The Evolution of Forelimb Morphology and Flight Mode in Extant Birds A

Home , Allometry, Anhinga, Antarctic prion, Ascension frigatebird, Australasian gannet, Australian pelican

The Evolution of Forelimb Morphology and Flight Mode in Extant Birds

A dissertation presented to

the faculty of

the College of Arts and Sciences of Ohio University

In partial fulfillment

of the requirements for the degree

Doctor of Philosophy

Erin L. R. Simons

August 2009

This dissertation titled

The Evolution of Forelimb Morphology and Flight Mode in Extant Birds

ERIN L. R. SIMONS

has been approved for

the Department of Biological Sciences

and the College of Arts and Sciences by

Patrick M. O'Connor

Associate Professor of Anatomy

Benjamin M. Ogles

Dean, College of Arts and Sciences

ABSTRACT

SIMONS, ERIN L. R., Ph.D., August 2009, Biological Sciences

The Evolution of Forelimb Morphology and Flight Mode in Extant Birds (221 pp.)

Director of Dissertation: Patrick M. O'Connor

The research presented herein examines the morphology of the wing skeleton in

the context of different flight behaviors in extant birds. Skeletal morphology was

examined at several anatomical levels, including the whole bone, the cross-sectional

geometry, and the microstructure. Ahistorical and historical analyses of whole bone

morphology were conducted on densely-sampled pelecaniform and procellariiform birds.

Results of these analyses indicated that the external morphology of the carpometacarpus, more than any other element, reflects differences in flight mode among pelecaniforms. In

addition, elements of beam theory were used to estimate resistance to loading in the wing

bones of fourteen species of pelecaniform. Patterns emerged that were clade-specific, as

well as some characteristics that were flight mode specific. In all pelecaniforms

examined, the carpometacarpus exhibited an elliptical shape optimized to resist bending

loads in a dorsoventral direction. Moreover, birds that utilize flapping exhibited distal

elements that were more elliptical than other flight modes, perhaps pertaining to the

higher frequency of loading. Soaring birds exhibited wing elements with near-circular

cross-sections and higher polar moments of area than in the flap and flap-gliding birds,

suggesting shapes optimized to offer increased resistance to torsional loads. Congruent

results between pelecaniform and procellariiform birds (two distantly related groups)

indicate the presence of general trends in the structure and function of the avian wing

4 skeleton. In a separate analysis, bone microstructure of the forelimb elements of select pelecaniform, procellariiform, and falconiform taxa was examined. Data on the degree of primary vascular canal laminarity (i.e., the orientation of vascular canal networks) were collected and used to test a series of hypotheses related to the relationship between skeletal microstructure and inferred wing loading during flight. The dynamic soaring birds exhibited significantly lower laminarity in the wing elements than the flapping and static soaring birds, a result that may be explained by the difference in loading pattern due to overall wing shape variation among the groups. Finally, mechanical testing at the whole bone level was performed on individual wing elements to test predictions derived from whole-bone, cross-sectional geometric, and histological studies. These results revealed that variation in stiffness (Young’s modulus) exists both among wing elements within a given species and among species that utilize different primary flight modes.

Specifically, the CMC and ulna were significantly stiffer than the humerus in all species, presumably to accommodate the loads transmitted through the flight feathers. In addition, the dynamic soaring albatross and continuous flapping cormorant exhibited stiffer wing elements than the static soaring pelican. In sum, differences detected in morphology and mechanical properties of avian wing elements do correspond with variation in primary flight mode and offer insight into the relationship between structure and function in the avian wing.

Approved: ______

Patrick M. O'Connor

Associate Professor of Anatomy

This work is dedicated to my husband,

Verne Simons

ACKNOWLEDGMENTS

I would first like to extend a special thank you to my dissertation advisor, Patrick

O’Connor, for his advice, insight, patience, and support. I wish to also thank my dissertation committee members, Audrone Biknevicius, Susan Williams, and Alycia

Stigall, for providing valuable perspectives and encouragement.

I would like to specifically thank Larry Witmer and Ryan Ridgely for assistance at the Ohio University microCT facility; Patrick O’Connor, Robert Hikida, Tobin

Hieronymus, Susan Williams, and Andrew Lee for assistance with histological preparation and analysis; and Betty Sindelar and Susan Williams for assistance with mechanical testing.

In addition, discussions with Audrone Biknevicius, Lisa Noelle Cooper, Joseph

Daniel, David Dufeau, Joseph Eastman, Michael Habib, Jennifer Hancock, Jennifer

Herman, Tobin Hieronymus, Casey Holliday, Dawn Holliday, Angela Horner, Michael

Jorgensen, Andrew Lee, Emanuel de Margerie, Eric McElroy, Donald Miles, Molly

Morris, Patrick O’Connor, Biren Patel, Steve Reilly, Willem Roosenburg, Christopher

Ruff, Verne Simons, Betty Sindelar, Nancy Stevens, Alycia Stigall, Susan Williams,

Larry Witmer, have greatly aided this research.

I am indebted to the following curators and collection managers for access to specimens in their collections: P. Sweet and P. Hart (American Museum of Natural

History), S. Rogers and B. Livezey (Carnegie Museum of Natural History), D. Willard

(Field Museum of Natural History), J. Dean and S. Olson (National Museum of Natural

History).

I also thank W. and B. Fox (Pelican Harbor Seabird Station, Miami, FL), C.

Rehkemper and B. Zaun (Kauai National Wildlife Refuge Complex, HI), A. Freiman, S.

Krause, M. Smith, and W. Smith (Sweetbriar Nature Center, NY), and Willem

Roosenburg for avian specimens used during the course of this work.

This research was supported by grants from the Ohio University Graduate Student

Senate, the Ohio University Office of Research and Sponsored Programs (Student

Enhancement Award), an AMNH Collection Study Grant from the Department of

Ornithology, and the Ohio University College of Osteopathic Medicine (PMO).

Finally, I wish to thank my husband, Verne Simons, for his constant support, encouragement, sense of humor, and incredible knowledge of how animals (and mechanical things) work. I am grateful to our families and friends for their endless encouragement.

TABLE OF CONTENTS

Page

Abstract ...... 3

Acknowledgments...... 6

List of Tables ...... 12

List of Figures ...... 14

Preface ...... 17

Literature Cited ...... 26

Chapter 1: Forelimb skeletal morphology and flight mode evolution in pelecaniform birds

...... 32

Abstract ...... 32

Introduction ...... 33

Materials and Methods ...... 35

Results ...... 40

Allometric Analyses ...... 40

Multivariate Analyses ...... 41

Discussion ...... 42

Allometric Aspects of the Pelecaniform Forelimb Skeleton ...... 42

Multivariate Analyses: A Map to Flight Mode Evolution in Pelecaniforms ...... 44

Conclusion ...... 48

Literature Cited ...... 50

Chapter 2: Cross-sectional geometry of the forelimb skeleton and flight mode in pelecaniform birds ...... 63

Abstract ...... 63

Introduction ...... 64

Materials and Methods ...... 67

Results ...... 69

Discussion ...... 70

Patterns Common to All Pelecaniforms ...... 70

Flight Mode Patterns ...... 73

Literature Cited ...... 78

CHAPTER 3: Whole bone and cross-sectional morphology of the wing skeleton in

procellariiform seabirds: implications for differences in flight behavior ...... 91

Abstract ...... 91

Introduction ...... 92

Materials and Methods ...... 96

Whole-Bone Morphometrics ...... 98

Cross-Sectional Geometry ...... 100

Results ...... 102

Allometric Analyses ...... 102

Multivariate Analyses ...... 102

Cross-Sectional Geometry ...... 105

Discussion ...... 106

Allometry of the Forelimb Skeleton ...... 106

Multivariate Analyses ...... 107

Patterns in Cross-Sectional Geometry ...... 111

Conclusion ...... 114

Literature Cited ...... 115

CHAPTER 4: BONE MICROSTRUCTURE, Primary vascular canal orientation and flight mode in birds ...... 137

