Morphological and Behavioral Development in a Top North American Carnivore, the Coyote

Home , Fossa (animal)

MORPHOLOGICAL AND BEHAVIORAL DEVELOPMENT IN A TOP NORTH AMERICAN CARNIVORE, THE COYOTE

Suzanne La Croix

A DISSERTATION

Submitted to Michigan State University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Zoology

2011

ABSTRACT

MORPHOLOGICAL AND BEHAVIORAL DEVELOPMENT IN A TOP NORTH AMERICAN CARNIVORE, THE COYOTE

Suzanne La Croix

Young animals must resolve the conflicting demands of survival and growth, ensuring that they can function while developing towards their adult form. The ontogenetic relationship between cranium and mandible is of special interest because these two parts of the feeding apparatus must function in concert while both change along steep growth trajectories. Further, phenotypic plasticity in the feeding apparatus warrants investigation because variation in the emergent phenotype of the food processing apparatus has obvious implications for form and function, and important consequences for survival.

This dissertation investigated morphological and behavioral development of coyotes. First, I examined ontogenetic changes in skull shape and the relationship between cranium and mandible, relative to key life-history events. In coyotes, there was synchrony of growth between cranium and mandible, and asynchrony of mandibular development; these patterns were also characteristic of spotted hyenas.

However, coyotes had a much less protracted development than that in hyenas, and coyotes were handicapped relative to adults for a much shorter time. Morphological development did not predict life-history events in these two carnivores as it has in rodents. Second, I investigated the ontogeny of feeding performance and feeding biomechanics in coyotes. The results showed that the development of feeding

performance was asynchronous with that of both feeding biomechanics and skull

morphology and that this pattern was also characteristic of spotted hyenas. This

developmental asynchrony suggests that a certain minimum threshold of physical

growth and development, together with the associated development of

biomechanics, are required to produce effective mastication. Third, I examined

developmental plasticity in feeding behavior and feeding apparatus morphology by

documenting variation in adult phenotypes due to experimental manipulation of diet.

Variation in early bone processing opportunities lead to differences in adult skull

shape, size and mastication musculature and this variation in morphology mediated

the relationship between early diet and adult feeding performance. Fourth, I

examined phenotypic plasticity in feeding apparatus morphology and feeding

biomechanics through a comparison of captive and wild coyote skulls drawn from

the same geographical area. Skull shape, skull size and length, and feeding

biomechanics exhibited phenotypic plasticity in response to captivity. I

demonstrated environmental effects on adult form, and such differences in form

have implications for function.

Collectively, the chapters in this dissertation provide a great deal of insight

into developmental relationships between skull morphology and feeding behavior in

Canis . Our results support the idea that developmental processes exhibit phenotypic plasticity, being sensitive to environmental effects. This results in variation in ontogenetic outcomes and has implications for both adult form and function.

SUZANNE LA CROIX

2011

ACKNOWLEDGEMENTS

My doctoral research follows from my life-long fascination with canids and my undergraduate regret of never having taken a course in zoology. In 1992, I remedied the regret and was set upon a new life path - a scientific career that featured ethological research on canids. To start at the beginning, then, I must thank

Dr. Nancy Seefelt, my graduate student colleague at both Central Michigan

University and Michigan State University (MSU), who made my introduction to Dr.

Kay Holekamp in 2001. Having found a synergy with Kay, I soon moved to East

Lansing with my entourage of seven dogs and set to work.

For this opportunity and for all her support throughout my doctoral research, I thank Kay Holekamp. Kay is an outstanding mentor who generously shares her vision while providing us with the tools to develop our own. She is a rigorous, passionate, and intelligent scientist who leads by example. Her scholarly achievements and professional demeanor inspire us to develop complex research questions and to network across academic disciplines. She is also understanding, encouraging and loyal; her faith in our abilities to reach our prescribed goals never wavers and she is first at the finish line to offer her sincerest congratulations. Kay has impacted me in so many ways, personally and professionally, that words are inadequate to express what a privilege it is to count her among my confidents and my colleagues. For all this, I am appreciative and grateful.

My other committee members, Drs. Laura Smale, Barbara Lundrigan and

Joseph Vorro, have also generously offered me their support and insights at every

v critical stage of my doctoral program. Without their encouragement and wisdom, this dissertation would surely be a different product. Laura Smale gave scientific basis to my interests in development periods and all things ontogenetic. Barb Lundrigan encouraged my efforts to create a one-of-a-kind ontogenetic series of known-age coyote skulls and set me on the path of geometric morphometrics. And Joe Vorro brought fresh perspectives and anatomical expertise to my research while offering great life advice at no extra charge.

During my time at MSU, I have had the opportunity to share a lab with Sarah

Benson-Amram, Pat Bills, Andy Booms, Katy Califf, Leslie Curren, Stephanie

Dloniak, Anne Engh, Andy Flies, David Green, Julia Greenberg, Sarah Jones, Joe

Kolowski, Sarah Lansing, Eva-Maria Muecke, Wiline Pangle, Kate Shaw, Jennifer

Smith, Greg Stricker, Eli Swanson, Jaime Tanner, Kevin Theis, Russ Van Horn,

Page Van Meter, Aaron Wagner, Sofi Wahaj, Heather Watts, and Marc Wiseman.

Our extended lab family included Terri McElhinny and the members of the

McAdam’s lab. I have enjoyed their camaraderie, insight and encouragement and thank them for all their suggestions and assistance along the way. In addition, I would like to thank Casey Leonard and Gwen Webster, MSU undergraduates, and

Andrea Beaudet, a promising high school student, for their personal assistance in collecting data.

This work could never have been accomplished without the assistance and direct support of the USDA APHIS Wildlife Services National Wildlife Research

Center Logan Field Station in Millville, Utah. I am grateful for their interest in this project and for their tireless efforts. Dr. John Shivik, Doris Zemlicka, Jeff Schultz,

Stacey Brummer, Patrick Darrow, and many other staff members at the Logan Field

Station provided unparalleled support in facilitating my access to research animals

and in managing animal care during my study. Further, I would like to thank the

Utah State University community for welcoming me to their canid luncheons,

especially Drs. Fred Knowlton, Mike Jaeger, Eric Gese and Lynne Gilbert-Norton.

Their camaraderie and insights into coyotes were most valuable to me.

The success of this project has relied greatly on collaborative input. I am

especially indebted to Dr. Miriam Zelditch of the University of Michigan. Her

knowledge of geometric morphometrics provided the foundation for my analyses of

skull growth, development and plasticity. Further, Miriam’s insights and mentoring

have molded my critical thinking skills and made me a better scientist.

I would also like to thank Laura Abraszinskas and Paula Hildebrandt, of MSU

Museum, who provided the technical expertise and manpower for creating a world

class ontogenetic collection of coyote skulls. In addition, Drs. David Long and

Matthew Parsons of MSU’s Department of Geological Sciences provided equipment

and supervision for the lyophilization of mastication muscles. Finally, I would like to

thank the entire staff of the MSU Zoology Department; it is their administrative

support and daily doses of cheer that permit us to successfully navigate the hurdles

to degree completion.

I thank MSU for their generous support during all phases of my dissertation

work. The College of Natural Science awarded me Graduate Summer Research and

Dissertation Completion Fellowships. I received Research Enhancement and Travel

Fellowships from the Graduate School. The Department of Zoology provided me

vii with a John R. Shaver Award and Research Grants. This work was also made possible by National Science Foundation grants awarded to Kay Holekamp

(IOB0618022 and IOS-0819437). Without this financial support, none of this research would have been possible.

Finally, I would like to express my gratitude to my family. The unwavering support of my husband, Geoff Barker, and our daughter, Brooke, have permitted me to pursue and complete this research. My parents, too, have assisted whenever possible to help me attain my personal goals. And of course, our dogs were my loyal companions throughout.

viii

TABLE OF CONTENTS

LIST OF TABLES……………………………………………...…………………...……….x

LIST OF FIGURES…………………………………………………………………..……xiv

CHAPTER 1: GENERAL INTRODUCTION…………………………………..…………………...…..…1

CHAPTER 2: ONTOGENETIC RELATIONSHIPS BETWEEN CRANIUM AND MANDIBLE IN COYOTES AND HYENAS………………...………………………………………..…….13 Introduction………………………………………………………….…….………..14 Methods…………………………………………………………………..………...18 Results…………………………………………………………………..………….28 Discussion……………………………………………………………..………...…42

CHAPTER 3: ONTOGENY OF FEEDING PERFORMANCE AND BIOMECHANICS IN COYOTES AND HYENAS………………...………………………………………………………..….50 Introduction………………………………………………………………………...51 Methods…………………………………………………………………….....……55 Results………………………………………………………………....……..…….65 Discussion………………………………………………………………….....……82

CHAPTER 4: EFFECTS OF EARLY BONE PROCESSING ON FEEDING PEFORMANCE, SKULL MORPHOLOGY, AND BIOMECHANICS IN COYOTES…………………………..….93 Introduction…………………………………………………………………….…..93 Methods……………………………………………………………………….....…95 Results………………………………………………………………….………....109 Discussion………………………………………………………………………...126

CHAPTER 5: DIFFERENCES IN SKULL MORPHOLOGY AND FEEDING BIOMECHANICS BETWEEN WILD AND CAPTIVE COYOTES……….……………………………..…131 Introduction……………………………………………………………………..…131 Methods…………………………………………………………………….….….132 Results..………………………………………………………………………..….143 Discussion…………………………………………………………….…………..152

APPENDIX…………..……………………………………………………………..……..159

LITERATURE CITED……………………………………………………………..……..178

LIST OF TABLES

Table 2.1. Life history stages of Canis latrans and the corresponding developmental milestones and ages in weeks. Sample sizes are given for skulls photographed for analyses in ventral cranium ( V), lateral cranium ( L), and mandible ( M) views.……..19

Table 2.2. Comparisons between ontogenetic trajectories of successive life history stages of Canis latrans for the ventral cranium, lateral cranium, and mandible views. Comparisons were made between successive life history stages only in cases where both single stages exhibited significant age-related change in shape. Ontogenetic trajectories of two successive stages were determined to differ significantly when the angle between the vectors of the two stages exceeded that obtained by resampling (400 times) within each stage (younger and older). If the observed angle between stages exceeded the 95% confidence interval of the two within-stage ranges, the difference was judged to be statistically significant, here indicated in bold font…….31

Table 2.3. Best-fitting models for the measures of size (centroid size) and shape (Procrustes distance) maturity. The AIC weight evaluates relative goodness-of-fit by balancing the distance between model and data by degrees of freedom. AC refers to serial autocorrelation among residuals of the model (“ns” indicates there was no statistically significant serial autocorrelation)………………………………………..….36

Table 2.4. Estimates, based on best fitting model, for asymptotic size, A; growth rate constant, k; age (in weeks) at the curve inflexion point, To; and age (in weeks) at adult maturity, M; 95% confidence intervals given in parentheses…………….….37

Table 2.5. Estimates, based on best fitting model, for asymptotic size, A; growth rate constant, k; age (in weeks) at the curve inflexion point, To; and age (in weeks) at adult maturity, M; 95% confidence intervals given in parentheses………………..38

Table 2.6. Relative maturity of size (CS/A) and relative maturity of shape (PD/A) at post-natal ages, based on parameters of the logistic model. CS is centroid size, PD is Procrustes distance, and A is the corresponding Asymptotic value for shape or size...... …………………………………………...………..39

Table 3.1. Best-fitting models for the measures of feeding performance and skull biomechanics maturation. The AIC weight evaluates relative goodness-of-fit by balancing the distance between model and data by degrees of freedom. AC refers to serial autocorrelation among residuals of the model (“ns” indicates there was no statistically significant serial autocorrelation). Percent variance explained (% Var) by the best-fit model is also given…………………………………………………..……….64

Table 3.2. Estimates of maturation for: feeding performance, based on the best- fitting German Gompertz model, for asymptotic feeding performance speed (in secs), A; initial growth rate, k; decay of the growth rate, b; and age (in weeks) at adult maturity, M; maximum zygomatic arch breadth (ZAB), based on the best-fitting monomolecular model, for asymptotic size (in mm), A; growth rate constant, k; age (in weeks) at the onset of growth, To; and age (in weeks) at adult maturity, M; and relative bite strength, based on the best-fitting logistic model, for asymptotic value, A; growth rate constant, k; age (in weeks) at the curve inflexion point, To; and age (in weeks) at adult maturity, M. Ninety-five percent confidence intervals are given in parentheses……………………………………………………………...…………………76

Table 3.3. Age at maturation (95% of adult value), as estimated from best-fit models; feeding performance and biomechanics (this study) and skull size and shape (Ch. 2, La Croix, et. al., 2011)…………………..……………………...………...77

Table 3.4. Relative maturity of feeding performance, maximum zygomatic arch breadth (ZAB), and relative bite strength, based on parameters of its best-fit model. Relative maturity is calculated by dividing a model’s predicted value at a given age by the corresponding asymptotic value………………………………………………....81

Table 3.5. Mean values of mechanical advantage of the temporalis, for two age groups of coyotes, and relative maturity of the younger group (indicated in bold), as calculated by dividing mean younger group mechanical advantage by that of the older (adult) group…………………………………………………….…………………...81

Table 4.1. Results of ANOVA for effects of litter, age, and treatment (bone males vs. no-bone males vs. no-bone females) on whole body mass among animals aged 565 days; SS is the sum of squares, df is the degrees of freedom, MS is the mean square, F is the test statistic, p is the probability of statistical significance, and * indicates statistically significant differences…………………………….……………..120

Table 4.2. Summary of the analyses of differences in feeding performance, skull morphology, feeding biomechanics and muscle mass between male coyotes provided with early access to bone processing opportunities (Bone) and those without (No-bone). Statistically significant differences (p < 0.05) between the groups are indicated, *; differences that were not statistically significant are indicated, “ns;” differences that are indicative of a trend, 0.05 < p < 0.10 are also indicated.……………………………………………………………….………..….…….128

Table 5.1. Sample sizes of wild and captive coyotes for analyses of skull shape; coyotes do not exhibit sexual shape dimorphism for the skull, therefore, specimens of known and unknown sex were pooled for these analyses…………………..…...141

Table 5.2. Sample sizes of wild and captive coyotes for analyses of skull shape, skull size, skull length, mechanical advantage of the temporalis, maximum zygomatic arch breadth (ZAB) and relative bite strength…………………………….141

Table 5.3. Results of MANOVA for effects of sex and environment (wild vs. captive) on coyote skull size, measured as centroid size (CS), for three views of the skull; SS is the sum of squares, df is the degrees of freedom, MS is the mean square, F is the test statistic, p is the probability of statistical significance, and * indicates statistically significant differences……………………………………………………………...….…144

Table 5.4. Mean centroid size (CS) for three views of the skull for wild and captive coyotes of both sexes………………………………………………………………...….145

Table 5.5. Results of MANOVA for effects of sex and environment (wild vs. captive) on coyote skull length, mechanical advantage of the temporalis, maximum zygomatic arch breadth, and relative bite strength; SS is the sum of squares, df is the degrees of freedom, MS is the mean square, F is the test statistic, p is the probability of statistical significance, and * indicates statistically significant differences. …………………………………………………………….…………………146

Table 5.6. Mean skull length (mm), mechanical advantage of the temporalis, maximum zygomatic arch breadth (ZAB)(mm), and relative bite strength for captive and wild coyotes of both sexes…………………………………………………………154

Appendix Table A.2.1. Canis latrans specimens……………………….…..…….…160

Appendix Table A.2.2. Description of landmarks for each view………….…....….162

Appendix Table A.2.3. Relative fit of the eight growth models fitted to centroid size. The AIC weight evaluates relative goodness-of-fit by balancing the distance between model and data by degrees of freedom. AC refers to serial autocorrelation among residuals of the model (statistically significant are indicated by an asterisk). The AIC is not applied to models with significant AC. The model judged best is in bold type………………………………………………………………………….…………..…164

Appendix Table A.2.4. Relative fit of the eight growth models fitted to the measure of developmental maturity (Procrustes distance). The AIC weight evaluates relative goodness-of-fit by balancing the distance between model and data by degrees of freedom. AC refers to serial autocorrelation among residuals of the model (statistically significant are indicated by an asterisk). The AIC is not applied to models with significant AC. The model judged best is in bold type………….…….165

Appendix Table A.3.1. Canis latrans specimens……………………….…….……..166

Appendix Table A.3.2. Canis latrans feeding performance subjects………..….…167

Appendix Table A.3.3. Canis latrans bite strength subjects………………..………169

xii

Appendix Table A.3.4. Relative fit of the eight growth models fitted to the feeding performance, maximum zygomatic arch breadth and relative bite strength. The AIC weight evaluates relative goodness-of-fit by balancing the distance between model and data by degrees of freedom. AC refers to serial autocorrelation among residuals of the model (statistically significant are indicated by an asterisk). The AIC is not applied to models with significant AC. The model judged best is in bold type…………………………………………………………………….………………..…170

Appendix Table A.4.1. Canis latrans specimens………………..……………….….171

Appendix Table A.4.2. Description of landmarks for each view…….…..……..….172

Appendix Table A.5.1. Canis latrans specimens, captive………………….………174

Appendix Table A.5.2. Canis latrans specimens, wild….………………….….……175

Appendix Table A.5.3. Description of landmarks for each view……….…...... ….176

xiii

LIST OF FIGURES

Figure 2.1. Changes in cranium size and shape are visually dramatic among coyotes in successive life history stages…..……………………………………..……..16

Figure 2.2. Landmarks (numbered) and semi-landmarks ( o) shown on the cranium (a. ventral, b. lateral) and mandible (c.) of an 18 month old coyote, Canis latrans . Descriptions of each landmark and semi-landmark are provided in Appendix Table A.2.2………...……………………………………………………………………...……….23

