. 2009. Application of thin-plate spline transformations to finite element
models, or, how to turn a bog turtle into a spotted turtle to analyze both. Evolution
63:1348-1355.

Strathman, R. R. 1978. Progressive vacating of adaptive types during the Phanerozoic.
Evolution 32:907-914.

Tanner, J. B., M. Zelditch, B. L. Lundrigran, and K. E. Holekamp. 2010. Ontogenetic
change in skull morphology and mechanical advantage in the spotted hyena
(Crocuta crocuta). Journal of Morphology 271:353-365.
386

Tedford, R. H., X. Wang, and B. E. Taylor. 2009. Phylogenetic systematics of the North
American fossil Caninae (Carnivora: Canidae). Bulletin of the American Museum
of Natural History 325:1-218.

Thomas, R. D. K., and W.-E. Reif. 1993. The skeleton space: A finite set of organic
designs. Evolution 47:341-360.

Tseng, Z. J. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96:51-67.

Tseng, Z. J., M. Antón, and M. J. Salesa. 2011. The evolution of the bone-cracking model
in carnivorans: Cranial functional morphology of the Plio-Pleistocene cursorial
hyaenid Chasmaporthetes lunensis (Mammalia: Carnivora). Paleobiology 37:140-
156.

Tseng, Z. J., and W. J. Binder. 2010. Mandibular biomechanics of Crocuta crocuta,
Canis lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia).
Zoological Journal of Linnean Society 158:683-696.

Tseng, Z. J., and D. Stynder. 2011. Mosaic functionality in a transitional ecomorphology:
skull biomechanics in stem Hyaeninae compared to modern South African
carnivorans. Biological Journal of the Linnean Society 102:540-559.

Tseng, Z. J., and X. Wang. 2010. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology 271:1386-1398.

Tseng, Z. J., and X. Wang. In press. Do convergent ecomorphs evolve through
convergent morphological pathways? Cranial shape evolution in fossil hyaenids
and borophagine canids (Carnivora, Mammalia). Paleobiology.

Turnbull, W. D. 1970. Mammalian masticatory apparatus. Fieldiana: Geology 18:149-
356.

Turner, A., M. Antón, and L. Werdelin. 2008. Taxonomy and evolutionary patterns in the
fossil Hyaenidae of Europe. Geobios 41:677-687.

Van Valkenburgh, B. 1988. Trophic diversity in past and present guilds of large
predatory mammals. Paleobiology 14:155-173.

387
Van Valkenburgh, B. 1989. Carnivore dental adaptations and diet: a study of trophic
diversity within guilds Pp. 410-436 in J. L. Gittleman, ed. Carnivore behavior,
ecology, and evolution. Cornell University Press, New York.

Van Valkenburgh, B. 1996. Feeding behavior in free-ranging, large African carnivores.
Journal of Mammalogy 77:240-254.

Van Valkenburgh, B. 1999. Major patterns in the history of carnivorous mammals.
Annual Review of Earth and Planetary Science 27:463-493.

Van Valkenburgh, B. 2007. Déjà vu: the evolution of feeding morphologies in the
Carnivora. Integrative and Comparative Biology 47:147-163.

Van Valkenburgh, B., T. Sacco, and X. Wang. 2003. Chapter 7. Pack hunting in Miocene
borophagine dogs: evidence from craniodental morphology and body size.
Bulletin of the American Museum of Natural History 279:147-162.

Van Valkenburgh, B., X. Wang, and J. Damuth. 2004. Cope's Rule, hypercarnivory, and
extinction in North American canids. Science 306:101-104.

Vaughan, T. A., J. R. Ryan, and N. J. Czaplewski. 2000. Mammalogy. Saunders College
Publishing, Fort Worth.

Wainwright, P. C. 2007. Functional versus morphological diversity in macroevolution.
Annual Review of Ecology, Evolution, and Systematics 38:381-401.

Wang, X. 1994. Phylogenetic systematics of the Hesperocyoninae (Carnivora: Canidae).
Bulletin of the American Museum of Natural History 221:1-207.

Wang, X., R. H. Tedford, and M. Antón. 2008. The Dog Family, Canidae, and Their
Evolutionary History. Columbia University Press, New York.

Wang, X., R. H. Tedford, and B. E. Taylor. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243:1-391.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

Werdelin, L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
Pp. 582-624 in J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution.
Cornell University Press, New York.

388
Werdelin, L., and N. Solounias. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30:1-104.

Wright, S. 1932. The roles of mutation, inbreeding, crossbreeding and selection in
evolution. Proceedings of the XI International Congress of Genetics 1:356-366.

Wroe, S., P. D. Clausen, C. McHenry, K. Moreno, and E. Cunningham. 2007. Computer
simulation of feeding behaviour in the thylacine and dingo as a novel test for
convergence and niche overlap. Proceedings of the Royal Society B: Biological
Sciences 274:2819-2828.

Wroe, S., T. L. Ferrara, C. McHenry, D. Cumoe, and U. Chamoli. 2010. The
craniomandibular mechanics of being human. Proceedings of the Royal Society B:
Biological Sciences 277:3579-3586.

Wroe, S., C. McHenry, and J. J. Thomason. 2005. Bite club: comparative bite force in big
biting mammals and the prediction of predatory behaviour in fossil taxa.
Proceedings of the Royal Society of London, Series B 272:619-625.

Wroe, S., and N. Milne. 2007. Convergence and remarkably consistent constraint in the
evolution of carnivore skull shape. Evolution 61:1251-1260.

389
Chapter Eight Appendix 1. Hyaenids and extinct canids used in the study. Institutional
abbreviations: AMNH, American Museum of Natural History, New York; F:AM, Frick
Collection, American Museum of Natural History, New York; HMV, Hezheng
Paleozoology Museum, Gansu, China; IVPP, Institute of Vertebrate Paleontology and
Paleoanthropology, Beijing, China; LACM, Natural History Museum of Los Angeles
County, California; MCZ, Museum of Comparative Zoology, Harvard University,
Massachusetts; MVZ, Museum of Vertebrate Zoology, University of California,
California; PPHM, Plains-Panhandle Museum, Texas; UAMZ, University of Alberta
Museum of Zoology, Alberta, Canada; UCMP; University of California Museum of
Paleontology, Berkeley, California.

Canidae: Aelurodon ferox: F:AM27346, 61746, 61757; A. mcgrewi: F:AM61778; A.
taxoides: F:AM61781; Borophagus secundus: UCMP30101; Canis dirus: LACM2077;
Desmocyon matthewi: AMNH49177; Epicyon haydeni: PPHM1100; Epicyon saevus,
AMNH8305; Hesperocyon gregarius: UCMP65380; Paraenhydrocyon josephi,
F:AM54115, MCZ2102; Phlaocyon leucosteus, AMNH8768; Protomarctus optatus,
F:AM61156.

Hyaenidae: Adcrocuta eximia: F:AM28-L233, 35-B216, 41-L339; Crocuta crocuta:
LACM30655, MVZ124188, 124259, 165159-165163, 165165, 165167-165177, 165179-
165182, 173733-173734, 173736-173740, 174743, 173751, 173759, 173762-173764,
173768, 173771-173773, 175801, 184088-184089, 4823; Hyaena hyaena: LACM31264;
Hyaenictitherium wongi: AMNH23032; Proteles cristata: LACM60619, MVZ117841,
UAMZ10470.

Percrocutidae: Dinocrocuta gigantea: HMVX0361, HMVM0358, IVPPV15649.
390
Chapter Eight Appendix 2. Theoretical models and their parameters. D:L, skull depth to
length ratio; W:L, skull width to length ratio; elements: number of four-noded tetrahedral
finite elements in model; SE, skull strain energy (in Joules); adjSE, strain energy adjusted
by model volume (Dumont et al. 2009); Fout, output bite force (in Newtons); MA,
mechanical advantage; S.T., solution time required for FEA (in minutes).

Model name D:L W:L Elements SE (J)
adjSE
(J) Fout (N) MA
S.T.
(min)
J022611T38 0.33 0.42 1,139,265 3.39 3.30 7653.53 0.19 63
J022611T39 0.33 0.55 1,177,734 3.10 3.05 7347.61 0.18 90
J022611T40 0.33 0.69 1,138,486 2.67 2.63 6999.04 0.18 73
J022611T41 0.33 0.83 1,086,419 3.27 3.21 6737.50 0.17 78
J022611T42 0.33 0.97 1,047,996 3.02 2.93 6442.83 0.16 65
J022611T43 0.33 1.11 1,007,072 2.83 2.72 5932.73 0.15 43
J022611T33 0.38 0.42 1,146,486 4.20 4.10 8960.00 0.23 108
J021411T01 0.38 0.55 1,203,707 3.75 3.75 8635.51 0.22 97
J030211T48 0.38 0.69 1,153,383 3.67 3.70 8313.37 0.21 70
J021511T03 0.38 0.83 1,103,659 3.61 3.64 7966.53 0.20 57
J021711T08 0.38 0.97 1,072,618 3.01 3.02 7768.39 0.20 64
J021711T09 0.38 1.11 1,044,292 3.33 3.32 7401.19 0.19 44
J022611T34 0.51 0.42 1,095,971 5.81 5.62 9436.23 0.24 77
J021711T10 0.51 0.55 1,124,466 5.31 5.32 9919.35 0.25 69
J021811T14 0.51 0.69 1,108,403 4.16 4.22 9768.02 0.25 66
J021811T15 0.51 0.83 1,095,138 3.60 3.67 9275.19 0.23 72
J021811T16 0.51 0.97 1,094,398 3.51 3.57 9116.86 0.23 72

391
Chapter Eight Appendix 2 continued
J021811T17 0.51 1.11 1,069,458 2.99 3.03 8496.61 0.21 47
J022611T35 0.59 0.42 1,030,327 6.17 5.91 10263.06 0.26 67
J021711T11 0.59 0.55 1,077,924 5.96 5.94 10127.58 0.25 67
J030211T46 0.59 0.69 1,081,675 5.05 5.13 10029.38 0.25 70
J022011T21 0.59 0.83 1,103,739 5.82 5.95 9428.57 0.24 74
J030211T47 0.59 0.97 1,129,459 3.97 4.06 9717.49 0.24 68
J021911T18 0.59 1.11 1,096,699 4.54 4.64 9269.83 0.23 75
J030311T49 0.65 0.42 995,611 6.00 5.68 10898.32 0.27 56
J021811T12 0.65 0.55 1,049,741 6.62 6.53 11338.75 0.28 48
J022111T26 0.65 0.69 1,085,106 4.71 4.74 10571.10 0.27 63
J022111T25 0.65 0.83 1,127,189 5.74 5.85 10200.98 0.26 81
J022011T23 0.65 0.97 1,182,948 4.96 5.07 9366.17 0.24 86
J022011T19 0.65 1.11 1,134,838 5.04 5.16 10320.85 0.26 74
J030211T45 0.73 0.42 962,202 7.09 6.63 10939.08 0.27 45
J021811T13 0.73 0.55 1,032,173 7.93 7.76 11725.13 0.29 49
J022111T28 0.73 0.69 1,092,655 6.09 6.10 11245.83 0.28 62
J022111T27 0.73 0.83 1,167,406 5.54 5.63 10570.20 0.27 80
J030211T44 0.73 0.97 1,248,480 5.45 5.56 10679.10 0.27 100
J021611T06 0.73 1.11 1,186,504 5.61 5.74 10949.00 0.27 90

392
Chapter Eight Appendix 3. Finite element models of actual fossil and extant carnivorans
analyzed in the study. Abbreviations as in Chapter Eight Appendix 2.

Model name D:L W:L Elements SE (J)
adjSE
(J) Fout (N) MA
Crocuta crocuta 0.44 0.66 1,173,850 4.49 3.87 10743.08 0.27
Parahyaena brunnea 0.43 0.65 537,232 4.35 3.82 10601.93 0.27
Ikelohyaena abronia 0.38 0.63 1,188,737 5.30 4.50 8503.25 0.21
Chasmaporthetes lunensis 0.48 0.66 1,121,102 4.94 4.93 9387.11 0.24
Ictitherium sp. 0.38 0.55 1,203,707 3.75 3.75 8635.51 0.22
Borophagus secundus 0.53 0.71 920,143 3.49 3.16 9280.70 0.23
Epicyon haydeni 0.44 0.71 937,061 2.75 2.77 7706.61 0.19
Microtomarctus conferta 0.42 0.66 948,735 3.07 2.79 7508.95 0.19
Canis lupus 0.38 0.58 679,111 5.79 4.46 9121.44 0.23
Lycaon pictus 0.41 0.67 1,134,781 4.24 3.72 6668.69 0.17
Mesocyon coryphaeus 0.35 0.62 1,115,623 4.35 3.92 8982.51 0.23
Dinocrocuta gigantea 0.44 0.74 1,532,146 6.38 6.66 10551.26 0.26

393
Figure 8.1. Convergent evolution of skull shapes in borophagine canids (A-B) and
hyaenids (C-D) as observed from two-dimensional geometric morphometrics analyses. A,
C, dorsal views; B, D, lateral views. Illustrations of skull show the measurements of
width to length (W:L) and depth to length (D:L) taken from theoretical and actual skull
shapes. Modified from Tseng and Wang (In press).

394
Figure 8.2. Theoretical models generated by geometric modification of an Ictitherium
skull. The hybrid morphospace occupied by the 36 models spanned D:L ratios from 0.33
to 0.73 and W:L ratios from 0.42 to 1.11. Theoretical skull shapes are shown in rostral-
lateral view.

395
Figure 8.3. Construction of the functional landscape from theoretical morphologies. W:L
and D:L are plotted on the x- and y-axes, respectively. The functional properties
mechanical advantage (MA) and skull strain energy (SE, in joules) are plotted on the z-
axis. A, D, three-dimensional plots of the data points from analysis of theoretical models;
B, E, the wireframe mesh overlaid and interpolated using the theoretical models; C, F, the
theoretical models removed, leaving the mesh representing the functional landscapes for
MA and SE, respectively.

396
Figure 8.4. Distribution of actual species on the functional landscapes. A, D, distribution
of hyaenids (dark circles) and fossil canids (light circles) on two-dimensional contour
plots of MA and SE, respectively. Lines are isoclines. B, E, distribution of hyaenid and
canid species on the three-dimensional functional landscapes for MA and SE,
respectively. C, F, the pathways occupied by the hyaenid (shaded) and canid (outlined)
lineages on the MA and SE landscapes, respectively. Note continuous climb on the MA
landscape and shifting towards shallower slopes on the SE landscape.

397
Figure 8.5. Locations of optimal functional in the hybrid morphospace. Small shaded
squares represent theoretical models matched by existing hyaenid and canid species
(small unshaded square shows position of insectivorous Proteles cristata). Suboptimal
regions are shown in larger squares. Regions marked by parenthesized labels represent
optimal areas not occupied by actual species.

398
Figure 8.6. Theoretical and actual MA and SE values. A. distribution of theoretical
models overlaid with values from FE models of actual species, all scaled by total surface
area. B, distribution of theoretical and actual models, the latter scaled by condylobasal
length of the skull. Red triangles indicate the theoretical pathway traveled by actual
species on the functional landscape. The positions of hyaenid (darker shade) and canid
(lighter shade) groupings are shown as ovals in (B). Species abbreviations (hyaenids):
Ccr, Crocuta crocuta; Clu, Chasmaporthetes lunensis; Iab, Ikelohyaena abronia; Ict,
Ictitherium sp.; Pbr, Parahyaena brunnea; Pcr, Proteles cristata. Canids: Bor,
Borophagus secundus; Can, Canis lupus; Epi, Epicyon haydeni; Lpi, Lycaon pictus; Mes,
Mesocyon coryphaeus; Mic, Microtomarctus conferta. Percrocutid: Dgi, Dinocrocuta
gigantea.

399
Figure 8.7. Stress distributions on the FE skull models of actual fossil and extant species.
A, Dinocrocuta gigantea; B, Crocuta crocuta; C, Parahyaena brunnea; D, Ikelohyaena
abronia; E, Chasmaporthetes lunensis; F, Ictitherium sp.; G, Proteles cristata; H,
Epicyon haydeni; I, Borophagus secundus; J, Microtomarctus conferta; K, Canis lupus; L,
Lycaon pictus; M, Mesocyon coryphaeus. Phylogenetic relationships for hyaenids (A-G)
based on Werdelin and Solounias (1991), and for canids (H-M) based on Wang (1994),
Wang et al. (1999), and Tedford et al. (2009). Skulls are scaled to the same length for
ease of presentation.

400
Figure 8.8. Distributions of modern North American and East African carnivoran species
on the MA (A), SE (B), and MA:SE (C-D) landscapes. Arrows indicate pathways of
evolution for hyaenids (light arrows) and borophagine canids (dark arrows). Species
distributions of modern carnivoran species are plotted as solid contours. Peaks on the
MA:SE landscape (D) represent optimized theoretical skull shapes that are either realized
(light shade) or unoccupied (dark shade, with question mark). (D) corresponds with
Figure 8.5.

401
Table 8.1. List of actual models and species measurements of hyaenids and canids used in
the study. For list of specimen numbers see Chapter Eight Appendix 1.
†
extinct taxon.

Actual models:
Hyaenidae Canidae
Crocuta crocuta Lycaon pictus
Parahyaena brunnea Canis lupus
Ikelohyaena abronia
†
Borophagus secundus
†

Chasmaporthetes lunensis
†
Epicyon haydeni
†

Ictitherium sp.
†
Microtomarctus conferta
†

Proteles cristata Mesocyon coryphaeus
†

Additional measurements:
Crocuta crocuta (n=45) Canis dirus (n=1)
†

Adcrocuta eximia (n=3)
†
Borophagus secundus (n=1)
†

Hyaena hyaena (n=1) Epicyon haydeni (n=1)
†

Hyaenictitherium wongi (n=1)
†
Epicyon saevus (n=1)
†

Proteles cristatus (n=3) Aelurodon ferox (n=3)
†

Aelurodon mcgrewi (n=1)
†

Percrocutidae Aelurodon taxoides (n=1)
†

Dinocrocuta gigantea (n=3)
†
Protomarctus optatus (n=1)
†

Phlaocyon leucosteus (n=1)
†

Desmocyon matthewi (n=1)
†

Paraenhydrocyon josephi (n=2)
†

Hesperocyon gregarius (n=1)
†

402
Table 8.2. List of modern North American and East Africa carnivoran species used to
construct contour of carnivoran distribution. For specimen lists see Tseng and Wang (In
press-b).

North America n East Africa n
Alopex lagopus 10 Acinonyx jubatus 2
Canis latrans 10 Atilax paludinosus 5
Canis lupus 11 Bdeogale crassicauda 5
Gulo gulo 5 Canis aureus 8
Lynx canadensis 5 Caracal caracal 1
Lynx rufus 4 Civettictis civetta 4
Martes pennanti 9 Crocuta crocuta 45
Mephitis mephitis 4 Felis sylvestris 7
Mustela frenata 7 Genetta rubiginosa 13
Neovison vison 2 Herpestes sanguineus 8
Procyon lotor 3 Hyaena hyaena 1
Puma concolor 9 Ichneumia albicauda 2
Taxidea taxus 6 Ictonyx striatus 3
Urocyon cinereoargenteus 1 Lycaon pictus 8
Ursus americanus 10 Mellivora capensis 1
Ursus arctos 9 Nandinia binotata 5
Vulpes vulpes 11 Otocyon megalotis 7
Panthera leo 24
Panthera pardus 7
Proteles cristatus 3
403
Chapter Nine: Enamel microstructure versus macrostructural evolution

This chapter has been published as:

Tseng, Z. J. 2011. Variation and implications of intra-dentition HSB pattern in fossil
hyaenids and canids (Carnivora, Mammalia). Journal of Vertebrate Paleontology
31(5):1163-1167. doi:10.1080/02724634.2011.602161

A copy of the accepted manuscript begins on the next page.
404
Variation and implications of intra-dentition HSB pattern in fossil
hyaenids and canids (Carnivora, Mammalia)

ZHIJIE JACK TSENG,
1,2 1
Department of Vertebrate Paleontology, Natural History
Museum of Los Angeles County, 900 Exposition Boulevard, Los Angeles, California
90007, U.S.A., jtseng@nhm.org;
2
Integrative and Evolutionary Biology Program,
Department of Biological Sciences, University of Southern California, Los Angeles,
California 90089, U.S.A.