Abstract ...... 137

Introduction ...... 138

Hypotheses Examined in Study ...... 142

Materials and Methods ...... 144

Histological Preparation ...... 145

Vascular Organization ...... 146

Results ...... 147

Among Quadrants Analysis ...... 147

Among Elements Analysis ...... 147

Among Flight Mode Analysis ...... 148

Remodeling of Avian Cortical Bone ...... 149

Discussion ...... 149

Evidence of Bone Functional Adaptation in Wing Element Microstructure ...... 149

Evolution of Microstructure in Response to Flight Mode ...... 151

Haversion Remodeling in Wing Elements ...... 154

Implications of Growth Dynamics for Bone Microstructure ...... 155

Literature Cited ...... 157

Chapter 5: Mechanical properties of the wing skeleton of birds utilizing different primary flight modes ...... 174

Abstract ...... 174

Introduction ...... 174

Materials and Methods ...... 178

Results ...... 179

Among Elements Analysis ...... 180

Among Flight Modes Analysis ...... 180

Discussion ...... 180

Literature Cited ...... 186

Chapter 6: Summary and future directions ...... 198

Appendix A: Supplementary information for Chapter 1 ...... 204

Appendix B: Supplementary information for Chapter 2 ...... 208

Appendix C: Supplementary information for Chapter 3 ...... 211

Appendix D: Supplementary information for Chapter 4 ...... 218

Appendix E: Supplementary information for Chapter 5 ...... 221

LIST OF TABLES

Page

Table 1–1: Body mass means, wingspan ranges, and primary flight modes of

pelecaniform taxa examined ...... 57

Table 1–2: Summary of reduced major axis (RMA) regression results of skeletal metrics

on geometric mean (GM – body size) for pelecaniform birds ...... 58

Table 2–1: List of pelecaniform taxa scanned and flight mode groups ...... 82

Table 2–2: Species means of cross-sectional geometric parameters for each of the three

main forelimb bones in pelecaniform birds ...... 83

Table 3–1: Body mass means, wingspan ranges, and primary flight modes of

procellariiform taxa examined ...... 122

Table 3–2: Summary of reduced major axis (RMA) regression results of skeletal metrics

on geometric mean (GM – body size) for procellariiform birds ...... 123

Table 3–3: List of procellariiform taxa scanned and flight mode groups ...... 124

Table 3–4: Species means of cross-sectional geometric parameters for each of the three

main forelimb bones in procellariiform birds ...... 125

Table 4–1: Pooled laminarity index (LI) for each of the three main wing elements of

species examined in this study ...... 162

Table 4–2: Mean Laminarity Index (LI) of each of the quadrants for the three elements of

the three main species sampled ...... 164

Table 5–1: Mean Young’s Modulus (E) of the humerus, ulna, and carpometacarpus of

species in study ...... 189

Table A–1: Pelecaniform taxa used in whole bone analyses ...... 204

Table A–2: Species means for measured morphometric variables and calculated

geometric mean (in mm) of pelecaniform taxa ...... 206

Table B–1: Pelecaniform taxa used in cross-sectional geometry analysis ...... 208

Table C–1: Procellariiform taxa used in whole bone analyses ...... 211

Table C–2: Species means for measured morphometric variables and calculated

geometric mean (in mm) of procellariiform taxa ...... 214

Table D–1: Laminarity index (LI) for each quadrant for each element from all specimens

in bone microstructure study ...... 218

Table E–1: Young’s Modulus (E) of the humerus, ulna, and carpometacarpus for species

in mechanical testing study ...... 221

LIST OF FIGURES

Page

Figure 1–1: Composite phylogeny of pelecaniform taxa used in whole bone study ...... 59

Figure 1–2: Representative log-log reduced major axis (RMA) regression plots for

pelecaniform birds ...... 60

Figure 1–3: Plots of first and second canonical discriminant functions from ahistorical

and historical multivariate analyses ...... 61

Figure 2–1: Composite phylogeny of pelecaniform taxa used in cross-sectional geometry

study ...... 84

Figure 2–2: Example cross sections of the humerus, ulna, and carpometacarpus of four

species, representing the four main flight mode categories ...... 85

Figure 2–3: Pooled species means of cross-sectional geometric parameters for the

humerus, ulna, and carpometacarpus of pelecaniform taxa ...... 86

Figure 2–4: Imax/Imin ratio of the carpometacarpus, ulna, and humerus for the four flight

modes (static soar, dynamic soar, flap, flap-glide) ...... 87

Figure 2–5: Length standardized polar moment of area (J/L) of the humerus for the four

flight modes (static soar, dynamic soar, flap, flap-glide) ...... 88

Figure 2–6: Relative cortical area (CA/TA) of the humerus for the four main flight modes

(static soar, dynamic soar, flap, flap-glide) ...... 89

Figure 2–7: Illustration of wing anatomy of Morus bassanus (OUVC 10587) and

Pelecanus occidentalis (OUVC 10586) in ventral view ...... 90

Figure 3–1: Example wing profiles of select procellariiform taxa ...... 126

Figure 3–2: Seabird supertree of procellariiform taxa included in whole bone study, with

example cross sections of CMC, ulna, and humerus for each taxa included in the

cross-sectional geometry study ...... 127

Figure 3–3: Representative log-log reduced major axis (RMA) regression plots for

procellariiform birds ...... 129

Figure 3–4: Plots of first and second principle components from ahistorical and historical

multivariate analyses ...... 130

Figure 3–5: Classification and Regression Tree (CART) Analysis of whole bone

procellariiform size-corrected species mean data ...... 132

Figure 3–6: Pooled species means of cross-sectional geometric parameters for the

humerus, ulna, and carpometacarpus of procellariiform taxa ...... 133

Figure 3–7: Imax/Imin ratio of the carpometacarpus, ulna, and humerus for the four flight

mode groups (dynamic soar, flap-glide1, flap-glide2, underwater flap) ...... 134

Figure 3–8: Length standardized polar moment of area (J/L) of the carpometacarpus,

ulna, and humerus for the four flight mode groups (dynamic soar, flap-glide1,

flap-glide2, underwater flap) ...... 135

Figure 3–9: Relative cortical area (CA/TA) of the carpometacarpus, ulna, and humerus

for the four flight mode groups (dynamic soar, flap-glide1, flap-glide2,

underwater flap) ...... 136

Figure 4–1: Examples of the four types of primary canal orientations: longitudinal,

circular, radial, and oblique ...... 165

Figure 4–2: A biomechanical model of the avian wing ...... 166

Figure 4–3: Sampling protocol for histological sections generated in study ...... 168

Figure 4–4: Example histology sections of humeri exhibiting a range of primary vascular

canal orientations ...... 169

Figure 4–5: Laminarity indices (LI) for individual bones in the three focal species ....171

Figure 4–6: Humeral Laminarity Index (LI) for all species in the study ...... 172

Figure 4–7: Example sections showing secondary (Haversion) remodeling of avian

cortical bone ...... 173

Figure 5–1: Schematic of three-point bending set-up used in study ...... 190

Figure 5–2: Load (N) – Extension (mm) curves for a representative individual of the three

focal species ...... 191

Figure 5–3: Young’s modulus (E) of the humerus, ulna, and carpometacarpus of three

species that utilize different primary flight modes ...... 192

Figure 5–4: Young’s modulus (E) of the three species that utilize different primary flight

modes ...... 194

Figure 5–5: Young’s modulus (E) of the carpometacarpus (CMC) of all species in study

...... 196

Figure 5–6: Schematic illustration of wing anatomy ...... 197

PREFACE

The skeleton of vertebrates serves many important functions, including support, protection, production of red blood cells, and as a mineral reservoir. Primarily, bone must be strong and stiff enough to effectively resist stresses without excessive deformation or breakage during movements and when encountering unintentional loads (e.g., the impact when falling to the ground). Limb bones, in particular, act as levers and move by the contraction and relaxation of muscles. Thus, bone provides the framework for muscle- induced movement, whether in the context of running, flying, ventilating or masticating.