Figure 2.3. Ontogenetic changes from a linear regression of shape on log(age) for the cranium (a. ventral, b. lateral) and mandible (c.). Vectors on landmarks and semi-landmarks show the direction and magnitude of change from the youngest to the oldest specimens after centroid size is scaled to the same size for each specimen.……………………………………………………………….……….…..……..29

Figure 2.4. Ventral view of the cranium showing the linear regression of shape on log(age) for each life history stage. Vectors show the direction and magnitude of change from the youngest to the oldest specimens in that life history stage after centroid size is scaled to the same size for each specimen. No statistically significant change occurred during the subadult period, so this stage is not shown……….…..32

Figure 2.5. Lateral cranium (a-e) and mandible (f-i) views of the linear regression of shape on log(age) for each life history stage. Vectors and notation are as in Fig. 2.4.………………………………………………………………………………..…………33

Figure 2.6. Timeline illustrating the relative maturity in size (a.) and shape (b.) for the ventral and lateral views of the cranium and the mandible for coyotes in relation to major life history events. There is no significant difference between males and females in rate of maturation of skull size or skull shape……………………….…….40

Figure 2.7. Timeline illustrating the age at maturation for coyote skull morphology in relation to major life history events. Male and female skull maturation occurs within the same week for each morphological measure; the latest maturation age is indicated for each measure by a downward-pointing arrow………………..………...41

Figure 2.8. Comparison of maturation timing of cranium (ventral and lateral views) and mandible between coyotes and spotted hyenas in relation to major life history events. Maturation for coyote (this study) is indicated with a dashed line, and for spotted hyenas (Tanner et. al., 2010) with a solid line………………………….……..47

xiv

Figure 3.1. Ontogenetic variation in cranial and mandibular size and shape among coyotes ( Canis latrans ) aged, from top, 1 day, 6 weeks, 12 weeks, and 26 weeks is visually dramatic…………………………………………………………………………...52

Figure 3.2. Traditional linear measures of the skull used to calculate mechanical advantage of the temporalis, and to estimate bite strength. The in-lever arm length is the distance between the mandibular condyle and the dorsal tip of the coronoid process. The out-lever arm length is the distance between the mandibular condyle and the bite point (here, the dorsal tip of the mandibular carnassial tooth, M1). Maximum zygomatic arch breadth was a proxy for muscle mass in estimating bite strength.……………………………………………………...……………………………..61

Figure 3.3. Growth plots of coyote feeding performance for females (a.) and males (b.) as measured by consumption time, in seconds, for a 32g dog biscuit. At six weeks of age, coyote pups of both sexes were unable to process the biscuit. Only data performance for animals under 81 weeks of age are shown here.………..…..66

Figure 3.4. Plot of bite strength, measured in Newtons (N), obtained using a piezoelectric transducer, for male and female coyotes………………………….……68

Figure 3.5 . Schedule of tooth eruption and replacement (including post-canine teeth) in coyotes. The presence of each deciduous (D) or Adult (A) tooth is indicated (a. cranium, b. mandible). All coyote skull specimens examined at each age displayed the specified tooth configuration……………………...………………..……69

Figure 3.6. Plot of mechanical advantage of the temporalis for coyotes, by age (a) and comparison of mean mechanical advantage of the temporalis between coyotes with primarily deciduous dentition and those with primarily adult dentition (b.). Only data for coyotes aged less than 100 weeks are shown in (a.), while data for all ages are included in (b.)…………………………………………..………………………...…..71

Figure 3.7. Growth plots for the components of the mechanical advantage of the temporalis: the in-lever arm length (distance between the mandibular condyle and the dorsal tip of the coronoid process) and out-lever arm length (distance between the mandibular condyle and the bite point) (a.); regression of out-lever arm length on in-lever arm length (b.); and plot of the residuals from that regression (c.). Only data for coyotes aged less than 100 weeks are shown in (a.), while data for all ages are included in (b.) and (c.)………………………………………………………..….…..73

Figure 3.8. Plots of maximum zygomatic arch breadth (ZAB), as measured from skulls (a.), and relative bite strength, as estimated by models (b.), by age, in weeks, for coyotes. Only data for coyotes aged less than 100 weeks are shown……..…...74

Figure 3.9. Timeline illustrating the relative maturity of coyote feeding performance to that of feeding biomechanics (a.), skull size (b.), and skull shape (c.), in relation to major life history events. Relative maturity for biomechanics (this study) and skull morphology (Ch. 2, La Croix et. al., 2011) is indicated with a dashed line, and for feeding performance (this study) with a solid line. Maturation of mechanical advantage of the temporalis (a.) does not follow a traditional growth pattern, but is, instead maintained within two discrete ranges throughout ontogeny, one for younger animals and one for older animals (see Fig. 3.5.a.); here, we have diagrammed the relative maturity of the younger group as calculated by dividing the mean younger group mechanical advantage by that of the older (adult) group (see Table 3.5.)…..78

Figure 3.10. Timeline illustrating the age at maturation for coyote feeding performance, biomechanics and skull morphology in relation to major life history events. Maturation for skull morphology (Ch. 2., La Croix et. al., 2011) is indicated with a dashed line, and for feeding performance and biomechanics (this study) with a solid line. Where maturation age differed between sexes for a measure, the latest maturation age is diagrammed…………………………………….……………..………84

Figure 3.11. Comparison of maturation timing of feeding performance and biomechanics for coyotes and spotted hyenas in relation to major life history events. Maturation for coyotes (this study, Ch. 2, and La Croix et. al., 2011) is indicated with a dashed line, and for spotted hyenas (Tanner et. al., 2010 and Tanner, unpublished data) with a solid line. The latest maturation age is diagrammed for all measures...89

Figure 4.1. Landmarks (numbered) and semi-landmarks ( o) shown on the cranium (a. ventral, b. lateral) and mandible (c.) of an 18 month old coyote, Canis latrans . Descriptions of each landmark and semi-landmark are provided in Appendix Table A.4.2.….……………...…………………………………………………………………...103

Figure 4.2. Traditional linear measures of the skull used to compare skull length, calculate mechanical advantage of the temporalis, and estimate bite strength. Skull length is the distance between the anterior-most point on the maxilla and the posterior-most point on the nuchal crest (a.). The in-lever arm length is the Moment Arm of the Temporalis which is the distance between the mandibular condyle and the dorsal tip of the coronoid process (b.). The out-lever arm length is the distance between the mandibular condyle and the bite point (here, the highest cusp of the mandibular M1, the carnassial tooth) (b.). Maximum zygomatic arch breadth was a proxy for muscle mass in estimating bite strength and was measured as the widest point on the zygomatic arches (c.)…..…………………………………………….…..107

xvi

Figure 4.3. Comparisons of mean feeding performances between coyotes with early bone chewing experience (Bone), and those without (No-bone) for standardized food objects: a) rawhide chew twist, b) 32g dog biscuit, and c) beef shank. Bone chewing animals, on average, consumed rawhide chew twists and biscuits faster and reduced beef shanks by a greater percentage. The percent reductions in beef shank mass achieved in the reduction test (c.) are diagrammed for individual animals (d.) with a line connecting each matched pair of littermates.…………………………………………………………..…………………….110

Figure 4.4. Diagrams of differences in skull shape between coyotes with early bone chewing access (Bone), and those without (No-bone), for the cranium (a. ventral view, b. lateral view) and mandible (c.). Vectors on landmarks and semi-landmarks show the direction and magnitude of change from the control (no-bone) animals’ landmarks to the bone chewing animal’s landmarks after centroid size is scaled to the same size for each specimen.…………………………………………………...…113

Figure 4.5. Comparison of crania between matched pairs of littermate coyotes at 18 months old: a) animals without access to early bone chewing opportunities; b) animals with early bone chewing opportunities. Greater relative skull length, posterior to the orbit, and enhanced sagittal and nuchal crest development visually distinguish animals that chewed bones from those that did not in this view of the skull.………………………………………………………...…………………..…………115

Figure 4.6. Comparison of mandibles between matched pairs of littermate coyotes, at18 months old: a) animals without access to early bone chewing opportunities; b) animals with early bone chewing opportunities. Enhanced condylar and angular processes and greater horizontal ramus depth, below the carnassials, visually distinguish animals that chewed bones from those that did not in this view of the mandible………………………………………………………………………………..…116

Figure 4.7. Comparisons of skull centroid size (CS), between coyotes with early bone chewing access (Bone), and those without (No-bone), for the cranium (a. ventral view, b. lateral view), and the mandible (c.). Comparison of skull length, measured in mm, between the two groups (d.). Results show significant differences between the two groups (*) for cranial size (CS), in the lateral view (b.), and for skull length (d.) ……………………………………………………….……………………..…117

Figure 4.8. Comparisons for measures of feeding biomechanics between coyotes with early bone chewing access (Bone), and those without (No-bone): a) mechanical advantage of the temporalis; b) maximum zygomatic arch breadth (ZAB) (mm); and c) relative bite strength, calculated as the product of mechanical advantage of the temporalis and maximum ZAB…………………………..…………118

xvii

Figure 4.9 . Plots of whole body mass (kg) for animals at 28, 90, 120, and 565 days of age by litter group (a.); treatment groups are indicated, bone treatment animals (▲), no-bone males ( ), and no-bone females ( o). Comparisons of whole body mass (kg) among diet groups at 28 days of age (b.), and at 585 days of age (c.); statistically significant differences between treatment groups are indicated (*)...…121

Figure 4.10. Comparisons of cranial (a.) and mandibular mass (b.), measured in grams, between coyotes with early bone chewing opportunities (Bone), and those without (No-bone)…………………………………….………………………………..…123

Figure 4.11. Comparisons of mean mastication muscle dry mass (a.) and mastication muscle dry mass, normalized by dividing by whole body mass (b.), between coyotes with early bone chewing opportunities (Bone), and those without (No-bone)…….……………………………………………………………………………125

Figure 5.1. Landmarks (numbered) and semi-landmarks ( o) shown on the cranium (a. ventral, b. lateral) and mandible (c.) of an 18 month old coyote, Canis latrans . Descriptions of each landmark and semi-landmark are provided in Appendix Table A.5.3…………………………..……………………………………………...……………136

Figure 5.2. Traditional linear measures of the skull used to compare skull length, calculate mechanical advantage of the temporalis, and estimate bite strength. Skull length is the distance between the anterior-most point on the maxilla and the posterior-most point on the nuchal crest (a.). The in-lever arm length is the distance between the mandibular condyle and the dorsal tip of the coronoid process (b.). The out-lever arm length is the distance between the mandibular condyle and the bite point (here, the highest cusp of the mandibular M1, the carnassial tooth) (b.). Maximum zygomatic arch breadth was a proxy for muscle mass in estimating bite strength and was measured as the widest point on the zygomatic arches (c.)..….140

Figure 5.3. Comparisons of skull centroid size (CS), between captive and wild coyotes of both sexes, for the cranium (a. ventral view, b. lateral view), and the mandible (c.). Comparison of skull length, measured in mm, between the groups (d.). Statistically significant differences are indicated (*)………………………..…..147

Figure 5.4 . Deformation of change in skull shape between captive and wild coyotes, for the cranium (a. ventral view, b. lateral view) and mandible (c.). Vectors on landmarks and semi-landmarks show the direction and magnitude of change from the captive animals’ landmarks to the wild animals’ landmarks after centroid size is scaled to the same size for each specimen……………………………………...……150

xviii

Figure 5.5. Comparisons of the cranium, in ventral and lateral views, and the mandible, between captive (left) and wild (right) coyotes. Differences in skull shape that are easily visible and that distinguish wild coyotes from captive ones, include narrowing and flattening along the anteroposterior axis of the cranium, enhanced sagittal and nuchal crest development, and broadening of the coronoid and articular processes……………………………………………………….……………………...…151

Figure 5.6. Comparisons of measures of feeding biomechanics, between captive and wild coyotes of both sexes, including mechanical advantage of the temporalis (a.), maximum zygomatic arch breadth measured in millimeters (b.), and relative bite strength (c.). Statistically significant differences are indicated (*)…..………….…..153

xix

Chapter ONE

GENERAL INTRODUCTION

“Not infrequently, performance characteristics, measured as maximal speed or endurance, make the difference between eating and being eaten” (Tenney, 1967).

Thus, performance translates into individual survival and fitness (Arnold, 1983;

Garland Jr. and Losos, 1994). This is true within each life stage of the organism: neonatal, juvenile, subadult and adult. However, the specific set of physiological and behavioral challenges that must be addressed at each life stage is unique.

Young animals must resolve the conflicting demands of survival and growth, ensuring that they can function as infants or juveniles while developing towards their adult form (Herrel and Gibb, 2006; Monteiro et al., 1999; Wainwright and Reilly,

1994). How developing animals resolve these time-critical challenges of survival and growth can illuminate the relationship between form and function.

During ontogeny, feeding behavior and morphology are subject to developmental constraints that may yield a form that is sub-optimal for solving the food acquisition and ingestion problems confronted by the organism in its early life stages. Indeed, it has previously been shown that juvenile vertebrates are limited in their feeding capabilities due to small size, inexperience and lack of mechanical advantage of the jaws (Binder and Van Valkenburgh, 2000; Herrel and Gibb, 2006;

Tanner et al., 2010). In some species, ontogenetic variation in the skull yields juveniles and subadults with bite strength ability that is considerably less than that of adults (Binder and Van Valkenburgh, 2000; Christiansen and Adolfsen, 2005;

Erickson et al., 2003a; Tanner et al., 2010).

1 The ontogenetic relationship between cranium and mandible is of special interest because these two parts of the feeding apparatus must function in concert while both change along steep growth trajectories. Not only does each part exhibit its own ontogenetic change but also their ontogenetic coordination is necessary to ensure continuous functionality. Thus, understanding the relationship between the cranium and mandible during ontogeny is basic to understanding the function of the developing feeding apparatus, especially the way in which young mammals negotiate the challenges posed by competition with adults.

Young mammals do enjoy a brief period of nutritional dependence on the parents; however, at weaning, juveniles must be capable of processing adult foods or adopt an alternative diet (Herrel and Gibb, 2006; Monteiro et al., 1999; Wainwright and Reilly, 1994). The extent to which morphology limits juvenile feeding performance and determines the youngest ages at which adult performance can be achieved, is unknown for most mammals (Erickson et al., 2003b; Herrel and O'Reilly,

2006; Tanner, 2007; Tanner et al., 2010; Thompson et al., 2003a). Carnivores, especially, may face tremendous challenges during ontogeny as they attempt to provision themselves with food and compete with adults despite their immature morphology.

It is phenotypic plasticity, or sensitivity of developmental processes to environmental influences, that enables ontogenies to produce multiple outcomes.

Phenotypic plasticity in the feeding apparatus warrants investigation here because variation in the emergent phenotype of the food processing apparatus has obvious implications for form and function, and important consequences for survival.

2 This dissertation focuses on morphological and behavioral development in a

top North American carnivore, the coyote, Canis latrans . Coyotes are endemic to

North America and are the most numerous large carnivore despite constant

persecution by humans over the last several centuries. In the absence of other large

predators, like wolves, coyotes have expanded their range over the last 150 years

from an historical nucleus in the American southwest to a nearly cosmopolitan

coverage of the continent. As omnivorous generalists, coyotes demonstrate

remarkable behavioral plasticity in consuming small and large prey, living in pairs or

packs, and colonizing rural and urban environments (Bekoff, 1977; Bekoff and Gese,

2003). Their generalized carnivore morphology makes them appropriate for

comparisons with more highly specialized carnivores, such as bone-cracking

hyenas.

My research questions about the relationships among skull growth and

development, feeding performance, and phenotypic plasticity in morphology and

behavior arose from Jaime Tanner’s research on spotted hyenas. Indeed, my

coyote project builds on her investigations of morphological and behavioral

development in spotted hyenas, part of the Michigan State University Mara Hyena

Project, a long term research project addressing the behavioral ecology of spotted

hyenas in the Masai Mara National Reserve, Kenya.

Spotted hyenas ( Crocuta crocuta ) are morphologically and behaviorally specialized carnivores that are capable of cracking open bones of large diameter.

They possess specialized skulls and dentition, as well as immense mastication muscles, which permit them to generate and sustain bite forces that can crack open

3 and process even large ungulate femurs. These impressive animals live in a unique

female-dominated social system wherein they are subject to intense intra-specific

feeding competition. This species has evolved an extremely protracted juvenile

period, even among large carnivores, with weaning occurring at 14 months of age

and first parturition at three years of age. It is this protracted juvenile period that provides the time necessary for the development of the specialized skull that will permit individuals to compete as adults within hyena society. In order to gain perspective on the ontogeny of the hyena skull, and for evaluating skull growth and development in relation to life history milestones in hyenas, a comparative study was needed. As spotted hyenas are the most abundant large carnivore on the African continent, Dr. Holekamp was considering the most abundant large carnivores on the

North American continent. And then, serendipitously, a unique opportunity presented itself in the form of a colony of captive coyotes.

The National Wildlife Research Center (NWRC) is a major facility within the

U.S. Department of Agriculture (USDA) Animal and Plant Health Inspection Service's

(APHIS) Wildlife Services (WS) program, which maintains a captive colony of coyotes at its Logan Field Station. The 165-acre field station, located in Millville,

Utah, houses a colony of over 100 coyotes under semi-natural conditions. Research at this facility has generally focused on predator control with respect to livestock depredation. More recently, research there has detailed the ecology and behavior of coyotes in an effort to identify new management techniques and strategies, especially non-lethal tools.