405
Chapter Nine Introduction

Cenozoic mammalian carnivores evolved stereotypical morphologies that can be
correlated to adaptations for specific dietary capabilities (Van Valkenburgh 2007,
Werdelin 1996b). The diagnosis of such ecomorphologies is based on a suite of
morphological characters of the teeth and skeleton, using a presumed relationship
between form and function (Lauder 1995, Van Valkenburgh 1985, 1988). Among the
morphological indicators of functional capability in extinct carnivorans, a series of
studies on the enamel microstructure of carnivoran teeth have shown a well-established
link between enamel prism decussation patterns and diet in modern carnivorans, and their
equivalents in the fossil record (Ferretti 2007b, Rensberger and Stefen 2006, Rensberger
and Wang 2005, Stefen 1994, 1997, Tomes 1906). In this study, new data are presented
on evolutionary patterns revealed by examination of enamel microstructure using optical
light, highlighting aspects of intra-dentition variation of enamel microstructure in extinct
hyaenid and canid carnivorans.
A general survey of enamel microstructure in carnivorans by Stefen (1994, 1997)
demonstrated the widespread occurrence of structural modification and a gradient of
complexity in the arrangement and decussation of enamel prisms. Layers (or bands) of
prisms that decussate in opposing orientations are called Hunter-Schreger Bands (HSB).
The three-dimensional organization of enamel prisms that occur in carnivorans can be
categorized into three major types of HSB: undulating, acute-angled, and zig-zag (Stefen
1997, 1999). These types were defined by Stefen (1997) based on the approximate angles
406
of the enamel bands, measured as the angle between the boundaries of a given enamel
prism fold: >140º for undulating, 140º-70º for acute-angled, and 70º-50º (or smaller) for
zig-zag HSB. Each type represents a grade along which the teeth possessing them are
adapted for consumption of hard foods; undulating HSB is the plesiomorphic condition,
whereas extensive zig-zag HSB is found in specialized bone-cracking ecomorphologies
(Stefen 1997). There is a close association between zig-zag HSB in premolars and bone-
consumption in modern spotted hyaenas (Rensberger and Stefen 2006, Stefen and
Rensberger 2002). Convergently evolved zig-zag HSB are observed in extinct
borophagine canids (Rensberger and Wang 2005, Stefen 1999). Specialized HSB patterns
have also been observed in other mammal groups such as creodonts, condylarths, and
entelodonts (Stefen 1997).
Previous studies of HSB adaptation in fossil hyaenids and canids, and indeed for
most other carnivorans, focused on only a few tooth positions and reported broad trends
of specialization across lineages (Ferretti 2007b, Stefen 1994, 1999). Others reported on
variation of HSB orientations within a single tooth (Rensberger and Stefen 2006,
Rensberger and Wang 2005). Even though this latter variability of intra-tooth HSB has
been recognized by many workers, the variability of HSB across different tooth positions
within the dentition of a given taxon has not been dealt with in more detail. Whereas
broad trends of HSB changes through the evolutionary history of entire lineages can
contribute evidence regarding patterns of changing diets and paleoecology, the pattern by
which different teeth in the dentition evolved specialized HSB can further reveal potential
selective drivers in the evolving functionality of the heterodont carnivoran dentition.
407
Focusing on such dentition-level patterns, which have been previously recognized but not
emphasized (Stefen 1999, Stefen and Rensberger 1999), this report describes the intra-
dentition variation of HSB enamel microstructure within two carnivoran lineages.
Enamel microstructure patterns of fossil hyaenids and canids were studied to demonstrate
the utility of examining entire dentitions in revealing trends and timing of structural
adaptations that complement more generalized surveys of higher-level lineages and
groups.
Institutional Abbreviations— AMNH, American Museum of Natural History,
New York, USA; BM, Natural History Museum, London, UK; F:AM, Frick Collection,
American Museum of Natural History; HMV, Hezheng Paleozoological Museum, Gansu,
China; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese
Academy of Sciences, Beijing, China; LACM, Natural History Museum of Los Angeles
County, California, USA; LACM-CIT, California Institute of Technology collection,
housed in the Natural History Museum of Los Angeles County; PMU, Evolutionsmuseet,
University of Uppsala, Sweden; PPHM, Panhandle-Plains Historical Museum, Texas,
USA; UCMP, University of California Museum of Paleontology, Berkeley, California,
USA; USNM, National Museum of Natural History (Smithsonian), Washington DC,
USA.

408
Chapter Nine Materials and Methods

Specimens examined are listed in Chapter Nine Appendix. A total of 280
dentitions and isolated teeth were examined. The study sample included species of
hesperocyonine and borophagine canids, as well as fossil hyaenids. All specimens were
thoroughly cleaned with cotton swabs and acetone or ethyl alcohol. Teeth were then
examined with a stereomicroscope at 5-40x magnification with an oblique light source
(Koenigswald 1980, Stefen 1997). HSB patterns are visible as alternating light and dark
bands that represent different enamel orientations, which vary in their refractive
properties. HSB pattern was recorded at three different levels in each tooth; the bottom,
middle, and top thirds of the tooth crown (Fig. 9.1). Crown examination was divided into
three regions based on previous observation of base-to-tip changes in enamel
microstructure within a single tooth, often with a transitional region in the middle part of
the crown (Stefen and Rensberger 2002). The separation was made qualitatively, relative
to the total unworn crown height of each specimen. The major HSB pattern (undulating,
acute-angled, or zig-zag) within each third of the tooth was recorded as representative of
that region. Very worn specimens, in which the top third or two-thirds of the tooth crown
were worn away, were included in the analysis only when complemented by more
complete specimens of the same genus and tooth position. Data were composed on the
genus level from complete dentitions where available, and from composite dentitions
otherwise. Eight HSB patterns were ranked in their order of inferred specialization, from
no specialization (undulating in all three regions of the crown) to complete zig-zag
409
specialization (zig-zag in all three regions). No attempts were made to further divide the
zig-zag HSB patterns by their folding angle (Ferretti 2007b). Phylogenetic relationships
for fossil canids are based on Wang (1994) and Wang et al. (1999). Those for hyaenids
are based on Werdelin and Solounias (1991).

Chapter Nine Results

The majority of hesperocyonine canid teeth examined have unspecialized,
undulating HSB (Fig. 9.2A). Intermediate, acute-angled HSB is found in the premolars
and molars of four genera (Paraenhydrodon, Mesocyon, Cynodesmus, and Sunkahetanka).
Mesocyon and Sunkahetanka also showed evidence of intermediate HSB pattern in the
canine tooth. Enhydrocyon showed acute-undulating HSB in the incisors and canines, but
not in the premolars or molars.
Borophagine canids from Archaeocyon to Paracynarctus displayed no zig-zag
HSB specialization; acute-angled HSB appeared in the posterior premolars and first
molars only (Fig. 9.2A). Zig-zag HSB appeared first in Microtomarctus at the premolars
and molars, and eventually spread to the canines and incisors in Epicyon and Borophagus.
Ancestral genera of the Borophagina subtribe, Paratomarctus, Carpocyon, and
Protepicyon, had acute-angled but not zig-zag HSB. None of the borophagine canids
examined had full zig-zag HSB in the first premolar, except for Epicyon.
Among fossil hyaenids, the Plioviverrops specimen available for study showed
unspecialized HSB, and the entire dentition was probably without acute-angled or zig-zag
410
HSB, as found by Ferretti (2007b). Tungurictis spocki had acute-angled HSB at the
carnassials (P4/m1), but is otherwise unspecialized in the cheek dentition. Specimens of
Ictitherium showed presence of zig-zag HSB throughout the premolars, molars, and also
at the canines (Fig. 9.2B). Species of Hyaenictitherium are comparably specialized, with
variation mainly in the degree of acute-angled HSB pattern in the incisors. Zig-zag HSB
is similarly present in the premolars and molars of Lycyaena, but more widespread in its
sister genus Chasmaporthetes (Fig. 9.2B). The robust hyaenines Pachycrocuta,
Adcrocuta, and Crocuta all have zig-zag HSB across the dentition, except for variation in
the first premolar (Fig. 9.2B).

Chapter Nine Discussion

Stefen (1999) noted the variability of within-tooth transition from undulating to
zig-zag HSB in the dentition of living and fossil canids, and generalized two major trends:
one type showed identical transition patterns throughout the entire dentition, the other
showed more zig-zag HSB initially in the posterior premolars, and also in the incisors.
Results from the current study showed that the patterns of transition and the amount of
zig-zag HSB present is associated with the degree of craniodental specialization for
durophagy previously recognized in borophagine canids and hyaenids (Fig. 9.2). The
pattern in both lineages shows that zig-zag HSB tend to appear first in the posterior
premolars and molars and almost never in the anterior dentition before other teeth.
411
Furthermore, in the hesperocyonine genera examined, none reached a degree of zig-zag
HSB specialization seen in borophagine canids and hyaenids.
Even though unavailable to this study, the premolars and molars of the early
hyaenid Protictitherium gaillardi have been examined by Stefen and Rensberger (1999),
and only undulating HSB was observed. Following the trends of intra-dentition zig-zag
HSB appearance observed in the current study, the incisors of P. gaillardi likely had
undulating HSB. Stefen and Rensberger (1999) also reported that zig-zag HSB was
present with occasional undulating HSB bands at the base of the crown throughout the
dentition of modern Crocuta, Parahyaena, and Hyaena, which were not examined here.
These previously observed patterns are consistent with the new data analyzed, and
suggest that fully zig-zag HSB tooth crowns are not only present in the most derived
fossil hyaenids and borophagine canids, they are also more widespread in the dentition of
those taxa.
The fourth premolar (P4) and first molar (m1), which are used by modern
hypercarnivores for shearing meat and cracking bone, were an innovation of carnivorans
that had a largely conserved function throughout much of their evolutionary history
(Savage 1977). Accordingly, the carnassial pair (P4 and m1) and surrounding teeth tend
to evolve zig-zag HSB before other teeth in the dentition. In the case of borophagine
canids and hyaenids, increased consumption of bone in their diets may have played a role
in generating selective pressure for stronger premolars and molars. Such transition
towards full zig-zag HSB in the dentition appeared relatively abruptly in the dentition of
the hyaenid Ictitherium and less so in borophagine canids Aelurodon, Epicyon, and
412
Borophagus (Fig. 9.2). Even though the derived, putative bone-crackers in these two
lineages have zig-zag HSB throughout the majority of the teeth in the dentition, the
appearance of zig-zag HSB per se does not seem to be correlated to robust craniodental
morphology. This observation was made previously by Stefen and Rensberger (1999) and
Ferretti (2007b), and the current study shows that this phenomenon is coupled with
presence of zig-zag HSB in those earlier taxa that is restricted to the posterior premolars
and molars only. This is especially apparent in the fossil hyaenids, where
Chasmaporthetes and the robust hyaenines probably had larger capacity for bone-
consumption compared to Ictitherium and Hyaenictitherium (Tseng 2009, Tseng et al.
2011a, Werdelin and Solounias 1991); nonetheless, the carnassial teeth in the latter group
were already specialized microstructurally with zig-zag HSB. Functional studies of
transitional hyaenid species indicate that bone-cracking ability, along with its suite of
craniodental morphological features exemplified by the modern spotted hyena Crocuta
crocuta, were not symplesiomorphic features of hyaenines (the group containing all
highly robust hyaenids). The change in skull shape and size had been quite gradual, and
skulls of transitional and ancestral species probably had intermediate or mosaic
functionality for bone-cracking (Tseng and Stynder In Press, Tseng and Wang In press-b).
In borophagine canids, functional capability for bone-cracking estimated using finite
element modeling showed that Microtomarctus did not share the strong and efficient
skulls of Epicyon and Borophagus (Tseng and Wang 2010); however, in the current
analysis, the premolars and molars of Microtomarctus clearly show the presence of zig-
zag HSB (Fig. 9.2A). It is likely that some degree of mosaic evolution also occurred
413
within borophagine canids, as HSB specialization responded to increased durophagy
initially, with subsequent modification of the overall craniodental morphology.
The disjoint evolution of specialized zig-zag HSB and robust craniodental
morphology in hyaenids, and to an extent in borophagine canids, indicates that HSB
specialization was a more flexible evolutionary feature than overall skull shape or tooth
shape. Cranial and dental morphology changed over time with a stereotypical pattern of
repeated (or iterative) evolution (Holliday and Steppan 2004, Van Valkenburgh et al.
2004). This mode of evolution often led to specialized hypercarnivores that have
restricted morphological disparity (Holliday and Steppan 2004). Zig-zag HSB
occurrences in the two lineages examined, on the other hand, do not show an exclusive
link to bone-cracking hypercarnivore taxa that are identified by superficial craniodental
morphology (e.g. Epicyon, Borophagus, Pachycrocuta, Crocuta). In the specific case of
durophagous ecomorphologies, increasingly folded HSB patterns would have provided
incremental functional advantage to carnivorans incorporating more hard food, by
reinforcing the dentition and preventing catastrophic failure of teeth by halting micro-
cracks within HSB regions (Chai et al. 2009). Even though HSB adaptation is linked with
the robust craniodental morphology in modern spotted hyenas (Rensberger and Wang
2005), the benefits and structural strength of zig-zag HSB teeth as an independent
mechanical adaptation is not obliterated in absence of comparably efficient skeletal
features (Rensberger 1995). This adaptation in enamel microstructure by itself could have
imparted some advantage for durophagy, even in absence of robust skeletal morphology.
Such interpretation is supported by the fact that zig-zag HSB appeared first in the
414
posterior premolars and molars in fossil hyaenids and borophagine canids, in the teeth
that were eventually adopted for bone-cracking in later, more specialized
ecomorphologies (Kruuk 1972b, Werdelin 1989). Therefore it is hypothesized here that
the flexible patterns of HSB specialization allow carnivorans to adapt rapidly to changing
dietary environments, and such flexibility allowed species to consume harder foods
without locking into a macroevolutionary ratchet mode of morphological evolution
observed on the more inclusive scale of skull and tooth shape changes (Van Valkenburgh
2007). One example where such incremental adaptation could allow species to respond to
environmental degradation is seen in the fossil canids of Rancho La Brea (Van
Valkenburgh 2009, Van Valkenburgh and Hertel 1993). In several specimens of Canis
dirus examined, full zig-zag HSB are present in the cheek dentition, even though the
skull of C. dirus is similar to that of the modern Canis lupus, when compared to the bone-
cracking Crocuta crocuta (Tseng 2009). Zig-zag HSB have also been identified in
modern Canis latrans, which are more omnivorous but consume bones of their smaller
prey (Stefen 1999). Evidence of individual variation in intra-dentition HSB distribution
(e.g. upper and lower dentition of Lycyaena, Fig. 9.2) further supports the evolutionary
flexibility hypothesis.
A limitation of examination of enamel microstructure from the surface using a
light-optical method is that this technique does not capture the overall complexity of
prism arrangement, which can vary from the enamel-dentine junction (EDJ) to the
surface of the crown (Rensberger and Wang 2005, Stefen 1997). Sectioning of the
specimens is required to study the internal prism morphology, and such destructive
415
sampling is difficult to conduct for broader-scale comparisons of entire dentitions of
fossil species. In addition, one important aspect which requires more in-depth study is the
association of HSB patterns with enamel thickness within a tooth, across the dentition, as
well as across an entire lineage. The orientation and bending of HSB near shear facets of
the cheek dentition may also contribute to increased functionality for durophagy, as
demonstrated in a range of fossil and modern hyaenids (Rensberger and Stefen 2006,
Stefen and Rensberger 1999).
Lastly, there are probably other features of the dentition, perhaps even in the skull,
that exhibit similarly flexible modes of evolution as proposed here for HSB. There has
been little research on whether finer-scale and coarser-scale morphological characters
differ in their evolutionary potential, even though the intrinsic variance of larger skull
characters have been shown to be higher than smaller ones in mammals (Palmeirim 1998).
Functional demands of durophagy act on multiple levels (skull shape, bone histology,
tooth shape, enamel microstructure, etc.) at the same time, but it is unclear whether the
tempo and mode of the evolutionary responses are comparable or not. Features such as
intra-dentition HSB patterns may allow microevolutionary adaptations to enhance fitness
“under the radar” of the larger context of the macroevolution ratchet, which has been
shown to act strongly in extremely specialized carnivoran lineages. Potential interactions
of such morphological characters remain to be explored in more depth.

416
Chapter Nine Acknowledgments

The author thanks X. Wang for guidance and help with museum access and loans.
The following people are acknowledged for their help in accessing the respective
collections: J. Galkin, J. Meng (AMNH); A. Currant, C. Delmer, R. Pappa (BM); S. Chen,
W. He (HMV); Q. Li, Z.D. Qiu, Z.X. Qiu (IVPP); G. Takeuchi (LACM); J-O. Ebbestad,
V. Berg-Madsen (PMU); J. Indeck (PPHM); P. Holroyd, S. Tomiya (UCMP); M. Carrano
(USNM). X. Ni (AMNH) helped with stereomicroscope photography. J. Liu (University
of Alberta) provided comments and much support in spirit. T. Konishi (Royal Tyrrell
Museum) provided stimulating discussion. The editor and reviewers provided helpful
comments. Research was funded by a National Science Foundation Graduate Research
Fellowship and a Doctoral Dissertation Improvement Grant (DEB-0909807).

417
Chapter Nine References
Chai, H., J. J.-W. Lee, P. Constantino, P. W. Lucas, and B. Lawn. 2009. Remarkable
resilience of teeth. Proceedings of the National Academy of Sciences 106:7289-
7293.

Ferretti, M. P. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience 100:41-52.

Holliday, J. A., and S. J. Steppan. 2004. Evolution of hypercarnivory: the effect of
specialization on morphological and taxonomic diversity. Paleobiology 30:108-
128.

Koenigswald, W. v. 1980. Schmelzmuster und Morphologie in den Molaren der
Arvicolidae (Rodentia). Adhandlungen der Senckenbergischen naturforschenden
Gesellschaft 539:1-129.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior; in G. B.
Shaller (ed.), Wildlife Behavior and Ecology Series. The University of Chicago
Press, Chicago, 335 pp.

Lauder, G. V. 1995. On the inference of function from structure; pp. 9-18 in J. J.
Thomason (ed.), Functional Morphology in Vertebrate Paleontology. Cambridge
University Press, New York.

Palmeirim, J. M. 1998. Analysis of skull measurements and measurers: can we use data
obtained by various observers? Journal of Mammalogy 79:1021-1028.

Rensberger, J. M. 1995. Determination of stresses in mammalian dental enamel and their
relevance to the interpretation of feeding behaviors in extinct taxa; pp. 151-172 in
J. J. Thomason (ed.), Functional morphology in vertebrate paleontology.
Cambridge University Press, New York.

Rensberger, J. M., and X. Wang. 2005. Microstructural reinforcement in the canine
enamel of the hyaenid Crocuta crocuta, the felid Puma concolor, and the late
Miocene canid Borophagus secundus. Journal of Mammalian Evolution 12:379-
403.

Rensberger, J. M., and C. Stefen. 2006. Functional differentiations of the microstructure
in the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A
278:149-162.

Savage, R. J. 1977. Evolution in carnivorous mammals. Palaeontology 20:237-271.

418
Stefen, C. 1994. Zahnschmelzdifferenzierungen bei Raubtieren: [Carnivora im Vergleich
zu Vertretern der Creodonta, Arctocyonidae, Mesonuchidae, Entelodontidae
(Placentalia), Thylacoleonidae, Dasyuridae und Thylaacinidae (Marsupialia)].
Ph.D. thesis/dissertation, Rheinischen Friedrich-Wilhelms-Universtat zu Bonn.

Stefen, C. 1997. Chapter 7. Differentiations in Hunter-Schreger bands of carnivores; pp.
123-136 in W. v. Koenigswald and P. M. Sander (eds.), Tooth enamel
microstructure. A.A. Balkema, Rotterdam.

Stefen, C. 1999. Enamel microstructure of recent and fossil Canidae (Carnivora:
Mammalia). Journal of Vertebrate Paleontology 19:576-587.

Stefen, C., and J. M. Rensberger. 1999. The specialized structure of hyaenid enamel:
description and development within the lineage - including percrocutids. Scanning
Microscopy 13:363-380.

Stefen, C., and J. M. Rensberger. 2002. The specialized enamel structure of hyaenids
(Mammalia, Hyaenidae): description and development within the lineage -
including percrocutids. Zoologische Abhandlungen 52:127-147.

Tomes, C. S. 1906. On the minute structure of the teeth of creodonts, with special
reference to their suggested resemblance to marsupials. Proceedings of the
Zoological Society of London 1906:45-58.

Tseng, Z. J. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96:51-67.

Tseng, Z. J., and X. Wang. 2010. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology 271:1386-1398.

Tseng, Z. J., and D. Stynder. 2011. Mosaic functionality in a transitional ecomorphology:
skull biomechanics in stem Hyaeninae compared to modern South African
carnivorans. Biological Journal of the Linnean Society 102:540-559.