Muscles must have a rigid skeleton for the forces to act against. In addition, bones of the skull and ribcage provide a strong layer of protection for vital internal organs. Finally, bones provide important resources, such as red blood cells and calcium and phosphate ions, to the rest of the body. The evolution of bone tissue has also clearly played a role in the overall body size that vertebrates have achieved, allowing animals to attain amazing size in the terrestrial (e.g., sauropod dinosaurs), aquatic (e.g., blue whale) and aerial environments (e.g., giant pterosaurs such as Quetzalcoatlus). This size range greatly eclipses the largest terrestrial and flighted arthropods, another animal group in which individuals can reach relatively large size. Increased size has also allowed for enhanced cephalization and the consequential increase in brain size that may be at the root of complex behaviors exhibited by relatively few clades where animals have achieved notable multi-cellular size. Clearly, the skeleton is vital for vertebrates and many aspects of the biology of bone are important for understanding the relationship between skeletal form and function.

Bone is a mineralized connective tissue that is characterized by its stiffness and strength. Bone consists mainly of the protein collagen, hydroxyapatite mineral crystals, and water (Martin et al., 1998; Boskey, 2001; Currey, 2002). Bone first begins to develop in the embryo and can be produced by one of two different methods (Marks and Hermey,

1996; Eroschenko, 2005). Endochondral ossification involves bony replacement of an existing cartilage model. In contrast, during intramembranous ossification, the bone develops from the connective tissue mesenchyme. Although bones can be formed by these two different methods, they have roughly similar histological organization

(Eroschenko, 2005). Several different types of cells are found within bone tissue.

Osteoblasts are responsible for the formation of bone (Marks and Hermey, 1996; Martin et al., 1998; Currey, 2002; Eroschenko, 2005). These cells secrete osteoid, which is later mineralized. Osteoblasts that have become surrounded by bone matrix are termed osteocytes. Osteocytes communicate with other cells through processes that extend through an extensive canalicular network (Marks and Hermey, 1996; Nijweide et al.,

1996; Martin et al., 1998; Eroschenko, 2005). Such processes maintain contact with surrounding osteocytes through gap junctions, which allow the passage of small molecules between cells (Nijweide et al., 1996; Eroschenko, 2005). Osteoclasts are cells that resorb bone tissue (Marks and Hermey, 1996; Martin et al., 1998; Currey, 2002;

Eroschenko, 2005). These three main bone cells work together to both model bone during its initial formation and to remodel bone later in life. Modeling occurs during growth of a bone, when it is increasing in both length and diameter. Specifically, bone material is added and removed to shape the geometry of the bone (Martin et al., 1998). During this

19 process, the actions of the osteoblasts and osteoclasts are independent from one another

(Martin et al., 1998). In contrast, remodeling occurs in response to damage (i.e., microcracks) or in response to loading conditions experienced by the bone (Martin et al.,

1998; Currey, 2002). The remodeling is mediated by osteocytes embedded in the bone tissue that become activated in response to strain and/or damage (Lanyon, 1993).

Remodeling generally does not affect the overall size or shape of the bone, but rather removes and replaces bone in specific places through the coordinated actions of osteoclasts and osteoblasts (Martin et al., 1998). Primary bone is composed of tissue that has not been remodeled and retains its primary structure. Remodeling results in secondary bone, consisting of secondary (or Haversion) osteons.

Bone also provides important life-history information by recording the growth of an organism and providing evidence of specific life stages. For example, the long bones of some female birds can change rapidly in response to hormonal changes associated with ovulation and egg laying (e.g., Bloom et al., 1941; Dacke et al, 1993). A thick layer of medullary bone is deposited that acts as a reservoir of calcium for the shelling of eggs.

The slow regular deposition of bone throughout the life of an organism, however, is an attribute that has been used in both extant and extinct taxa for documenting growth strategies. Lines of arrested growth (LAGs), which are periodically laid down in bone, are commonly used to assess life-history traits such as age at sexual maturity, growth rate, and longevity (Peabody, 1961). LAGs depend on environmental rhythms and are present in both extant (e.g., amphibians, birds, and mammals) and extinct (e.g., dinosaurs) vertebrates (e.g., Hemelaar and Van Gelder, 1979; Horner et al., 1999; Padian

20 et al., 2001; Castanet et al., 2004). Other aspects of the microstructure of bone, such as the bone tissue type present, can be used as indicators of bone growth rate. This has been shown recently in a study of king penguins, a species of bird that exhibits a unique growth pattern specific to the environment in which it lives (de Margerie et al., 2004). In this species, periods of rapid growth are recognizable in the bone as a distinct tissue type

(de Margerie et al., 2004). Another method used predominantly in mammals is age assessment based on the number of secondary osteons present. Secondary osteon number has been found to accurately predict age at death in humans (e.g., Thompson and Galvin,

1983; Ericksen, 1991; Lynnerup et al., 2006; Paine and Brenton, 2006). Long-lived mammals, such as humans, become susceptible to certain skeletal diseases in the final stage of skeletal ontogeny. Osteoporosis, in particular, results from both genetic and environmental factors and can lead to a decrease in bone density and strength (Gross et al., 1996).

To understand how and why bone responds to load in the way it does, an explanation of some basic materials properties concepts is necessary. The material properties of bone are affected by the porosity, mineralization, organization of the collagen fibers, presence of fatigue microdamage, and presence of secondary osteons

(Martin et al., 1998; Currey, 2002). The properties of bone material itself are determined by conducting a load-extension test. Given amounts of load are applied to the tissue while documenting the amount of resulting deformation (extension). The load and extension during a mechanical test initially exhibit a linear relationship. This is described as the elastic region, as the bone can return to its original shape once the load is removed.

Eventually, however, a yield point is reached whereby bone behaves plastically and will not return to its original shape. Finally, with the continued application of load, fracture or failure of the bone will occur (An et al., 2000; Frankel and Nordin, 2001).

The mechanical properties of the whole bone, however, depend not just on the properties of the bone material itself, but also on the distribution of the material (or bone shape) (Currey, 2002). Therefore, the standardized variables stress and strain are typically used in the context of bone loading (Einhorn, 1996; An et al., 2000; Frankel and Nordin,

2001). Stress is the amount of load per unit area and the strain characterizes the change in shape of the object being loaded. In addition, Young’s modulus is a standardized measure of the stiffness of the bone (Frankel and Nordin, 2001). Combinations of load create different stresses on bones that result in specific deformations. For example, forces directed toward each other result in compressional stress, whereas forces pulling away from each other result in tensional stress (Frankel and Nordin, 2001). These are examples of axial stresses that can occur on a bone. Bending of a bone will produce a combination of compression and tension on different surfaces of the bone. In addition, a twisting

(torsional) load will create a shear stress on a bone (Frankel and Nordin, 2001). During normal movement, most skeletal elements may experience a combination of these different types of stresses, but some bones likely experience predominant loading regimes in a habitual activities such as terrestrial quadrupedal locomotion and mastication.

Bones exhibit amazing adaptability, especially during growth, and are capable of responding to external loads. The ability of bone tissue to change its structure in response to dynamic mechanical loading has traditionally been defined as Wolff’s Law, or now

22 more commonly, as bone functional adaptation (Wolff, 1892; Ruff et al., 2006).