4 We discovered that the Logan Field Station also housed a collection of 182 coyote skulls, all over the age of 8 months, which had been used in previous research (Knowlton and Whittemore, 2001). Most of these skulls were from known- age individuals, but many were not suitable for the analyses here, as they were damaged. Between 2004 and 2006, an additional 154 known-age coyote specimens were made available to us, and through the tireless efforts of Paula Hildebrandt, specimen preparator at the Michigan State University Museum, we assembled a unique collection of 147 coyote skulls, of known age, that formed an ontogenetic series ranging from neonates to individuals over 13 years of age. Two-thirds (101) of the specimens were under six months of age thus spanning the period of growth and development for the coyote skull. This unique collection of 147 skulls was accessioned into the Michigan State University Museum collection and is, to the best of our knowledge, a one-of-a-kind ontogenetic series of known-age carnivores.

What makes this collection even more unique is that specimen data include exact birth and death dates, as well as parentage.

At the Logan Field Station, the animal housing included two blocks of chain- link fenced kennels in a covered pole barn. These kennels provided an ideal setting for efficiently conducting feeding performance tests on individually housed coyotes.

Over a two year period, we were able to conduct feeding performance tests on adult coyotes and to conduct repeated tests of feeding performance on coyote pups as they matured. The captive setting also provided an opportunity not afforded when working with animals in the wild. That is, within the kennel setting, we were able to experimentally manipulate the diets of 24 coyote pups, and subsequently conduct

5 feeding performance tests on those animals at 18 months of age. Because the

facility elected not to maintain these animals as part of their breeding colony, these

same animals were immediately provided to us for dissection of mastication muscles

and preparation of skulls. This afforded us a unique opportunity to concurrently

analyze diet-related feeding performance and feeding apparatus morphology in the

same individuals. An experimental manipulation like this had not previously been

accomplished in any large carnivore.

From the very beginning, this project has been a collaborative effort. All the

data chapters in this dissertation were generated in collaboration with Dr. John

Shivik of the Logan Field Station, Dr. Miriam Zelditch, of the University of Michigan,

and other MSU investigators. Thus, this dissertation is the result of a great deal of

effort and insight from a number of collaborations. Indeed, each chapter was

prepared as a manuscript for publication with some of those collaborators.

Therefore, each of the following data chapters is written in the first person plural to

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

6 OVERVIEW OF CHAPTERS

I begin, in CHAPTER TWO, by examining growth and development of the cranium and mandible, using our unique ontogenetic series of known-age coyote skulls.

Here, I analyzed ontogenetic changes in the shapes of each part, and the relationship between them, relative to key life-history events. I found that both cranial and mandibular development in coyotes conform to general mammalian patterns, but that each also exhibits temporally and spatially localized maturational transformations, yielding a complex relationship between growth and development of each part, as well as complex patterns of synchronous growth and asynchronous development between parts. One major difference I found between the cranium and the mandible was that the cranium changes dramatically in both size and shape over ontogeny whereas the mandible undergoes only modest shape change. I found that the cranium and mandible are synchronous in growth, reaching adult size at the same life-history stage; but while growth and development are synchronous for the cranium, they are not for the mandible. I also found that the synchrony of growth between cranium and mandible, and asynchrony of mandibular development, as seen in coyotes, was also characteristic of a highly specialized carnivore, the spotted hyena. However, coyotes had a much less protracted development than that in hyenas, and coyotes were handicapped relative to adults for a much shorter time. I found that morphological development did not predict life-history events in these two carnivores, which is contrary to what had been reported for two rodent species (Myers et al., 1996). Finally, I concluded that adult functional demands cannot be satisfied by the morphology characterizing earlier life-history stages. This

7 chapter was recently accepted for publication in Journal of Morphology, where it is currently in press.

In CHAPTER THREE, I consider the ontogeny of feeding performance and feeding biomechanics in coyotes, in relation to what I had discovered about the ontogeny of skull growth and development. Here, the parts of the feeding apparatus and its associated biomechanics must maintain functional integrity to meet the feeding needs of juveniles even as the relationship between parts must change to meet the demands imposed on adults. I concurrently examined the ontogenetic relationships of feeding performance, dentition, and feeding biomechanics, relative to key life-history events, using our unique ontogenetic series of skulls and feeding performance data. I found that the development of feeding performance was asynchronous with development of both feeding biomechanics and skull morphology; feeding performance lagged during ontogeny despite surprisingly large early mechanical advantage of the temporalis, due in part, to early relative maturity of mandibular shape. In coyotes, feeding performance and biomechanics, like skull morphology, matured well after weaning at 6 weeks of age. Late maturation of bite strength and feeding performance was mediated by ongoing and continued growth of the temporalis muscles as measured by maximum zygomatic arch breadth. Thus, while mandibular growth patterns were important to early mechanical advantage of the temporalis, it was the ongoing development of the primary mastication muscles, the temporalis, that had an acute effect on the maturation of bite strength and ultimately feeding performance and fitness. I also found that males and females may resolve developmental conflicts differently, as females traded earlier maturity for

8 smaller maximum zygomatic arch breadth, lower relative bite strength and inferior feeding performance, compared to males. Again, I found that the asynchrony of feeding performance development seen in coyotes was also characteristic of spotted hyenas, but that coyotes had a much less protracted development, being

handicapped relative to adults for a much shorter time. This developmental

asynchrony between feeding performance and morphological measures suggests

that a certain minimum threshold of physical growth and development, together with

the associated development of biomechanics, are required to produce effective

mastication.

In CHAPTER FOUR, I examine developmental plasticity in feeding behavior

and feeding apparatus morphology by documenting variation in adult phenotypes

due to experimental manipulation of diet. Here, coyote pups were assigned to two

feeding groups, one of which had access to bones, while the other did not. Through

feeding performance trials and a parallel geometric morphometric analysis of skulls

at 18 months of age, I documented differences in form and function between animals

that received early bone processing opportunities and those that did not. I found

that access to early bone processing opportunities improved adult feeding

performance; coyotes that chewed bones showed faster consumption speeds for

large dog biscuits and rawhide chew twists, and greater reduction of shank bones, in

timed trials. I also found that the skull shape of animals that processed bones was

significantly different from that of animals that did not chew bones; the cranial and

mandibular differences were clearly visible. These differences included larger

sagittal and nuchal crests, basicranial narrowing along the anterio- posterior axis of

9 the skull, and larger areas for muscle attachment at the masseteric fossa and anterior to the articular process, in the bone chewing group. The only aspects of feeding biomechanics that differed between diet groups, were maximum zygomatic arch breadth and mastication muscle mass. Thus, I found that coyote feeding performance, feeding apparatus morphology, and feeding biomechanics exhibited plastic responses to variation in early diet opportunities. Further, I showed that variation in early bone processing opportunities lead to differences in adult skull shape, size and mastication musculature; it is this variation in morphology that mediates the relationship between early diet and adult feeding performance. Thus, early experience can put individual animals on different ontogenetic trajectories, and this may have important consequences for adult form and function in ways that influence fitness.

Finally, in CHAPTER FIVE, I examine phenotypic plasticity in feeding apparatus morphology and feeding biomechanics through a comparison of captive and wild coyote skulls drawn from the same geographical area. I found that skull shape, skull size and length, and feeding biomechanics exhibited phenotypic plasticity in response to captivity. Cranial and mandibular shape were significantly different between captive and wild individuals with wild specimens exhibiting dramatically enhanced skull structures for muscle attachment including larger surface area for muscle attachment at the sagittal and nuchal crests, and broader coronoid and articular processes, compared to captive specimens. Wild coyotes here, like captive coyotes previously studied (Ch. 2), exhibited sexual dimorphism in skull size but not in skull shape. Wild coyotes did not have significantly larger

10 craniums, nor did they have larger maximum zygomatic arch breadth, than captive

coyotes, although wild females did have larger mandibles than captive females.

Indeed, females were disadvantaged by captivity with significantly smaller

mandibles, mechanical advantage of the temporalis, and relative bite strength

compared to wild females. Males, however, were not significantly impacted by

captivity for measures of skull size and biomechanics and captive males had

significantly larger maximum zygomatic arch breadth than wild males. These

differences in skull morphology and feeding biomechanics have implications for

captive management of mammals as well as for interpretations of data obtained from

captive populations. By examining differences in skull shape, skull size and feeding

biomechanics between captive and wild animals, I demonstrated environmental

effects on adult form, and such differences in form have implications for function.

Thus, environmental variation can lead to variation in morphology, which may have

important consequences for function in ways that influence fitness.

Collectively, the chapters in this dissertation provide a great deal of insight

into developmental relationships between morphology and behavior in Canis . Our results suggest that development is a prolonged process in the generalist coyote, although not as protracted as in the highly specialized spotted hyena. Further, our findings demonstrate that there are developmental constraints in coyotes, which, as in hyenas, affect their performance capabilities. Juvenile coyotes are handicapped in their feeding performance long after weaning occurs, although this duration is shorter than that for the more specialized hyenas. Further, our results support the idea that developmental processes exhibit phenotypic plasticity, being sensitive to

11 environmental effects. This results in variation in ontogenetic outcomes and has implications for both adult form and function. However, additional studies on carnivores are still needed to provide a richer context for interpreting the causes, consequences and functional bases for differences between morphological and behavioral ontogenetic patterns and for interpreting the relationships between life histories and the maturation of morphology and behavior. Further, more work is clearly needed to delineate the effects of diet and of captive environments on skull shape among other carnivores.

12 Chapter TWO

La Croix, S, Holekamp, KE, Shivik, JA, Lundrigan, BL, Zelditch, ML, 2011. Ontogenetic relationships between cranium and mandible in coyotes and hyenas. Journal of Morphology 272(6), in press.

13 Chapter TWO

ONTOGENETIC RELATIONSHIPS BETWEEN CRANIUM AND MANDIBLE IN COYOTES AND HYENAS

INTRODUCTION

The coordination between form and function is of particular interest during ontogeny because each life-history stage represents a temporal ecological niche, and animals must be adapted to that niche to survive to the next stage. Moreover, because developing organisms continually change their morphologies, they must reconcile the conflict between demands of survival within the current life-history stage with those of maturation as they negotiate the transitions through succeeding stages

(Herrel and Gibb, 2006; Monteiro et al., 1999; Wainwright and Reilly, 1994). The ontogenetic relationship between cranium and mandible is of special interest because these two parts must function in concert while both change along steep growth trajectories. Not only does each part exhibit its own ontogenetic change but also their ontogenetic coordination is necessary to ensure continuous functionality.

Understanding the ontogenetic changes in mandibular shape and their integration with the cranium is basic to understanding the function of the developing feeding apparatus especially the way in which young mammals negotiate the challenges posed by competition with adults.

Young mammals do not compete directly with adults for food early in postnatal development but they must often be able to do so after weaning. During their initial postnatal development, the skull undergoes tremendous changes in both size and shape (Fig. 2.1.), and the cranium and mandible must function to satisfy the

14 feeding needs of a juvenile, which are not necessarily those of an adult. To

investigate the ontogeny of the feeding apparatus, and the relationships among its

components, a complete description is needed of both the developing cranium and

mandible. Although there are numerous descriptions of the developing cranium,

there are presently few detailed descriptions of the ontogeny of mandibular shape

(Boughner and Dean, 2008; Cardini and Tongiorgi, 2003; Christiansen, 2008), much

less of the two parts taken together (but see, Cardini and O'Higgins, 2005).

That cranial shape maturity predicts the timing of life-history and

developmental milestones has already been shown for two rodent species with

disparate life-history strategies (Mus and Sigmodon spp., Zelditch et al., 2003). It is

not clear whether cranial shape maturity is equally predictive in other mammalian

orders because only one comparable study has been published (Order Carnivora,

Tanner et al., 2010); patterns of growth and development may be as myriad as life-

history strategies. Further, it is not known whether mandibular shape maturity can

likewise predict life-history scheduling. Understanding the relationship between

development of the feeding apparatus (the skull) and life-history schedules is

important because it has obvious implications for fitness.

We capitalized on a unique opportunity to assemble a large ontogenetic

series of known-age skulls of coyotes ( Canis latrans ) and provide detailed descriptions of the ontogenetic allometry of a non-domesticated canid utilizing shape-based geometric morphometrics. Our investigation considers the ontogeny of the cranium and mandible concurrently, which is vital to understanding their temporal relationship and integration across post-natal development. These

Figure 2.1. Changes in cranium size and shape are visually dramatic among coyotes in successive life history stages.

16 descriptive data serve as a useful contrast with those obtained from more specialized carnivores (e.g., spotted hyena, Tanner et al., 2010), a baseline for studies of variation within the genus Canis (e.g., Kieser and Groeneveld, 1992;

Wayne, 1986), and a groundwork for future studies of the ontogeny of coyote feeding performance (see Ch. 3).

Here, we first describe the ontogenetic trajectories for age-related change in shape of the coyote cranium and mandible across the entire lifespan. Next, we investigate whether these ontogenetic trajectories are simple or composed of multi- phase stages, and describe any temporally restricted shape change. Non-linear growth models are fitted to our data in order to describe cranial and mandibular growth (i.e., changes in size over ontogeny) and development (i.e., changes in shape over ontogeny), examine sexual dimorphism, and assess growth and development within the context of the coyote’s life history. We then describe the ontogenetic patterns and temporal coincidence of skull growth and development including their synchrony, or lack thereof, and implications for functionality of the feeding apparatus. Finally, we contrast these patterns of development in coyotes with those in a highly specialized carnivore, the spotted hyena ( Crocuta crocuta ), to uncover more generalized patterns of skull development and examine their relationship to life-history events.

17 MATERIALS AND METHODS

Coyote life-history stages

The annual cycle of coyote life-history events begins with mating during January,

February and March (Bekoff, 1977). Following a 63 day gestation, pups are born beginning in March (Bekoff, 1977). Weaning of pups begins by four weeks of age and is usually completed by six weeks of age (F. Knowlton, personal communication) (5-7 weeks in Bekoff, 1977; Snow, 1967). The onset of deciduous tooth replacement occurs by 12 weeks of age, and eruption of the adult dentition is complete by 26 weeks of age (Bekoff, 1977; SLC, personal observation). Pups are initially provisioned by both parents, and during the summer the juveniles develop greater independence and self-reliance (Harrison and Gilbert, 1985; Harrison et al.,

1991). Subadults of either sex may emigrate in the fall of the year; first breeding in non-persecuted populations typically occurs at 22 months of age although both sexes are physiologically capable of breeding at 10 months of age (Bekoff, 1977;

Bekoff and Gese, 2003).

Here, we divide the coyote’s life history into five stages (Table 2.1.) based on documented life-history milestones. The pre-weaning stage lasts from birth until weaning, at six weeks of age (Bekoff, 1977). The post-weaning stage lasts from weaning until the onset of deciduous tooth replacement, at approximately 12 weeks of age (Bekoff, 2001). The juvenile stage lasts from the onset of deciduous tooth replacement until eruption of adult dentition is complete, which occurs by 26 weeks of age (SLC, personal observation). The subadult stage starts with the complete eruption of adult dentition at 26 weeks of age and ends with first parturition,

Life history Developmental Age Sample size stage milestones (weeks) V L M Pre-weaning Birth to weaning 0.1 to 6 34 33 25

Post-weaning Weaning to onset of deciduous 6 to 12 30 30 26 tooth replacement Juvenile Onset to complete tooth 12 to 26 27 28 27 replacement & adult tooth eruption Subadult Completion of tooth replacement 26 to 93 25 26 27 & eruption to first parturition

Adult First parturition to death 93 to 694 26 26 25

19 at approximately 93 weeks of age (Bekoff and Gese, 2003). The Adult stage extends from first parturition to death, which may occur at 13 or more years.

Among canids, coyotes are usually categorized as omnivorous generalists,

with young coyotes capitalizing on that portion of the adult diet that is most easily

obtained or subdued (Andelt et al., 1987; Arjo et al., 2002; Bowen, 1978; Clark,

1972; Gese et al., 1996; Hernandez et al., 2002; Hidalgo-Mihart et al., 2001;

Johnson, 1978). Following weaning, juvenile coyotes are initially provisioned by

their parents; subsequently, they ingest mostly small mammals, vegetation,

invertebrates, and birds whereas adults consume larger mammals, fewer

invertebrates and few birds (Hawthorne, 1970). Fruit and insect ingestion by

coyotes occurs most frequently in the summer, which corresponds temporally with a

coyote pup’s increasing independence (Young et al., 2006).

Specimens

An ontogenetic series of 187 coyote skulls (102 males and 85 females), all of known

age, was used in this study; sample sizes are shown in Table 2.1. Individuals

ranged in age from one day to 13.3 years, including 79 specimens under the age of

six months. All specimens were prepared by, and catalogued into either the

Michigan State University Museum collection, or the United States Department of

Agriculture/APHIS/Wildlife Services National Wildlife Research Center’s Logan Field

Station collection in Millville, UT (Appendix Table A.1.).

Specimens were collected incidentally from a captive colony of coyotes

comprised of animals of wild-caught and colony parentage and maintained at the

Logan Field Station between 1979 and 2006. Within the ontogenetic series, 179

20 specimens were captive born, including all specimens under the age of one year, while eight were wild born. All animals were captive reared, and although the current study did not investigate whether the observed ontogenetic patterns are equivalent to those found in the wild, rearing conditions at the Logan Field Station were designed to be as natural as possible. All animals were maintained on a commercially produced wet food diet designed for fur bearing animals and containing chicken (beaks, feet, and feathers) and grain. USDA/APHIS/WS/ NWRC

IACUC approved the study protocol QA-1179.

Exact dates of death were known for all individuals and exact dates of birth were known for all captive-born animals. For animals obtained from the wild as pups, dates of birth were estimated by the original collectors based on deciduous tooth eruption and den observations – those estimated dates of birth were used here. Collector and animal records for all specimens are maintained at the Logan

Field Station.