Tseng, Z. J., and X. Wang. 2011. Do convergent ecomorphs evolve through convergent
morphological pathways? Cranial shape evolution in fossil hyaenids and
borophagine canids (Carnivora, Mammalia). Paleobiology 37(3):470-489.

Tseng, Z. J., M. Antón, and M. J. Salesa. 2011. The evolution of the bone-cracking model
in carnivorans: Cranial functional morphology of the Plio-Pleistocene cursorial
hyaenid Chasmaporthetes lunensis (Mammalia: Carnivora). Paleobiology 37:140-
156.
419

Van Valkenburgh, B. 1985. Locomotor diversity within past and present guilds of large
predatory mammals. Paleobiology 11:406-428.

Van Valkenburgh, B. 1988. Trophic diversity in past and present guilds of large
predatory mammals. Paleobiology 14:155-173.

Van Valkenburgh, B. 2007. Déjà vu: the evolution of feeding morphologies in the
Carnivora. Integrative and Comparative Biology 47:147-163.

Van Valkenburgh, B. 2009. Costs of carnivory: tooth fracture in Pleistocene and Recent
carnivorans. Biological Journal of the Linnean Society 96:68-81.

Van Valkenburgh, B., and F. Hertel. 1993. Tough times as La Brea: tooth breakage in
large carnivores of the late Pleistocene. Science 261:456-459.

Van Valkenburgh, B., X. Wang, and J. Damuth. 2004. Cope's Rule, hypercarnivory, and
extinction in North American canids. Science 306:101-104.

Wang, X. 1994. Phylogenetic systematics of the Hesperocyoninae (Carnivora: Canidae).
Bulletin of the American Museum of Natural History 221:1-207.

Wang, X., R. H. Tedford, and B. E. Taylor. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243:1-391.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

Werdelin, L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective;
pp. 582-624 in J. L. Gittleman (ed.), Carnivore behavior, ecology, and evolution.
Cornell University Press, New York.

Werdelin, L., and N. Solounias. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30:1-104.

Submitted February, 2011; accepted June 24, 2011

420
Chapter Nine Appendix. Specimens used in the study.

Hesperocyoninae: Hesperocyon gregarius: LACM-CIT 100 (left and right), 94 (left and
right), 621, 1400 (left and right), 1444, 1526, 4917 (left and right), LACM 4932 (left and
right), 4933 (left and right), USNM 15937 (left and right), 450576, 450577;
Paraenhydrocyon josephi: AMNH 6910, F:AM 27753; P. robustus: F:AM 12884;
Mesocyon brachyops: LACM-CIT 1242, 1346, 1819 (left and right), LACM 5265; M.
coryphaeus: UCMP 1383 (left and right), 1307 (left and right), 1165 (left and right);
“Mesocyon” temnodon: F:AM 102381; Sunkahetanka geringensis: LACM 9406, 15910
(left and right); Osbornodon iamonensis: USNM 8836; O. renjiei: F:AM 63316

Borophaginae: Archaeocyon pavidus: F:AM 63970, 50338, LACM-CIT 1338, UCMP
76652 (left and right); Otarocyon macdonaldi: F:AM 38986; Rhizocyon oregonensis:
UCMP 79365; Phlaocyon annectens: F:AM 49006, 50299; P. leucosteus: AMNH 8768; P.
latidens: UCMP 76296; Cormocyon sp.: LACM 2739, 5424 (left and right), 2743;
Paracynarctus kelloggi: UCMP 11474; ?Cynarctus marylandica: USNM 15561;
Metatomarctus sp.: UCMP 12604; Metatomarctus sp. A: UCMP 38290; Microtomarctus
conferta: LACM-CIT 1232 (left and right); Protomarctus optatus: F:AM 61270;
Tephrocyon sp.: UCMP 19460, 19461; Tomarctus hippophaga: F:AM 24270, 61156,
LACM-CIT 774; Tomarctus sp.: LACM 34061, UCMP 24291, 24292 (left and right);
Aelurodon mcgrewi: F:AM 27153, 8307; A. asthenostylus: F:AM 28355; A. ferox:
USNM 352360 (left and right), V523, UCMP 32241; Paratomarctus temerarius: F:AM
421
Chapter Nine Appendix Continued

50146; P. euthos: F:AM 61101, 67121, Paratomarctus sp.: LACM 1377 (left and right);
Carpocyon webbi: F:AM 27366B, 61335, 61336; Protepicyon raki: F:AM 61738;
Epicyon saevus: LACM 59697 (left and right), 59813 (left and right), 127794, F:AM
61418, 67396, USNM 128; E. haydeni: LACM 143519, 127790 (left and right), PPHM
1100, F:AM 61498, 61476, 61552, USNM 127; Borophagus sp.: LACM 34060, 62702,
62703, UCMP 30626; B. pugnator: F:AM 61662; B. parvus: F:AM 108396, LACM
62698; B. secundus: 61690-5, three isolated AMNH specimens, UCMP 30492, 30476; B.
littoralis: LACM 16734 (left and right)

Caninae: Canis dirus: LACM-CIT 1919, 1920 (left and right), 2077 (left and right).

Hyaenidae: Plioviverrops sp.: F:AM 96607; Tungurictis spocki: F:AM 26600;
Thalassictis sp.: IVPP 65001; T. robusta: BM M8983 (left and right), M8984, M8986-
M8989; Ictitherium sp.: F:AM China 4-L94; HMV 0163 (left and right), 0432 (left and
right); I. viverrinum: PMU M26-M27 (left and right), M3706, M3773, M3774;
Hyaenictitherium hyaenoides: AMNH 26371, HMV 0550 (left and right), 0552 (left and
right), 0553, 0556, 0560, 0573 (left and right), 0169 (left and right), IVPP V14737 (left
and right), V14738, PMU Loc. 49 Ex. 19, M3849 (left and right), M3853-M3854 (left
and right), M40, M7186; H. wongi: F:AM China 11-L112, HMV 0751 (left and right),
IVPP Baode Ex. 2, 30, PMU M28-M30 (left and right), M31-M32 (left and right),
422
Chapter Nine Appendix Continued

M3707-M2709 (left and right), M7140, M7141, M7144, M7150, M8142-M8143 (left and
right), M8183-M8185 (left and right), M8192, M8193, M8202; Lycyaena sp.: F:AM
China 38-B296, 45-L400, 26-B47, IVPP V2923; L. chaeretis: BM M8978-M8979 (left
and right); L. dubia: PMU M3856 (left and right); L. macrostoma: BM M13179, M15703;
Chasmaporthetes lunensis: IVPP V15162; C. ossifragus: USNM 10223; C. sp.: AMNH
26369, LACM 74046, LACM-CIT 164 (left and right); AMNH 26370; Pachycrocuta
brevirostris: AMNH 27756, 27757, HMV C0066 (left and right), C0193 (left and right),
IVPP 193013, V13932 (left and right); Adcrocuta eximia: AMNH 26373, 26374, BM
M8966 (left and right), M8967 (left and right), M8968 left and right), M8969 (left and
right), M8970 (left and right), M8971, M8972, M9041, F:AM China 10-L4, HMV 0543,
0577 (left and right), 1435 (left and right), PMU M3859-M3860 (left and right), M55
(left and right), M56 (left and right), M57 (left and right); Crocuta crocuta honanensis:
IVPP L.21625; C. crocuta spelaea: BM M177, M581, M595-M599, M697, M750, M797,
M812, M854, M856, M959, M1080, M1082, M5594, M14173, M14274, M15434,
M16693, M16694, M16696, M16702, M16706, M17982, M17987, M17988, M18273,
M18274, M18982 (three specimens), M18982A, M18982C, M34347, M34348, M44716,
M50020; C. c. ultima: IVPP V2003, V15160, V15163, V15164 (left and right).
423
Figure 9.1. Light microscope examination of enamel Hunter-Schreger Bands (HSB). A,
Regions of tooth crown recorded for HSB pattern in this study (top, middle, bottom),
shown on lower third premolar of Osbornodon renjiei, F:AM 63316; B, Example of
transitional acute-angle HSB to zig-zag HSB, dotted circle indicates the sharp-angled
HSB folding that characterizes full zig-zag HSB; C, Example of undulating HSB with
semi-horizontal light and dark enamel prism bands. Scale bar equals 1 mm for part A
only.

424
Figure 9.2. Intra-dentition variation in HSB patterns in A, fossil hesperocyonine and
borophagine canids and B, hyaenids. Three-letter codes in legend indicate the HSB
pattern represented by each of the three regions of the tooth crown examined, with first
letter representing tip of crown, and third letter bottom of crown. Inferred HSB in incisors
is based on the most specialized HSB that does not supersede cheek tooth specializations,
or from known HSB in one or more teeth in the incisor group. I, incisors; C, canines; P,
premolars; M, molars (lower case indicates lower jaw). Silhouettes on right represent
dorsal skull shapes and relative sizes of some of the genera examined, in order to
demonstrate the disjoint evolution of zig-zag HSB and large, robust skulls typical of
bone-cracking specialists.

425
Chapter Ten: Microstructural evolution and paleodiet indicators

This chapter is currently in press as:

Tseng, Z. J. in press. Connecting Hunter-Schreger Band microstructure to enamel
microwear features: New insights from durophagous carnivores. Acta Palaeontologica
Polonica. doi:10.4202/app.2011.0027.

A copy of the accepted manuscript begins on the next page.
426
Connecting Hunter-Schreger Band microstructure to enamel
microwear features: New insights from durophagous carnivores

ZHIJIE JACK TSENG

Chapter Ten Abstract

Several recent studies have clarified the linkage between microwear features and diet
among living carnivorans, but it is still unclear whether previously interpreted
evolutionary trends for dietary specialization, based on examination of enamel
microstructure, are consistent with such insights from microwear analysis. This study
examined the relationship between microwear and microstructure features using a sample
of fossil hyaenids and canids. Hunter-Schreger Bands (HSB) and microwear features
were examined at the same magnification level using optical stereomicroscopy. Multiple
trials conducted on each specimen showed higher variance of smaller (<0.03 mm)
microwear features compared to large (>0.03 mm) features. The number of pits exhibited
a significant positive relationship with more derived HSB in both p4 and m1 teeth; fossil
teeth with derived HSB possessed microwear features similar to patterns found in modern
Spotted Hyenas. Microscopic scratches were not as closely associated with HSB patterns,
but large scratches were more tightly linked to HSB than smaller ones on p4. An
examination of evolutionary trends in HSB specialization in the two carnivoran lineages
showed that derived HSB patterns evolved prior to the highly robust craniodental
427
characteristics typical of later bone-cracking ecomorphologies. Therefore, the increase of
hard food in the diet of less specialized hyaenids and canids was accompanied by a
mosaic mode of evolution, with microstructural changes preceding key macrostructural
morphological adaptations.

Key words: Borophagine, Canidae, Hyaenidae, durophagy, bone-cracking, Miocene,
Cenozoic

Zhijie Jack Tseng [jtseng@nhm.org] Integrative and Evolutionary Biology Program,
Department of Biological Sciences, University of Southern California, Los Angeles,
California 90089, U.S.A., and Department of Vertebrate Paleontology, Natural History
Museum of Los Angeles County, 900 Exposition Boulevard, Los Angeles, California
90007, U.S.A.

428
Chapter Ten Introduction

Carnivoran enamel microstructure is arranged into Hunter-Schreger Bands (HSB),
specializations of which have evolved convergently across the order (Stefen 1997).
Increased folding of HSB into sharp-angled, “zig-zag” patterns is thought to provide
structural support and biomechanical advantages in the teeth of species that also possess
craniodental specializations for durophagy (Ferretti 1999, Ferretti 2007b, Stefen 1999,
Stefen and Rensberger 1999). In those durophagous ecomorphologies, enamel
microstructural changes are accompanied by macrostructural changes in morphology
such as more robust premolars, deeper skulls, and relatively shorter rostra (Tseng and
Wang In press-b, Van Valkenburgh 2007, Werdelin 1989). Inferences of paleoecology
from morphological features in durophagous species, however, have seldom been linked
with more direct indicators of paleodiet.
Dental microwear features, the patterns of abrasion on teeth from food and other
ingested particles, have been studied as direct indicators that reveal diets of extinct
animals (Peters 1982, Walker et al. 1978). Under paleoecological contexts, microwear
analyses have been applied to a number of vertebrate groups, including dinosaurs
(Williams et al. 2009), crocodilians (Osi and Weishampel 2009), fishes (Purnell et al.
2006), and a wide range of mammals: marsupials (e.g. Robson and Young 1990),
primates (e.g. Daegling and Grine 1994, Scott et al. 2005), rodents (e.g. Rodrigues et al.
2009), sirenians (Domning and Beatty 2007), carnivorans (Anyonge 1996, Goillot et al.
2009, Hagura and Onodera 1987, Peigne et al. 2009, Schubert et al. 2010, Van
429
Valkenburgh et al. 1990), xenarthrans (Green 2009), and most commonly, ungulate
mammals (e.g. Merceron et al. 2004, Kaiser and Brinkmann 2006, Sanson et al. 2007,
Semprebon and Rivals 2007, Joomun et al. 2008, Merceron et al. 2010, among others).
Compared to the large number of studies that interpret paleodiets using microwear
analysis, less commonly has the link between enamel microstructure and microwear
patterns been explored (Maas 1991, Teaford et al. 1996). This study focuses on the
correlation between enamel microstructure and enamel microwear patterns among
durophagous mammalian carnivores, using fossil canids and hyaenids as a case study.
Canidae and Hyaenidae are two carnivoran families that have convergently
evolved bone-cracking ecomorphologies (Werdelin 1996b). Specialized enamel
microstructure in modern spotted hyenas is linked with their bone-rich diet (Kruuk 1972b,
Rensberger and Stefen 2006, Rensberger and Wang 2005). Similar microstructural
specializations are observed in the fossil records of hyaenids (Ferretti 2007b) and
borophagine canids (Stefen 1999). However, it is not known whether microstructural
adaptations for increased structural strength are actually associated with microwear
indicators of more hard food in the diet. Previous microwear studies on durophagous
carnivores found that spotted hyenas possess a larger number of microscopic pits relative
to scratches on their teeth, and also fewer narrow microwear features when compared to
non-durophagous species (Van Valkenburgh et al. 1990). Therefore, I test the hypothesis
that HSB specialization is linked to increasing number of large microwear features and
also more pits relative to scratches.
430
Institutional abbreviations ―AMNH, American Museum of Natural History,
New York, USA; F:AM, Frick Collection, American Museum of Natural History; HMV,
Hezheng Paleozoology Museum, Gansu, China; IVPP, Institute of Vertebrate
Paleontology and Paleoanthropology, Beijing, China; LACM, Natural History Museum
of Los Angeles County, California, USA; MVZ, Museum of Vertebrate Zoology,
University of California, Berkeley, California, USA; PPHM, Plain-Panhandle Museum,
Texas, USA; UAMZ, University of Alberta Museum of Zoology, Alberta, Canada;
UCMP, University of California Museum of Paleontology, Berkeley, California, USA;
USNM, National Museum of Natural History (also NMNH), Smithsonian Institution,
Washington D.C., USA.

Chapter Ten Material and Methods

Microwear analysis—Both microwear and microstructure methods used in this
study were non-invasive, and therefore could be applied across both fossil and extant
specimens without destructive sampling. 64 p4 specimens and 75 m1 specimens were
examined in the study. The lower fourth premolar (p4) is used to crack bones, and the
lower first molar (m1; carnassial) is used to both cut flesh and crack bones in the
durophagous Spotted Hyena (Kruuk 1972). Specimens included fossil hyaenids and
canids, as well as the extant Spotted Hyena Crocuta crocuta and African Hunting Dog
Lycaon pictus (Chapter Ten Appendices 1-2). All specimens were cleaned with cotton
swabs and either acetone or ethanol solutions. Specimens were then examined under an
431
optical stereomicroscope to make sure the enamel surface was free of adhesives or
sediment matrix. Molds were made with a vinyl polysiloxane material (3M ESPE Dental
Products, St. Paul, MN, USA) through a dispensing gun as described in Solounias and
Semprebon (2002). All specimens were molded twice, with the first mold as a cleaning
layer, which was discarded. Second molds were then cast using high-resolution,
spectrally clear, optical grade epoxy resin (EPO-TEK 301, Epoxy Technology Inc.,
Billerica, MA, USA). The casts were evacuated in a vacuum chamber for 20 minutes, and
then left to set at room temperature for five days.
Examination for microwear features was done using an optical stereomicroscope
at 30x magnification (compared to 35x in Solounias and Semprebon 2002). Microwear
features were recorded inside an 1 x 1 mm
2
area, which was delineated on the labial edge
of the p4 and m1 wear facet using a crossed stage micrometer scale fitted inside the
eyepiece of the microscope (Fig. 10.1). Care was taken to examine only areas on the
enamel portion of the tooth crown, as opposed to the dentine area present on the shear
facet of the lower molar (Fig. 10.1); weathering and laboratory preparation can easily
create artificial features on the dentine, which is a softer material than enamel. Features
were either scored as scratches (i.e. length:width ratio > 4:1) or pits (i.e. length:width
ratio < 4:1). At 30x, all features with widths smaller than the width between two
micrometer ticks (<0.03 mm) were scored as small, and others larger than 0.03 mm were
scored as large. As such, four categories of features were analyzed: small versus large
scratches, and small versus large pits (Fig. 10.2).
432
All specimens were examined three times by the author, with the interval between
any two trials on the same specimen ranging from four days to six months apart.
Analyses were done in ~4 hour sittings, with random sampling of specimens in no
particular order of examination. No attempts were made to analyze identical 1 x 1 mm
2
areas on the tooth during each trial, as the trials were intended to include a sample of
different areas on the labial face of the wear facets. It should be noted that the methods
employed in this study were aimed to address the specific question of the connection
between microwear signals and enamel microstructure in durophagous hyaenids and
canids; results from other microwear studies should not be compared directly to the data
in this study. Furthermore, results obtained from replication of the methods used here in
studies of other taxa also should not be directly compared to results in the current study,
as such comparisons are potentially affected by other taxon-specific factors with yet
unknown influence on microwear patterns (see discussion).
Microstructure analysis—All of the fossil specimens examined in the
microwear analysis were also analyzed for Hunter-Schreger Band specialization (Stefen
1997). Specimens with lighter colored enamel preservation and clean surfaces can be
examined directly with an oblique light source, illuminating HSB patterns without the
necessity of sectioning the specimens (Koenigswald 1980, Stefen and Rensberger 1999).
Only specimens that fit such criteria were chosen for both microwear and microstructure
analyses. At the time of cleaning and molding, the original specimens were analyzed with
an optical stereomicroscope at 30x magnification. HSB patterns were recorded at three
levels on the tooth crown: the top, middle, and bottom 33% of the crown (Fig. 10.3). The
433
most commonly found (> 50% of the region) HSB in each region was recorded as
representative of that region; if two HSB categories are equally common in a given
region, the more derived HSB pattern is recorded. As previously categorized by Stefen
(1997), the most derived HSB pattern in carnivorans is zig-zag HSB, followed by acute-
angled undulating HSB, and then undulating HSB (Fig. 10.3B-D).
To obtain a context for evolutionary changes in HSB patterns across different
tooth positions in fossil hyaenids and canids, additional specimens were examined for
HSB only (Chapter Ten Appendix 3). The entire lower dentition, in addition to p4 and
m1, was subject to HSB analysis as described above. Where possible, complete dentaries
were used, supplemented by isolated specimens as appropriate (Chapter Ten Appendix 3).
An abbreviation system was created to refer to different degrees of HSB specialization: a
single letter representing HSB pattern at each of the three levels of the crown, with “z”
for zig-zag, “a” for acute-angled undulating, and “u” for undulating HSB (e.g. “zau” for a
tooth with zig-zag HSB at the top 33% of the crown, acute-undulating HSB at the middle,
and undulating HSB at the bottom of the crown).