Experimental data have shown that increasing or changing loads on limb elements will result in new bone growth and provides the most solid evidence for bone functional adaptation. For example, such experiments have shown that when one of the paired forelimb elements (i.e., radius or ulna) is removed from a terrestrial quadruped, extensive growth results in the remaining bone in the plane of the removed element after experimental animals are allowed to resume locomotor activities (Goodship et al., 1979;

Lanyon et al., 1982). In addition, by functionally isolating a limb element, it has been shown that removal of load will also elicit a response, resulting in increased remodeling and bone resorption (Lanyon and Rubin, 1984; Rubin and Lanyon, 1984). The bone isolation technique developed by Rubin and Lanyon (1984) eliminates the natural loading of the bone and is able to document remodeling in response to artificial loads or the lack of loading. The functionally isolated turkey ulna retains its nutritive connection and after loading exhibits a substantial increase in new bone formation and when left unloaded shows extensive resorption. Rubin and Lanyon have also demonstrated that bone responds maximally to dynamic loading, with minimal response to continuous static loads (Lanyon and Rubin 1984). Indeed, evidence suggests that even very low magnitude loads, as long as they are high frequency, will elicit response in bone (Rubin et al., 2001). Noninvasive studies also illustrated the local response of bone to loading.

Umemura et al. (1997) trained rats to jump between 5 and 100 times per day and found significant increases in tibial and femoral mass between jumpers and the controls. Only five jumps per day were sufficient to increase the mass and resistance to breaking of the

23 bones; additional jumps per day only showed slight improvement. In addition, humans experiencing long periods of weightlessness (i.e., astronauts) or immobilization exhibit significant bone loss (e.g., Minaire et al, 1974; Whedon, 1984) Fortunately, resistance exercise training has been found to be effective in reducing the amount of bone mineral density lost during disuse in humans (Shackelford et al., 2004). These experimental results clearly suggest that skeletal bone responds to mechanical loading situations.

However, it is important to note that responses attributed to functional adaptation may also be due to the response of bone to other stimuli such as healing, or simply to the plasticity of a bone during growth (Bertram and Swartz, 1991).

Although experimental manipulation of bone through disuse, bone isolation, and overloading has indeed documented local response in bone growth and/or resorption, the overall form of bone is considered to be genetically determined (Currey, 2002). The shape and structure of bones today is the result of natural selection acting on past organisms. The first evidence of skeletal remains in the fossil record is from the early

Cambrian period (~550 million years ago). At that time, animals began to form skeletons from many materials, including calcium phosphates, calcium carbonate, and silica (Carter and Beaupré, 2001). Mineralized tissues are preferentially preserved in the fossil record and thus provide our main evidence for the appearance and evolution of bone throughout the geological history of Earth (Carter and Beaupré, 2001). The agnathans (jawless fishes) from the late Cambrian were the earliest chordates that possessed bone, specifically dermal bone (Northcutt and Gans, 1983; Benton, 1997; Shu et al., 2003). By the Devonian (~415 million years ago), endochondral ossification was the primary

24 mechanism for long bone development, whereas intramembranous ossification formed the skull and girdle bones in most tetrapods (Carter and Beaupré, 2001). In modern organisms, much of the main structure of the skeleton has developed before significant external loads from locomotion are applied to it (Gould, 2002), suggesting that at least some aspects of skeletal form result from Darwinian adaptation by natural selection, The size and shape of bones, as well as processes, protuberances, flanges, tubercles, and tuberosities on bones, can offer some indication of the magnitude and/or direction of load and/or muscle-action on bones, allowing functional interpretation of fossils (e.g., Jungers and Minns, 1979; Van Valkenburgh, 1987; Fleagle and Meldrum, 1988; Van

Valkenburgh et al., 2004).

In summary, the skeleton provides a unique opportunity to investigate the interaction between functional and Darwinian adaptation. In this dissertation, I examine the form of the forelimb skeleton of birds at several different anatomical levels and investigate how this relates to function, or flight behavior. In general, the avian wing is a tetrapod forelimb that has been exapted for flight. There are more than 10,000 species of extant birds spread throughout in virtually all known environments around the globe. The majority of species occupy the energetically expensive aerial environment, a fact that underscores the necessity for economy of design. The external shape of a bird’s wing is directly tied to the way it flies and it is assumed that the underlying skeletal support of the wing has also been modified to meet the demands of flight. However, this basic assumption has not been explicitly tested in a comparative context, nor have putative relationships among different levels of skeletal organization (e.g. from whole bone shape

25 to the structural organization within bone tissue) and inferred loading regimes been examined in extant birds. This dissertation will specifically examine both functional adaptation (e.g., the relationship between bone shape/structure and resistance to load) and

Darwinian adaptation (e.g., the clade-wide analyses of the relationship between bone shape and flight style) in several groups of extant birds.

Literature Cited

An, Y.H., Barfield, W.R., Draughn, R.A. 2000. Basic concepts of mechanical property

measurement and bone biomechanics. In: An, Y.H., Draughn, R.A. (Eds.)

Mechanical Testing of Bone and the Bone-Implant Interface. Boca Raton, FL:

CRC Press.

Benton, M.J. 1997. Vertebrate Paleontology, 2nd edition. London, UK: Chapman & Hall.

Bertram, J.E.A., Swartz, S.M. 1991. The ‘law of bone transformation’: a case of crying

Wolff? Biological Reviews 66: 245-273.

Bloom, W., Bloom, M.A., McLean, F.C. 1941. Calcification and ossification. Medullary

bone changes in the reproductive cycle of female pigeons. Anatomical Record 81:

443-475.

Boskey, A.L. 2001. Bone mineralization. In: Cowin, S.C. (Ed.) Bone Mechanics

Handbook, 2nd edition. Boca Raton, FL: CRC Press.

Carter, D.R., Beaupré, G.S. 2001. Skeletal Function and Form: Mechanobiology of

skeletal development, aging, and regeneration. New York, NY: Cambridge

University Press.

Castanet, J., Croci, S., Aujard, F., Perret, M., Cubo, J., de Margerie, E. 2004. Lines of

arrested growth in bone and age of estimation in a small primate: Microcebus

murinus. Journal of Zoology, London 263: 31-39.

Currey, J.K. 2002. Bones: structure and mechanics. Princeton, NJ: Princeton University

Press.

Dacke, C.G., Arkle, S., Cook, D.J., Wormstone, I.M., Jones, S., Zaidi, M., Bascal, Z.A.

1993. Medullary bone and avian calcium regulation. Journal of Experimental

Biology 184: 63-88.

Einhorn, T.A. 1996. Biomechanics of bone. In: Bilezikian, J.P., Raisz, L.G., Rodan, G.A.

(Eds.) Principles of Bone Biology. San Diego, CA: Academic Press.

Ericksen, M.F. 1991. Histologic estimation of age at death using the anterior cortex of the

femur. American Journal of Physical Anthropology 84: 171-179.

Eroschenko, V.P. 2005. Atlas of Histology with Functional Correlations, 10th edition.

Baltimore, MD: Lippincott Williams & Williams.

Fleagle, J.G., Meldrum, D.J. 1988. Locomotor behavior and skeletal morphology of two

sympatric pitheciine monkeys, Pithecia pithecia and Chiropotes satanas.

American Journal of Primatology 16: 227-249.

Frankel, V.H., Nordin, M. 2001. Biomechanics of Bone. In: Nordin, M., Frankel, V.H.

(Eds.) Basic Biomechanics of the Musculoskeletal System. Baltimore, MD:

Lippincott Williams & Williams.

Goodship, A.E., Lanyon, L.E., MacFie, H. 1979. Functional adaptation of bone to

increased stress. Journal of Bone and Joint Surgery 61-A: 539-546.

Gould, S.J. 2002. The Structure of Evolutionary Theory. Cambridge, MA: The Belknap

Press of Harvard University Press.