Ontogenetic shape change

Ontogenetic changes in cranial and mandibular growth and development were analyzed by geometric morphometrics (Rohlf and Slice, 1990; Zelditch et al., 2004).

Each cranium was photographed in ventral and lateral views. In ventral view, the skull was oriented with the palate parallel to the photographic plane, and in lateral view, the skull was oriented with the mid-sagittal plane parallel to the photographic plane. Mandibles were photographed in lateral view, oriented with the longest axis of the mandible parallel to the photographic plane; only mandibles with intact symphyses were photographed. To assess measurement error, an initial sample of

21 121 crania was photographed in ventral and lateral views at three different times by the same observer. Measurement error was determined to be negligible, accounting for only 1.2% of the variation in the sample. The remaining 66 specimens were thus photographed and digitized only once; mandibles were photographed and digitized only once. Twenty-seven landmarks visible on photographs of the ventral cranium provide the data for the analysis of shape in that view (Fig. 2.a.). Landmarks alone cannot fully capture the dorsal curve of the lateral cranium or the mandibular ramus so these were analyzed using a combination of landmarks and semi-landmarks.

Semi-landmarks, unlike landmarks, are not discrete anatomical loci that can be recognized as homologous points, and they contain less information than landmarks because their spacing along the curve is arbitrary. However, semi-landmarks make it possible to study complex curving morphologies where landmarks are sparse.

Fourteen landmarks and 32 semi-landmarks were selected for the lateral cranium

(Fig. 2.b.), and 11 landmarks and 75 semi-landmarks were selected for the mandible

(Fig. 2.c.). Descriptions of these landmarks are contained in Appendix Table A.2.

Landmarks and semi-landmarks were digitized using tpsDig2.10 (Rohlf, 2005). For semi-landmarks, the curve-tracing function was applied and semi-landmarks were evenly spaced along the curves using the ‘‘resample’’ function of the curve-tracing tool.

22 Figure 2.2. Landmarks (numbered) and semi-landmarks ( o) shown on the cranium (a. ventral, b. lateral) and mandible (c.) of an 18 month old coyote, Canis latrans . Descriptions of each landmark and semi-landmark are provided in Appendix Table A.2.2.

23 to minimize the Procrustes distance from the mean shape; we use this criterion for

sliding semi-landmarks because the Procrustes distance is the metric underlying the

general theory of shape. According to this method, the tangent to the curve at each

semi-landmark is estimated and then each semi-landmark is slid toward the normal

of its respective tangent, minimizing the overall difference from the reference

(Andresen et al., 2000; Bookstein et al., 2002; Sampson et al., 1996). Following

superimposition, semi-landmarks can be used in conventional shape analyses

provided that statistical tests take into account that they have only one degree of

freedom. For the ventral cranium, bilaterally homologous landmarks were reflected

and averaged after the Procrustes superimposition because bilaterally homologous

landmarks are not independent of each other; to ease interpretation of the visual

results, they are shown as whole skulls. To quantify skull size, we used centroid

size, the square root of the summed squared distances from each landmark to the

geometric center of the object. Superimposition of landmarks was done using

CoordGen6h (Sheets, 2009); semi-landmarks were superimposed in SemiLand

(Sheets, 2003). Reflection and averaging of bilateral landmarks was done in Sage

(Marquez, 2007).

To analyze ontogenetic change in shape across the entire lifespan as well as

within single life-history stages, shape was regressed on age; age was log-

transformed because most shape change occurs early in ontogeny. The relationship

between shape and age was tested for statistical significance using a generalized

Goodall’s F test (Rohlf, 2007; Sheets, 2003), which measures the ratio of explained to unexplained variation in units of Procrustes distance. Adult coyotes are known to

24 be sexually dimorphic in size; adult females are up to 20% smaller than males

(Bekoff, 1977; Bekoff and Gese, 2003). We assessed whether the ontogeny of

shape is also sexually dimorphic using MANCOVA, with sex as the main factor and

age as the covariate. Because sexual shape dimorphism was not statistically

significant in any view, we pooled males and females in the analyses of shape.

Results are depicted using the thin-plate spline function (Bookstein, 1991).

Using linear regression to analyze the whole of ontogeny presumes that the

same shape changes occur over all ontogenetic phases. To test the hypothesis that

ontogeny is more complex than that, we subdivided our sample into five life-history

stages (Table 2.1.), and compared the ontogenetic trajectories of successive stages

(so long as statistically significant age-related shape change was present within both

single stages). Comparisons were done by measuring the angle between the

vectors of ontogenetic change; should both point in the same direction, the angle

between them is 0 °. The statistical analysis tested the null hypothesis that the angle between the vectors was no greater than expected by chance. That was done by comparing the angle between successive life-history stages to the range of angles that could be obtained by resampling with replacement (400 times), as this is a bootstrap procedure, within a single stage. Should the angle between vectors exceed the 95% confidence intervals for both stages, the difference was judged statistically significant; such a difference does not imply that ontogeny changed direction at the boundary between stages. Regression was done in Regress6

(Sheets, 2003); comparisons between vectors were done in VecCompare (Sheets,

2003).

25 Rates and timing of growth and maturity

To determine the age at which skull growth and development attained adult values,

we used a series of non-linear growth models to find the one that best fit the data for

size and shape. Eight models, Chapman-Richards, Logistic, Monomolecular,

Gompertz, German Gompertz, Von Bertalanffy, Quadratic, and Linear were fitted to

the skull size and shape data and assessed for their relative goodness-of-fit using

Aikaike Information Criterion (AIC) (following Zelditch et al., 2003); the model with

the lowest AIC value was judged to be the best so long as the residuals were not

serially autocorrelated. Several models were excluded because they induced

autocorrelations among residuals in one or more of the analyses (Appendix A.3. and

A.4.). Of those that remained, several fit equally well. We chose the logistic model

(following Gaillard et al., 1997) because this model not only fit the data well in all

views, but it was the only model that did not induce autocorrelations among

residuals in any of the analyses. This model was the basis for estimating rates and

timing of growth and development:

k(To-t) x(t) = A/{1+e }

Where x(t) is the measurement of interest at time t, A is the asymptotic maturity, k is the rate of growth (rate of approach to asymptotic value), and T0 is the age at the curve inflexion point (where growth has attained 50% of asymptotic value)

(Gaillard et al., 1997; Zelditch et al., 2003). We report age at maturity as the estimated age at which the measurement of interest reaches 95% of asymptotic

(adult) value. Data for individuals above the 95% breakpoint were subsequently regressed on age to ensure that age had no further significant impact on the

26 measurement of interest. Evaluation of growth models and estimation of parameters

were performed using GrowChoice (Sheets, 2003).

To analyze rates and timing of growth, we used centroid size as our measure

of size; analyses of rates and timing of development were done using the Procrustes

distance between a particular specimen and the average for the youngest

specimens (Zelditch et al., 2003). We used this measure because it does not

require that species (or ontogenetic stages) have the identical trajectory for shape,

unlike a more conventional procedure which compares rates and timings along a

single ontogenetic trajectory (e.g., Alberch et al., 1979), nor does it divide the

inherently multidimensional data of shape into a collection of one-dimensional

parameters analyzed one at a time (e.g., McKinney and McNamara, 1991). Instead,

the rate of shape change is estimated by the distance traveled along a species-

specific shape trajectory relative to time (Gould, 1977; Tanner et al., 2010; Zelditch et al., 2003). A large Procrustes distance for a particular individual could mean that

the individual is oddly shaped rather than mature, but we found no major outliers in

our sample.

27 RESULTS

Changes in shape across the entire life span

Statistically significant and visually dramatic ontogenetic changes in skull shape

occur between 1 day and 13 years of age in all three views (Fig. 2.3.). For all views,

the null hypothesis of isometric growth can be rejected (p < 0.01). In ventral view,

the dominant feature of shape change is the relative narrowing of the skull along the

anteroposterior axis and lengthening of the rostrum (Fig. 2.3.a.). In lateral view, the

cranial profile becomes less rounded and the nuchal and occipital crests expand

posterodorsally (Fig. 2.3.b.). The dominant features of mandible ontogeny are the

relative expansion of the angular process and a marked anterodorsal reorientation of

the coronoid process, resulting in a taller and more vertically oriented process in

adults than neonates (Fig. 2.3.c.).

Changes in shape within life-history stages

Shape changes substantially throughout pre-weaning, post-weaning, and juvenile

stages for both cranial views; the null hypothesis of isometry can be rejected for

those stages (p < 0.01). Isometry can also be rejected for the adult lateral view (p =

0.011), but not for the adult ventral view (p = 0.060) or the subadult cranial views (p

> 0.05). For the mandible, age-related changes in shape are statistically significant

in all but the adult life-history stage (p < .001). Comparisons between successive

stages that exhibit statistically significant changes in shape reveal the complexity of

ontogeny (Table 2.2.). In all cases, the differences between stages are statistically

significant and in some the angles are exceptionally large, especially in the case of

the mandible, where the angles for pre- vs. post-weaning, and juvenile vs.

29 subadult, exceed 90 °. It is clear from this that the dominant features of a complete ontogeny (as described above) cannot capture the ontogenetic dynamics of shape; we therefore describe the changes occurring within each stage.

During the pre-weaning stage (Fig 2.4.a., Fig. 2.5.a.), nearly half of the variance in cranial shape is related to age (ventral view: 46.6%, lateral view: 46.1%).

In this first stage, the cranium changes shape dramatically as the initially bulbous neonate skull undergoes a tremendous elongation of rostrum and palate, and the basicranium narrows relative to its length and diminishes in relative height. Further, the frontal and parietal bones align dorsally, creating a continuous dorsal curve, visible in the lateral view of the cranium. During the subsequent post-weaning stage

(Fig. 2.4.b., Fig. 2.5.b.), when feeding is accomplished with deciduous teeth, shape continues to change, but age accounts for a much lower proportion of the variance for ventral (30.1%) than for lateral view (51.9%). Rostral elongation continues, as does relative narrowing of the basicranium and flattening of the brain case, all of which further reduce the infant skull’s bulbosity. Specific to the post-weaning stage, the basicranium lengthens relative to skull length as evidenced by the posterior displacement of the landmarks for the jugular foramen and the posterior point of the foramen magnum. In the next interval, the juvenile stage (Fig. 2.4.c., Fig. 2.5.c.), the deciduous teeth are replaced by adult dentition. The proportion of ventral cranial shape accounted for by age here (29.1%) is similar to that seen in the previous stage, but a greater proportion of the variance is accounted for by age in lateral view

(63%). During this stage, the general pattern of cranial elongation and dorso-

Age-related change in shape Skull view Life history stages compared Between Younger Older (in degrees) (in degrees) (in degrees) Ventral Pre- vs. post-weaning 34.6 15.5 20.9 Post-weaning vs. juvenile 40.4 26.6 25.4 Lateral Pre- vs. post-weaning 46.7 20.3 14.3 Post-weaning vs. juvenile 26.3 16.9 12.1 Mandible Pre- vs. post-weaning 110.7 57.9 59.0 Post-weaning vs. juvenile 73.9 52.8 46.7 Juvenile vs. subadult 97.1 45.3 53.7

Figure 2.5. Lateral cranium (a-e) and mandible (f-i) views of the linear regression of shape on log(age) for each life history stage. Vectors and notation are as in Fig. 2.4.

33 ventral flattening continues, but now there is also expansion of the zygomatic arches

(evident from the posterolateral displacement of the landmark at the jugal/squamosal suture), as well as enlargement of the sagittal crest, which lengthens and broadens, creating an increased area for origin of the temporal muscles. No statistically significant changes in cranial shape were detected during the subadult stage.

However, in the adult stage (Fig. 2.4.e., Fig. 2.5.e.), cranial shape in the lateral view changed significantly, with age accounting for 9.5% of the variance while cranial shape in the ventral view was nearly significant (p = 0.060). Over this interval, shape change is mostly subtle except for a distinct widening of the zygomatic arches.

Age consistently explains little of the variance in mandibular shape (9.4%,

10.8%, 16.3%, and 11.8% for pre-weaning, post-weaning, juvenile, and subadult, respectively). During the pre-weaning stage (Fig. 2.5.f.), the anterior portion of the mandible body elongates and deepens and the vertical ramus dramatically broadens. In the subsequent, post-weaning stage (Fig. 2.5.g.), the mandible body and ramus (anterior to the condyloid process) broaden and deepen and the condyloid process expands posteriorly. During the next, juvenile stage (Fig. 2.5.h.), the coronoid process lengthens and broadens, while the vertical ramus elongates, becoming more perpendicular in orientation relative to the base of the ramus. During the subadult stage (Fig. 2.5.i.), there is an overall flattening and reduction in ventral curvature along the body of the ramus, and ongoing growth at the extremes of the coronoid and condyloid processes. And as noted above, no statistically significant changes in shape are detected during the adult stage for the mandible.

34 Rates and timing of growth and maturity

The best-fitting model for both growth and development of the cranium and mandible is logistic growth (Table 2.3.; see also, Appendix Tables A.2.3. and A.2.4.).

Asymptotic adult size ( A) differs significantly between male and female coyotes

(Table 2.4.), so each sex was analyzed separately when estimating the age at which size and shape reached adult maturity (95% of the asymptotic values). Among males, for both ventral and lateral views of the cranium, adult size is reached at 22.5 weeks, and for the mandible at 23 weeks. Females reach adult size within one week of males for all views of the skull (22 weeks). The age at which shape reaches adult maturity is not as consistent among views (Table 2.5.). For the ventral skull, adult shape is reached at 17.2 weeks in males and 18 weeks in females, but the lateral skull takes three to four weeks longer, maturing by 21 weeks in both sexes. The mandible takes even longer, reaching adult shape at 28.2 weeks in both sexes.

Placed in the context of milestones regularly recorded in mammalian life- history studies, the cranium and mandible follow similar growth patterns (Table 2.6.,

Fig. 2.6.a.). Degree of maturation is similar at birth, and all views attain over half of their adult size by weaning (ventral view, 55% for males, 56% for females; lateral view, 60% for both sexes; mandible, 51% for males, 52% for females). By the onset of tooth replacement at 12 weeks, all have attained approximately 80% of the adult size (ventral view, 77% for males, 78% for females; lateral view, 79% for males, 80% for females; mandible, 74% for males, 76% for females). The crania and mandibles reach maturity up to three weeks before the adult dentition completes eruption at 26 weeks (Fig. 2.7.).

Female Male Best-Fit AIC AIC AC % Var AC % Var Model Weight Weight Size: Ventral Cranium Logistic 0.2186 ns 0.986 0.2307 ns 0.989 Lateral Cranium Logistic 0.1943 ns 0.969 0.1942 ns 0.978 Lateral Mandible Logistic 0.2197 ns 0.985 0.2363 ns 0.989 Shape: Ventral Cranium Logistic 0.2208 ns 0.974 0.2132 ns 0.973 Lateral Cranium Logistic 0.2172 ns 0.947 0.2379 ns 0.960 Lateral Mandible Logistic 0.1475 ns 0.750 0.1446 ns 0.690

Ventral Cranium Lateral Cranium Lateral Mandible Sex A k To M A k To M A k To M a b c d ♀ 273.095 0.168 4.501 22.0 559.760 0.159 3.249 22.0 642.526 0.182 5.525 22.0 (270.080- ( 0.155- (4.177 - (554.843- (0.141- (2.747 - (635.530- (0.163 - (5.207-

276.285) 0.184 ) 4.780) 564.564) 0.181) 3.690) 651.312) 0.205) 5.899) a b c d ♂ 286.303 0.167 4.831 22.5 584.289 0.155 3.515 22.5 680.097 0.172 5.925 23.0 (283.636- (.156- (4.594 - (580.070- (0.143- (3.150 - (673.873- (0.160 - (5.660-

289.005) .177) 5.067) 588.179) 0.168) 3.855) 686.512) 0.185) 6.189)

a Asymptotic size is significantly different between male and female ventral crania, t = 6.3767, p = 1.81e-010 b Asymptotic size is significantly different between male and female lateral crania, t = 7.5942, p = 3.0864e-014 c Asymptotic size is significantly different between male and female lateral mandibles, t = 7.5377, p = 4.774e-014 d To for growth is highly suggestive of a trend for a difference between male and female lateral mandibles, t = 1.9323, p = .05332

Table 2.5. Estimates, based on best fitting model, for asymptotic shape, A; growth rate constant, k; age (in weeks) at the curve inflexion point, To; and age (in weeks) at adult maturity, M; 95% confidence intervals given in parentheses.

Ventral Cranium Lateral Cranium Lateral Mandible Sex A k To M A k To M A k To M ♀ 0.124 0.270 6.991 18.0 0.179 0.271 9.671 21.0 0.042 0.118 2.794 28.2 (0.122- (.237- (6.634- (0.175- (0.218- (8.975- (0.040- (0.069- (-1.450-

0.127) .308) 7.378) 0.182) 0.346) 10.426) 0.045) 0.18) 4.950) ♂ 0.122 0.284 6.608 17.2 0.182 0.259 9.319 21.0 0.040 0.112 1.434 28.2 (0.120- (0.256- (6.343- (0.179- (0.225- (8.901- (0.038- (0.074- (-1.946-

0.124) 0.318) 6.899) 0.184) 0.294) 9.792) 0.041) 0.163) 3.728)

Note: No significant difference between males and females for A, k or To for ventral cranium shape, lateral cranium shape or lateral mandible shape.