Chapter Ten Results

Microwear features exhibited different degrees of intra-tooth variation across the
three trials conducted on each specimen; large microwear features (>0.03 mm) tend to
have lower variance than smaller (<0.03 mm) features. In addition, m1 tend to have more
variable counts in microwear features compared to p4 (Fig. 10.4). For all subsequent
434
comparisons of microwear feature counts, mean values of the trials were used to
represent each specimen.
Specimens containing at least 33% zig-zag HSB had a larger range of small and
large pit counts compared to teeth with less specialized HSB; however, this is true for
only p4, but not m1 (Fig. 10.5A-B, E-F). Number of small scratches showed no trend
across different HSB categories, either in p4 or m1 (Fig. 10.5C, G). Large scratches in p4
specimens with all zig-zag HSB were more numerous than in all less derived teeth, and
the former approaches the count in Crocuta; no such pattern was present in m1 specimens
(Fig. 10.5D, H).
In order to achieve comparable sample sizes for statistical analysis, HSB
categories were combined into three bins, and analyses of variance (ANOVA) were
conducted on the categories along with the modern Crocuta sample (Table 10.1, Fig.
10.6). Number of pits (both large and small) was significantly different across HSB
categories on p4, and the increase was incremental (Fig. 10.6A-B). Number of pits was
also significantly different on m1, but the number of large pits among fossil specimens
did not approach the range observed in modern Crocuta (Fig. 10.6F). Among scratches,
only large scratches on p4 were significantly different across categories; such increase
was incremental as observed for large pits on p4 (Fig. 10.6D).
In the evolutionary context, derived zig-zag HSB (zau to zzz) were not observed
among the hesperocyonine canids examined; in borophagine canids zig-zag HSB were
observed in the m1 of Microtomarctus and all subsequent genera (Fig. 10.7A). Zig-zag
HSB spread throughout the premolar dentition in more derived borophagines, and in
435
Epicyon, specimens with all zig-zag HSB across p1-m2 were observed. Zig-zag HSB
were present in the premolars and molar of the hyaenid Ictitherium, and in most of the
subsequent hyaenid genera (Fig. 10.7B). Cheek tooth dentitions in hyaenids became
reduced in the derived, larger-bodied genera, in which p1 and m2 were lost. The
remaining cheek teeth in those genera retained full zig-zag HSB patterns.

Chapter Ten Discussion

Microwear analysis on the p4 and m1 teeth of fossil hyaenids and canids showed
a general pattern of increasing number as well as variability of large and small pits with
more derived HSB (Fig. 10.5A-B, E, F; Fig. 10.6A-B, E, F). These findings are
consistent with those of Van Valkenburgh et al. (1990), who found relatively higher
quantity of pits and higher variability of microwear features in carnivorans with more
hard food in their diets. These trends have been shown to be correlated to increasing HSB
specialization in this study, supporting the initial hypothesis that derived HSB is
associated with increasing number of pits and variability of features.
These results are also consistent with previous functional interpretations of the
derived zig-zag HSB in allowing a more durophagous diet (Stefen and Rensberger 1999).
The intensely folded layers of enamel prisms of zig-zag HSB is thought to function in
resisting the large stresses incurred from consumption of hard food such as bone; in
addition, the layered structure of HSB can also shorten or halt crack propagation in
instances of local enamel failure (Chai et al. 2009). Accordingly, zig-zag HSB evolved
436
convergently not only in fossil canids and hyaenids, but also in other carnivoran lineages
with tendencies to consume hard food items (Stefen 2001).
On the other hand, the number and variability of scratches showed no clear trend
with increasing HSB specialization (Fig. 10.5C-D, G-H; Fig. 10.6C-D, G-H). The only
results in this aspect that supported the hypothesis of larger features with HSB
specialization are the number of large scratches on p4 specimens (Fig. 10.6D). Feature
orientation, which was not analyzed in this study, could distinguish HSB patterns on the
basis of scratch distribution. Differences in the proportion of parallel scratches are
expected not only in teeth that occlude more precisely (e.g. carnassial teeth) versus teeth
that occlude in between other teeth (e.g. carnivoran second to fourth premolars), but also
in the teeth of durophagous species on which scratches are more randomly distributed
(Van Valkenburgh et al. 1990). Nevertheless, the overlap in number of large and small
scratches between extant Crocuta and Lycaon specimens indicate that such counts do not
differentiate between modern bone-cracking and soft-meat hypercarnivores in the context
of this study (Fig. 10.5C-D, G-H). Therefore, the trends that are present across different
categories of HSB specialization are more in line with microwear features (i.e. large pits)
that are clearly associated with durophagy, supporting the interpretation of evolutionary
specialization of HSB as being in close association with increase of hard food in the diet
(Ferretti 2007b, Goillot et al. 2009, Stefen 1999).
If we apply this close association between a hard-food rich diet and increasingly
derived HSB to the examination of evolutionary patterns of HSB change in fossil
hyaenids and canids, the resulting trend shows that zig-zag HSB in both lineages
437
appeared prior to the large-bodied bone-cracking ecomorphologies typically recognized
from craniodental morphology (Fig. 10.7). The borophagine genera Aelurodon, Epicyon,
and Borophagus, and the hyaenid genera Pachycrocuta, Adcrocuta, and Crocuta are
commonly categorized as bone-crackers (Van Valkenburgh and Koepfli 1993, Wang et al.
1999, Werdelin 1989). However, zig-zag HSB evolved no later than in the canid
Microtomarctus and the hyaenid Ictitherium (Fig. 10.7). Such a disjoint in the timing of
the appearance of robust craniodental morphology, and a more continuous increase of
hard food in the diet implied by HSB specialization, indicate a mosaic mode of character
evolution. In other words, the suite of morphological features present in the derived bone-
cracking canids and hyaenids, including both microstructural and macrostructural
characteristics, evolved in stages. Microstructural changes occurred prior to
macrostructural ones, as a response to increasingly durophagous diets inferred from
microwear patterns (Figs. 10.6-7). Such a mode of adaptive change may be pervasive
especially in the gradual evolution of hyaenids, as evidence of mosaic morphological
evolution have also been found in overall craniodental morphology, exemplified by the
intermediate biomechanical adaptations of the transitional bone cracker Ikelohyaena
abronia (Tseng and Stynder 2011).
Interestingly, p4 and m1 teeth shared similar numbers of microscopic scratch
features, but m1 teeth tend to have a much larger number of both large and small pits (Fig.
10.5A-B, E-F). Larger quantities of microwear features on the m1 may reflect the
important function of the tooth as a main flesh-shearing tool, which is also used by extant
Crocuta crocuta in cracking hard food items (Van Valkenburgh 1996). It is also likely
438
that at least some of the microscopic pits resulted from corrosion by stomach acids, as
regurgitation is relatively common in Crocuta crocuta (Kruuk 1972, Van Valkenburgh
1996). Nevertheless, microwear patterns from both p4 and m1 support a close
relationship to HSB specialization, and evolutionary trends show that the entire cheek
dentition was eventually specialized for increased durophagy in borophagine canids and
hyaenids (Fig. 10.7). Tooth crown attrition in both modern hyaenids and derived fossil
hyaenids and borophagine canids shows that all cheek teeth can be heavily worn,
indicating their use in durophagy (personal observation).
There have been few published studies exploring the relationship between
microstructure and microwear patterns. Maas (1991) conducted experiments to examine
the relationship between prism structure, shear force, and abrasive particle size; she found
that the size and quantity of microscopic scratches can be confounded by the type of
prism arrangement present in the enamel of a given species. Teaford et al. (1996) found
that enamel crystallite and prism microstructure and microwear data provided
complementary information that allowed a more nuanced reconstruction of the paleodiets
of Oligocene anthropoids. However, no study has specifically looked at microwear and
microstructure across types of Hunter Schreger Bands. The findings from the current
study, even though narrow in scope, provide evidence that such microstructure-
microwear comparisons at the HSB level are also informative. Even so, care should still
be taken in interpreting or extrapolating the results of this study to other durophagous
taxa; dental wear patterns observed on the enamel likely resulted from complex
interactions among body size, masticatory musculature, behavior, etc., in addition to
439
morphology and microstructure. The interactions among these physical and physiological
factors still need to be clarified with carefully designed studies.
Multiple trials of microwear analysis conducted in this study showed that larger
features tend to be less variable within each tooth across examinations than smaller
features (Fig. 10.4). Smaller features may be more variable simply because a larger
number of them can occur within the defined areas sampled on each tooth. Such a trend
may also reflect an inherent source of error based on the inability of the observer to
discern increasingly miniscule features (Palmeirim 1998); software-based microwear
analytical methods have been proposed in order to make counting of features more
objective (Scott et al. 2005, Ungar 1995). Application of a software-based approach may
clarify the high variability and lack of correlation of microscopic scratches to HSB
patterns found by the current study, but the general agreement of results with previous
microwear studies of carnivorans indicates that a consistent trend in microwear features is
present, regardless of the type of method employed (see below). Along the same lines,
agreement of results based on different levels of magnification (SEM, confocal
microscopy, high-magnification stereomicroscopy, low-magnification stereomicroscopy)
shows that dietary signals for durophagy are present across multiple levels of scale, at
least in carnivorans (Van Valkenburgh 1990, Goillot et al. 2009, Schubert et al. 2010, this
study).
More recently, the application of microwear texture analysis to the carnassial
teeth of Crocuta crocuta has provided promising results for correlating bone
consumption with enamel surface complexity (Schubert et al. 2010). Results from texture
440
analysis supported the main findings of Van Valkenburgh et al. (1990), which were based
on Scanning Electron Microscope images; taken together, the durophagous C. crocuta
tend to show larger number of pits, a wider range of feature sizes, as well as more
variable scratch directions relative to the main occlusal axis (Schubert et al. 2010, Van
Valkenburgh et al. 1990). Results from the current study are consistent with these
findings, and indicate that HSB microstructure specialization is likely to be positively
correlated to additional characteristic features of durophagous taxa found using SEM and
texture analysis as well. One caveat to be noted here is that this and other studies cited
above were all conducted on carnivoran mammals, therefore caution should be taken
when discussing durophagy in other taxa. Microwear patterns for similar diets may differ
in unrelated taxa, and such differences need to be quantified before any direct
comparisons are made.
In conclusion, a sample of fossil hyaenid and canid specimens were examined for
HSB microstructure and microwear features, and a connection between the number of
pits and the degree of HSB specialization was found. The correlation between HSB and
microscopic scratches was not as strong; in general, patterns on p4 are more pronounced
than those on m1. The first appearance of zig-zag HSB in borophagine canids and
hyaenids occurred prior to the evolution of robust craniodental morphology. The
association between increasing HSB specialization and more microwear features
indicative of a durophagous diet both indicate that hard food was already being
incorporated in the diet of less specialized hyaenids and borophagines. Therefore, the
441
present adaptations for durophagy seen in Spotted Hyenas have evolved in a mosaic
manner, with microstructural adaptations appearing before macrostructural modifications.

Chapter Ten Acknowledgments

I thank X. Wang and my Ph.D. committee for discussion and guidance. J. Galkin,
J. Kelly, and J. Meng helped with access to and cleaning agents for studying the AMNH
collection; P. Holroyd and S. Tomiya (UCMP), C. Conroy and E. Lacey (MVZ), R.
Purdy and M. Carrano (NMNH), J. Indeck (PPHM), Z. Qiu and Z. Liu (IVPP), and S.
Chen and W. He (HMV) all assisted in the study of collections in their care. J. Liu
(University of Alberta) helped with access to the UAMZ collection for microstructure
photography; M.V.H. Wilson (University of Alberta) provided microscopes for analysis.
B. Beatty, J. Rensberger, and the editor provided constructive comments that improved
the manuscript. This research was funded by a National Science Foundation (U.S.)
Graduate Research Fellowship and a Doctoral Dissertation Improvement Grant (DEB-
0909807), and an AMNH Collection Study Grant.
442
Chapter Ten References
Anyonge, W. 1996. Microwear on canines and killing behavior in large carnivores: saber
function in Smilodon fatalis. Journal of Mammalogy 77 (4): 1059-1067.

Chai, H., Lee, J.J.-W., Constantino, P., Lucas, P.W., and Lawn, B. 2009. Remarkable
resilience of teeth. Proceedings of the National Academy of Sciences 106 (18):
7289-7293.

Daegling, D.J. and Grine, F.E. 1994. Bamboo feeding, dental microwear, and diet of the
Pleistocene ape Gigantopithecus blacki. South African Journal of Science 90:
527-532.

Domning, D.P. and Beatty, B.L. 2007. Use of tusks in feeding by dugongid sirenians:
Observations and tests of hypotheses. Anatomical Record 290 (6): 523-538.

Ferretti, M.P. 1999. Tooth enamel structure in the hyaenid Chasmaporthetes lunensis
lunensis from the late Pliocene of Italy, with implications for feeding behavior.
Journal of Vertebrate Paleontology 19 (4): 767-770.

Ferretti, M.P. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience 100: 41-52.

Goillot, C., Blondel, C., and Peigne, S. 2009. Relationships between dental microwear
and diet in Carnivora (Mammalia) - Implications for the reconstruction of the diet
of extinct taxa. Palaeogeography Palaeoclimatology Palaeoecology 271: 13-23.

Green, J.L. 2009. Dental microwear in the orthodentine of the Xenarthra (Mammalia) and
its use in reconstructing the palaeodiet of extinct taxa: the case study of
Nothrotheriops shastensis (Xenarthra, Tardigrada, Nothrotheriidae). Zoological
Journal of Linnean Society 156: 201-222.

Hagura, N. and Onodera, S. 1987. A secular and experimental study of the microwear of
the carnassial teeth in dogs (Canis familiaris). Journal of the Mammal Society of
Japan 12 (1-2): 41-55.

Joomun, S.C., Hooker, J.J., and Collinson, M.E. 2008. Dental wear variation and
implications for diet: An example from Eocene perissodactyls (Mammalia).
Palaeogeography Palaeoclimatology Palaeoecology 263: 92-106.

Kaiser, T.M. and Brinkmann, G. 2006. Measuring dental wear equilibriums- the use of
industrial surface texture parameters to infer the diets of fossil mammals.
Palaeogeography Palaeoclimatology Palaeoecology 239: 221-240.

443
Koenigswald, W.v. 1980. Schmelzmuster und Morphologie in den Molaren der
Arvicolidae (Rodentia). Adhandlungen der Senckenbergischen naturforschenden
Gesellschaft 539: 1-129.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior. 335 pp.
The University of Chicago Press, Chicago.

Maas, M.C. 1991. Enamel Structure And Microwear An Experimental Study Of The
Response Of Enamel To Shearing Force. American Journal of Physical
Anthropology 85 (1): 31-50.

Merceron, G., Blondel, C., Brunet, M., Sen, S., Solounias, N., Viriot, L., and Heintz, E.
2004. The Late Miocene paleoenvironment of Afghanistan as inferred from dental
microwear in artiodactyls. Palaeogeography Palaeoclimatology Palaeoecology
207: 143-163.

Merceron, G., Escarguel, G., Angibault, J.-M., and Verheyden-Tixier, H. 2010. Can
dental microwear textures record inter-individual dietary variations? PLoS ONE 5
(3): e9542.

Osi, A. and Weishampel, D.B. 2009. Jaw mechanism and dental function in the late
Cretaceous basal eusuchian Iharkutosuchus. Journal of Morphology 270 (8): 903-
920.

Palmeirim, J.M. 1998. Analysis of skull measurements and measurers: can we use data
obtained by various observers? Journal of Mammalogy 79 (3): 1021-1028.

Peigne, S., Goillot, C., Germonpre, M., Blondel, C., Bignon, O., and Merceron, G. 2009.
Predormancy omnivory in European cave bears evidenced by a dental microwear
analysis of Ursus spelaeus from Goyet, Belgium. Proceedings of the National
Academy of Sciences 106 (36): 15390-15393.

Peters, C.R. 1982. Electron-optical microscopic study of incipient dental microdamage
from experimental seed and bone crushing. American Journal of Physical
Anthropology 57: 283-301.

Purnell, M.A., Hart, P.J.B., Baines, D.C., and Bell, M.A. 2006. Quantitative analysis of
dental microwear in threespine stickleback: a new approach to the analysis of
trophic ecology in aquatic vertebrates. Journal of Animal Ecology 75 (4): 967-977.

Rensberger, J.M. and Stefen, C. 2006. Functional differentiations of the microstructure in
the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A 278
(1-6): 149-162.

444
Rensberger, J.M. and Wang, X. 2005. Microstructural reinforcement in the canine enamel
of the hyaenid Crocuta crocuta, the felid Puma concolor, and the late Miocene
canid Borophagus secundus. Journal of Mammalian Evolution 12 (3-4): 379-403.

Robson, S.K. and Young, W.G. 1990. A comparison of tooth microwear between an
extinct Marsupial predator, the Tasmanian tiger Thylacinus cynocephalus
(Thylacinidae) and an extant scavenger, the Tasmanian devil Sacrophilus harrisii
(Dasyuridae: Marsupialia). Australian Journal of Zoology 37: 575-589.

Rodrigues, H.G., Merceron, G., and Viriot, L. 2009. Dental microwear patterns of extant
and extinct Muridae (Rodentia, Mammalia): ecological implications.
Naturwissenschaften 96: 537-542.

Sanson, G.D., Kerr, S.A., and Gross, K.A. 2007. Do silica phytoliths really wear
mammalian teeth? Journal of Archaeological Science 34: 526-531.

Schubert, B.W., Ungar, P.S., and DeSantis, L.R.G. 2010. Carnassial microwear and
dietary behaviour in large carnivorans. Journal of Zoology 280 (3): 257-263.

Scott, R.S., Ungar, P.S., Bergstrom, T.S., Brown, C.S., Grine, F.E., Teaford, M.F., and
Walker, A. 2005. Dental microwear texture analysis shows within-species diet
variability in fossil hominins. Nature 436: 693-695.

Semprebon, G. and Rivals, F. 2007. Was grass more prevalent in the pronghorn past? An
assessment of the dietary adaptations of Miocene to Recent Antilocapridae
(Mammalia: Artiodactyla). Palaeogeography Palaeoclimatology Palaeoecology
253: 332-347.

Solounias, N. and Semprebon, G. 2002. Advances in the reconstruction of ungulate
ecomorphology with application to early fossil equids. American Museum
Novitates 3366: 1-49.

Stefen, C. 1997. Chapter 7. Differentiations in Hunter-Schreger bands of carnivores. In:
W.v. Koenigswald, and P.M. Sander (eds.), Tooth enamel microstructure, 123-
136. A.A. Balkema, Rotterdam.

Stefen, C. 1999. Enamel microstructure of recent and fossil Canidae (Carnivora:
Mammalia). Journal of Vertebrate Paleontology 19 (3): 576-587.

Stefen, C. 2001. Enamel structure of arctoid Carnivora: Amphicyonidae, Ursidae,
Procyonidae, and Mustelidae. Journal of Mammalogy 82 (2): 450-462.

445
Stefen, C. and Rensberger, J.M. 1999. The specialized structure of hyaenid enamel:
description and development within the lineage - including percrocutids. Scanning
Microscopy 13 (2-3): 363-380.

Teaford, M.F., Mass, M.C., and Simons, E.L. 1996. Dental microwear and microstructure
in early Oligocene primates from the Fayum, Egypt: Implications for diet.
American Journal of Physical Anthropology 101 (4): 527-543.

Tseng, Z.J. and Stynder, D. 2011. Mosaic functionality in a transitional ecomorphology:
skull biomechanics in stem Hyaeninae compared to modern South African
carnivorans. Biological Journal of the Linnean Society 102: 540-559.

Tseng, Z.J. and Wang, X. 2011. Do convergent ecomorphs evolve through convergent
morphological pathways? Cranial shape evolution in fossil hyaenids and
borophagine canids (Carnivora, Mammalia). Paleobiology 37 (3): 470-489.

Ungar, P.S. 1995. A semiautomated image analysis procedure for the quantification of
dental microwear II. Scanning 17: 57-59.

Van Valkenburgh, B. 1996. Feeding behavior in free-ranging, large African carnivores.
Journal of Mammalogy 77 (1): 240-254.

Van Valkenburgh, B. 2007. Déjà vu: the evolution of feeding morphologies in the
Carnivora. Integrative and Comparative Biology 47 (1): 147-163.

Van Valkenburgh, B. and Koepfli, K.-P. 1993. Cranial and dental adaptations to
predation in canids. Symposium of the Zoological Society of London 65: 15-37.

Van Valkenburgh, B., Teaford, M.F., and Walker, A. 1990. Molar microwear and diet in
large carnivores: inferences concerning diet in the sabretooth cat, Smilodon fatalis
Journal of Zoology 222: 319-340.

Walker, A., Hoeck, H.N., and Perez, L. 1978. Microwear of mammalian teeth as
indicator of diet. Science 201: 908-910.

Wang, X., Tedford, R.H., and Taylor, B.E. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243: 1-391.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15: 387-401.

446
Werdelin, L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
In: J.L. Gittleman (ed.), Carnivore behavior, ecology, and evolution, 582-624.
Cornell University Press, New York.

Williams, V.S., Barrett, P.M., and Purnell, M.A. 2009. Quantitative analysis of dental
microwear in hadrosaurid dinosaurs, and the implications for hypotheses of jaw
mechanics and feedings. Proceedings of the National Academy of Sciences 106
(27): 11194-11199.