Gross, C., Eccleshall, T.R., Feldman, D. 1996. Vitamin D receptor gene alleles and

osteoporosis. In: Bilezikian, J.P., Raisz, L.G., Rodan, G.A. (Eds.) Principles of

Bone Biology. San Diego, CA: Academic Press.

Hemelaar, A.S.M., Van Gelder, J.J. 1979. Annual growth rings in phalanges of Bufo bufo

(Anura, Amphibia) from the Netherlands and their use in age determination.

Netherlands Journal of Zoology 30: 129-135.

Horner, J.R., de Ricqlès, A., Padian, K. 1999. Variation in dinosaur skeletochronology

indicators: implications for age assessment and physiology. Paleobiology 25: 295-

304.

Jungers, W.L., Minns, R.J. 1979. Computed tomography and biomechanical analysis of

fossil long bones. American Journal of Physical Anthropology 50: 285-290.

Lanyon, L.E. 1993. Osteocytes, strain detection, bone modeling and remodeling.

Calcified Tissue International 53: S102–S107.

Lanyon, L.E., Rubin, C.T. 1984. Static vs. dynamic loads as an influence on bone

remodeling. Journal of Biomechanics 17(12): 897-905.

Lanyon, L.E., Goodship, A.E., Pye, C.J., MacFie, J.H. 1982. Mechanically adaptive bone

remodeling. Journal of Biomechanics 15: 141-154.

Lynnerup, N., Frohlich, B., Thomsen, J.L. 2006. Assessment of age at death by

microscopy: Unbiased quantification of secondary osteons in femoral cross

sections. Forensic Science International 159S: S100-S103. de Margerie, E., Robin, J.P., Verrier, D., Cubo, J., Groscolas, R., Castanet, J. 2004.

Assessing a relationship between bone microstructure and growth rate: a

fluorescent labeling study in the king penguin chick (Aptenodytes patagonicus).

Journal of Experimental Biology 207: 869-879.

Marks Jr., S.C., Hermey, D.C. 1996. The structure and development of bone. In:

Bilezikian, J.P., Raisz, L.G., Rodan, G.A. (Eds.) Principles of Bone Biology. San

Diego, CA: Academic Press.

Martin, R.B., Burr, D.B., Sharkey, N.A. 1998. Skeletal Tissue Mechanics. New York,

NY: Springer-Verlag.

Minaire, P., Meunier, P., Edouard, C., Bernard, J., Courpron, P., Bourret, J. 1974.

Quantitative histological data on disuse osteoporosis. Calcified Tissue Research

17: 57-73.

Nijweide, P.J., Burger, E.H., Nulend, J.K., Van der Plas, A. 1996. The osteocyte. In:

Bilezikian, J.P., Raisz, L.G., Rodan, G.A. (Eds.) Principles of Bone Biology. San

Diego, CA: Academic Press.

Northcutt, R.G., Gans, C. 1983. The genesis of neural crest and epidermal placodes:

a reinterpretation of vertebrate origins. The Quarterly Review of Biology 58: 1-

28.

Padian, K., de Ricqlès, A.J., Horner, J.R. 2001. Dinosaurian growth rates and bird

origins. Nature 412: 405-408.

Paine, R.R., Brenton, B.P. 2006. Dietary heath does affect histological age assessment: an

evaluation of the Stout and Paine (1992) age estimation equation using secondary

osteons from the rib. Journal of Forensic Science 51: 489-492.

Peabody, F.E. 1961. Annual growth zones in living and fossil vertebrates. Journal of

Morphology 108: 11-62.

Rubin, C.T., Lanyon, L.E. 1984. Regulation of bone formation by applied dynamic loads.

The Journal of Bone and Joint Surgery 66-A: 397-402.

Rubin, C.T., Turner, A.S., Bain, S., Mallinckrodt, C., McLeod, K. 2001. Low mechanical

signals strengthen long bones. Nature 412: 603-604.

Ruff, C., Holt, B., Trinkaus, E. 2006. Who’s afraid of the big bad Wolff?: “Wolff’s Law”

and bone functional adaptation. American Journal of Physical Anthropology 129:

484-498.

Shackelford, L.C., LeBlanc, A.D., Driscoll, T.B., Evans, H.J., Rianon, N.J., Smith, S.M.,

Spector, E., Feeback, D.L., Lai, D. 2004. Resistance exercise as a countermeasure

to disuse-induced bone loss. Journal of Applied Physiology 97: 119-129.

Shu, D.-G., Morris, S.C., Han, J., Zhang, Z.-F., Yasui, K., Janvier, P., Chen, L., Zhang,

X.-L., Liu, J.-N., Li, Y., Liu, H.-Q. 2002. Head and backbone of the Early

Cambrian vertebrate Haikouichthys. Nature 421: 526-529.

Thompson, D.D., Galvin, C.A. 1983. Estimation of age at death by tibial osteon

remodeling in an autopsy series. Forensic Science International 22: 203-211.

Umemura, Y., Ishiko, T., Yamauchi, T., Kurono, M., Mashiko, S. 1997. Five jumps per

day increase bone mass and breaking force in rats. Journal of Bone and Mineral

Research 12, 1480-1485.

Van Valkenburgh, B. 1987. Skeletal indicators of locomotor behavior in living and

extinct carnivores. Journal of Vertebrate Paleontology 7: 162-182.

Van Valkenburgh, B., Wang, X., Damuth, J. 2004. Cope’s rule, hypercarnivory, and

extinction in North American canids. Science 306: 101-104.

Whedon, G.D. 1984. Disuse osteoporosis: physiological aspects. Calcified Tissue

International 36: S146-S150.

Wolff, J. 1892. Das gesetz der transformation der knochen. A. Hirshwald, Berlin.

CHAPTER 1: FORELIMB SKELETAL MORPHOLOGY AND FLIGHT MODE

EVOLUTION IN PELECANIFORM BIRDS

Abstract

The total length and mid-shaft diameters of wing elements of 50 species of pelecaniform birds were examined to investigate how forelimb skeletal morphology varies with body size and flight mode within this group. Pelecaniforms were assigned to flight mode categories based on primary habitual behaviors (soar, flap-glide, continuous flap). Allometric and discriminant function analyses were conducted on wing element variables in both historical (using independent contrasts) and ahistorical contexts. Results of this study indicate that when phylogenetic relationships are taken into account, only the length of the ulna scaled with positive allometry, whereas all other variables exhibit isometry. These results differ from the ahistorical allometric analysis. Discriminant function analysis significantly separated the flight mode groups (Wilk’s λ = 0.002, p <

0.00001), with only six individuals from two species (from total n = 284) misclassified.

Results of historical canonical variates analysis support the ahistorical DFA analysis and identified two carpometacarpal (CMC) variables as important for separating the flight mode groups: dorsoventral CMC diameter and total CMC length. The CMC is that portion of the forelimb skeleton that serves as the attachment point for the primary flight feathers, and thus, that portion of the airfoil surface that mediates detailed flight control.

Its morphology, more than any other element, reflects differences in flight mode in pelecaniforms. Results of this study indicate that, in pelecaniforms, wing bones generally exhibit isometry (with exception of the ulna) and do possess specific morphologies

33 reflective of the demands associated with different types of aerial locomotor specialization.

Introduction

Wing shape is related to many aspects of avian biology, including signaling displays, feeding behavior and/or foraging mode, and locomotor potential. Traditionally, two parameters, wing loading (body weight/wing area) and aspect ratio (wing span2/wing area), have been used to describe whole wing size and shape, and have been shown to correlate with different flight behaviors in birds (Savile, 1957; Warham, 1977; Norberg,

1985; 1995). More recently, new methods such as thin-plate spline relative warp analysis

(Brewer and Hertel, 2007) have been introduced to more thoroughly quantify whole wing shape and investigate its relationship with specific flight behaviors. It is less clear, however, how the underlying skeletal structure has been modified to accommodate different modes of flight. With ~10,000 species living in virtually all known environments, birds represent a natural experiment for examining the relationship between form and function of the vertebrate skeleton.