Ventral Cranium Lateral Cranium Lateral Mandible Age Female Male Female Male Female Male (weeks) CS/A PD/A CS/A PD/A CS/A PD/A CS/A PD/A CS/A PD/A CS/A PD/A Birth 0.319 0.132 0.309 0.133 0.372 0.070 0.367 0.081 0.271 0.420 0.266 0.456 4 0.478 0.309 0.465 0.324 0.529 0.180 0.519 0.199 0.433 0.536 0.418 0.562 8 0.642 0.568 0.629 0.600 0.681 0.390 0.667 0.414 0.611 0.648 0.588 0.662 12 0.779 0.795 0.767 0.825 0.802 0.651 0.788 0.667 0.763 0.746 0.739 0.749 16 0.873 0.920 0.865 0.938 0.885 0.845 0.874 0.851 0.869 0.825 0.849 0.819 20 0.931 0.972 0.926 0.981 0.936 0.942 0.928 0.943 0.932 0.885 0.918 0.871 24 0.964 0.991 0.961 0.996 0.965 0.980 0.960 0.980 0.966 0.926 0.957 0.909 28 0.981 0.998 0.979 0.999 0.981 0.994 0.978 0.994 0.983 0.955 0.978 0.936 32 0.990 0.999 0.989 0.999 0.990 0.999 0.988 0.999 0.992 0.974 0.989 0.954

In striking contrast to their synchronous patterns of growth, cranial and

mandibular shape show very disparate ontogenetic patterns. At birth, cranial shape

is conspicuously immature, while mandibular shape is remarkably mature (Table

2.6., Fig. 2.6.b.). Over the first two life-history stages (pre-weaning and post-

weaning), cranial shape changes rapidly, such that, by the onset of tooth

replacement, the cranium has progressed roughly 80% of the way to its mature

shape for features visible in ventral view (83% for males; 80% for females) and 65%

for features visible in lateral view (67% for males; 65% for females). Mandibular

shape changes much more gradually, and although maturity at the onset of tooth

replacement falls between the two cranial views (75% for both sexes), the mandible

does not reach adult shape until several weeks after the cranium; indeed, maturity of

the mandible is not attained until two weeks after the adult dentition is in place (Fig.

2.7.).

DISCUSSION

At birth, the coyote cranium is small and immature; both size and shape change dramatically over ontogeny. The coyote mandible, like the cranium, is also small at birth. However, unlike the cranium, the shape of the mandible is not immature at birth, and it changes only modestly in shape over ontogeny. In its cranial development, the coyote skull follows the general mammalian pattern

(Moore, 1981): bulbous at birth with subsequent elongation of the cranium relative to its width, flattening of the lateral profile, and broadening of the zygomatic arches.

The ontogeny of the mandible also follows a general mammalian pattern: an

elongation of the horizontal body relative to its height, vertical reorientation and broadening of the upright ramus and coronoid process, and augmentation of the condyloid and angular processes. But the ontogenies of both parts are more complex than suggested by these general patterns; for each part there are features that change during only a specific phase of ontogeny. For example, the basicranium and nuchal crest are displaced posteriorly between weaning and the onset of deciduous tooth replacement (Fig. 2.4.b., Fig. 2.5.b.), whereas the zygomatic arches lengthen, relative to the basicranium, and the sagittal crest takes shape, between the onset of deciduous tooth replacement and the completion of adult dentition eruption (Fig. 2.4.c., Fig. 2.5.c.). For the mandible, there is an elongation and deepening of the anterior-most part of the horizontal mandible body between birth and weaning (Fig. 2.5.f.), whereas there is an elongation and deepening of the posterior parts of the body of the mandible and ramus (anterior to the condyloid process) between weaning and the onset of deciduous tooth replacement (Fig.

2.5.g.). These age-specific patterns are distinctive enough to make ontogenetic trajectories during successive life-history stages significantly different from one another.

The cranium and the mandible are synchronous in their growth, keeping pace with one another as they enlarge, and reaching adult size at the same life-history stage (Fig. 2.6.). The cranium itself is also synchronous with respect to growth and development; i.e., cranial size and shape show similar temporal patterns and reach adult levels at the same life-history stage. In contrast, whereas mandibular growth keeps pace with cranial growth, the developmental patterns evident in mandibular

and cranial shape exhibit striking differences. The mandible is relatively mature in shape at birth, but despite this early maturity, it develops more slowly and attains adult shape later than the cranium (Fig. 2.7.). Any mismatch between the growth and maturation of different parts belonging to a single functional complex can be disruptive unless function itself is also changing. When function is changing, however, constancy of the relationships between parts would not be expected. The functional relationships between cranium and mandible (as well as among regions within each part) do change over postnatal mammalian ontogeny. An obvious example is that suckling and mastication make different demands with respect to occlusion between upper and lower teeth. We might generally expect changes in the relationships among parts unless the demands of adult function are satisfied by the infant morphology. It is, therefore, not surprising that the coyote cranium and mandible each display the complex ontogeny described above, or that the mandible and cranium are asynchronous in development if not in growth.

Adult shape of the coyote feeding apparatus is attained long after the transition to solid food at six weeks of age: cranial shape does not mature for another 11-14 weeks, and mandibular shape, for another 22 weeks. At weaning, cranial shape is as little as 30% mature, whereas the mandible is already 61% mature, having been more mature at birth. Thus, juvenile coyotes are far from mature at weaning, so we would expect that their feeding performance at weaning would be poor relative to that of adults. Further, the ongoing maturation of the skull parts suggests that adult function is not achieved by the morphology characteristic of the post-weaning stage. Synchrony of cranial and mandibular development may be

unnecessary except to the extent that it affects growth and orientation of the masticatory muscles. Bone growth plays a pivotal role in orienting muscles and their action lines, in that regional bone growth differentially stretches the periosteum, with muscles being carried along with it via their periosteal attachments (Carlson, 1983;

Frankenhuis-van den Heuvel et al., 1992; Grimm and Katele, 1979). Additionally, bones must grow to accommodate enlarging muscles, so we might anticipate synchrony in regions where masticatory muscles originate and insert, such as the sagittal crest (visible only in lateral view) and mandible. Indeed, our results show that association between mandibular development and sagittal crest; mandibular development in the subadult stage is marked by localized changes in condylar and coronoid shape and these changes are concurrent with localized cranial changes, particularly in the sagittal crest, that continue through adulthood (Fig. 2.5.). The influence of developing secondary and primary dentition on the synchrony of cranial and mandibular shape development remains unexplored, but tooth development and occlusion might well affect the synchrony of cranial and mandibular development.

It is not presently known whether the asynchrony shown here in coyote cranial and mandibular shape development reflects a more generalized mammalian pattern because only one other study, on the spotted hyena , Crocuta crocuta

(Tanner et al., 2010), has concurrently examined ontogenetic patterns of cranial and mandibular shape change in relation to life history. Hyenas, unlike coyotes, are capable as adults of cracking open large bones using craniodental adaptations for durophagy that include a vaulted forehead, a large sagittal crest, massive zygomatic arches and robust premolars with crack-resistant enamel. Another distinction

between hyenas and coyotes may be related to those morphological demands of durophagy: the hyena’s life history is notable for its protracted period of maternal dependence even in relation to other large carnivores (Watts et al., 2009). Skull ontogeny in the spotted hyena, like that in coyotes, demonstrates relative synchrony in cranial and mandibular growth but asynchrony in cranial and mandibular shape development, cranial size and shape maturation and mandibular size and shape maturation (Tanner et al., 2010). What differs strikingly between coyotes and hyenas are both the sequence of adult size and shape maturation events and their timing within the life history (Fig. 2.8.). In the coyote, mandibular shape matures after cranial shape, and mandibular shape is the last skull measure to mature. By contrast, in the spotted hyena, cranial shape maturation is dramatically delayed, occurring long after mandibular shape maturation. In addition, all coyote skull maturation is completed before or shortly after the eruption of the adult dentition and long before reproductive maturity, but in the spotted hyena, skull maturation (except for cranial shape in the lateral view) occurs at or after reproductive maturity. Like the coyote skull, the spotted hyena skull matures long after weaning, and despite a late age-at-weaning in spotted hyenas (compared to other carnivores), the skull of a just- weaned spotted hyena is still notably immature (ventral, 73.5% mature; lateral,

90.3%; mandible, 83.5%, Tanner, unpublished data). Delayed skull maturation in the spotted hyena, especially late cranial maturation and delayed feeding performance maxima, may be explained as costs of the adult ability to crack open large bones.

Indeed, the dominant feature of post-weaning shape change in a spotted hyena skull is the development of the bony areas of muscle insertion: zygomatic arches expand

and the sagittal and nuchal crests develop (Tanner et al., 2010). For coyotes and

other mammalian carnivores lacking such specialized adult function, we might thus

expect a more modest delay in skull maturation and achievement of feeding

performance maxima than is seen in spotted hyenas. We have demonstrated the

former here; future work is needed to determine whether a comparable delay in

maturation also exists regarding achievement of feeding performance maxima, and

whether maturation schedules influence dietary choices.

Overall, comparison of ontogenetic patterns in coyotes and spotted hyenas

reveals that both species share a synchronous pattern of maturation in growth

between cranium and mandible, but temporally mismatched patterns of maturation in

shape between cranium and mandible. That morphological development does not

predict life-history events in these two carnivores distinguishes them from the two

rodents studied to date; in those two rodents, one precocial, the cotton rat

(Sigmodon fulviventer ), and the other altricial, the house mouse ( Mus musculus domesticus ), cranial shape maturity predicted life-history scheduling (Zelditch et al.,

2003). In contrast, coyote skull maturity occurs well before sexual maturity, and

spotted hyena skull maturity occurs well after.

By looking at growth and development in the cranium and mandible

concurrently, examining the relationship between their maturation patterns, and

placing both within the context of life history, we can identify a synchronous pattern

of maturation in growth across parts as well as a temporally mismatched pattern of

shape maturation. It is likely that these patterns are complicated by the interaction

between localized features in one part of the skull with those of another (such as the

cranial origins and mandibular insertions of masticatory muscles). Clearly, additional studies on carnivores are needed to provide a richer context for interpreting the causes, consequences and functional bases for differences between cranial and mandibular ontogenetic patterns. Of particular importance for understanding the links between morphology, feeding performance, and life history is the protracted period of cranial development in carnivores, which makes juveniles handicapped relative to adults for a significant portion of their life history. Our findings demonstrate that an adult’s functional demands cannot be satisfied by the morphology characterizing earlier life-history stages, and this has significant consequences for life-history evolution whenever juveniles must compete with adults.

Chapter THREE

La Croix, S, Zelditch, ML, Shivik, JA, Lundrigan, BL, Holekamp, KE. Ontogeny of feeding performance and biomechanics in coyotes and hyenas. Journal of Zoology, London submitted.

Chapter THREE

ONTOGENY OF FEEDING PERFORMANCE AND BIOMECHANICS IN COYOTES AND HYENAS

INTRODUCTION

During ontogeny, each life-history stage represents a temporal ecological niche to which animals must be adapted in order to survive to the next stage. Because developing organisms are continually changing their morphologies, they must reconcile the conflict between demands of survival within the current life-history stage, with those of maturation, as they negotiate the transitions through succeeding stages. Understanding how developing animals resolve these time critical challenges of growth and survival can illuminate the relationship between form and function. Behavior and morphology may be subject to developmental constraints that can yield a form that is sub-optimal for solving the problems confronted by the organism.

During early postnatal development, the young mammal’s feeding apparatus undergoes tremendous changes in both size and shape, and the cranium and mandible must function to satisfy the feeding needs of a juvenile, even as it develops toward an adult form (Ch. 2, La Croix et al., 2011) (Fig. 3.1.). In some species, ontogenetic changes in the skull yield juveniles and subadults with bite strength ability that is considerably less than that of adults (Binder and Van Valkenburgh,

2000; Christiansen and Adolfsen, 2005; Erickson et al., 2003a). Even though young mammals enjoy a brief period of nutritional dependence on the parents, at weaning, juveniles must be capable of processing adult foods or adopt a different diet

Figure 3.1. Ontogenetic variation in cranial and mandibular size and shape among coyotes ( Canis latrans ) aged, from top, 1 day, 6 weeks, 12 weeks, and 26 weeks is visually dramatic.

(Herrel and Gibb, 2006; Monteiro et al., 1999; Wainwright and Reilly, 1994).

Carnivores, especially, face tremendous challenges during ontogeny as they attempt to provision themselves with food and compete with adults despite their immature morphology. Small changes in morphology can have profound effects on an animal’s functional capabilities (Koehl, 1996). The period of deciduous tooth eruption and replacement presents a challenge to early food processing that is unique to young mammals. It has been suggested that the mechanical nature of deciduous teeth (e.g. greater brittleness and sharpness) may somewhat offset the mechanical disadvantage that juveniles encounter with their smaller muscle mass

(Binder and Van Valkenburgh, 2000). Even so, these teeth, which are sized to fit into the much smaller pre-weaning feeding apparatus, undergo replacement because they no longer meet the functional demands imposed by the succeeding life-history stage.

Recent work on the spotted hyena (Tanner et al., 2010) showed that juveniles and subadults in this durophagous species remain handicapped for an extended period of time after weaning due to protracted development of a robust feeding apparatus capable of cracking open large bones. It remains unclear to what extent, and for how long, the developing feeding apparatus handicaps juveniles in other, less specialized, carnivore species, when they compete with adults. In contrast to spotted hyenas, coyotes show no specialized adaptations for durophagy and they consume relatively little bone in their diet (Andelt et al., 1987; Arjo et al., 2002). With the behavioral plasticity to consume small and large prey, live in pairs or packs, and colonize rural and urban environments, the coyote’s diet and lifestyle are

opportunistic and generalist (Bekoff, 1977; Bekoff and Gese, 2003). Their generalized carnivore morphology makes them appropriate for comparisons with more highly specialized carnivores, such as hyenas.

To investigate the extent to which coyotes might be handicapped as juveniles, we took advantage of a unique opportunity to assess the ontogeny of feeding performance in known-age coyotes, and examine feeding biomechanics with an ontogenetic series of known-age coyote skulls. Both samples were drawn from the same population. Here, we describe the ontogeny and maturation of feeding performance, tooth eruption and replacement schedules, and measures of feeding biomechanics in coyotes. We fit non-linear growth models to our data to describe development of feeding performance and biomechanics over the course of ontogeny, examine sexual dimorphism, and assess development within the context of the coyote’s life history. We then describe the ontogenetic patterns and temporal coincidence of feeding performance and biomechanics including their synchrony, or lack thereof, and implications for functionality of the feeding apparatus. Finally, we contrast these patterns of development in coyotes with those in spotted hyenas, to uncover more generalized patterns of development and examine their relationship to life-history events. Understanding relationships among biomechanics, life-history schedules, and ontogeny of feeding performance is important because these relationships have obvious implications for fitness.

METHODS

Subjects and specimens

All data for this study were collected from known-age individuals housed at the

USDA Wildlife Services National Wildlife Research Center’s Logan Field Station in

Millville, UT.

Feeding performance trials were conducted with coyotes from a captive colony comprised of animals of wild-caught and colony parentage, maintained at the

Logan Field Station between 2004 and 2006. A total of 44 pups (26 males, 18 females) from 11 litters participated in the trials; three of the litters (20 animals) were wild-born in 2004, while eight of the litters (24 animals) were captive-born in 2005

(Appendix Table A.3.1.). All animals were captive reared, and although the current study did not investigate whether the observed ontogenetic patterns are equivalent to those found in the wild, rearing conditions at the Logan Field Station were designed to be as natural as possible. All animals were maintained on a commercially produced wet food diet designed for fur bearing animals which contained chicken (beaks, feet, and feathers) and grain. Prior to weaning, pups also received a milk replacement powder commercially produced for puppies and reconstituted with water; during weaning, pups received a pea-sized kibble version of the wet food diet, moistened with water. Pups younger than six weeks were maintained in litter groups. Following weaning at approximately six weeks, pups were housed in either small groups of littermates (two or three individuals) or singly, until four months, when all animals were subsequently housed singly. Captive-born pups were initially housed indoors in 0.6m x 1.2m wire dog kennels with plastic floor

pans while wild-born pups were maintained in a wire enclosure in an open pole-barn

with gravel flooring. Following weaning, animals were moved to 3.7 m x 1.2 m

2 chain-link fenced enclosures with a concrete floor and containing a 0.6m round plastic den box within the pole barn. Animals had no access to bones, other foods, or chew toys except during performance trials.

Feeding biomechanics were investigated using an ontogenetic series of 187 coyote skulls (102 males and 85 females), all of known age (see Ch. 2), also obtained from the Logan Field Station. Individuals ranged in age from one day to

13.3 years, including 79 specimens under the age of 6 months. Specimens were collected incidentally from captive animals of wild-caught and colony parentage that were maintained at the Logan Field Station between 1979 and 2006. All skulls were catalogued into either the Michigan State University Museum collection, or the

United States Department of Agriculture/APHIS/Wildlife Services National Wildlife

Research Center’s Logan Field Station collection in Millville, UT (Appendix Table

A.3.2.). Within the ontogenetic series, 179 specimens were captive-born, including all specimens under the age of one year, while eight were wild-born. Exact dates of death were known for all individuals and exact dates of birth were known for all captive-born animals. For animals obtained from the wild as pups, dates of birth were estimated by the original collectors based on deciduous tooth eruption and den observations, and those estimated dates of birth were used here. Collector and animal records for all specimens are maintained at the Logan Field Station.

USDA/APHIS/WS/NWRC IACUC approved the study protocol QA-1179.