447
Chapter Ten Appendix 1. Premolar specimens used in the study. SL, large scratches; SS,
small scratches; PL, large pits; PS, small pits.

Species Specimen # HSB SL.mean SS.mean PL.mean PS.mean
Lycaon pictus MVZ117806 - 1.00 14.67 1.00 1.00
MVZ124258 - 1.00 7.00 1.33 4.67
MVZ184089 - 1.33 12.33 0.00 0.67
MVZ4842 - 2.33 11.00 0.00 0.67
Cynarctus sp. F:AM27543 uuu 2.67 6.67 0.33 1.33
F:AM49405 uuu 2.00 2.67 0.00 0.00
Phlaocyon sp. F:AM49006 uuu 0.00 6.33 0.00 0.00
Phlaocyon leucosteus F:AM8768 uuu 1.00 4.67 1.33 1.33
Phlaocyon minor F:AM27578 uuu 0.00 4.67 0.00 0.00
Tomarctus
brevirostris F:AM61121 uuu 2.00 1.67 2.00 0.67
F:AM61127 uuu 1.33 11.00 6.00 5.33
Tomarctus
hippophagus F:AM27232 uuu 0.67 2.00 0.67 0.33
F:AM27505 uuu 2.33 15.00 0.00 1.33
F:AM61213 uuu 3.33 17.33 2.33 2.33
Hyaenictitherium
wongi F:AM4-L34 aau 0.33 17.00 0.00 0.67
Ictitherium sp. F:AM10-L26 aau 2.00 20.00 0.33 1.67
F:AM11-L131 aau 3.33 18.33 0.33 1.33
Aelurodon ferox F:AM61746 zau 3.67 16.00 2.00 5.33
F:AM61749 zau 1.33 5.00 2.33 3.33
F:AM61768 zau 0.00 5.67 3.67 3.67
448
Chapter Ten Appendix 1 Continued
F:AM67372 zau 2.67 8.00 2.67 2.33
Aelurodon taxoides UCMP33478 zau 2.33 10.00 1.00 5.33
Chasmaporthetes
lunensis F:AM99786 zau 2.00 16.33 1.00 4.00
Lycyaena chaeretis F:AM26-B47 zau 2.67 17.33 2.00 7.33
F:AM56-L560 zau 0.33 6.67 0.00 0.00
Hyaenictitherium
wongi F:AM11-L114 zau 1.00 14.33 0.67 3.67
F:AM42-L357 zau 4.00 16.67 0.67 1.00
F:AM93-B1000 zau 3.00 11.67 2.67 4.67
F:AMB-L15 zau 1.00 13.33 0.00 6.00
Adcrocuta eximia F:AM26372 zza 0.33 13.67 2.00 3.00
Aelurodon ferox F:AM70624 zza 1.67 5.67 6.67 3.67
Chasmaporthetes
lunensis F:AM99788 zza 1.00 14.67 0.67 0.00
Epicyon haydeni F:AM61524 zza 2.67 11.00 3.00 3.33
F:AM61531 zza 1.33 11.33 3.00 7.00
Adcrocuta eximia F:AM57-L533 zzz 1.00 12.33 1.33 0.00
Epicyon haydeni F:AM61494A zzz 4.00 7.67 2.00 7.67
F:AM61532 zzz 3.00 9.33 5.00 5.00
F:AM61540 zzz 5.33 2.67 4.33 4.67
USNM127 zzz 4.00 16.00 1.00 12.00
Pachycrocuta
brevirostris F:AM107781 zzz 6.33 3.67 1.00 0.00
Pliocrocuta perrieri F:AM107767 zzz 2.67 18.67 1.00 2.00
449
Chapter Ten Appendix 1 Continued
Crocuta crocuta MVZ165160 - 3.67 13.67 4.67 7.67
MVZ165163 - 4.67 16.67 2.67 4.33
MVZ165165 - 6.00 14.33 5.00 2.67
MVZ165166 - 4.33 10.33 2.00 2.67
MVZ165169 - 4.67 9.33 4.00 6.00
MVZ165170 - 1.33 16.00 1.33 2.33
MVZ165175 - 2.33 12.33 3.67 4.67
MVZ165176 - 3.00 12.67 4.00 5.00
MVZ165179 - 4.67 16.33 2.00 3.00
MVZ165181 - 3.67 12.33 2.00 8.67
MVZ165182 - 1.33 8.67 2.00 4.33
MVZ173733 - 4.00 13.00 4.67 8.67
MVZ173734 - 1.67 20.00 4.67 4.67
MVZ173743 - 2.00 11.33 5.00 6.33
MVZ173744 - 2.00 8.33 0.00 4.33
MVZ173746 - 5.00 11.67 3.00 4.33
MVZ173747 - 3.00 12.67 6.00 12.00
MVZ173756 - 4.00 10.00 4.00 8.67
MVZ173761 - 2.00 10.33 0.00 2.67
MVZ173768 - 3.00 11.00 3.33 5.33
MVZ173770 - 3.33 11.33 1.67 6.67
MVZ173771 - 2.33 9.67 2.00 5.67
MVZ184551 - 3.00 13.00 3.00 6.67

450
Chapter Ten Appendix 2. Molar specimens used in the study. Borophagus specimens are
corresponding upper fourth premolars.

Species Specimen # HSB SL.mean SS.mean PL.mean PS.mean
Lycaon pictus MVZ117806 - 1.33 8.67 3.33 12.00
MVZ124258 - 2.33 7.33 0.33 17.33
MVZ184089 - 3.00 8.33 1.67 16.00
MVZ4842 - 2.67 6.33 1.00 4.33
Cynarctus sp. F:AM27543 uuu 3.00 3.67 1.33 0.00
F:AM49405 uuu 0.00 2.00 0.00 0.67
Phlaocyon sp. F:AM49006 uuu 0.00 5.00 0.00 6.33
Phlaocyon
leucosteus F:AM8768 uuu 2.67 10.33 0.00 7.00
Phlaocyon minor F:AM27578 uuu 0.67 9.33 0.00 5.00
Tomarctus
brevirostris F:AM61121 uuu 3.00 3.67 5.00 0.33
F:AM61127 uuu 1.67 4.00 7.33 1.00
Tomarctus
hippophagus F:AM27232 uuu 0.67 4.00 0.67 4.00
F:AM27505 uuu 2.67 4.67 0.67 5.00
F:AM61213 uuu 1.00 11.00 3.67 11.67
Hyaenictitherium
wongi F:AM4-L34 auu 1.33 15.67 0.00 1.67
Ictitherium sp. F:AM10-L26 aau 1.67 15.67 0.00 0.67

F:AM11-
L131 aau 2.00 13.33 1.00 5.33
Aelurodon ferox F:AM61746 zau 2.67 9.00 2.33 5.33
451
Chapter Ten Appendix 2 Continued
F:AM61749 zau 0.00 5.33 1.33 0.00
F:AM61768 zau 2.67 3.00 3.00 7.33
F:AM67372 zau 3.33 7.00 5.00 3.33
Aelurodon taxoides UCMP33478 zau 2.00 12.67 3.33 0.00
Chasmaporthetes
lunensis F:AM99786 zau 6.33 11.00 0.67 1.67
Lycyaena chaeretis
F:AM56-
L560 zau 0.00 13.67 0.00 7.33
Hyaenictitherium
wongi
F:AM11-
L114 zau 1.67 11.67 0.67 0.67

F:AM42-
L357 zau 3.00 13.33 2.00 6.67

F:AM93-
B1000 zau 2.00 13.67 1.33 3.33
F:AMB-L15 zau 0.33 8.33 0.00 3.00
Adcrocuta eximia F:AM26372 zza 2.00 17.00 0.00 8.33
Aelurodon ferox F:AM70624 zza 2.67 5.00 3.00 5.67
Chasmaporthetes
lunensis F:AM99788 zza 1.67 11.67 1.67 3.00
Lycyaena chaeretis F:AM26-B47 zza 0.33 4.00 3.33 6.33
Epicyon haydeni F:AM61524 zza 3.33 9.67 3.00 8.00
F:AM61531 zza 2.67 7.33 2.67 7.33
Adcrocuta eximia
F:AM57-
L533 zzz 1.33 12.33 0.00 0.00

452
Chapter Ten Appendix 2 Continued
Borophagus
secundus UCMP30102 zzz 0.00 13.00 1.67 18.67
UCMP30103 zzz 3.00 17.00 2.00 17.67
UCMP30116 zzz 3.00 8.33 2.00 22.00
UCMP30129 zzz 2.33 10.00 4.33 21.33
UCMP30130 zzz 2.33 9.33 4.00 13.33
UCMP30131 zzz 1.67 8.33 3.00 9.00
UCMP30132 zzz 1.67 6.67 1.67 16.67
UCMP30136 zzz 1.33 9.33 2.67 14.33
UCMP30478 zzz 1.00 10.67 1.00 7.67
UCMP30479 zzz 2.33 7.00 4.33 13.00
UCMP30482 zzz 0.33 6.00 4.33 30.00
UCMP30484 zzz 0.33 9.00 2.67 27.00
UCMP30486 zzz 1.00 10.00 2.67 24.33
UCMP30668 zzz 1.67 6.00 4.00 11.00
Epicyon haydeni F:AM61494A zzz 2.67 9.67 2.67 3.00
F:AM61532 zzz 2.67 8.00 6.67 10.00
F:AM61540 zzz 4.00 8.00 3.33 5.67
USNM127 zzz 7.00 6.00 2.00 8.00
Pachycrocuta
brevirostris F:AM107781 zzz 2.33 2.67 1.33 0.00
Pliocrocuta perrieri F:AM107767 zzz 1.00 12.33 2.00 6.33
Crocuta crocuta MVZ165160 - 2.33 22.00 1.33 4.67
MVZ165163 - 1.33 1.67 9.00 18.00
MVZ165165 - 1.33 6.33 5.00 10.67
453
Chapter Ten Appendix 2 Continued
MVZ165166 - 1.33 7.67 2.33 6.00
MVZ165169 - 1.00 5.33 6.67 6.67
MVZ165175 - 0.67 7.33 7.33 7.33
MVZ165176 - 2.67 7.00 3.67 8.00
MVZ165179 - 1.67 15.33 9.67 17.67
MVZ165181 - 1.00 15.00 4.00 21.67
MVZ165182 - 0.00 15.00 3.00 6.33
MVZ173733 - 1.67 9.33 2.67 7.67
MVZ173743 - 3.00 15.33 3.00 5.33
MVZ173744 - 2.33 10.33 2.00 3.33
MVZ173746 - 2.00 5.00 15.33 11.33
MVZ173747 - 1.67 4.67 5.33 6.33
MVZ173756 - 0.67 5.00 8.33 8.33
MVZ173761 - 2.00 4.00 4.00 4.67
MVZ173770 - 5.00 7.67 4.00 8.67
MVZ173771 - 2.00 2.67 8.00 14.67
MVZ184551 - 3.00 4.67 9.00 9.00

454
Chapter Ten Appendix 3. Specimens examined for evolution of Hunter-Schreger Band
enamel microstructure.

Hesperocyoninae: Hesperocyon gregarius: LACM-CIT100 (left and right), 94 (left and
right), 621, 1400 (left and right), 1444, 1526, 4917 (left and right), LACM4932 (left and
right), 4933 (left and right), USNM15937 (left and right), 450576, 450577;
Paraenhydrocyon josephi: AMNH6910, F:AM27753; P. robustus: F:AM12884;
Mesocyon brachyops: LACM-CIT1242, 1346, 1819 (left and right), LACM5265; M.
coryphaeus: UCMP1383 (left and right), 1307 (left and right), 1165 (left and right);
“Mesocyon” temnodon: F:AM102381; Sunkahetanka geringensis: LACM9406, 15910
(left and right); Osbornodon iamonensis: USNM8836; O. renjiei: F:AM63316

Borophaginae: Archaeocyon pavidus: F:AM63970, 50338, LACM-CIT1338,
UCMP76652 (left and right); Otarocyon macdonaldi: F:AM38986; Rhizocyon
oregonensis: UCMP79365; Phlaocyon annectens: F:AM49006, 50299; P. leucosteus:
AMNH8768; P. latidens: UCMP76296; Cormocyon sp.: LACM2739, 5424 (left and
right), 2743; Paracynarctus kelloggi: UCMP11474; ?Cynarctus marylandica:
USNM15561; Metatomarctus sp.: UCMP12604; Metatomarctus sp. A: UCMP38290;
Microtomarctus conferta: LACM-CIT1232 (left and right); Protomarctus optatus:
F:AM61270; Tephrocyon sp.: UCMP19460, 19461; Tomarctus hippophaga: F:AM24270,
61156, LACM-CIT774; Tomarctus sp.: LACM34061, UCMP24291, 24292 (left and
right); Aelurodon mcgrewi: F:AM27153, 8307; A. asthenostylus: F:AM28355; A. ferox:
USNM352360 (left and right), V523, UCMP32241; Paratomarctus temerarius:
455
Chapter Ten Appendix 3 Continued

F:AM50146; P. euthos: F:AM61101, 67121, Paratomarctus sp.: LACM1377 (left and
right); Carpocyon webbi: F:AM27366B, 61335, 61336; Protepicyon raki: F:AM61738;
Epicyon saevus: LACM59697 (left and right), 59813 (left and right), 127794,
F:AM61418, 67396, USNM128; E. haydeni: LACM143519, 127790 (left and right),
PPHM1100, F:AM61498, 61476, 61552, USNM127; Borophagus sp.: LACM34060,
62702, 62703, UCMP30626; B. pugnator: F:AM61662; B. parvus: F:AM108396,
LACM62698; B. secundus: 61690-5, three isolated AMNH specimens, UCMP30492,
30476; B. littoralis: LACM16734 (left and right)

Hyaenidae: Plioviverrops sp.: F:AM96607; Tungurictis spocki: F:AM26600; Thalassictis
sp.: IVPP 65001; Ictitherium sp.: F:AM China 4-L94; HMV 0163 (left and right), 0432
(left and right); Hyaenictitherium hyaenoides: AMNH26371, HMV 0550 (left and right),
0552 (left and right), 0553, 0556, 0560, 0573 (left and right), 0169 (left and right), IVPP
V14737 (left and right), V14738; H. wongi: F:AM China 11-L112, HMV 0751 (left and
right), IVPP Baode Ex. 2, 30; Lycyaena sp.: F:AM China 38-B296, 45-L400, 26-B47,
IVPP V2923; Chasmaporthetes lunensis: IVPP V15162; C. ossifragus: USNM10223; C.
sp.: AMNH26369, LACM74046, LACM-CIT164 (left and right); Palinhyaena reperta:
AMNH26370; Pachycrocuta brevirostris: AMNH27756, 27757, HMV C0066 (left and
right), C0193 (left and right), IVPP 193013, V13932 (left and right); Adcrocuta eximia:
AMNH26373, 26374, F:AM China 10-L4, HMV 0543, 0577 (left and right), 1435 (left
456
Chapter Ten Appendix 3 Continued

and right); Crocuta crocuta honanensis: IVPP L.21625; C. c. ultima: IVPP V2003,
V15160, V15163, V15164 (left and right).
457
Figure 10.1. Tooth positions examined in the study. Black square outlines indicate
approximate size of area examined during each trial. Note the exposed areas of dentine
on the shear facet of m1; all trials were done on the enamel portion of the teeth only.

458
Figure 10.2. Examples of microwear features examined. A. Labial (buccal) wear facet on
p4 of Crocuta crocuta, showing a typical specimen with moderate tooth crown attrition.
B. Examples of small (thin) scratches. C. Examples of large (thick) scratches. D.
Examples of small pits. E. Examples of large pits. F. p4 crown surface of Lycaon pictus,
note paucity of microwear features. Scale bars represent 1 mm.

459
Figure 10.3. Method of enamel microstructure analysis. A. Three regions of the tooth
crown were examined for Hunter-Schreger Bands (HSB), representing top, middle, and
bottom thirds of the crown. One of three types of HSB was recorded for each region: B.
Examples of a region with mostly (> 50%) zig-zag HSB. C. Region with acute-angled
undulating HSB (note that some zig-zag HSB is also present, e.g. indicated by a dotted
circle). D. Undulating HSB.

460
Figure 10.4. Histograms of variance among specimen microwear trials in p4 (A-D) and
m1 (E-H). A, E, small scratches; B, F, large scratches; C, G, small pits; D, H, large pits.
Plot ranges for p4 and m1 were adjusted to show maximum spread of data, respectively.

461
Figure 10.5. Box plots of microwear features across HSB categories in p4 (A-D) and m1
(E-H) specimens. Boxes represent inter-quartile ranges, horizontal lines within boxes are
medians; vertical lines show upper and lower limits, and asterisks represent outliers.

462
Figure 10.6. Plots of mean values and 95% confidence intervals for binned HSB
categories in p4 (A-D) and m1 (E-H) specimens. Mean values are connected in fossil
samples to show trend.

463
Figure 10.7. Intra-dentition evolution of HSB microstructure in: A. Fossil Canidae and B.
Hyaenidae. Phylogenies for fossil canids based on Wang (1994) and Wang et al. (1999),
and for hyaenids based on Werdelin and Solounias (1991). Progressively more derived
HSB patterns are indicated by darker shades of grey.

464
Table 10.1. ANOVA of microwear features according to categories of HSB specialization.
Statistical testing was done on binned category data (as in Fig. 10.6).

p4 vs. HSB m1 vs. HSB
F
3,59
p F
3,70
p
small pits 8.33 <0.001* 8.13 <0.001*
large pits 5.42 0.002* 11.46 <0.001*
small scratches 0.99 0.405 0.47 0.702
large scratches 4.72 0.005* 0.57 0.637

465
Chapter Eleven: Conclusion and Synthesis

This chapter is being prepared for publication as:

Tseng, Z. J. in prep. Dissecting the bone-cracking model in carnivorans to understand the
evolution of feeding specializations.