The majority of bird species occupy the energetically expensive aerial environment. Flying animals must produce enough lift to support their weight, and generate enough thrust to counteract the drag imposed by viscosity. This underscores the necessity for economy of design, a theme repeated numerous times over the course of avian evolutionary history. Many birds must continuously flap their wings to create enough lift to stay aloft. Whereas continuous level flapping flight is an effective form of locomotion, some birds are able to utilize energy-saving strategies such as gliding and

34 soaring to minimize the cost of flight (Norberg, 1985; Rayner, 1988; Rayner et al., 2001).

During gliding, a bird is not actively flapping and is therefore either losing altitude or speed. In contrast, soaring is a type of flight that allows a bird to maintain or even increase altitude without flapping. Soaring birds use moving air currents to gain potential or kinetic energy, making this the least energetically expensive type of flight (Norberg,

1985).

Several researchers have investigated how skeletal morphology relates to whole wing shape. General scaling studies indicate that individual wing bone and total wing length are positively allometric (Bochenski and Bochenski, 1993; Olmos et al., 1996;

Nudds, 2007), but such studies have not been examined with reference to different flight modes. The Brachial Index (BI), a ratio of humerus to ulna length, has been shown to vary with both whole wing shape and wing kinematics (Rayner and Dyke, 2003; Nudds et al., 2004; Nudds et al., 2007). Specifically, Nudds and colleagues (2004) identified significant differences in BI at all taxonomic levels, thereby suggesting the existence of an underlying functional signal. Other researchers have also incorporated the length of the carpometacarpus and compared the proportions of all three main limb elements in birds with those of theropod dinosaurs, pterosaurs, and bats (Middleton and Gatesy,

2000; McGowan and Dyke, 2007; Gatesy and Middleton, 2007). Such studies typically include a wide range of species and have only recently begun to take the effects of phylogeny into account (e.g., Nudds, 2007; Nudds et al., 2007). Moreover, such studies are generally restricted to an examination of element length, and thus may fail to fully characterize whole bone morphology (e.g., bone diameter, robusticity, etc.). The

35 objectives of this study are to (1) examine the scaling relationships of individual forelimb elements and (2) investigate how the morphology of the forelimb skeleton varies among flight modes within pelecaniform birds, through the use of ahistorical and historical univariate and multivariate analyses.

Materials and Methods

Pelecaniformes are a wide-ranging group of marine birds consisting of: the

Pelecanidae (pelicans), Sulidae (gannets and boobies), Phaethonidae (tropicbirds),

Phalacrocoracidae (cormorants), Fregatidae (frigatebirds), and Anhingidae (anhingas and darters). Pelecaniforms range in size from the small tropicbirds (mean body mass: 334 g) to the large pelicans (mean body mass: 9000 g) (Dunning 1993) (Table 1–1). Although the monophyly and specific ingroup relationships of pelecaniforms have been in a state of flux (e.g., Sibley and Ahlquist, 1990; Van Tuinen et al., 2001; Fain and Houde, 2004;

Hackett et al., 2008), for the comparative analysis conducted in this study, a monophyletic composite phylogeny (Figure 1-1) of the group was assembled based on the following studies: Sibley and Ahlquist (1990), Siegel-Causey (1988), Friesen and

Anderson (1997), Kennedy and Spencer (2004), and Livezy and Zusi (2007). Four species (Pelecanus crispus, Pelecanus philippensis, Phalacrocorax carunculatus, and

Phalacrocorax albiventer) sampled as part of this study were not included in any previous phylogenetic analysis. As such, these species were intercalated as soft polytomies into the composite phylogeny near species inferred to be closely related based on natural history information (Purvis and Garland, 1993; Nelson, 2005).

Genera within pelecaniforms were initially assigned to one of three primary flight mode groups based on information on flight behavior obtained from the literature: Soar

(including Pelecanus, Fregata, Morus, and Anhinga), Flap-glide (Sula and Phaethon), and Flap (Phalacrocorax) (Table 1–1) (e.g., Ashmole, 1971; Pennycuick, 1972; Schnell,

1974; Nelson, 1978; Pennycuick, 1983; Johnsgard, 1993; Hertel and Ballance, 1999;

Weimerskirch et al., 2003). Taxa were included in a flight mode category if they habitually utilize that flight behavior. Unfortunately, quantified flight behavior data (i.e. time budgets for different flight modes) were unavailable for these taxa; therefore observational data were used as a best estimate of primary flight mode. The flightless cormorant (Phalacrocorax harrisi) was excluded from all functional analyses.

Significantly, the soaring group did not meet the assumption of multivariate normality required for some analyses. Although each of the species initially assigned to this category use various types of soaring during locomotion, there are apparent differences in other aspects of their behavior, such as foraging technique that may impact morphology of the flight apparatus. This is in contrast to the Flap and Flap-glide groups, which are more uniform in foraging behavior. For example, pelicans and frigatebirds both utilize thermal soaring (soaring in rising air currents), but whereas pelicans largely forage on the surface of shallow water, frigatebirds are dedicated marine birds that feed on the wing. In addition, two of the groups in the Soar category utilize underwater foraging techniques. Anhingas are specialized for stealthy underwater prey capture and gannets perform spectacular plunge-dives and may even use their wings to extend the depth of their dive (Bourne, 1976; Johnsgard, 1993; Garthe et al., 2000; Nelson, 2005). Therefore,

37 based on these other relevant behaviors, the soaring group was divided into three groups:

Soar1 (Pelecanus), Soar2 (Anhinga and Morus), and Soar3 (Fregata) for all multivariate analyses.

Measurements were taken on 321 skeletal specimens from 50 species (mean of 6 individuals per species) of pelecaniforms, representing ~76% of species diversity in the group, and three outgroup species (Balaeniceps rex, Cochlearius cochlearius, Scopus umbretta) from collections at the American Museum of Natural History (AMNH), the

Carnegie Museum of Natural History (CM), the Field Museum of Natural History

(FMNH), the National Museum of Natural History (NMNH), and the Ohio University

Vertebrate Collections (OUVC). Nine forelimb measurements were obtained for each specimen: total bone length, craniocaudal and dorsoventral mid-shaft diameters of the humerus, ulna, and carpometacarpus (CMC). Anatomical orientation of wing bones

(craniocaudal vs. dorsoventral) was determined based on a fully extended wing position.

Measurements were taken with digital calipers (Mitutoyo Digimatic calipers).

Anatomical reference points and orientations used standardized nomenclature as outlined in Nomina Anatomica Avium (Baumel and Witmer, 1993).

A geometric mean (GMb) was established as a proxy for body size from five

additional measurements: femur length, synsacral length, sternal length, sternal width,

and height of sternal keel (Mosimann, 1970; Mosimann and James, 1979; Niemi, 1985).

The species means of the calculated GMb were significantly correlated with the general

species means of body mass from Dunning Jr. (1993) (p < 0.0001, r2 = 0.91); therefore

the GMb is appropriate for use in regression analyses. To remove the effects of size in

38 multivariate analyses, the log10 transformed body size estimate (GMb) was subtracted

from each log10 transformed variable. However, exploratory PCA analysis indicated all

variables remained highly loaded on PC1, suggesting that whereas GMb represents total

body size, another size factor (presumably total wing size) was not adequately removed.

Therefore an additional geometric mean (GMw) was calculated from the nine wing element variables collected and used instead of GMb to control for the effect of total wing

size in multivariate analyses. Exploratory PCA analysis indicated that GMw effectively

removed wing size.