Feeding performance

Feeding performance trials were conducted from 2004 to 2006 using repeated measures on 44 coyotes between the ages of six weeks and 80.3 weeks. Pups born in 2004 were tested at 14 week intervals; the testing interval was later adjusted and pups born in 2005 were tested every eight weeks, in order to document performance at as many ages as possible. Feeding performance trials were carried out in home enclosures 18-24 hours after the animals’ last feedings. On the day of testing, animals that did not consume a pre-test biscuit (a 4g, Iams brand puppy dog biscuit) within five minutes of delivery, with the Tester (SLC) present in the kennel block, were excused from the test. Qualifying animals were then presented with a standardized food item, a 32g, Iams brand, adult large dog biscuit. Three trials were conducted serially, with at least a 60 second delay between finishing ingestion of one biscuit and delivery of the next; animals that failed to consume an entire biscuit within 10 minutes were excused from further testing for that day. Each trial was videotaped in natural light using a Sony Handycam Vision CCD-TRV65 NTSC –

VideoHi8TR Steady Shot with 72X digital zoom. The video camera was mounted on a tripod outside the subject’s kennel with a viewpoint 36” above the ground. The time (in seconds) required to consume the biscuit was later calculated from these videotapes. Our measure of feeding performance was consumption time, defined as the sum of the periods of continuous and sustained chewing, including breaking up of the biscuit into pieces that could be swallowed. Timing began with the first audible crunch of the biscuit into more than one piece and concluded when the biscuit was completely consumed. Only the best consumption time from each

animal’s three consecutive biscuit trials was used in feeding performance analyses, provided that best time was for a biscuit completely processed within the 10 minute trial period; two females, aged 20 weeks, and one female, aged 28 weeks, were excluded from the analysis as they failed to completely consume a biscuit within 600 seconds. All six-week old pups, which were unable to bite even the 4g test biscuit into pieces, were assigned a best consumption time of 600 seconds for the 32g biscuit to facilitate the development of growth curves for feeding performance.

Direct Measurements of Bite Strength

We obtained direct measurements of bite strength using a bite force transducer

(Kistler, Amherst, NY) from 12 animals at Logan Field Station (see Appendix Table

A.3.3.). The transducer was baited with meat and hand-held by the investigator

(SLC). Direct measurement of bite strength required an animal’s willingness to approach the investigator at the kennel fence and voluntarily bite down on the transducer. When a coyote voluntarily bit down, a piezoelectric plate embedded in the transducer measured the force generated in Newtons, and this was recorded from a handheld charge amplifier. Similar devices have been used with wild hyenas

(Tanner, 2007), captive hyenas (Binder, 1998; Binder and Van Valkenburgh, 2000), captive short-tailed opossums (Thompson et al., 2003b), and with both captive and wild American alligators (Erickson et al., 2003a).

Tooth eruption and replacement

As is characteristic of most mammals, coyotes are diphyodont, meaning they have two sets of teeth during their lifetime. During ontogeny, feeding performance may be affected by the loss of these deciduous teeth and the eruption of adult teeth. Here,

the ontogenetic series of known age skulls was visually examined for tooth eruption

and replacement patterns. A tooth that protruded above its bony socket was

counted as an erupted tooth. In addition, the teeth of 24 coyote pups (from the

feeding performance trials) were examined (by SLC) between the ages of four

weeks and 12 weeks; in these living animals, a tooth that protruded above the gum

was counted as an erupted tooth. To document patterns of tooth eruption and

replacement, we recorded the age at which all specimens or live animals evidenced

eruption of the deciduous or adult tooth of interest.

Mechanical advantage and bite strength

Mammalian mastication, including that of carnivores, has been described by various

authors (Hiiemae, 1985; Langenbach, 2001; Schumacher, 1985; Simpson, 1978;

Turnbull, 1970; Van Valkenburgh, 1989; Weijs, 1994). Carnivore jaw movement is

mostly restricted to a hinge-like action in a single plane, and has been modeled as a

modified Class I lever (Turnbull, 1970). By modeling the jaw as a lever, it becomes

possible to assess feeding ability and to calculate the mechanical advantage of the

feeding apparatus (Greaves, 1983; 1985; Radinsky, 1981; Smith, 1993; Thomason,

1991) . In addition, relative bite strength can be inferred by estimating both the

mechanical advantage of the primary masticatory muscles (here, the temporalis) and

The length of the in-lever arm is measured as the distance between the

muscle insertion point and the mandibular condyle, and the length of the out-lever

arm is measured as the distance between the mandibular condyle and the bite point

on the mandible (Fig. 3.2.). Adductor muscle size is estimated from the maximal

width across the zygomatic arches (Binder, 1998; Gittleman and Van Valkenburgh,

1997; Radinsky, 1981) . Over ontogeny, as the length of the in-lever arm increases

relative to that of the out-lever arm, mechanical advantage increases (Gittleman and

Van Valkenburgh, 1997; Hurov et al., 1988). Here, the length of the in-lever arm for the primary masticatory muscle, the temporalis, was measured as the distance between the dorsal tip of the coronoid process and the mandibular condyle, and the length of the out-lever arm was measured as the distance between the mandibular condyle and the bite point (Fig. 3.2.). The bite point was the highest cusp of the mandibular first molar; this carnassial tooth is the largest tooth in the mouth.

Measurements of the skull were obtained using a digital caliper accurate to

0.01 mm; measurements were taken in triplicate by two different observers, and then

averaged. Following Radinsky (1981), mechanical advantage was calculated as the

in-lever arm length divided by the out-lever arm length. Evaluation of relationships

among the measurements comprising mechanical advantage was accomplished by

regression of the out-lever on the in-lever and by examination of the residuals;

statistical analyses were performed using STATISTICA, version 8.0 (StatSoft, Inc.,

2007, www.statsoft.com). Relative bite strength was calculated by multiplying

mechanical advantage of the temporalis by maximum zygomatic arch breadth (ZAB)

(Fig. 3.2., the widest point on the skull) (Binder, 1998; Hildebrand, 1984; Radinsky,

1981).

Age and timing of maturation

To determine the age at which feeding performance and biomechanical variables

(mechanical advantage, maximum ZAB and relative bite strength) attained adult values, we used a series of non-linear growth models to find the one that best fit the data. Coyotes are known to exhibit sexual size dimorphism as adults (Bekoff, 1977), so we considered male and female data separately when fitting growth curves. Eight models (Chapman-Richards, Logistic, Monomolecular, Gompertz, German

Gompertz, Von Bertalanffy, Quadratic, and Linear) were fitted to the data for each feeding performance and biomechanical variable and assessed for their relative goodness-of-fit using Aikaike Information Criterion (AIC) (following Zelditch et al.,

2003). The model with the lowest AIC value was judged to be the best so long as the residuals were not serially autocorrelated. Some models were excluded because they induced autocorrelations among residuals in one or more of the analyses; where several models fit equally well, the model with the highest AIC weight was selected as best (Appendix Table A.3.4.).

For feeding performance, we chose the Gompertz model as formalized by

Fiorello & German (1997), herein referred to as the German Gompertz model)

(following Zelditch et al., 2003), because this model fit the data well for both sexes and was the only model that did not induce autocorrelations among residuals for this variable (Table 3.1., Appendix Table A.3.4.). This model was the basis for estimating the parameters of feeding performance maturation:

–k e -bt x(t) = Ae

Where x(t) is feeding performance at time t, A is the asymptotic maturity, k is

the initial rate of growth (initial rate of approach to asymptotic value), and b is the

decay of the growth rate (Gaillard et al. 1997, Zelditch et al. 2003).

No growth model could be fit to the data for mechanical advantage of the

temporalis for either sex due to autocorrelations among the residuals.

For maximum ZAB breadth, we chose the monomolecular model (following

Gaillard et al. 1997) because it not only fit the data well for both sexes, but also had

the highest AIC weight of the several models that fit equally well (Table 3.1.,

Appendix Table A.3.4.). This model was the basis for estimating the parameters of

maximum ZAB maturation:

k(To-t) x(t) = A{1-e }

Where x(t) is maximum ZAB at time t, A is the asymptotic maturity, k is the

rate of growth (rate of approach to asymptotic value), and T0 is the age at which growth begins (Gaillard et al., 1997; Zelditch et al., 2003).

For relative bite strength, we chose the logistic model (following Gaillard et al., 1997) because this model fit the data well for both sexes and it had a higher AIC weight than the general Chapman-Richards model (Table 3.1., Appendix Table

A.3.4.). This model was the basis for estimating the parameters of relative bite strength maturation:

k(To-t) x(t) = A/{1+e }

Table 3.1. Best-fitting models for the measures of feeding performance and skull biomechanics maturation. The AIC weight evaluates relative goodness-of-fit by balancing the distance between model and data by degrees of freedom. AC refers to serial autocorrelation among residuals of the model (“ns” indicates there was no statistically significant serial autocorrelation). Percent variance explained (% Var) by the best-fit model is also given. Female Male AIC AIC Best-Fit Model AC % Var AC % Var Weight Weight Feeding Performance German Gompertz 1.000 ns 0.862 1.000 ns 0.918 Biomechanics: Maximum ZAB Monomolecular 0.4486 ns 0.992 0.4619 ns 0.991 Relative Bite Strength Logistic 0.7082 ns 0.978 0.7287 ns 0.976

Where x(t) is the relative bite strength at time t, A is the asymptotic maturity, k

is the rate of growth (rate of approach to asymptotic value), and T0 is the age at the curve inflexion point (where growth has attained 50% of asymptotic value) (Gaillard et al., 1997; Zelditch et al., 2003).

For all variables, we report age at maturity as the estimated age at which the measurement of interest reaches 95% of asymptotic (adult) value. Data for individuals above the 95% breakpoint were subsequently regressed on age to ensure that age had no further significant impact on the measurement of interest.

Evaluation of growth models and estimation of parameters were performed using

GrowChoice (Sheets 2003).

RESULTS

Feeding performance

Six week old coyotes were unable to bite a 32g biscuit into pieces (Fig. 3.3.). After twelve weeks of age, however, animals successfully consumed the biscuit within 600 seconds. As expected, coyote feeding performance improved over ontogeny, as evidenced by reduced biscuit consumption times (Fig. 3.3.). Males showed faster processing times than females, at most ages (Fig. 3.3.).

Measured Bite Strength

Although we recorded 40 bites on the transducer from 12 coyotes, for a number of reasons, we were unable to collect sufficient bite strength data to document early ontogenetic changes in bite strength. First, some animals, especially juveniles, were simply unwilling to approach the transducer in proximity to the investigator. Second,

animals willing to interact with the transducer would frequently remove the meat covering, test the transducer gently with their teeth, and then cease interacting with the device altogether. Further, it was clear that some animals biting the transducer were not biting down as hard as they could; some used their incisors while others used their carnassials. One individual, from which the greatest bite forces were repeatedly measured, did execute highly motivated bouts of tug-o-war with the transducer held by the investigator. Even so, bite strength measures, collected from

12 individuals between the ages of 52 and 414 weeks, showed that older animals exhibited greater bite strength compared to younger animals, and that males bit more forcefully than females of the same age (Fig. 3.4.). A maximum bite strength of 708N was obtained from our oldest subject, a male aged 414 weeks.

Tooth eruption and replacement

An examination of the ontogenetic series of coyote skulls revealed that eruption of the deciduous teeth begins at approximately two weeks of age (Fig. 3.5.). This concurs with existing literature on coyote tooth eruption: upper canines (Day 14), lower canines and upper incisors (Day 14-15), and lower incisors (Day 16) (Bekoff,

2001; Bekoff and Jamieson, 1975). Further, we observed that by weaning at six weeks of age, the coyote pup dental formula is as follows: I3/3 C1/1 P2/2 M1/1 for

28 deciduous teeth total (Fig. 3.5.). The specific teeth present in the six week old pup are dI1, dI2, dI3, dC1, dP2, dP4, and dM1 in the upper jaw and di1, di2, di3, dc1, dp2, dp4, and dm1 in the lower jaw. Note that dP4/dm1 constitute the deciduous carnassial pair. The first permanent teeth, P1/p1, erupt at eight weeks of age and are fully erupted by 12 weeks of age. Replacement of the deciduous

Figure 3.4. Plot of bite strength, measured in Newtons (N), obtained using a piezoelectric transducer, for male and female coyotes.

Figure 3.5. Schedule of tooth eruption and replacement (including post-canine teeth) in coyotes. The presence of each deciduous (D) or Adult (A) tooth is indicated (a. cranium, b. mandible). All coyote skull specimens examined at each age displayed the specified tooth configuration.

incisors begins by 12 weeks, with lower incisors lost first, followed by upper incisors; adult incisors are fully present by 18 weeks of age. Between 14 and 20 weeks, the remaining deciduous teeth, canines, premolars and molars are replaced, and adult teeth, P3/p3 and M2/m2, erupt. By 21 weeks of age, coyotes lack only the adult m3 in the lower jaw; this tooth erupts during the twenty-first week. By 26 weeks of age, coyotes have attained their complete adult dental formula of I3/3, C1/1, P4/4, M2/3 for 42 teeth total.

Mechanical advantage and bite strength

The ontogeny of mechanical advantage of the temporalis did not follow a traditional growth pattern in coyotes, unlike feeding performance and other biomechanical measures. Instead, mechanical advantage data formed two discrete clouds, one containing smaller values that represented young animals, and the other containing larger values representing older animals (Fig. 3.6.a.). The transition between these two clouds of data points occurred in the age range during which the deciduous carnassials and most of the post-canine teeth were being replaced. Indeed, the mechanical advantage in the younger group with deciduous dentition was significantly lower than in the older group of animals with adult dentition (Fig. 3.6.b.).

We found that mechanical advantage for coyote pups aged two to 14 weeks ranged from 0.313 to 0.420, with some two week old pups having a mechanical advantage equivalent to pups that were 10 weeks older. The mechanical advantage for coyotes over the age of 21 weeks ranged from 0.451 to 0.559, with animals aged 21 weeks exhibiting a mechanical advantage equivalent to animals twice that age.

Figure 3.6. Plot of mechanical advantage of the temporalis for coyotes, by age (a.) and comparison of mean mechanical advantage of the temporalis between coyotes with primarily deciduous dentition and those with primarily adult dentition (b.). Only data for coyotes aged less than 100 weeks are shown in (a.), while data for all ages are included in (b.).

Examination of the biomechanical components of mechanical advantage of the temporalis showed that the ontogenies for both the in-lever arm length and the out-lever arm length were characterized by early rapid growth (Fig. 3.7.a.).

However, while out-lever arm length increased more quickly than did in-lever arm length at the earliest ages (Fig. 3.7.a., b.), the latter continued to grow after the length of the out-lever arm stabilized (Fig. 3.7.b., c.). Furthermore, the lever arm lengths for animals under 14 weeks of age came from animals with primarily deciduous dentition, whereas those for older animals came from animals with adult dentition; specifically, the endpoint of the out-lever arm length measure shifted from the deciduous carnassial tooth to the newly erupted adult carnassial tooth. Also, the out-lever arm underwent an unusually large increase in length at 12-14 weeks of age which, in a punctuated increase, brought its length within the range of animals over six weeks older, even though it was measured to the deciduous rather than the adult carnassial.

Early growth in maximum ZAB (Fig. 3.8.a.) and relative bite strength

(Fig. 3.8.b.) exhibited patterns similar to those for feeding performance and lever arm lengths: they underwent rapid early growth before asymptotic maturation was achieved by 30 (maximum ZAB) and 31.5 weeks (relative bite strength) of age.

Maximum ZAB increased especially quickly at the youngest ages.

Ages at maturity

The German Gompertz growth model, the best-fitting model for feeding performance

(Table 3.1., Appendix Table A.3.4.), indicated that adult feeding performance was not attained until 34.6 weeks in males and 36 weeks in females; asymptotic feeding

performance was achieved significantly faster by males than females (p = 0.001)

(Table 3.2.). The monomolecular growth model, the best-fitting model for Maximum

ZAB (Table 3.1., Appendix Table A.3.4.), is typified by extremely rapid growth at the youngest ages. Despite explosive early growth, maximum ZAB did not reach maturity until 26.9 weeks in females and 30.4 weeks in males; asymptotic maximum

ZAB was significantly larger for males (p < 0.001), but growth rate was significantly faster for females (p = 0.006) (Table 3.2.). The model that best fit relative bite strength, the logistic model (Table 3.1., Appendix Table A.3.4.), estimated maturation for relative bite strength at 29.1 weeks for females and 31.5 weeks for males; asymptotic relative bite strength was significantly greater for males than females (p < 0.001). Females were significantly younger than males at the growth curve inflexion point, by which they had reached 50% of their adult relative bite strength (p = 0.005) (Table 3.2.).

Maturation and life history

Feeding performance and biomechanics reach maturity long after weaning at six weeks of age (Table 3.3.); further, during the course of ontogeny, feeding performance maturation lags considerably behind biomechanical maturation (Fig.

3.9.a.), as well as skull size and shape maturation (Fig. 3.9.b., c.). At the age of weaning, feeding performance for males is extremely immature at only 16% of adult values; likewise, for females, it is only 23% of the adult values (Table 3.4.). Six weeks later, at the onset of tooth replacement, feeding performance among females has improved to 42% of the adult values, while males show an even greater improvement to 47% of the adult values. By the time adult dentition is complete at

Table 3.2. Estimates of maturation for: feeding performance, based on the best-fitting German Gompertz model, for asymptotic feeding performance speed (in secs), A; initial growth rate, k; decay of the growth rate, b; and age (in weeks) at adult maturity, M; maximum zygomatic arch breadth (ZAB), based on the best-fitting monomolecular model, for asymptotic size (in mm), A; growth rate constant, k; age (in weeks) at the onset of growth, To; and age (in weeks) at adult maturity, M; and relative bite strength, based on the best-fitting logistic model, for asymptotic value, A; growth rate constant, k; age (in weeks) at the curve inflexion point, To; and age (in weeks) at adult maturity, M. Ninety-five percent confidence intervals are given in parentheses.