A copy of the manuscript begins on the next page

466
Dissecting the bone-cracking model in carnivorans to understand the
evolution of feeding specializations

Zhijie Jack Tseng
1,2

1
Integrative and Evolutionary Biology Program, Department of Biological Sciences,
University of Southern California, Los Angeles, California, and
2
Department of
Vertebrate Paleontology, Natural History Museum of Los Angeles County, Los Angeles,
California 90007; jtseng@nhm.org

467
Chapter Eleven Abstract

The Cenozoic evolution of carnivoran mammals exhibits numerous cases of
morphological convergence in the context of ecomorphologies, stereotypical
morphologies that represent unique ecological adaptations. Such examples demonstrate
the iterative adaptation of distantly related clades to a few lucrative hypercarnivore (or
meat specialist) niches. To explore the mechanistic explanations underlying convergent
evolution of ecomorphologies, this paper documents and reviews recent advances in our
understanding of feeding specializations in one particular hypercarnivore niche, the bone-
crackers. Bone-cracking specialists evolved at least three times in Carnivora, in the
hyaenid, percrocutid, and borophagine canid lineages. These studies in the evolutionary
changes of skull shape, enamel microstructure, enamel microwear, and craniodental
biomechanics of hyaenids and borophagine canids show that the suite of adaptive
morphological characters commonly found in the bone-cracking functional complex
evolved convergently in an ordered and mosaic manner. Microstructural changes in the
enamel were related to increased durophagy as inferred from microwear analysis,
followed by continuous skull shape changes toward increased robustness and strength.
Subsequently, skull stress dissipation patterns became adapted to handle mechanical
demands of durophagy, followed by a split into even more mechanically efficient
terminal species versus body size specialists. An updated definition of bone-cracking
specialization is presented, and implications for more general understanding of feeding
specialization are discussed. An ordered evolutionary sequence of adaptive traits in a
468
functional complex represents a flexible mode of evolution that can accommodate
different degrees of specialization in increasingly durophagous lineages, and may serve to
explain similar adaptations in other carnivorans and non-carnivoran mammals.
469
Chapter Eleven Introduction

The Cenozoic evolutionary history of carnivoran mammals (species within the
Order Carnivora) is one of clade replacement (both active and passive) and iterative
evolution of ecomorphs to fill morphospaces (Van Valkenburgh 1999). The overall
pattern of carnivoran evolution, especially those of large-sized (> 7 kg) species, can be
understood as an essentially continuous filling and replacement of the lucrative
hypercarnivore niche (Van Valkenburgh 1991, 1999). A major mechanism that has been
proposed as a driver of these repeating trends is the “macroevolutionary ratchet”; the
combination of enlarging body size (Cope’s Rule), lack of evolutionary reversals (Dollo’s
Law), and ever-increasing specialization of the feeding and locomotive apparatus have
been identified as likely factors contributing to the demise of aging carnivoran lineages
and replacement by others (Holliday and Steppan 2004, Van Valkenburgh 1999, 2007,
Van Valkenburgh et al. 2004). Within these trends, the iterative (repetitive) evolution of
hypercarnivorous ecomorphs (meat specialists) resulted in prominent cases of
convergence: cat-like, wolf-like, and hyena-like species from distantly-related lineages
appeared throughout the Cenozoic (the past 65 million years) on different continents
(Martin 1989, Van Valkenburgh 2007, Werdelin 1996b). Such examples of convergence
indicate the presence of generalized processes by which carnivorans and other mammals
adapt to changing environments. Therefore they are instructive as case studies of general
constraints on evolution and adaptation.
470
In contrast with studies of modern carnivoran community and guild interactions,
however, the evolution of ecomorphs throughout the Cenozoic represent long-term
patterns and processes, and not static categories. Under the backdrop of trans-continental
mammal dispersals permitted by occasional formation of land bridges across major
continents (Eurasia, North America, and South America), many carnivoran groups
became more widespread, or were out-competed as incumbents, during the Cenozoic
(Qiu 2003, Webb 1976). These groups also occupied different ecological niches over
time. One particular ecomorph which established themselves in Eurasia and North
America with relatively high success (i.e. they became diverse and abundant) is
represented by lineages leading to hyena-like hypercarnivores (Martin 1989, Werdelin
1996b). These ecomorphs have been defined as bone-crackers, which utilize their
hypertrophied premolars for cracking open prey bones (Werdelin 1989).
Bone-crackers evolved at least three times in the Order Carnivora, in the hyaenid,
percrocutid, and borophagine canid lineages (Van Valkenburgh 2007, Wang et al. 1999,
Werdelin 1989)(Fig. 11.1). They represent a stereotypical feeding ecomorph whose
adaptive morphological characteristics are as obvious as those of sabertooths and insect
specialists (Ewer 1973). Bone-crackers tend to have robust and hypertrophied premolars
and skulls, with deep zygomatic arches and mandibular rami (Biknevicius and Ruff 1992,
Werdelin 1989); they have specialized enamel microstructure which function in resisting
masticatory stresses associated with bone consumption (Rensberger 1995, Stefen 1999,
Stefen and Rensberger 1999), caudally extended frontal sinus which serves a stress-
dissipating function (Joeckel 1998, Tanner et al. 2008, Tseng 2009), and these species
471
also tend to be large in body size (Wang et al. 1999, Werdelin and Solounias 1991). This
suite of adaptive morphological characters are embodied by the modern spotted hyena
Crocuta crocuta in the form of an integrated functional complex, with which individuals
are able to feed on prey bones as large as those of rhinos and giraffes (Kruuk 1972b).
This paper synthesizes recent studies that focus on the evolutionary patterns and
processes behind the evolution of adaptive characters in this particular functional
complex (Table 11.1), and discusses the implications for understanding more generalized
patterns of specialization in carnivorous mammals.

The relationship between biological form and function
In the biologist’s discussion of form and function, the emphasis has been placed
on the current role of form in performing a function that is adaptive (Admunson and
Lauder 1994). This usage can help to elucidate form-function relationships which are
relevant to the biology and behavior of extant organisms. However, when a deep-time
perspective is presented by documentation of patterns of macroevolutionary change
(which relies almost exclusively on paleontological data), such biological thinking does
not cover additional possibilities with which complex traits may have evolved. A simple
theoretical exercise demonstrates this point (Fig. 11.2): a complex of morphological traits
which work together to perform some function in an extant taxon could evolve as (1) a
single modular complex, (2) a complex containing several modules, (3) gradually through
accumulation of component traits (either ordered or unordered), or through some other
combination of evolutionary histories. What appears to be a collection of harmonious
472
traits that work together to perform some adaptive function may have evolved at separate
times in different species over the lifespan of a lineage (Dawkins 1986). Without the
evolutionary context provided by phylogenies containing both living and extinct relatives,
the identification of such patterns remains difficult, if not impossible.
Therefore, it is quite important and necessary that a study of the evolution of form
and function be applied within a phylogenetic context of living and extinct species of a
given lineage, in this case of convergently evolved ecomorphs. Fossil data represent an
exclusive source of information for morphological evolution. The utilization of
evolutionary histories of the major bone-cracking carnivoran lineages is a common theme
in studies which are discussed below. Bone-cracking function only makes sense in light
of the evolutionary patterns of changes in craniodental form recorded by the geologic
record.

Hyaenids and borophagine canids
Hyenas and dogs are distant relatives (Fig. 11.1); they last shared a common
ancestor approximately 43 Ma (million years ago) at the split of the feliform and
caniform clades within the Order Carnivora (Wesley-Hunt and Flynn 2005). Their
evolutionary histories were largely played out on separate continents: Eurasia for hyenas
and North America for dogs (Wang et al. 2008, Werdelin and Solounias 1991). More
specifically, the specialized bone-cracking species in each lineage never overlapped in
their distribution, even though non-bone-cracking canines arrived in Europe from North
473
America as early as 8-7 Ma (Crusafont-Pairó 1950), and cursorial hyenas arrived in North
America from Asia during the Plio-Pleistocene (Berta 1981, Kurtén and Werdelin 1988).
Werdelin (1989) articulated the still-accepted interpretation of convergent bone-
cracking hyenas and dogs as a case of trade-off between constraint and adaptation.
Hyenas evolved bone-cracking premolars which did not include the carnassial teeth,
allowing retention of the plesiomorphic (ancestral) function of those teeth as meat-
shearing tools. The fossil record of dogs (borophagine canids to be specific) contained
bone-cracking species which evolved at the expense of shearing function in their
carnassials; those shearing teeth showed similar wear as the non-carnassial premolar teeth
of hyenas (Wang et al. 1999, Werdelin 1989). Both bone-cracking lineages evolved
large-bodied specialists that had stereotypically robust and dome-shaped skulls which
supposedly served similar functions in stress dissipation (Werdelin 1989).
It is with these interpretations of convergence and stereotypical characteristics
that we understand bone-cracking ecomorphs. There has been ample evidence to
demonstrate the function and benefits of the complex of morphological characters in
modern bone-crackers (Table 11.1). Furthermore, connections have been made to the
carnivoran fossil record, and extinct bone-cracking ecomorphs have also been identified
(Van Valkenburgh 2007). Recent studies suggest, however, these morphological
characters have a more complicated history of evolution than was previously realized
(Table 11.2). The list of characters used in diagnosing likely bone-cracking ecomorphs
actually evolved at somewhat different times in different species of each lineage. This
becomes apparent only after breaking down the major characters in the functional
474
complex, and tracing their distinctive evolutionary histories in unrelated lineages. These
evolutionary trends represent a unique example of sequential character evolution in a
functional complex, supported by convergent evidence.

Skull shape evolution in hyenas and borophagines
The evolutionary changes that occurred in the feeding apparatus, the skull, of the
hyena and borophagine lineages were qualitatively described in previous systematic
works (Wang et al. 1999, Werdelin and Solounias 1991). Distinct ecomorphological
categories have even been proposed for hyaenids (Werdelin and Solounias 1996b) and
borophagines (Wang et al. 1999). Directional trends of ever-increasing size and
robustness of the craniodental system are recognized in both lineages, as more derived
species evolved hypercarnivorous adaptations (Van Valkenburgh et al. 2004, Werdelin
and Solounias 1991). Recent quantitative analyses using geometric morphometrics
methods on both lineages confirmed these trends, and also showed that, as a whole, skull
shape evolution towards bone-crackers in both lineages were gradual and parallel (Tseng
and Wang 2011). These analyses were also able to pick up signals of the adaptive
constraint in borophagine canids proposed by Werdelin (1989), showing the lack of
overlap in the shortening of the rostrum and reduction of the posterior dentition in
borophagines and hyaenids. Borophagine canids were constrained in their extent of
rostrum reduction because of their unreduced cheek dentition. Even so, the evolution of
skull shape in both lineages was largely unaffected otherwise by their respective
phylogenetic histories, reflecting the importance of functional demands in their
475
transformation (Tseng and Wang 2011). It has been made clear that species from wolf-
like to hyena-like ecomorphs in the two lineages are parts of a continuum of
morphological change, and such change appears to be more or less unidirectional,
complementing body size increases (Tseng and Wang 2011, Van Valkenburgh et al.
2004).
As shown in the next section, this mode of convergent evolution was punctuated
by enamel microstructural adaptations which did not correspond closely in their detailed
trends to the gradual shape changes in the lineages. Adaptive changes in the dentition
appeared before highly specialized skull characteristics evolved. A similar mechanism is
observed in sabertoothed carnivorans, where elongate canines first appeared in species
that did not show the characteristic skull shape of the later, derived forms (Slater and Van
Valkenburgh 2008). Instead of a coordinated change towards more robust shape and
stronger teeth, the evolution of enamel microstructure for bone-cracking appeared to have
preceded highly robust skull shapes. Nonetheless, the similarity in evolutionary pathways
in the hyaenid and borophagine canid lineages demonstrates that cranial shape
convergence is a phenomenon already established prior to the appearance of the terminal
bone-cracking specialists. Instead of a sudden occurrence of bone-cracking skull shapes,
this was a drawn-out process of cumulative adaptation over the history of both lineages.

Enamel microstructure evolution and microwear analysis
Unlike skull shape, the evolution of enamel microstructure in hyaenids and
borophagines progressed from being unspecialized to having specialized microstructure
476
throughout the dentition through a short transition (Ferretti 2007b, Tseng 2011).
Carnivoran enamel has structural patterns that result from layers of enamel prisms
arranging themselves in opposing directions, creating patterns called Hunter-Schreger
Bands (HSBs)(Stefen 1994, Tomes 1906). The convergent evolution of different HSBs
across Carnivora has been well documented (Stefen 1997, 1999, 2001, Stefen and
Rensberger 2002), and has been linked to functional differentiation related to durophagy
(Chai et al. 2009, Rensberger 1995, Rensberger and Stefen 2006, Rensberger and Wang
2005). The most derived HSB, called zig-zag HSB, is found in both hyaenids and
borophagine canids (Stefen 1999, Stefen and Rensberger 2002). A progression of
increasingly folded HSBs in the teeth of hyaenids shows that evolution in the lineage has
been quite directional towards durophagy (Ferretti 1999, Ferretti 2007b). More recent
examination of intra-dentition HSB variation in hyaenids and borophagine canids further
highlighted the fact that zig-zag HSBs tend to evolve in the carnassial teeth first, and
became widespread through the dentition in more derived taxa (Tseng 2011). This trend
did not overlap with the appearance of large, robust skulls, however. Zig-zag HSBs first
appeared in non-specialized species such as Ictitherium and Microtomarctus, which had
neither the skull morphology nor the biomechanical adaptation which allowed them to
process bones to the degree possible in the most specialized bone-crackers (Tseng 2011,
Tseng and Wang 2010).
Another microscopic signature on mammalian enamel which can provide
information on dietary changes is the microscopic features on the enamel surface, termed
microwear (Solounias and Semprebon 2002). These superficial features result from the
477
scratching and pitting of the tooth enamel by abrasive and hard food items. Linkage
between microwear patterns and diets has been well-established for herbivorous
mammals such as ungulates (Merceron et al. 2004, Sanson et al. 2007, Solounias and
Semprebon 2002), but such correlations have only recently been studied in more depth
for carnivorans (Goillot et al. 2009, Schubert et al. 2010, Van Valkenburgh et al. 1990).
Both microstructure and microwear can give indications of the amount of durophagy in
different hyaenid and borophagine species, but only one study has analyzed the
correlation between the two sources of data (Tseng in press-a). Tseng (in press-a) found a
statistically significant relationship between the number of large and small microscopic
pits and the degree of HSB specialization in fossil hyaenids and borophagine canids;
teeth with presence of zig-zag HSBs only at the tips of the crown nevertheless already
show increased quantities of pits, which are indicative of durophagy in modern hyenas
(Goillot et al. 2009, Van Valkenburgh et al. 1990). Taken together, it appears that earlier,
non-robust species in both the hyaenid and borophagine lineages evolved microstructural
reinforcement, as well as preserved microwear evidence, for durophagy prior to the
evolution of robust craniodental morphology.

Evolution of skull biomechanics
Demonstrating yet another aspect of evolutionary change, the skull biomechanics
of several extinct species have now been studied using computer models that reconstruct
craniodental function (Rayfield 2007, Ross 2005). Even given uncertainties in boundary
conditions and other modeling parameters, this method can still provide comparisons that
478
reveal relative functional capabilities of mammalian jaws, after accounting for likely
sources of variation in analytic outcomes (Tseng et al. 2011b). Complementing
traditional studies using principles of mechanics (Biknevicius and Ruff 1992), recent
studies using finite element analysis (FEA) have been conducted on fossil and modern
hyaenids (Tanner et al. 2008, Tseng et al. 2011a, Tseng and Stynder 2011), percrocutids
(Tseng 2009, Tseng and Binder 2010), and borophagine canids (Tseng and Wang 2010).
FEA, a technique used previously in the engineering disciplines for modeling and
conducting material tests, has been readily applied in biology and paleontology for
understanding biomechanic questions. Bone-cracking specializations are highly amenable
to such biomechanical analyses because of the relatively extreme functional demands for
performing such a task. What emerges from these studies is the comparable skull
mechanics (after adjusting for size) of modern Crocuta, the percrocutid Dinocrocuta, and
the borophagine Borophagus, all purported bone-crackers. The dome-shaped forehead,
which is accentuated by a caudally extended frontal sinus, functions as a smooth stress-
dissipating pathway in Crocuta and Dincrocuta, and to a lesser extent in Borophagus
(Tseng 2009, Tseng and Wang 2010). On the other hand, species that were once thought
to have less specialized skulls, such as the extinct hyaenid Chasmaporthetes, were shown
to also have comparably structured skull biomechanics as Crocuta (Tseng et al. 2011a).
Furthermore, a study of the extinct, transitional bone-cracker Ikelohyaena showed that
derived skull stress distribution patterns evolved before the higher mechanical advantage
and bite strength of the most derived bone-crackers, which would have evolved with
further body size increase (Tseng and Stynder 2011).
479
More in-depth analysis of the comparative datasets in each of the above studies
suggests that mosaic evolution is also a theme in the evolution of craniodental
biomechanics. An interpretation proposed by Tseng et al. (2011a) for Chasmaporthetes
indicates that the combination of a cursorial post-cranial skeleton with a high-performing
bone-cracking skull in this taxon represents a mix of apomorphic (unique and novel
adaptations such as increased cursoriality) and plesiomorphic (ancestral or inherited
adaptation such as a skull adapted for durophagy) traits. The cursorial hyaenids were
capable of pursuit hunting, but their bone-cracking adaptations also allowed for
scavenging as well as more complete processing of prey carcasses compared to modern
pursuit predators. In light of enamel microstructural and microwear data discussed in the
previous section, it makes sense that the Chasmaporthetes clade had already been
durophagous in their diet prior to the evolution of their adaptation for running.
Durophagy had already been shown as part of the diet in earlier hyaenids such as
Ictitherium and Hyaenictitherium (Tseng 2011). Such evidence highlights the complexity
of the evolution of the functional complex for bone-cracking, where different
combinations of adaptive characters existed at different points in a lineage’s evolutionary
history. These findings also indicate that in some cases previous interpretations based on
qualitative morphological descriptions and evaluations do not match those based on
reconstruction of craniodental biomechanics. One solution to resolving such
inconsistencies is to generate a more informed definition of bone-cracking specializations
that encompasses all available data.

480
Combined insights from the different approaches
More traditional definitions and usage of “ecological morphology” are concerned
with morphological correlates that can be used as proxies for ecological niche, adaptation,
and specialization (Van Valkenburgh 1999, 2007, Werdelin 1996b). The patterns of
adaptation in bone-cracking hyaenids and canids, specifically, appeared to proceed more
or less directionally towards specialized skull shapes, mechanics, dental microstructures,
and microscopic dietary signatures (Ferretti 2007b, Tseng 2011, Tseng and Wang 2011).
However, it is not until a consilient approach is taken, by analyzing these different
datasets together, that a more complex picture of morphological evolution begins to take
shape.
Cranial shape, as mentioned above, evolved through parallel morphological
pathways in hyaenids and canids (Tseng and Wang 2011). Even though phylogenetic
constraints in the canid dentition and skull were recognized since Werdelin’s (1989)
analysis, new results from morphometric analyses showed that such constraint did not
alter the direction of pathways by which convergent ecomorphologies evolved. What was
affected was actually the extent with which each lineage traveled along the same
evolutionary pathways (Tseng and Wang 2011). Such findings support a robust
interpretation of ecomorphology based on evolution according to functional demands,
and less on lineage-specific phylogenetic constraint, which only limited the extent of
change that occurred. The larger context of skull shape changes within Carnivora also
supports this interpretation (Tseng, in prep.).
481
A second major realization concerns the mosaic manner in which the suite of
adaptive morphological characters evolved in terminal, specialized bone-cracking
borophagine canids and hyaenids. Cranial and mandibular biomechanical modeling
revealed clear-cut differences in bone-cracking versus obviously non-specialized species
(Tseng 2009, Tseng et al. 2011a, Tseng and Binder 2010, Tseng and Stynder 2011, Tseng
and Wang 2010), but also a more nuanced view of transitional species that possess
intermediate biomechanical performance (Tseng and Stynder 2011). These interpretations
of skull biomechanics, however, did not correspond very closely with enamel
microstructural (Tseng 2011) or microwear dietary signals (Tseng in press-a). The latter
aspects had closer correspondence between themselves than with craniodental
biomechanics. In other words, the incorporation of bone into the diet was followed by
microstructural adaptations before significant changes towards more robust skull
morphology allowed for improved biomechanical performance of mastication.
This interpretation highlights the role of the dentition and its microstructure in
corresponding with (and perhaps responding to) increased consumption of hard, abrasive
foods recorded by enamel microwear patterns. The disjoint evolution of macrostructural
biomechanics and enamel microstructure also suggests the need for a deeper
understanding of evolution operating at different scales by examination of morphology
and the fossil record at those levels. In this case, it is likely that some of the characters in
the bone-cracking functional complex (e.g. HSB adaptation) are able to respond to
selective pressure in a more flexible manner than overall craniodental shape which is
under multiple selective demands for sensory, protective, and masticatory functions (Fig.
482
11.3). Again, the mosaic mode of evolution in another prominent hypercarnivore
ecomorphology, that of sabertooth predators, indicate a shared pattern; initial evolution of
the dentition (in this case elongate canines) served a function which subsequently
required skull shape adaptations for enhanced performance (Antón et al. 2004, Salesa et
al. 2005, Slater and Van Valkenburgh 2008).
As it stands, much remains to be done in each of the methodologies utilized by
these recent studies, and additional insights are dependent upon these applications: finite
element modeling of skeletal components are improved by pairing with multibody
dynamics analysis of musculo-skeletal interactions (Moazen et al. 2008); geometric
morphometrics analyses can now benefit from three-dimensional landmark analyses
combined with reconstruction algorithms for incorporating incomplete but critical fossil
specimens (Wiley et al. 2005); enamel microstructure analysis can be improved by
obtaining structural properties from thin-sections and mechanical testing, and then
modeled using FEA (Rensberger and Stefen 2006); finally, microwear analyses have now
progressed into more objective and replicable confocal microscopy and surface
complexity analyses (Schubert et al. 2010). Nonetheless, the current repertoire of data
can be summarized into an updated interpretation of evolutionary trends in bone-cracking
specialization.