All variables were tested for phylogenetic signal using two methods: the Test for

Serial Independence (TFSI) in Phylogenetic Independence 2.0 (Abouheif, 1999) and by randomly re-shuffling the data across the tips of the tree 10,000 times and comparing the results to the measured squared length (number of character steps) in the given tree

(Blomberg et al., 2003; Laurin et al., 2004; Kriloff et al., 2008). The second method was performed in Mesquite 2.0 (Maddison and Maddison, 2007). Significant phylogenetic autocorrelation was present in all variables as determined by both methods (p < 0.0001) and justified the calculation of phylogenetic independent contrasts (Felsenstein, 1985;

Harvey and Pagel, 1991; Garland et al., 1992). Independent contrasts were calculated using the PDAP package (Midford et al., 2005) in Mesquite 2.0 (Maddison and

Maddison, 2007). As the phylogeny used in this study represents a composite of multiple analyses based on both molecular and morphological data, branch lengths were arbitrarily set using Pagel’s (1992) method (e.g., Lavin et al., 2008).

Two sets of analyses (allometric and discriminant function) were conducted on wing element variables and each set of analyses was completed in both historical and ahistorical contexts. First, an allometric analysis of individual wing elements was conducted using a Reduced Major Axis (RMA) regression model for each log10

transformed measurement, using log10GMb as a proxy for body size. RMA lines were

expressed with equations in the form of:

log(y) = b * log(x) + a

where y = variable of interest, a = y-intercept, b = slope, and x = body size. Under a geometric similarity model, isometry is indicated by slopes not significantly different from 1, positive allometry by slopes significantly greater than 1, and negative allometry by slopes significantly less than 1 (McMahon, 1975; Schmidt-Nielsen, 1984). The null hypothesis of b = 1 (isometry) was tested with a t-test (d.f. = n – 2, α = 0.05):

t = (b – 1)/SEb

where b = slope and SEb = standard error of slope (Zar, 1999). Ahistorical RMA was

done using RMA: Software for Reduced Major Axis Regression (Bohonak, 2002).

Phylogenetically-informed RMA analyses were completed using the PDAP package

(Midford et al., 2005) in Mesquite 2.0 (Maddison and Maddison, 2007). For slopes of

contrasts, the standard error of the OLS slope was used as the standard error of the RMA

slope following Nudds (2007).

Second, the utility of the wing element variables to predict known flight mode

groups was tested using Discriminant Function Analysis (DFA) and Classification and

Regression Tree Analysis (CART) in JMP version 7.0.1. DFA and CART were

40 performed on size-corrected variables (log10 transformed wing generated GMw for each

individual subtracted from each log10 transformed variable); prior probabilities

proportional to group size were specified. Historical DFA (canonical variates analysis)

for comparison was completed using the Rhetenor package (Dyreson and Maddison,

2001) in Mesquite 2.0 (Maddison and Maddison, 2007). The nexus file used for

comparative analyses is available from the author upon request.

Results

Allometric Analyses

Pelecaniform wing element lengths and diameters exhibit more than one

allometric pattern (Table 1–2, Figure 1–2). All pelecaniform and outgroup species were

included (n = 321 individuals); species means were generated (n = 53). Using an

ahistorical approach, all three humeral variables - total humeral length (HumL),

dorsoventral humeral diameter (HDdv), craniocaudal humeral diameter (HDcc) - as well

as total ulnar length (UlnL) and craniocaudal CMC diameter (CDcc) were positively

allometric. Four variables were isometric: both dorsoventral (UDdv) and craniocaudal

ulnar diameter (UDcc), total CMC length (CmcL), and dorsoventral CMC diameter

(CDdv). However, there was a significant amount of phylogenetic autocorrelation in

these variables (p < 0.0001). Therefore, a scaling analysis using independent contrasts

may better characterize the evolutionary pattern. Independent contrasts of species means

indicate that only one variable was “evolutionary” positively allometric: UlnL. All other

variables exhibited isometric relationships.

Multivariate Analyses

To investigate whether wing element variables can be used to predict flight mode, each pelecaniform genus was assigned to a flight mode group based on preferred locomotor behavior obtained from the literature. The Discriminant Function Analysis

(Figure 1–3A) significantly separated the five flight mode groups (Wilk’s λ = 0.002, p <

0.00001). Six individuals (from total n = 284) were misclassified. Five individuals of

Pelecanus (1 P. erythrorhynchos, 1 P. occidentalis, 3 P. conspicillatus) and one individual of Phaethon rubricauda were misclassified as members of the Flap group.

Additional individuals of each of these four species included in the analysis were classified correctly. The first discriminant function described 70.6% of the variance and was positively correlated with total ulnar length and negatively correlated with humeral diameter. An additional 23.7% of the morphological variance was associated with the second discriminant function, which was positively correlated with dorsoventral humeral diameter and negatively correlated with craniocaudal CMC diameter.

To validate the results of the DFA, a classification and regression tree (CART) analysis was performed. CART is a non-parametric method of partitioning the morphological data in a way as to optimize the splits between groups. Six splits were required to adequately partition the data (r2 = 0.783). Variables used to partition the data

were similar to those with high loadings in the DFA. Splits were based on dorsoventral

humeral diameter, ulnar length, humeral length, craniocaudal ulnar diameter, and two additional splits using humeral length. The flight mode groups remained consistent with

42 those recovered in the DFA, with the exception of the Flap-Glide group, which was split into three separate groups. A k-fold cross-validation was performed (r2 = 0.78).

As phylogenetic autocorrelation was present in the variables used in the DFA and

CART, a canonical variates analysis was performed on variable species means in the

Rhetenor package of Mesquite (Figure 1–3B). The historical canonical variates analysis indicated an identical separation of flight mode groups in canonical space as the ahistorical DFA analysis. However, in contrast to the previously described analyses,

dorsoventral CMC diameter was highly negatively loaded (~0.8) on the first function,

which described 67.45% of the morphological variance. Total CMC length was highly

positively loaded on the second function (~0.7), which described an additional 27.47% of the morphological variance. Along the first function, the Soar2 group (Anhinga and

Morus), Flap-Glide group (Phaethon and Sula), and Soar3 group (Fregata) all had a

CMC that was short relative to total wing size. Soar2 had a relatively large dorsoventral

CMC diameter, whereas the Soar3 had a relatively small dorsoventral CMC diameter.

The Flap (Phalacrocorax) and Soar1 (Pelecanus) groups had long CMC relative to total

wing size, with Pelecanus having slightly smaller relative dorsoventral CMC diameter. In the historical analysis, one species of pelican, P. conspicillatus, clustered with the Flap group, which is consistent with the results of the ahistorical DFA analysis.

Discussion

Allometric Aspects of the Pelecaniform Forelimb Skeleton

The allometric analyses performed here indicate that, when phylogenic

relationships are taken into account, only ulnar length exhibits positive allometry (Table

1–2). By comparison, humeral length, carpometacarpal length, and all wing element diameters exhibit isometry. Thus, the length of the ulna is increasing faster than would be expected with respect to overall body size. Previous analyses among a broad, synoptic survey of neognath birds have found proximal wing bone lengths to be positively allometric (e.g., Olmos et al., 1996; Nudds, 2007). The result of this detailed clade-level analysis is consistent with and supports results for the ulna, but not for the most proximal element, the humerus. In addition, no previous studies have examined the scaling relationships of the CMC or wing-bone diameters. The results of this study indicate that the distal element and all diameters exhibit isometry. Thus, in contrast to previously- conducted survey work (e.g., across multiple avian orders), positive allometry of wing bones is not as common to all elements or all dimensions of individual elements, suggesting that lower-level sampling-dense clade analyses are critical for detailed studies of the relationship between form and function. The notable differences between the ahistorical and historical RMA analyses also emphasize the importance of conducting the comparative analysis in an explicit phylogeny framework (e.g., Harvey and Pagel, 1991).