Feeding Performance Maximum ZAB Relative Bite Strength Sex A k b M A k To M A k To M a b c d e ♀ 137.998 -3.024 0.118 36.0 86.682 0.096 -4.275 26.9 44.131 0.134 7.312 29.1 (113.404- ( -4.498- (0.067- (85.781- ( 0.091- (-4.679- (43.375- (0.121- (6.666-

160.380) -2.325) 0.197) 87.506) 0.102) -3.907) 44.968) 0.150) 8.003) a b c d e ♂ 94.245 -4.061 0.129 34.6 92.554 0.086 -4.443 30.4 48.303 0.128 8.522 31.5 (80.491- (-5.577- (0.080- (91.606- (0.082- (-4.837- (47.402- (0.117- (7.919-

108.669) -3.104) 0.193) 93.472) 0.091) -4.109) 49.141) 0.138) 9.072)

a Asymptotic feeding performance speed is significantly different between females and males, t = 3.2464, p = 0.001 b Asymptotic maximum zygomatic arch breadth is significantly different between females and males, t = 9.0162, p < 0.001 c Growth rate constant, k, is significantly different between females and males, t = 2.7752, p = 0.006 d Asymptotic relative bite strength is significantly different between females and males, t = 6.5593, p < 0.001 e Age at the curve inflexion point, To is significantly different between females and males, t = 2.796, p = 0.005

Table 3.3. Age at maturation (95% of adult value), as estimated from best-fit models; feeding performance and biomechanics (this study) and skull size and shape (Ch. 2, LaCroix et al., 2011).

AGE AT MATURATION BEST-FIT MODEL (weeks) Female Male

Feeding Performance 36.0 34.6 German Gompertz Biomechanics: Maximum ZAB 26.9 30.4 Monomolecular Relative Bite Strength 29.1 31.5 Logistic Skull Size:

Ventral Cranium 22.0 22.5 Logistic Lateral Cranium 21.8 22.5 Logistic Lateral Mandible 21.7 23.0 Logistic Skull Shape: Ventral Cranium 18.1 17.2 Logistic Lateral Cranium 20.7 20.9 Logistic Lateral Mandible 28.2 28.2 Logistic

Figure 3.9. Timeline illustrating the relative maturity of coyote feeding performance to that of feeding biomechanics (a.), skull size (b.), and skull shape (c.) in relation to major life history events. Relative maturity for biomechanics (this study) and skull morphology (Ch. 2, La Croix et al., 2011) is indicated with a dashed line, and for feeding performance (this study) with a solid line. Maturation of mechanical advantage of the temporalis (a.) does not follow a traditional growth pattern, but is, instead maintained within two discrete ranges throughout ontogeny, one for younger animals and one for older animals (see Fig. 5.a.); here, we have diagrammed the relative maturity of the younger group as calculated by dividing the mean younger group mechanical advantage by that of the older (adult) group (see Table 3.5.).

26 weeks, feeding performance for both sexes has improved to approximately 86% of adult values but does not reach full maturity for another eight to 10 weeks (Fig.

3.9.a., Table 3.3.). Because data for the mechanical advantage of the temporalis did not fit a traditional growth curve, we were unable to model predictions of maturity, as we have for our other measures of feeding biomechanics. Instead, we calculated the relative maturity of mechanical advantage of the temporalis for the younger cloud of data by dividing its mean value by that of the older (adult) cloud of data. Mean value of mechanical advantage of the temporalis for younger females was 0.376, and for older females it was 0.508; males had greater mechanical advantage at both ages than did females, 0.383 among the younger males, and

0.516 among the older males (Table 3.5.). Relative maturity of mechanical advantage of the temporalis for young animals of both sexes is extremely mature at

74% of the adult value, especially in contrast to other biomechanical measures (Fig.

3.9.a.). Maturation of mechanical advantage occurs after deciduous tooth replacement but before eruption of adult dentition is complete (Fig. 3.9.a.) Maximum

ZAB for both sexes is only one third of adult size at birth, but doubles by the age of weaning six weeks later (Table 3.4.). By the onset of deciduous tooth replacement, rapid early growth in maximum ZAB is apparent, as females are already at 79% of the adult values and males are at 76% of the adult values (Fig. 3.9.a.).

Subsequently growth slows, with female coyotes reaching adult maximum ZAB one week after adult dentition is completely erupted; males mature three weeks after females (Table 3.3.). The maturation of relative bite strength among males lags slightly behind females throughout development, and reaches adult values after

Feeding Maximum Relative Bite

Performance ZAB Strength Age (weeks) Female Male Female Male Female Male 0 0.058 0.020 0.338 0.318 0.272 0.251 4 0.162 0.093 0.549 0.518 0.391 0.359 6 0.233 0.158 0.628 0.594 0.456 0.420 8 0.311 0.238 0.693 0.659 0.524 0.483 12 0.473 0.418 0.791 0.758 0.653 0.609 16 0.619 0.590 0.858 0.829 0.764 0.722 20 0.736 0.726 0.903 0.879 0.847 0.812 24 0.822 0.824 0.934 0.914 0.905 0.878 26 0.854 0.860 0.946 0.928 0.926 0.903 28 0.882 0.889 0.955 0.939 0.942 0.923 32 0.923 0.931 0.970 0.957 0.965 0.953 36 0.950 0.958 0.980 0.970 0.980 0.971

Table 3.5. Mean values of mechanical advantage of the temporalis, for two age groups of coyotes, and relative maturity of the younger group (indicated in bold), as calculated by dividing mean younger group mechanical advantage by that of the older (adult) group. Mechanical Advantage of the Temporalis Age (weeks) Female Male All Younger 0.1 – 14.3 0.376 0.383 0.380 Older 21.1 – 516.3 0.508 0.516 0.512 Relative Maturity 0.742 0.743 0.743 of Younger Group

females have matured. At birth, relative bite strength among males and females is approximately 26% of adult values, which is less mature than maximum ZAB, but more mature than feeding performance (Table 3.4., Fig. 3.9.a.). By the age of weaning, relative bite strength has improved to 46% of adult values for females and

42% of adult values for males. At the onset of deciduous tooth replacement, females continue to show greater maturity, at 65% of adult values, than do males, which are at 61% of adult values. With the completion of adult tooth eruption at 26 weeks, relative bite strength remains more mature for females, at 93% of adult values, than for males, which are at 90% of adult values. Females attain relative bite strength maturity three weeks after adult dentition is complete and two weeks before males reach maturity (Table 3.3.). Female coyotes attain their adult maximum ZAB and relative bite strength earlier than males, but they are disadvantaged in comparison to males with smaller maximum zygomatic arch breadths and lower relative bite strength; female feeding performance not only matures later than males, but it is also significantly lower than males (Table 3.2.).

DISCUSSION

Our results demonstrate that juvenile coyotes are handicapped by an immature food processing apparatus, relatively poor biomechanical abilities, and slower food processing times than those of adults. Coyote feeding performance and biomechanics mature long after weaning, which occurs at six weeks of age. Further, the development of feeding performance is asynchronous with development of both feeding biomechanics and skull morphology (Ch. 2, La Croix et al., 2011) (Fig. 3.9a.,

b., c.). Feeding performance, feeding biomechanics and skull morphology exhibit different temporal patterns of development, especially relative to early life history milestones. Skull size, mandibular shape and feeding biomechanics are dramatically more mature across ontogeny than is feeding performance, and these features achieve adult values much earlier in life. In particular, the relative maturity of mandibular shape during the youngest age intervals contributes to the interesting pattern of mature and constant mechanical advantage during early ontogeny; despite the early maturation of these features, feeding performance nevertheless lags behind. Cranial shape, the only measure aside from feeding performance that is extremely immature at birth, matures much more quickly than does feeding performance over the course of ontogeny. Indeed, feeding performance lags behind the relative maturity of all other measures considered here, and matures much later than most (Fig. 3.10.).

The developmental asynchrony shown here between feeding performance and morphological variables suggests that a certain minimum threshold of physical growth and development, together with the associated development of biomechanics, are required to produce effective mastication. In addition, feeding biomechanics exhibits interesting development patterns during ontogeny, suggesting adaptations in young animals to balance conflicting demands of their immediate need to process food with growth of the feeding apparatus towards an adult form.

There is also evidence that males and females resolve these conflicts differently, as

Figure 3.10. Timeline illustrating the age at maturation for coyote feeding performance, biomechanics and skull morphology in relation to major life history events. Maturation for skull morphology (Ch. 2, La Croix et al., 2011) is indicated with a dashed line, and for feeding performance and biomechanics (this study) with a solid line. Where maturation age differed between sexes for a measure, the latest maturation age is diagrammed.

females trade earlier maturity for smaller maximum zygomatic arch breadth, decreased relative bite strength, and diminished feeding performance, compared to males.

At weaning, coyote pups are unable to do more than nibble at the edges of a test biscuit. However, feeding performance soon improves dramatically, and 12 week old pups can process a 32g biscuit; their feeding performance has already improved to nearly half of adult values (Fig. 3.9.a.). However, by 12 weeks of age the biomechanical measures that contribute to mastication are already nearly 80% mature, so feeding performance remains relatively immature in comparison to these measures. Despite rapid early improvement in feeding performance, its maturation occurs late, relative to the maturation of feeding biomechanics, and especially late relative to achievement of maturity in skull size and shape (Fig. 3.10.).

During early ontogeny, the relative maturity of feeding biomechanics might lead one to expect more mature feeding performance, especially because mechanical advantage of the temporalis is maintained at approximately 74% of adult values during the first 14 weeks of life (Fig. 3.9.a.). Remarkably, early mechanical advantage is accomplished by an elegant maintenance of in- and out-lever arm length proportions over a developmental period wherein dramatic changes in cranial and mandibular size and shape occur (Ch. 2, La Croix et al., 2011) including tooth eruption and replacement. Isometric growth might be expected to maintain these lever arm length proportions, but we found that this was not the case (Fig. 3.7.b., c.).

Initially, growth in the length of the out-lever arm is greater than that of the in-lever arm; at 12 weeks, the distance between the deciduous bite point and the mandibular

condyle is already equivalent to the length of the adult out-lever arm. Subsequently, and in conjunction with ongoing tooth row elongation, the adult mandibular carnassial erupts slightly behind the deciduous carnassial and replaces it; the distance between the mandibular condyle and the adult bite point is conserved as the out-lever arm stops growing. In contrast, the in-lever arm length continues to increase, resulting in a shift in the mechanical advantage of the temporalis from the range of young animals to that of adults. By 21 weeks, growth of the in-lever arm length also ceases, resulting in maturation of the mechanical advantage of the temporalis. This early maturation age for mechanical advantage makes it temporally disjunct from the maturation of feeding performance, thus younger animals are beneficiaries of a unique punctuated growth pattern during the period when they are constrained by immature chewing muscles and tooth replacement. In the absence of this maturation pattern, feeding performance would undoubtedly fare worse, particularly during the challenging period immediately after weaning, when young animals must first feed themselves.

Conflicting demands between the immediate need to process food and the growth of the feeding apparatus toward an adult form were resolved differently by females and males. During ontogeny, growth of the maximum ZAB is significantly faster for females than for males, affording them early advantage as they achieve their adult maximum ZAB and adult relative bite strength sooner than males. The males however, continue to grow, albeit more slowly, and achieve larger maximum

ZABs and larger relative bite strengths. This pays off for males with faster feeding

performances that mature earlier than those of females who are slower at

processing food, and who achieve adult food-processing levels later.

Maturation of maximum ZAB, relative bite strength, and mandibular shape is

more temporally coincident with maturation of feeding performance than is skull size,

skull shape or mechanical advantage of the temporalis (Ch.2, La Croix et al., 2011)

(Fig. 3.10.). Because mechanical advantage of the temporalis matures so early (21

wks), it is the subsequent and ongoing growth in maximum ZAB, reflecting

augmentation of the temporalis muscles after 21 weeks of age that drives the

development of relative bite strength and ultimately, that of feeding performance.

The maturation of maximum ZAB by 30 weeks appears to set the foundation for the

maturation of relative bite strength by 32 weeks and maturation of feeding

performance by 36 weeks. It should be noted that maximum ZAB, used here as a

proxy for muscle strength in the calculation of relative bite strength, probably

underestimates mass of the primary mastication muscles, because it does not take

into account the substantial relative narrowing of the basicranium during coyote

growth and development; this change in skull shape yields a larger space within the

zygomatic arch for temporalis muscle deposition after skull shape reaches maturity

(Ch. 2, La Croix et al., 2011).

It is not presently known whether the asynchrony shown here among the

development of feeding performance, feeding biomechanics and skull morphology in

the coyote represents a more generalized mammalian pattern. In fact, only one other

study, on the spotted hyena , Crocuta crocuta (Tanner et al., 2010), has concurrently examined ontogenetic patterns of feeding performance, biomechanics and skull

shape change in relation to life history milestones. Hyenas, unlike coyotes, are capable of cracking open large bones as adults, using craniodental adaptations for durophagy that include a vaulted forehead, a large sagittal crest, massive zygomatic arches and robust premolars with crack-resistant enamel. Another distinction between hyenas and coyotes may similarly be related to morphological demands of durophagy; that is, the hyena’s life history is notable for its protracted period of maternal dependence relative to those in other large carnivores (Watts et al., 2009).

Previous research has shown that skull ontogeny in the spotted hyena, like that of coyotes, demonstrates relative synchrony in cranial and mandibular growth but asynchrony in cranial and mandibular shape development, cranial size and shape maturation, and mandibular size and shape maturation (Tanner et al. 2010, Ch. 2,

La Croix et al., 2011). The most striking differences between coyotes and hyenas, however, are the sequence of maturation events and their timing relative to key life history milestones (Fig 3.11.). Feeding performance in coyotes matures after the maturation of skull size and shape and is the last measure to mature. By contrast, in the spotted hyena, feeding performance maturation is dramatically delayed, occurring long after skull size maturation, but slightly before full maturation of skull shape (Watts et al., 2009). Patterns of maturation in regard to feeding biomechanics also differ between coyotes and hyenas. Whereas maturation of mechanical advantage of the temporalis coincides with maturation of skull size in both species, maximum ZAB matures before relative bite strength in coyotes, but this pattern is reversed in hyenas. Further, whereas maturation of maximum ZAB and relative bite strength are temporally coincident in coyotes, they are temporally disjunct in hyenas.

Figure 3.11. Comparison of maturation timing of feeding performance and biomechanics for coyotes and spotted hyenas in relation to major life history events. Maturation for coyotes (this study, Ch. 2, and La Croix et al., 2011) is indicated with a dashed line, and for spotted hyenas (Tanner et al., 2010 and Tanner, unpublished data) with a solid line. The latest maturation age is diagrammed for all measures.

Although use of maximum zygomatic arch breadth as a proxy for muscle strength may underestimate relative bite strength in coyotes, this problem would only be exacerbated in a hypercarnivore like the spotted hyena with its massive temporalis muscles.

Feeding performance in both coyotes and spotted hyenas matures long after weaning, suggesting that young individuals experience significant handicaps during feeding between weaning and the age at which performance maxima are achieved.

However, whereas all aspects of maturation in behavioral and morphological traits associated with feeding have reached full maturity by the age of sexual maturity in coyotes, this is not the case in spotted hyenas. Indeed, various morphological aspects of the feeding apparatus, such as maximum ZAB and adult skull shape, do not reach full maturity until long after puberty in spotted hyenas, and feeding performance maxima are not achieved until the mean age of first parturition.

Delayed feeding performance, biomechanics and skull maturation during ontogeny of spotted hyenas may be explained as trade-offs associated with the ability to crack open large bones in adulthood. The dominant feature of post-weaning shape change in the spotted hyena skull is the development of the bony areas of muscle insertion: zygomatic arches expand and the sagittal and nuchal crests develop (Tanner et al.,

2010). It has previously been shown for coyotes, which lack such specialized adult function as bone-cracking, that there is a more modest delay in skull maturation than that seen in spotted hyenas (Ch. 2, La Croix et al., 2011). Here, we have shown that a substantial delay in maturation also exists for the achievement of feeding performance and feeding biomechanics, but this delay is shorter in the less-

specialized species. Future work remains to determine whether maturation schedules influence diet choices in coyotes and other carnivores.

The temporal coincidence of weaning and adult tooth eruption may imply that juveniles are capable of effectively competing for food with conspecifics at this time

(Biknevicius and Van Valkenburgh, 1996). However, neither the results presented here nor the data from hyenas (Binder, 1998; Binder and Van Valkenburgh, 2000;

Tanner, 2007) support that hypothesis. Instead, our data suggest that juvenile coyotes are handicapped in their feeding performance, biomechanics, and skull morphology for up to seven months, between weaning at six weeks and feeding performance maturation at 36 weeks. Even though bi-parental provisioning may occur after weaning in coyotes, as it does in other canids, provisioning stops long before juveniles reach their adult performance capabilities. Thus, young animals must feed themselves after weaning and adult tooth eruption despite being considerably disadvantaged relative to the adults with which they must compete for food. Such handicaps are apparent in the generalist coyote as well as in the highly specialized bone-cracking hyenas.

Factors other than biomechanical constraints, as seen in coyotes, or selective pressures associated with adult dietary demands, as seen in hyenas, may influence the maturation of feeding performance in these and other carnivores, and thus determine how long young animals remain handicapped during feeding competition.

First, feeding performance requires a functional relationship between the parts of the skull; for example, the cranium and the mandible must work in concert with each other to form an effective feeding apparatus. This feeding apparatus must remain

functional across ontogeny, while shape and size of both cranium and mandible are changing. Asynchronous development of cranium and mandible might delay maturation of feeding performance. Second, efficient food processing takes practice; the effects of learning were not considered here. Third, tooth development and strength may limit biting ability.