An updated classification of bone-cracking specialization
The combination of evidence from different lines of inquiry is hereby summarized
as a series of steps with which bone-cracking specialization occurred in carnivoran
483
ecomorphs. These updated categories for convergent hyaenid and canid bone-cracking
ecomorphs are outlined in Table 11.2. The most basal, unspecialized species are in
category A, in which species without obvious microstructural or macrostructural
adaptations are placed. Category B includes species that have evolved zig-zag HSBs,
which is coupled with increased durophagy signatures in the microwear analysis; skull
shape had already begun to gradually change towards more stereotypical bone-cracking
forms. Species in category C are transitional or less-robust forms that have widespread
full zig-zag HSB in the dentition; they may also have stereotypical skull shape that
dissipates stress more evenly, but their body size and skull robustness have not yet
reached the degree seen in Crocuta. In addition, they may have evolved or retained other
adaptations (e.g. cursoriality) which are superimposed onto the inherited ability for
durophagy. Category D, representing the most specialized bone-crackers, is split into two
sub-categories to distinguish the mechanics-based versus size-based emphasis that the
various species represent: typical robust hyaenines (such as the living hyenas) and the
hyena-like Borophagus are in the first category, with highly-robust skulls, fully zig-zag
HSB throughout the dentition, and large bite forces given their body size. The second
category includes the large percrocutid Dinocrocuta and the large borophagine canid
Epicyon, species that evolved very large body size, but without very efficient bites or
stress-dissipating skulls (Tseng and Wang 2010, Tseng in prep.). This last category of
bone-crackers may have relied on their large size in hunting, food processing, and
agonistic interactions (Deng and Tseng 2010, Johnston 1939)
484
Such a scenario of ordered evolutionary changes is summarized in Figure 11.3.
Increased selective pressure for durophagy are observed through microwear signals,
which is coupled with the evolution of zig-zag HSBs in the carnassials and other
posterior cheek teeth in early hyaenids and borophagine canids. Subsequently, continual
changes in skull shape, coupled with enlarging body size, represents a trend that is
repeatedly observed in hypercarnivorous clades (Van Valkenburgh 2007). The skull
became more robust, and its shape became adapted for dissipating stresses associated
with bone-cracking, in transitional species that have not yet achieved the necessary body
size and large bite force to become specialists. Finally, the bone-cracking specialists
either evolved comparable robustness and bite force to the modern Crocuta, or have the
large body size that enabled them to crack large bones in absence of very efficient skulls
(Fig. 11.3).

Implications for understanding feeding specializations
A mosaic mode of evolution is not a new concept; it has been studied and
identified in many lineages of organisms, and is a potentially widespread phenomenon
across different groups, life history characteristics, as well as during early cladogenesis
(Daeschler et al. 2006, DeBeer 1954, Ji et al. 1999, McHenry 1975, Stebbins 1983,
Winter and Oxnard 2001). However, in most of the discussions on mosaic evolution, only
character transitions within single lineages are documented (e.g. McHenry 1994), or the
mosaic characters involved in the adaptive complex are shown to evolve in an unordered
manner (e.g. Lazzari et al. 2008). An ordered, mosaic convergence of a functional
485
complex such as the one documented in hyenas and borophagine dogs does not currently
have clear parallels besides possibly sabertoothed hypercarnivores (Slater and Van
Valkenburgh 2008). Whereas most cases of mosaic evolution demonstrate, to different
extents, the independent evolutionary trajectories of components characters in a given
adaptive complex, only in cases such as the convergent, ordered evolution of bone-
cracking characteristics in hyenas and dogs can the relationship among different
functional characters be clarified. For example, a highly complex history of homoplasies
in HSB evolution across Carnivora as compared to relatively fewer species that have
been identified as bone-cracking ecomorphs based on overall craniodental morphology
suggests that more evolutionary flexibility is present in the former characteristic (Stefen
1997). The gradual evolutionary changes in skull shape and the more pronounced
differences in skull biomechanics in the most derived hyaenids and borophagine canids
each represent somewhat different patterns that overlay microstructural evolution (Tseng
2009, Tseng et al. 2011a, Tseng and Stynder 2011, Tseng and Wang 2010, 2011). The
convergent evolutionary patterns by which these characters evolved in hyenas and dogs
demonstrate that adaptation for durophagy follows predictable and sequential
mechanisms of change. These changes did not evolve as one package, as its modern
exemplar might lead one to propose that such a mechanism was at work.
Early proponents of morphological integration, the synchronized evolution and
development of functionally and/or genetically linked traits, have outlined a general rule
for how those “units” evolve within whole organisms. Cheverud (1982) summed it up as
“each part of an organism is formed so that the role it plays in the function of the whole is
486
performed harmoniously with respect to all other parts. An organism’s phenotype is an
organized, integrated, functional whole” (p. 499). However, the generality of such
explanations have not been tested specifically in carnivorans until more recently
(Goswami 2006b). Goswami’s (2006b) analyses indicated that phylogeny, not diet, was
significantly associated with morphological integration in feliform and canid carnivorans.
Such results indicate that skulls of hyaenids (feliform) and borophagines (caniform)
evolved within a larger context of morphological integration, which was controlled by
constraints imposed by higher-level phylogeny (i.e. not necessarily family-specific).
Coupled with the sequential, ordered pattern of craniodental specialization for durophagy
demonstrated by the preceding discussion, it appears that the morphological pathways
toward bone-cracking ecomorphs nevertheless occurred in a stereotypical manner within
such overarching constraints. Even though the suite of adaptive characteristics in the
modern hyenas has clear benefits in bone-cracking function (e.g. robust craniodental
morphology, specialized enamel microstructure, large body size and masticatory
musculature, large bite forces for a given body size; Table 11.1), their evolution was
neither modular nor integrated. Instead, increased demand for durophagy very likely
elicited a response to increase microstructural strength of the posterior premolars and
molars; this is followed by a continual change in skull shape towards a biomechanically
more stress-resistant skull shape. Subsequent increases in body size and robustness
allowed for larger bite forces and additional mechanical advantage which enhanced the
ability to consume larger and harder bones. The consistency with which different
characters evolved within the functional complex suggests that it may be a more widely
487
recognizable mechanistic explanation. Such a mechanism is a highly flexible one that
allows not only terminal specialists, but also incrementally specialized durophagous
lineages, to evolve given proper selective pressures. Therefore, this mosaic mode of
evolution characterized in hyaenids and borophagine canids can be tested in other
carnivoran lineages in which adaptation for durophagy did not proceed through the entire
sequence. Lineages that were “partially” adapted for increased durophagy can now be
categorized in the same system of definitions as the bone-cracking ecomorphs (Table
11.2).
Other than bone-cracking and sabertooth ecomorphologies, there have been few
other such demonstration of a convergent, sequentially evolved functional complex
among carnivorans or even other mammals. Studies of functional complexes dealing with
the human hand (Tocheri et al. 2005), powered flight in birds (Ostrom et al. 1999), and
even with the generalized carnivorous dentition of mammals (De Muizon and Lange-
Badre 1997) have all analyzed suites of characters, which make possible their respective
adaptive functions, as static collections of morphological traits. A similarly framed
approach is taken in many studies of adaptation in behavioral ecology, where
morphological complexes are studied for their current adaptiveness (Krebs and Davies
1993). Some have even proposed that functional complexes can be under active,
pleiotropic control (Dawson et al. 1986). The documentation of a functional complex, i.e.
the craniodental characteristics of specialized bone-crackers, which evolved convergently
in an ordered and sequential manner, highlights the importance of tracing adaptations
through well-resolved phylogenies that include extinct taxa. Traditional, more “static”
488
definitions of bone-crackers cite the suite of morphological characters that are easily
identified in terminal specialists (Van Valkenburgh 2007, Werdelin 1989, 1996b), when
in fact intermediates exist but are often not easily categorized in a way to reflect their
positions on a functional gradient. Distinct ecomorphological categories have been
proposed for hyaenids based on overall morphology (Werdelin and Solounias 1996b), but
the generality of such categories for borophagines and other carnivorans are not clear.
The concurrent examination of enamel microstructure, microwear, craniodental shape
and function provides data for a more generalized hypothesis of how bone-cracking
specialist lineages are expected to evolve (or cease to evolve at one of the earlier stages).
Such a hypothesis of an ordered array of adaptive character evolution may even be
applicable to durophagous clades in general (i.e. bamboo-feeding and other non-bone-
cracking adaptations to consuming hard foods), where evidence show overlapping
functional demands are present (Figueirido et al. 2011a, Figueirido et al. 2011b).
Even with this updated, more general definition of bone-cracking specialization,
many questions still remain: how strong and lasting does selective pressure have to be, in
order to “push” for the evolution of a bone-cracking specialist that has all of the adaptive
characters in such a functional complex? The tempo of the progression from the very first
hyaenids and borophagines canids with zig-zag HSBs in their carnassials, to the
appearance of the derived hyaenines and Borophagus with biomechanically adapted
skulls, remains to be documented in more detail. What is clear is that the entirety of this
series of evolutionary changes in both lineages was complete by the end of the Miocene
(Fig. 11.4). The establishment of widespread open habitats and the evolution of cursorial
489
herbivores are prominent trends during this time period, but their linkage to carnivoran
evolution is weak (Van Valkenburgh 1999). How does the evolution of bone-cracking
adaptations, proceeding by the mechanism outlined in the preceding sections, fit into the
broader context of paleo-community dynamics? Are these iterative mechanisms for
adaptation controlled by climatic regimes (Zachos et al. 2001)? From an entirely different
level of examination, the genetic mapping of the phenotypic traits in the bone-cracking
functional complex is essentially unknown; how are developmental changes in
microstructure initiated in the enamel on the macroevolutionary scale? How do
developmental regimes interact with elevated levels of durophagy in the ontogeny of
juvenile and subadult individuals to produce adult morphology in incrementally
durophagous lineages? Given the similar material properties of mammalian bone, hide,
and soft tissues over the course of the Cenozoic, one can expect these convergent
mechanisms for feeding specialization to be repeated in similar selective environments,
regardless of prey morphology (Van Valkenburgh 1999). How do these sequential modes
of adaptation affect our understanding of evolutionary flexibility and the
“macroevolutionary ratchet”? To be sure, mechanistic explanations for complex
evolutionary processes are only beginning to be analyzed with integrated approaches, and
many innovative studies remain to be conducted to realize its potentials.

490
Chapter Eleven Acknowledgments

I thank D. Bottjer, H. Flashner, J. McNitt-Gray, B. Van Valkenburgh, and X.
Wang for comments and input on this manuscript and other parts of my Ph.D. dissertation.
J. Liu for encouragement and support. Research for this paper was supported by a
University of Southern California Joint Initiatives Fellowship.
491
Chapter Eleven References
Admunson, R., and G. V. Lauder. 1994. Function without purpose: the uses of causal role
function in evolutionary biology. Biology and Philosophy 9:443-469.

Antón, M., M. J. Salesa, J. Morales, and A. Turner. 2004. First known complete skulls of
the scimitar-toothed cat Machairodus aphanistus (Felidae, Carnivora) from the
Spanish late Miocene site of Batallones-1. Journal of Vertebrate Paleontology
24(4):957-969.

Berta, A. 1981. The Plio-Pleistocene hyaena Chasmaporthetes ossifragus from Florida.
Journal of Vertebrate Paleontology 1(3-4):341-356.

Biknevicius, A. R., and C. B. Ruff. 1992. The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology
228:479-507.

Binder, W. J., and B. Van Valkenburgh. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252:273-283.

Chai, H., J. J.-W. Lee, P. Constantino, P. W. Lucas, and B. Lawn. 2009. Remarkable
resilience of teeth. Proceedings of the National Academy of Sciences
106(18):7289-7293.

Cheverud, J. M. 1982. Phenotypic, genetic, and environmental morphological integration
in the cranium. Evolution 36(3):499-516.

Crusafont-Pairó, M. 1950. El primer representante del género Canis en el Pontiense
eurasiatico (Canis cipio nova sp.). Boletin de la Real Sociedad Española de
Historia Natural (Geologia) 48:43-51.

Daeschler, E. B., N. H. Shubin, and F. A. J. Jenkins. 2006. A Devonian tetrapod-like fish
and the evolution of the tetrapod body plan. Nature 440:757-763.

Dawkins, R. 1986. The Blind Watchmaker. Longman Scientific & TEchnical, Harlow.

Dawson, M. R., R. K. Stucky, L. Krishtalka, and C. C. Black. 1986. Machaeroides
simpsoni, new species, oldest known sabertooth creodont (Mammalia), of the Lost
Cabin Eocene. Contributions to Geology, University of Wyoming, Special Paper
3:177-182.

De Muizon, C., and B. Lange-Badre. 1997. Carnivorous dental adaptations in
tribosphenic mammals and phylogenetic reconstruction. Lethaia 30(4):353-366.
492

DeBeer, G. 1954. Archaeopteryx and evolution. Advancement of Science 11:160-170.

Deng, T., and Z. J. Tseng. 2010. Osteological evidence for predatory behavior of the
giant percrocutid (Dinocrocuta gigantea) as an active hunter. Chinese Science
Bulletin 55(17):1790-1794.

Ewer, R. F. 1973. The carnivores. Cornell University Press, Ithaca.

Ferretti, M. P. 1999. Tooth enamel structure in the hyaenid Chasmaporthetes lunensis
lunensis from the late Pliocene of Italy, with implications for feeding behavior.
Journal of Vertebrate Paleontology 19(4):767-770.

Ferretti, M. P. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia,
Carnivora). Swiss Journal of Geoscience 100:41-52.

Figueirido, B., N. MacLeod, J. Krieger, M. Di Renzi, J. A. Perez-Claros, and P.
Palmqvist. 2011a. Constraint and adaptation in the evolution of carnivoran skull
shape. Paleobiology 37(3):490-518.

Figueirido, B., P. Palmqvist, J. A. Perez-Claros, and W. Dong. 2011b. Cranial shape
transformation in the evolution of the giant panda (Ailuropoda melanoleuca).
Naturwissenschaften 98:107-116.

Goillot, C., C. Blondel, and S. Peigne. 2009. Relationships between dental microwear and
diet in Carnivora (Mammalia) - Implications for the reconstruction of the diet of
extinct taxa. Palaeogeography Palaeoclimatology Palaeoecology 271:13-23.

Goswami, A. 2006. Morphological integration in the carnivoran skull. Evolution
60(1):169-183.

Holliday, J. A., and S. J. Steppan. 2004. Evolution of hypercarnivory: the effect of
specialization on morphological and taxonomic diversity. Paleobiology
30(1):108-128.

Ji, Q., Z.-X. Luo, and S.-a. Ji. 1999. A Chinese triconodont mammal and mosaic
evolution of the mammalian skeleton. Nature 398:326-330.

Joeckel, R. M. 1998. Unique frontal sinuses in fossil and living Hyaenidae (Mammalian,
Carnivora): description and interpretation. Journal of Vertebrate Paleontology
18(3):627-639.

Johnston, C. S. 1939. A skull of Osteoborus validus from the early middle Pliocene of
Texas. Journal of Paleontology 13(5):526-530.
493

Krebs, J. R., and N. B. Davies. 1993. An introduction to behavioural ecology. Wiley-
Blackwell, Oxford.

Kruuk, H. 1972. The spotted hyena: a study of predation and social behavior. The
University of Chicago Press, Chicago.

Kurtén, B., and L. Werdelin. 1988. A review of the genus Chasmaporthetes Hay, 1921
(Carnivora, Hyaenidae). Journal of Vertebrate Paleontology 8(1):46-66.

Lazzari, V., C. Charles, P. Tafforeau, M. Vianey-Liaud, J.-P. Aguilar, J.-J. Jaeger, J.
Michaux, and L. Viriot. 2008. Mosaic convergence of rodent dentitions. PLoS
ONE 3(10):e3607.

Martin, L. D. 1989. Chapter 19. Fossil history of the terrestrial Carnivora. Pp. 536-568. In
J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution. Cornell
University Press, New York.

McHenry, H. M. 1975. Fossils and the mosaic nature of human evolution. Science
190(4213):425-431.

McHenry, H. M. 1994. Tempo and mode in human evolution. Proceedings of the
National Academy of Sciences 91:6780-6786.

Meers, M. B. 2002. Maximum bite force and prey size of Tyrannosaurus rex and their
relationships to the inference of feeding behavior. Historical Biology 16:12.

Merceron, G., C. Blondel, M. Brunet, S. Sen, N. Solounias, L. Viriot, and E. Heintz. 2004.
The Late Miocene paleoenvironment of Afghanistan as inferred from dental
microwear in artiodactyls. Palaeogeography Palaeoclimatology Palaeoecology
207:143-163.

Moazen, M., N. Curtis, S. E. Evans, P. O'Higgins, and M. J. Fagan. 2008. Combined
finite element and multibody dynamics analysis of biting in a Uromastyx
hardwickii lizard skull. Journal of Anatomy 213:499-508.

Ostrom, J. H., S. O. Poore, and G. E. Goslow, Jr. 1999. Humeral rotation and wrist
supination: Important functional complex for the evolution of powered flight in
birds? Smithsonian Contributions to Paleobiology 89:301-309.

Qiu, Z.-x., ed. 2003. Dispersals of Neogene carnivorans between Asia and North America.
American Museum of Natural History, New York.

494
Rayfield, E. J. 2007. Finite element analysis and understanding the biomechanics and
evolution of living and fossil organisms. Annual Review of Earth and Planetary
Science 35:541-576.

Rensberger, J. M. 1995. Determination of stresses in mammalian dental enamel and their
relevance to the interpretation of feeding behaviors in extinct taxa. Pp. 151-172.
In J. J. Thomason, ed. Functional morphology in vertebrate paleontology.
Cambridge University Press, New York.

Rensberger, J. M., and C. Stefen. 2006. Functional differentiations of the microstructure
in the upper carnassial enamel of the spotted hyena. Palaeontographica Abt. A
278(1-6):149-162.

Rensberger, J. M., and X. Wang. 2005. Microstructural reinforcement in the canine
enamel of the hyaenid Crocuta crocuta, the felid Puma concolor, and the late
Miocene canid Borophagus secundus. Journal of Mammalian Evolution 12(3-
4):379-403.

Ross, C. F. 2005. Finite element analysis in vertebrate biomechanics. The Anatomical
Record Part A 283A:253-258.

Salesa, M. J., M. Antón, A. Turner, and J. Morales. 2005. Aspects of the functional
morphology in the cranial and cervical skeleton of the sabre-toothed cat
Paramachairodus ogygia (Kaup, 1832) (Felidae, Machairodontinae) from the
Late Miocene of Spain: implications for the origins of teh machairodont killing
bite. Zoological Journal of Linnean Society 144:363-377.

Sanson, G. D., S. A. Kerr, and K. A. Gross. 2007. Do silica phytoliths really wear
mammalian teeth? Journal of Archaeological Science 34:526-531.

Schubert, B. W., P. S. Ungar, and L. R. G. DeSantis. 2010. Carnassial microwear and
dietary behaviour in large carnivorans. Journal of Zoology 280(3):257-263.

Slater, G. J., and B. Van Valkenburgh. 2008. Long in the tooth: evolution of sabertooth at
cranial shape. Paleobiology 34(3):403-419.

Solounias, N., and G. Semprebon. 2002. Advances in the reconstruction of ungulate
ecomorphology with application to early fossil equids. American Museum
Novitates 3366:1-49.

Stebbins, G. L. 1983. Mosaic evolution: an integrating principle for the modern synthesis.
Experientia 39:823-834.

495
Stefen, C. 1994. Zahnschmelzdifferenzierungen bei Raubtieren: [Carnivora im Vergleich
zu Vertretern der Creodonta, Arctocyonidae, Mesonuchidae, Entelodontidae
(Placentalia), Thylacoleonidae, Dasyuridae und Thylaacinidae (Marsupialia)].
Ph.D. Rheinischen Friedrich-Wilhelms-Universtat zu Bonn.

Stefen, C. 1997. Chapter 7. Differentiations in Hunter-Schreger bands of carnivores. Pp.
123-136. In W. v. Koenigswald, and P. M. Sander, eds. Tooth enamel
microstructure. A.A. Balkema, Rotterdam.

Stefen, C. 1999. Enamel microstructure of recent and fossil Canidae (Carnivora:
Mammalia). Journal of Vertebrate Paleontology 19(3):576-587.

Stefen, C. 2001. Enamel structure of arctoid Carnivora: Amphicyonidae, Ursidae,
Procyonidae, and Mustelidae. Journal of Mammalogy 82(2):450-462.

Stefen, C., and J. M. Rensberger. 1999. The specialized structure of hyaenid enamel:
description and development within the lineage - including percrocutids. Scanning
Microscopy 13(2-3):363-380.

Stefen, C., and J. M. Rensberger. 2002. The specialized enamel structure of hyaenids
(Mammalia, Hyaenidae): description and development within the lineage -
including percrocutids. Zoologische Abhandlungen 52:127-147.

Tanner, J. B., E. R. Dumont, S. T. Sakai, B. L. Lundrigan, and K. E. Holekamp. 2008. Of
arcs and vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta
crocuta). Biological Journal of the Linnean Society 95:246-255.

Tanner, J. B., M. Zelditch, B. L. Lundrigran, and K. E. Holekamp. 2010. Ontogenetic
change in skull morphology and mechanical advantage in the spotted hyena
(Crocuta crocuta). Journal of Morphology 271(3):353-365.

Tocheri, M. W., A. Razdan, R. C. Williams, and M. W. Marzke. 2005. A 3D quantitative
comparison of trapezium and trapezoid relative articular and nonarticular surface
areas in modern humans and great apes. Journal of Human Evolution 49(5):570-
586.