By looking at the ontogeny of feeding performance and biomechanics concurrently, examining their maturation patterns, and placing both within the context of life history, we can shed light on the processes at work during the protracted period of morphological and behavioral development of carnivores.

Additional studies on carnivores are needed to provide a richer context for interpreting the relationships between life histories and the maturation of morphology and behavior, and life history. We have demonstrated here, that, while mandibular growth patterns are important to early mechanical advantage of the temporalis, it is the ongoing development of the primary mastication muscles, the temporalis that can have an acute effects on the maturation of bite strength and ultimately feeding performance and fitness.

Chapter FOUR

EFFECTS OF EARLY BONE PROCESSING OPPORTUNITIES ON FEEDING PERFORMANCE, SKULL MORPHOLOGY AND BIOMECHANICS IN COYOTES

INTRODUCTION

Phenotypic plasticity, indicating the sensitivity of developmental processes to environmental influences, enables ontogenies to produce multiple outcomes. The phenotypic plasticity of the feeding apparatus, especially the sensitivity of craniodental development to environmental influences, is of particular interest, because variation in the emergent phenotype of the food processing apparatus has important implications for survival.

Much interest has been shown in the effects of environment on cranial shape

(reviewed by Herring, 1993). In particular, there has been persistent interest in the effects of captivity, and, especially the effects of wild and captive diets, on skull shape (Hollister, 1917; Zuccarelli, 2004). Studies of diet-related variation in skull morphology have most often used rats (Barber et al., 1963; Grauer and Hurov, 1989;

Killaridis et al., 1992; Maki et al., 2002; Odman et al., 2008; Watt and Williams,

1951), but several have considered carnivores, animals at the highest trophic level

(wild and captive: lions, Zuccarelli 2004; foxes, Trut, 1999; and spotted hyenas,

Lundrigan, unpublished data). However, despite common acceptance that dietary consistency affects craniofacial and mandibular shape, to date, few studies have documented clear effects of early diet on morphology because they lacked adequate sample sizes or controls. In the rare case of a well-designed study with adequate sample size, diet-related variation in the skull was trivial, although statistically

significant (Peromyscus, Myers et al., 1996). Further, these earlier studies have not addressed how different ontogenetic outcomes in the morphology of the feeding apparatus affect variation in adult feeding behavior. Studies are needed showing how early variation in diet leads to multiple ontogenetic outcomes that have consequences for both feeding form and function among adults.

Here, we took advantage of a unique opportunity to concurrently examine developmental plasticity in feeding behavior, feeding apparatus morphology and feeding biomechanics, in the same individuals. We investigated variation in emergent adult phenotypes in an animal at the highest trophic level through an experimental manipulation of early bone chewing opportunities between two known- age groups of coyote pups. We first describe differences in adult feeding performance between individuals afforded opportunities to chew bones throughout ontogeny and those that were not. Next, we provide detailed descriptions of differences between treatment and control groups with respect to shape and size of both cranium and mandible among adults, utilizing landmark-based geometric morphometrics. Finally, we investigate how differences between groups with respect to feeding biomechanics and mastication musculature might explain the differences found in adult feeding performance. In this way, we show that environmental effects during ontogeny can lead to variation in adult morphology, which yields variation in adult behavior. Here, differences in diet during ontogeny lead to differences in the adult feeding apparatus, which yield differences in adult feeding performance.

METHODS

Coyote life history

Coyotes are the most abundant large carnivores in North America, and the top predator in many ecosystems. With the behavioral plasticity to consume small and large prey, live in pairs or packs, and colonize rural and urban environments, the coyotes’ diet and lifestyle are opportunistic and generalist (Bekoff, 1977; Bekoff and

Gese, 2003). In the absence of other top predators, they have altered their behavior to fill a myriad of ecological niches (Bekoff and Gese, 2003).

The annual cycle of coyote life history events begins with mating in January through March (Bekoff, 1977). Following a 63 day gestation, pups are born, beginning in March (Bekoff, 1977). Weaning of pups begins by four weeks of age and is usually completed by six weeks of age (F. Knowlton, personal communication) (5-7 weeks in Bekoff, 1977; Snow, 1967). The onset of deciduous tooth replacement occurs by 12 weeks of age, and eruption of the adult dentition is complete by 26 weeks of age (Bekoff, 1977; Ch. 2, La Croix et al., 2011). Pups are initially provisioned by both parents, but during the summer the juveniles develop greater independence and self-reliance (Harrison and Gilbert, 1985; Harrison et al.,

1991). Subadults of either sex may emigrate in the fall of the year; first parturition in non-persecuted populations typically occurs at 22 months of age although both sexes are physiologically capable of breeding at 10 months of age (Bekoff, 1977;

Bekoff and Gese, 2003).

Among canids, coyotes are usually categorized as omnivorous generalists, with young coyotes capitalizing on that portion of the adult diet that is most easily

obtained or subdued (Andelt et al., 1987; Arjo et al., 2002; Bowen, 1978; Clark,

1972; Gese et al., 1996; Hernandez et al., 2002; Hidalgo-Mihart et al., 2001;

Johnson, 1978). Following weaning, juvenile coyotes are initially provisioned by

their parents; subsequently, they ingest mostly small mammals, vegetation,

invertebrates, and birds whereas adults consume larger mammals, fewer

invertebrates and few birds (Hawthorne, 1970). Fruit and insect ingestion by

coyotes occurs most frequently in the summer diet, which corresponds temporally

with a coyote pup’s increasing independence (Young et al., 2006).

Subjects

All data were collected from known-age, captive-born coyotes housed at the USDA

Wildlife Services National Wildlife Research Center’s Logan Field Station in Millville,

UT between 2005 and 2006. Exact dates of birth and death were known for all

individuals; animal records are maintained at the Logan Field Station.

USDA/APHIS/WS/NWRC IACUC approved the study protocol QA-1179.

We selected two male coyote pups and one female pup, of similar mass, from

each of eight litters born within a 10 day period in 2005 (Appendix Table A.1.);

selection of animals was made when they were 28 days of age. Animals were then

hand reared in litter groups until weaned at six weeks of age, and subsequently

housed singly. Litter groups were housed indoors in 0.6m x 1.2m wire dog kennels

with plastic floor pans. Following weaning, animals were housed indoors for up to

2 six weeks in 0.6m aluminum kennels with wire floors; subsequently, they were moved outdoors to an open pole barn with 3.7m x 1.2m chain-link fence kennel

2 enclosures with concrete floors and containing 0.6m round plastic den boxes. All

animals were maintained on a commercially produced wet food diet designed for fur bearing animals; the wet food contained chicken (beaks, feet, and feathers) and grain. Prior to weaning, all pups received a milk replacement powder commercially produced for puppies and reconstituted with water; during weaning, all pups received a pea-sized kibble version of the wet food diet, moistened with water. Two control animals were humanely euthanized by veterinarians during the study for health reasons unrelated to this study.

Diet manipulation

At six weeks of age, one male pup from each litter was assigned to our treatment group, and the remaining two pups from each litter were assigned to the control group. Treatment animals subsequently received daily access to bones while control animals did not. Other chewing opportunities were minimized within the kennel environment; control animals had no access to bones, other foods, or chew toys except during feeding performance trials.

Treatment animals received continuous access to bones for chewing from six weeks of age until testing at 80 weeks. Bones were rotated on a weekly basis among the animals and replaced with new bones as needed. A weekly log for each treatment animal was maintained by animal care workers and confirmed, through animal observations and examinations of bones that were removed from kennels, that each treatment animal chewed the bones. A variety of bones were used during the study to encourage chewing. All bones were obtained from Merrick Pet Care,

Texas. These bones, which were produced from human-grade cows and pigs from the United States, and sheep from New Zealand, were processed with a basted

coating and no artificial ingredients. Young pups chewed beef rawhide bones and sheep femurs; older animals chewed pig femurs, cow ribs, cow femurs, and cow knuckle bones. Except in the rare instance where an entire sheep or pig femur was consumed, little nutritive value was gained from chewing the bones.

Preparation of skull and muscle specimens

Coyotes were sacrificed at 18 months of age and final body weights obtained. Entire heads were immediately removed by dissection at the caudal vertebrae and frozen for later dissection in the laboratory. Following thawing, the major cranial muscles

(masseter, temporalis, pterygoid and digastric) were removed from the skulls of six matched pairs of male siblings. The masseter muscles were removed bilaterally by dissecting along their origination points on the zygomatic arch and along their insertion points on the masseteric fossa of the mandible. The temporalis muscles, which lie superior to the zygomatic arch, were removed by dissection at their origination points along the sagittal, frontal, and nuchal crests of the cranium and the zygomatic arch, and at their insertion points on the coronoid process of the dentary.

The pterygoid muscles were removed bilaterally by dissecting along their origination points on the pterygoid, palatine, and sphenoid bones and their insertion points on the medial surface of the mandible, ventral to the mandibular foramen and condylar process. The digastrics muscles were removed by dissecting along their origination points on the jugular process of the occipital bone and their insertion points on the medial surface and ventral border of the mandible. The masseter, temporalis, pterygoid, and digastric muscles from one side of a skull were treated as a single mastication muscle complex; each specimen thus had a right and a left mastication

muscle complex. Next, left and right mastication muscle complexes were lyophilized and their dry mass recorded. The total muscle dry mass for each individual was calculated by summing the dry mass values for its left and right muscle complexes.

Skull specimens were then prepared at the Michigan State University Museum and catalogued into that collection (Appendix Table A.4.1.).

Feeding performance

Feeding performance trials, utilizing standardized food items commercially produced for the pet industry, were conducted on 18 month old coyotes to delineate any cumulative performance advantage that may have accrued during their exposure to bone processing opportunities. Individuals were tested in their home kennels.

Feeding Performance Test 1: Rawhide chew twist

Shearing-related feeding performance was tested using a Dentley’s brand rawhide chew twist, 130mm in length and 5mm in diameter. Consumption of the rawhide chew twist required the ability to cut through the rawhide. Tests were performed in home enclosures 18-24 hours after the last feeding. On the day of testing, animals that did not consume a pre-test rawhide (a 45mm rawhide chew twist) within five minutes of delivery, with the Tester (SLC) present in the kennel block, were excused from the test. Qualifying animals were then presented with a

130mm rawhide chew twist. Animals unable to consume the entire rawhide chew twist within 10 minutes were scored a processing time of 600 seconds. Each trial was videotaped in natural light using a Sony Handycam Vision CCD-TRV65 NTSC –

VideoHi8TR Steady Shot with 72X digital zoom. The video camera was mounted on a tripod outside the subject’s kennel with a viewpoint 36” above the ground. The

time, in seconds, required to consume the rawhide chew twist was later calculated from these videotapes. Our measure of feeding performance for the rawhide chew twist trials was consumption time, defined as the sum of the periods of continuous and sustained chewing including breaking up of the rawhide chew twist into pieces that could be swallowed. Timing began with the first audible crunching/slicing of the rawhide chew twist and concluded when it was completely consumed.

Feeding Performance Test 2: 32g dog biscuit

Crushing-related feeding performance was assessed using a 32g, Iams brand, adult large dog biscuit. Consumption of the 32g biscuit required the ability to crush and crack the biscuit, as opposed to shearing it, as in the rawhide chew twist test. Tests were performed in home enclosures 18-24 hours after the last feeding.

On the day of testing, animals that did not consume a pre-test biscuit (a 4g Iams brand puppy dog biscuit) within five minutes of delivery, with the Tester (SLC) present in the kennel block, were excused from the test. Qualifying animals were then presented with a 32g biscuit. Three trials were conducted serially, with at least a 60 second delay between finishing ingestion of one biscuit and delivery of the next; animals unable to consume an entire biscuit within 10 minutes were scored a processing time of 600 seconds and excused from further biscuit trials that day. As with the rawhide chew twist tests, biscuit trials were videotaped and the time, in seconds, required to consume the biscuit later calculated from those videotapes.

Our measure of feeding performance was consumption time, defined as the sum of the periods of continuous and sustained chewing including breaking up of the biscuit into pieces that could be swallowed. Timing began with the first audible crunch of

100

the biscuit into more than one piece and concluded when the biscuit was completely consumed.

Feeding Performance Test 3: Beef shank reduction

Bite-strength related feeding performance was tested in a 12 hour trial using

Merrick brand beef shanks which were smoked meaty beef bones that retained tendons and beef jerky and contained marrow. Reduction of the beef shank required greater mechanical forces of chewing than were needed in the biscuit and rawhide chew twist tests, in order to remove tendons and access the bone’s marrow.

This test was performed concurrently on all animals, in their home enclosures. At least six hours after the daily feeding, a beef shank was weighed and delivered to each animal’s kennel for 12 hours of bone processing access. The mass of each beef shank was recorded, prior to delivery, and upon its collection at the conclusion of the trial. Care was taken in collecting all pieces of the beef shank that had not been consumed by the animal during the trial. Our measure of bite-strength related feeding performance was the percent reduction in the mass of the test bone (pre-trial mass divided by post-trial mass).

Feeding performance analysis

Because adult coyotes exhibit sexual size dimorphism, with males being about 20% larger than females, we restricted our feeding performance contrasts to the data for matched pairs of males. Our matched contrast design involved samples with an n < 10 and repeated measures, therefore, non-parametric Wilcoxon Signed

Rank Tests were performed to determine if there were significant feeding performance differences between treatment and control groups. Statistical analyses

101

were performed by hand (following Lowry, 1998-2011) and plots were created using

STATISTICA, version 8.0 (StatSoft, Inc., 2007, www.statsoft.com).

Skull shape and size

Diet-related variation in cranial and mandibular shape were analyzed by geometric

morphometrics (Rohlf and Slice, 1990; Zelditch et al., 2004). Each cranium was

photographed in ventral and lateral views. In ventral view, the skull was oriented with

the palate parallel to the photographic plane, and in lateral view, the skull was

oriented with the mid-sagittal plane parallel to the photographic plane. Mandibles were photographed in lateral view, oriented with the longest axis of the mandible

parallel to the photographic plane. Twenty-seven landmarks visible on photographs

of the ventral cranium provide the data for the analysis of shape in that view (Fig.

4.1.a.). Landmarks alone cannot fully capture the dorsal curve of the lateral cranium

or the mandibular ramus so these were analyzed using a combination of landmarks

and semi-landmarks. Semi-landmarks, unlike landmarks, are not discrete

anatomical loci that can be recognized as homologous points, and they contain less

information than landmarks because their spacing along the curve is arbitrary.

However, semi-landmarks make it possible to study complex curving morphologies

where landmarks are sparse. Fourteen landmarks

and 32 semi-landmarks were selected for the lateral cranium (Fig. 4.1.b.), and 11

landmarks and 75 semi-landmarks were selected for the mandible (Fig. 4.1.c.).

Descriptions of these landmarks are contained in Appendix Table A.4.2. Landmarks

and semi-landmarks were digitized using tpsDig2.10 (Rohlf, 2005). For semi-

102

103

landmarks, the curve-tracing function was applied and semi-landmarks were evenly spaced along the curves using the ‘‘resample’’ function of the curve-tracing tool.

Landmarks were superimposed using Generalized Procrustes Analysis to remove variation in scale, position and orientation (Rohlf and Slice, 1990; Zelditch et al., 2004). Semi-landmarks require a specialized superimposition method because their spacing is biologically arbitrary. Semi-landmarks were superimposed to minimize the Procrustes distance from the mean shape; we use this criterion for sliding semi-landmarks because the Procrustes distance is the metric underlying the general theory of shape. According to this method, the tangent to the curve at each semi-landmark is estimated and then each semi-landmark is slid toward the normal of its respective tangent, minimizing the overall difference from the reference

(Andresen et al., 2000; Bookstein et al., 2002; Sampson et al., 1996). Following superimposition, semi-landmarks can be used in conventional shape analyses provided that statistical tests take into account that they have only one degree of freedom. For the ventral cranium, bilaterally homologous landmarks were reflected and averaged after the Procrustes superimposition because bilaterally homologous landmarks are not independent of each other; to ease interpretation of the visual results, they are shown as whole skulls. To quantify skull size, we used centroid size, the square root of the summed squared distances from each landmark to the geometric center of the object. Superimposition of landmarks was done using

CoordGen6h (Sheets, 2009); semi-landmarks were superimposed in SemiLand

(Sheets, 2003). Reflection and averaging of bilateral landmarks was done in Sage

(Marquez, 2007).

104

Although coyote skulls exhibit sexual size dimorphism, they do not exhibit sexual shape dimorphism (Ch. 2, La Croix et al., 2011), so we pooled the data for male and female control animals in our analyses of skull shape, as geometric morphometrics removes the effects of size when considering shape. Pairwise one- way MANOVA analyses, using the Goodall’s F test statistic, were performed to determine if there were significant skull shape differences between the treatment and control groups for each view of the skull. Results are depicted using the thin- plate spline function (Bookstein, 1991). Analyses and diagrams were performed in

Twogroup6h (Sheets, 2000). To determine whether there were significant diet- related differences in cranial and mandibular size, measured as centroid size (CS), we restricted the data to those for matched pairs of males, and conducted non- parametric Wilcoxon Signed Rank Tests. Skull length, a traditional linear measurement, was calculated from the photographs of craniums, in the lateral view, using the linear measurement tool in tpsDig2. Skull length was measured as the distance between the most anterior point on the maxilla and the most posterior point on the nuchal crest, in millimeters (Fig. 4.2.a.) As with centroid size, Wilcoxon

Signed Rank Tests were conducted for matched pairs of males to determine whether there were statistically significant differences between the two diet groups.

Statistical analyses were performed by hand (following Lowry, 1998-2011) and plots were created using STATISTICA, version 8.0 (StatSoft, Inc., 2007, www.statsoft.com).

105

Feeding biomechanics, mastication muscles and body mass