Tomes, C. S. 1906. On the minute structure of the teeth of creodonts, with special
reference to their suggested resemblance to marsupials. Proceedings of the
Zoological Society of London 1906:45-58.

Tseng, Z. J. 2009. Cranial function in a late Miocene Dinocrocuta gigantea (Mammalia:
Carnivora) revealed by comparative finite element analysis. Biological Journal of
the Linnean Society 96:51-67.

496
Tseng, Z. J. 2011. Variation and implications of intra-dentition Hunter-Schreger band
pattern in fossil hyaenids and canids (Carnivora, Mammalia). Journal of
Vertebrate Paleontology 31(5):1163-1167.

Tseng, Z. J. in press. Connecting Hunter-Schreger Band microstructure to enamel
microwear features: New insights from durophagous carnivores. Acta
Palaeontologica Polonica:doi: 10.4202/app.2011.0027.

Tseng, Z. J., M. Antón, and M. J. Salesa. 2011a. The evolution of the bone-cracking
model in carnivorans: Cranial functional morphology of the Plio-Pleistocene
cursorial hyaenid Chasmaporthetes lunensis (Mammalia: Carnivora).
Paleobiology 37(1):140-156.

Tseng, Z. J., and W. J. Binder. 2010. Mandibular biomechanics of Crocuta crocuta,
Canis lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia).
Zoological Journal of Linnean Society 158:683-696.

Tseng, Z. J., J. McNitt-Gray, H. Flashner, X. Wang, and R. Enciso. 2011b. Model
sensitivity and use of the comparative finite element method in mammalian jaw
mechanics: mandible performance in the Gray Wolf. PLoS ONE
6(4):doi:10.1371/journal.pone.0019171.

Tseng, Z. J., and D. Stynder. 2011. Mosaic functionality in a transitional ecomorphology:
skull biomechanics in stem Hyaeninae compared to modern South African
carnivorans. Biological Journal of the Linnean Society 102:540-559.

Tseng, Z. J., and X. Wang. 2010. Cranial functional morphology of fossil dogs and
adaptation for durophagy in Borophagus and Epicyon (Carnivora, Mammalia).
Journal of Morphology 271(11):1386-1398.

Tseng, Z. J., and X. Wang. 2011. Do convergent ecomorphs evolve through convergent
morphological pathways? Cranial shape evolution in fossil hyaenids and
borophagine canids (Carnivora, Mammalia). Paleobiology 37(3):470-489.

Van Valkenburgh, B. 1991. Iterative evolution of hypercarnivory in canids (Mammalia:
Carnivora): evolutionary interactions among sympatric predators. Paleobiology
17:340-362.

Van Valkenburgh, B. 1999. Major patterns in the history of carnivorous mammals.
Annual Review of Earth and Planetary Science 27:463-493.

Van Valkenburgh, B. 2007. Déjà vu: the evolution of feeding morphologies in the
Carnivora. Integrative and Comparative Biology 47(1):147-163.

497
Van Valkenburgh, B., M. F. Teaford, and A. Walker. 1990. Molar microwear and diet in
large carnivores: inferences concerning diet in the sabretooth cat, Smilodon fatalis
Journal of Zoology 222:319-340.

Van Valkenburgh, B., X. Wang, and J. Damuth. 2004. Cope's Rule, hypercarnivory, and
extinction in North American canids. Science 306(5693):101-104.

Wang, X., R. H. Tedford, and M. Antón. 2008. The Dog Family, Canidae, and Their
Evolutionary History. Columbia University Press, New York.

Wang, X., R. H. Tedford, and B. E. Taylor. 1999. Phylogenetic systematics of the
Borophaginae (Carnivora: Canidae). Bulletin of the American Museum of Natural
History 243:1-391.

Webb, S. D. 1976. Mammalian faunal dynamics of the great American interchange.
Paleobiology 2:220-234.

Werdelin, L. 1989. Constraint and adaptation in the bone-cracking canid Osteoborus
(Mammalia: Canidae). Paleobiology 15:387-401.

Werdelin, L. 1996. Chapter 17. Carnivoran ecomorphology: a phylogenetic perspective.
Pp. 582-624. In J. L. Gittleman, ed. Carnivore behavior, ecology, and evolution.
Cornell University Press, New York.

Werdelin, L., and N. Solounias. 1991. The Hyaenidae: taxonomy, systematics and
evolution. Fossils and Strata 30:1-104.

Werdelin, L., and N. Solounias. 1996. The evolutionary history of hyenas in Europe and
western Asia during the Miocene. Pp. 290-306. In R. L. Bernor, V. Fahlbusch,
and S. Rietschel, eds. Later Neogene European biotic evolution and stratigraphic
correlation. Columbia University Press, New York.

Wesley-Hunt, G. D., and J. J. Flynn. 2005. Phylogeny of the Carnivora: basal
relationships among the carnivoramorphans, and assessment of the position of
'Miacoidea' relative to Carnivora. Journal of Systematic Palaeontology 3(1):1-28.

Wiley, D. F., N. Amenta, D. A. Alcantara, D. Ghosh, Y. J. Kil, E. Delson, W. Harcourt-
Smith, F. J. Rohlf, K. St. John, and B. Hamann. 2005. Evolutionary morphing.
IEEE Visualization 5(1):1-8.

498
Winter, W. d., and C. E. Oxnard. 2001. Evolutionary radiations and convergences in the
structural organization of mammalian brains. Nature 409:710-714.

Zachos, J., M. Pagani, L. Sloan, E. Thomas, and K. Billups. 2001. Trends, rhythms, and
aberrations in global climate 65 Ma to present. Science 292:686-693.

499
Figure 11.1. Phylogenetic relationships of A. Canidae and B. Hyaenidae. Silhouettes
show representative skull shapes and sizes in each lineage. From Tseng and Wang (2011).

500
Figure 11.2. Four possible modes of evolution for a given functional complex of four
adaptive traits (represented by colors) present in an extant species (“taxon 4”). A.
modular with adaptation of all traits within the complex at a single stage in evolution, B.
two modules evolving at different times in the lineage’s phylogenetic history, C. gradual
accumulation of individual traits over the course of the lineage’s history in an ordered
manner, D. gradual accumulation of traits in an unordered manner. These scenarios are
neither exhaustive nor mutually exclusive, and are used for demonstration of concept.

501
Figure 11.3. A graphical depiction of the convergent specializing trend in the hyaenid and
borophagine canid lineages. A. increased demand for durophagy, B. microwear patterns
reflect increased durophagy (from Tseng, in press), C. Enamel microstructure adapts in
response to increased durophagy (from Tseng, 2011), D. Skull shape gradually evolves
towards stereotypical bone-cracking ecomorph (from Tseng and Wang, 2011), E.
Transitional species show stereotypical biomechanical stress dissipation patterns, but lack
in large bite forces and mechanical advantage (from Tseng and Stynder, 2011), F.
Specialized bone-cracking ecomorphs adapt through either biomechanics- or body size-
emphasized pathways (from Tseng, in prep.).

502
Figure 11.4. Bone-cracking specialization categories mapped onto stratigraphic
occurrence of A. hyaenids and B. borophagine canids. Data for A are from Werdelin and
Solounias (1991), with ecomorphological categories from Werdelin and Solounias
(1996b). B is modified from Tseng and Wang (2010). Colors indicate specialization
categories (Table 11.2): A (red), B (orange), C (green), D
1
(blue), and D
2
(purple). For
hyaenids, category D
2
is represented by the closely related percrocutid Dinocrocuta,
which are early late Miocene in age (not drawn).

503
Table 11.1. Adaptive characteristics of the modern spotted hyena Crocuta crocuta,
constituting the definition of a specialized bone-cracking ecomorph.

Trait Reference
Robust crania and mandible Kruuk 1972; Werdelin 1989; Tanner et al. 2010; Biknevicius & Ruff 1992
Robust premolars Kruuk 1972; Werdelin 1989; Werdelin & Solounias 1991
Caudally extended frontal sinus Joeckel 1998; Tanner et al. 2008; Tseng 2009
Specialized enamel microstructure Stefen & Rensberger 1999; Rensberger & Wang 2005; Ferretti 2007; Tseng 2011
Large body size Werdelin 1989; Werdelin & Solounias 1991
Large bite force given body size Binder & Van Valkenburgh 2000; Meers 2002
Characteristic microwear patterns Van Valkenburgh et al. 1990; Goillot et al. 2009; Schubert et al. 2010

504
Table 11.2. Modified categories representing degrees of bone-cracking specialization in
hypercarnivorous ecomorphs. Data come from studies on hyaenid and borophagine
canids.

Category Characteristic Taxa Reference
A non-durophagous, no zig-zag HSB early hyaenids, borophagines
Ferretti 2007; Tseng 2011; in
press
B increased durophagy, zig-zag HSB restricted
Ictitherium, Hyaenictitherium,
Microtomarctus, Tomarctus
Tseng 2011; in press
C stereotypical skull shape, zig-zag HSB widespread
Ikelohyaena, Chasmaporthetes,
Aelurodon?
Tseng & Stynder 2011; Tseng &
Wang 2011; Tseng et al. 2011;
Tseng 2011
D
1
skull robust, large body size, biomechanical adaptations
Crocuta, Pliocrocuta,
Pachycrocuta, Adcrocuta,
Borophagus
Werdelin 1989; Tanner et al.
2008; Tseng & Wang 2010,
2011; Tseng 2011
D
2
skull robust, very large body size, less mechanical advantage Dinocrocuta, Epicyon
Tseng 2009; Tseng & Binder
2010; Tseng & Wang 2010;
Tseng in prep.

505
Comprehensive References

Adams, D. C., F. J. Rohlf, and D. E. Slice. 2004. Geometric morphometrics: ten years of
progress following the 'Revolution'. Italian Journal of Zoology 71:5-16.

Admunson, R., and G. V. Lauder. 1994. Function without purpose: the uses of causal role
function in evolutionary biology. Biology and Philosophy 9:443-469.

Andersson, K. 2005. Were there pack-hunting canids in the Tertiary, and how can we
know? Paleobiology 31(1):56-72.

Antón, M., M. J. Salesa, J. Morales, and A. Turner. 2004. First known complete skulls of
the scimitar-toothed cat Machairodus aphanistus (Felidae, Carnivora) from the
Spanish late Miocene site of Batallones-1. Journal of Vertebrate Paleontology
24(4):957-969.

Antón, M., A. Turner, M. J. Salesa, and J. Morales. 2006. A complete skull of
Chasmaporthetes lunensis (Carnivora, Hyaenidae) from the Spanish Pliocene site
of La Puebla de Valverde (Teruel). Estudios Geologicos 62(1):375-388.

Anyonge, W. 1993. Body mass in large extant and extinct carnivores. Journal of Zoology
231:339-350.

Anyonge, W. 1996. Microwear on canines and killing behavior in large carnivores: saber
function in Smilodon fatalis. Journal of Mammalogy 77(4):1059-1067.

Arbour, V. M., and E. Snively. 2009. Finite element analyses of ankylosaurid dinosaur
tail club impacts. The Anatomical Record 292:1412-1426.

Arnold, S. J. 2003. Performance surfaces and adaptive landscapes. Integrative and
Comparative Biology 43(3):367-375.

Ashby, M. F., L. J. Gibson, U. Wegst, and R. Olive. 1995. The mechanical properties of
natural materials. I. Material property charts. Proceedings of the Royal Society of
London, Series A 450:123-140.

Ashman, R. B., G. Rosinia, S. C. Cowin, and M. G. Fontenot. 1985. The bone tissue of
the canine mandible is elastically isotropic. Journal of Biomechanics 18(9):717-
721.

Bazant, Z. P., and Z. Li. 1995. Modulus of rupture: size effect due to fracture initiation in
boundary layer. Journal of Structural Engineering 121(4):739-746.
506
Bell, P. E., E. Snively, and L. Shychoski. 2009. A comparison of the jaw mechanics in
hadrosaurid and ceratopsid dinosaurs using finite element analysis. The
Anatomical Record 292:1338-1351.

Benoit, M. 2006. Multiphasic allometric analysis in lions (Panthera leo): Life history
expressed through morphometrics. Journal of Vertebrate Paleontology
26(3):S41A.

Berta, A. 1981. The Plio-Pleistocene hyaena Chasmaporthetes ossifragus from Florida.
Journal of Vertebrate Paleontology 1(3-4):341-356.

Biknevicius, A. R. 1996. Functional discrimination in the masticatory apparatus of
juvenile and adult cougars (Puma concolor) and spotted hyenas (Crocuta crocuta).
Canadian Journal of Zoology 74:1934-1942.

Biknevicius, A. R., and S. R. Leigh. 1997. Patterns of growth of the mandibular corpus in
spotted hyenas (Crocuta crocuta) and cougars (Puma concolor). Zoological
Journal of Linnean Society 120:139-161.

Biknevicius, A. R., and C. B. Ruff. 1992. The structure of the mandibular corpus and its
relationship to feeding behaviors in extant carnivorans. Journal of Zoology
228:479-507.

Biknevicius, A. R., and B. Van Valkenburgh. 1996. Design for killing: Craniodental
adaptations of predators. Pp. 393-428. Carnivore behavior, ecology, and evolution,
Vol. 2.

Binder, W. J., E. N. Thompson, and B. Van Valkenburgh. 2002. Temporal variation in
tooth fracture among Rancho La Brea dire wolves. Journal of Vertebrate
Paleontology 22(2):423-428.

Binder, W. J., and B. Van Valkenburgh. 2000. Development of bite strength and feeding
behaviour in juvenile spotted hyenas (Crocuta crocuta). Journal of the Zoological
Society of London 252:273-283.

Bookstein, F. L. 1991. Morphometric tools for landmarks data. Cambridge University
Press, London.

Bottjer, D. J. 1998. Phanerozoic non-actualistic paleoecology. Geobios 30(7):885-893.

Bourke, J., S. Wroe, K. Moreno, C. McHenry, and P. D. Clausen. 2008. Effects of gape
and tooth position on bite force and skull stress in the dingo (Canis lupus dingo)
using a 3-dimensional finite element approach. PLoS ONE 3(5):e2200.
507
Boyko, A. R., R. H. Boyko, C. M. Boyko, H. G. Parker, M. Castelhano, L. Corey, J. D.
Degenhardt, A. Auton, M. Hedimbi, R. Kityo, E. A. Ostrander, J. Schoenebeck, R.
J. Todhunter, P. Jones, and C. D. Bustamante. 2009. Complex poplation structure
in African village dogs and its implications for inferring dog domestication
history. Proceedings of the National Academy of Sciences 106(33):13903-13908.

Bright, J. A., and E. J. Rayfield. 2011. The response of cranial biomechanical finite
element models to variations in mesh density. The Anatomical Record
294(4):610-620.

Buckland-Wright, J. C. 1978. Bone structure and the patterns of force transmission in the
cat skull (Felis catus). Journal of Morphology 155:35-62.

Butler, D. L., Y. Guan, M. D. Kay, J. F. Cummings, S. M. Feder, and M. S. Levy. 1992.
Location-dependent variations in the material properties of the anterior cruciate
ligament. Journal of Biomechanics 25(5):511-518.

Carbone, C., G. M. Mace, S. C. Roberts, and D. W. Macdonald. 1999. Energetic
constraints on the diet of terrestrial carnivores. Nature 402:286-288.

Chai, H., J. J.-W. Lee, P. Constantino, P. W. Lucas, and B. Lawn. 2009. Remarkable
resilience of teeth. Proceedings of the National Academy of Sciences
106(18):7289-7293.

Chen, G. F., and N. Schmidt-Kittler. 1983. The deciduous dentition of Percrocuta
Kretzoi and the diphyletic origin of the hyaenas (Carnivora, Mammalia).
Palaontologische Zeitschrift 57(1-2):159-169.

Cheverud, J. M. 1982. Phenotypic, genetic, and environmental morphological integration
in the cranium. Evolution 36(3):499-516.

Christiansen, P. 1999. Scaling of the limb long bones to body mass in terrestrial
mammals. Journal of Morphology 239:167-190.

Christiansen, P. 2008. Evolution of skull and mandible shape in cats (Carnivora: Felidae).
PLoS ONE 3(7):e2807.

Clough, R. W. 1960. The Finite Element Method in Plane Stress Analysis. Proceedings of
2nd ASCE Conference on Electronic Computation. Pittsburgh, PA.

Cowin, S. C. 1989. Bone mechanics. CRC Press, Inc., Boba Baton, FL.

508

Craven, B. A., T. Neuberger, E. G. Paterson, A. G. Webb, E. M. Josephson, E. E.
Morrison, and G. S. Settles. 2007. Reconstruction and morphometric analysis of
the nasal airway of the dog (Canis familiaris) and implications regarding
olfactory airflow. The Anatomical Record 290:1325-1340.

Creel, S., and N. M. Creel. 1995. Communal hunting and pack size in African wild dogs,
Lycaon pictus. Animal Behaviour 50:1325-1339.

Crusafont-Pairó, M. 1950. El primer representante del género Canis en el Pontiense
eurasiatico (Canis cipio nova sp.). Boletin de la Real Sociedad Española de
Historia Natural (Geologia) 48:43-51.

Crusafont-Pairó, M., and J. Truyols-Santonja. 1956. A biometric study of the evolution of
fissiped carnivores. Evolution 10:314-332.

Daegling, D. J., and F. E. Grine. 1994. Bamboo feeding, dental microwear, and diet of
the Paleistocene ape Gigantopithecus blacki. South African Journal of Science
90:527-532.

Daeschler, E. B., N. H. Shubin, and F. A. J. Jenkins. 2006. A Devonian tetrapod-like fish
and the evolution of the tetrapod body plan. Nature 440:757-763.

Darwin, C. 1859. On the origin of species by means of natural selection or the
preservation of favoured races in the struggle for life. John Murray, London.

Davis, D. 1955. Masticatory apparatus in the spectacled bear Tremarctos ornatus.
Fieldiana: Zoology 37:25-46.

Davis, J. L., S. E. Santana, E. R. Dumont, and I. Grosse. 2010. Predicting bite force in
mammals: two-dimensional versus three-dimensional lever models. Journal of
Experiemental Biology 213:1844-1851.

Dawkins, R. 1986. The Blind Watchmaker. Longman Scientific & TEchnical, Harlow.

Dawson, M. R., R. K. Stucky, L. Krishtalka, and C. C. Black. 1986. Machaeroides
simpsoni, new species, oldest known sabertooth creodont (Mammalia), of the Lost
Cabin Eocene. Contributions to Geology, University of Wyoming, Special Paper
3:177-182.

De Bonis, L., S. Peigne, A. Likius, H. T. Mackaye, P. Vignaud, and M. Brunet. 2007.
The oldest African fox (Vulpes riffautae n. sp., Canidae, Carnivora) recovered in
late Miocene deposits of the Djurab desert, Chad. Naturwissenschaften 94:575-
580.
509

De Muizon, C., and B. Lange-Badre. 1997. Carnivorous dental adaptations in
tribosphenic mammals and phylogenetic reconstruction. Lethaia 30(4):353-366.

DeBeer, G. 1954. Archaeopteryx and evolution. Advancement of Science 11:160-170.

Dechow, P. C., and W. L. Hylander. 2000. Elastic properties and masticatory bone stress
in the macaque mandible. American Journal of Physical Anthropology 112:553-
574.

Deng, T., and Z. J. Tseng. 2010. Osteological evidence for predatory behavior of the
giant percrocutid (Dinocrocuta gigantea) as an active hunter. Chinese Science
Bulletin 55(17):1790-1794.

Denton, D. A., J. R. Blair-West, M. J. McKinley, and J. F. Nelson. 1986. Problems and
paradigms: physiological analysis of bone appetite (Osteophagia). BioEssays
4(1):40-43.

Dessem, D. 1985. The transmission of muscle force across the unfused symphysis in
mammalian carnivores. Pp. 289-291. In H. R. Duncker, and G. Fleischer, eds.
Vertebrate Morphology. Gustav Fischer Verlag, New York.

Dessem, D. 1989. Interactions between jaw-muscle recruitment and jaw-joint forces in
Canis familiaris. Journal of Anatomy 164:101-121.

DeVault, T. L., J. Rhodes, Olin E., and J. A. Shivik. 2003. Scavenging by vertebrates:
behavioral, ecological, and evolutionary perspectives on an important energy
transfer pathway in terrestrial ecosystems. Oikos 102:225-234.

Domning, D. P., and B. L. Beatty. 2007. Use of tusks in feeding by dugongid sirenians:
Observations and tests of hypotheses. Anatomical Record 290(6):523-538.

Dumont, E. R., J. L. Davis, I. Grosse, and