RMA analyses performed included all pelecaniform species and did not separate them into functional groupings (i.e., flight mode categories). However, qualitative functional interpretation of these results is possible. Species that utilize energy saving techniques such as soaring and gliding (Pelecanus, Fregata, Anhinga, Morus, Sula,

Phaethon) tend to exhibit wing element lengths and diameters that are larger than expected for body size; i.e., they fall above the best fit line for all pelecaniforms (Figure

1–2). Alternately, species that flap continuously (Phalacrocorax) tend to fall below the

44 best fit line and exhibit lengths and diameters that are smaller than expected for all pelecaniforms. Previous studies have shown that whole wing morphology (i.e., feather contours) of birds that soar or glide tends to be longer (higher aspect ratio) and/or broader than the wings of flapping birds (Norberg, 1985; Norberg and Norberg, 1986; Rayner,

1988). This study indicates that the skeletal elements making up the support structure of the wing in pelecaniform birds are also relatively longer (in particular the ulna) and exhibit larger diameters, likely reflecting bone modifications that act in concert with feather elaborations related to flight mode.

Multivariate Analyses: A Map to Flight Mode Evolution in Pelecaniforms

The multivariate approaches used herein suggest that pelecaniform wing bone morphology indeed reflects the demands associated with different types of aerial locomotion. Specifically, external measurements (length and mid-shaft diameters) of the main wing elements provide good predictors of flight behavior within this group of birds.

Given the presence of significant phylogenetic autocorrelation within the variables examined, the results of the phylogenetically-informed canonical variates analysis

(Figure 1–3B) are here discussed in more detail. Specifically, the carpometacarpus

(CMC) is most important in maximizing the separation among flight style groups. The

CMC is the site of attachment for the primary flight feathers. All pelecaniforms have 11 primary feathers, with the first being vestigial, so differences in length of the CMC are not explained by a different number of feather attachment sites. However, there are large differences in the shape of the whole wing and especially of the wingtip (e.g., rounded,

45 pointed, slotted) among pelecaniforms (Table 1–1), which may be congruent with the observed osteological differences identified in this study.

Of the five flight mode groups included in this study, two are characterized by having a long CMC relative to wing size, the Flap (cormorants/shags) and Soar1

(pelicans). These two groups are also separated by the least amount of multivariate distance of all groups, even exhibiting the most overlap between groups (Figure 1–3B).

Five of six misclassified individuals in the entire model are Soar1 birds misclassified to the Flap group. Despite utilizing different primary flight modes, there is convergence in the relative length of the distal wing element and likewise a similarity in the shape of the whole wing of these birds. Both pelicans and cormorants/shags exhibit broad wings with rounded slotted wing tips. Slots between wing tip feathers help maintain lift and prevent stalling at slow speeds and also aid in fine control (Gill, 1995). Thus, the relatively long

CMC in these two groups may be related to the spacing of and increased control of the slotted primary feathers. Notably, the species most often misclassified in this analysis is the Australian pelican (Pelecanus conspicillatus). More than half (3 out of 5) of the individuals examined were misclassified into the Flap group in the DFA analysis. This species is confined to the Australian region and is widespread in the terrestrial wetlands, rivers, estuaries, and coastal waters (Johnsgard, 1993; Nelson, 2005), and thus, may be occupying a much more variable aerial environment. The repeated misclassification of this species suggests that it may be utilizing flapping behaviors convergent with cormorants, rather than the static soaring more typical of other pelicans. However, available observational accounts of this species do not report specific flight behavior

46 differences from other species of pelican. An interesting expansion of the current research would include a detailed behavioral study of birds in this group (in particular the

Australian pelican), specifically taking into account the amount of time engaged in different flight behaviors.

The soaring birds were initially split into three functional groups based on differences in foraging behavior. Results of these analyses indicate that the three functional groups also exhibit different CMC morphologies (Figure 1–3). The three Soar groups are isolated from each other based primarily on the first canonical function

(CDdv). Soar3 (frigatebirds) is isolated on the right side of the plot as having a small dorsoventral CMC diameter for its total wing size. Soar2 (anhingas and gannets) has a relatively large CDdv. Soar2 and Soar3 are also characterized by having a relatively short

CMC. Soar 1 (pelicans) and Soar3 (frigatebirds) both commonly utilize thermal soaring, however based on skeletal measures the two groups are widely separated in multivariate space and do not overlap on either axis indicating behavioral convergence. They also exhibit very different whole wing shapes. Frigatebirds have high aspect ratio wings, with the length generated largely from the forearm elements (ulna and radius) and long primary feathers (Nelson, 2005). Frigatebird wings are also highly pointed, whereas pelican wingtips are more rounded with slotting present. In addition, differences in the dorsoventral diameter of the CMC in these three soaring groups may reflect differences in the loading environment of the CMC. Elliptically shaped cross sections are typically interpreted to represent a higher resistance to bending in a preferred direction. Thus, differences in morphology of the midshaft cross section at least suggest that the bones

47 may be shaped to accommodate different primary loading regimes. The mid-shaft of the

CMC in the gannet/anhinga group is slightly elliptical, with the major axis of the ellipse oriented dorsoventrally; a circularity index (ratio of dorsoventral diameter to craniocaudal diameter) of external measurements for these species is 1.07. In contrast, the frigatebird

CMC is slightly elliptical with the major axis oriented craniocaudally and a circularity index of 0.94. Although both of these groups exhibit CMC circularity indices that are quite close to 1 (nearly circular), they are significantly different from each other (p <

0.0001) and in opposite planes. These results suggest the CMC in these two groups are being loaded in different ways, which in turn may be related to the angle by which the primary feathers are attached to the bone.

The FlapGlide group (tropicbirds and boobies) is characterized by a relatively short CMC with a dorsoventral diameter that is neither relatively small nor large; it overlaps on the second axis (CMC length) with Soar2 and Soar3, and on the first axis

(dorsoventral CMC diameter) with Soar1. The FlapGlide group is one of two functional groups in this study that consists of more than one distantly related genus overlapping in morphospace. This grouping, along with a parallel grouping of anhingas and gannets in the Soar2 category, supports the hypothesis that wing element morphology is related to a primary flight mode. Tropicbirds and boobies are relatively distantly related (within

Pelecaniformes) yet display convergence in both general whole wing shape (but see

Brewer and Hertel, 2007) and wing element morphology. Conversely, based on skeletal metrics, boobies and gannets, although members of the same family (Sulidae) are significantly separated in multivariate space. The CMC of gannets is relatively short and

48 exhibits a larger dorsoventral diameter than that of boobies. In addition to utilizing a different flight mode, gannets also differ from boobies in their ability to use underwater wing-propelled swimming (Bourne, 1976). Because water is more viscous than air, a wing moving through it necessarily encounters higher resistance, and thus, will incur higher loads on the forelimb skeleton (Bourne, 1976). The proximal element is usually the largest and most robust of the wing, and therefore would be predicted to make up most of the length of the wing in an underwater flapper. Shorter distal wing elements may then be advantageous for underwater flapping. This is consistent with interpretations of differences in humerus and ulna length (BI) in gannets and boobies by previous studies

(Bourne, 1976; Rayner and Dyke, 2003).

Conclusion

The high-resolution sampling of pelecaniform birds conducted herein has demonstrated that the morphology of individual forelimb elements exhibits wide variation, corresponding in turn to differences in body size, primary flight mode, whole wing shape, and in some instances, foraging behavior. The skeletal element that makes up the main support of the antebrachium (ulna) exhibits positive allometry, a result that is congruent with more general studies conducted at higher taxonomic, but much less densely-sampled, levels. However, in contrast to such general studies, the humerus of pelecaniforms exhibits isometry. Morover, this study represents the first allometric analysis of the distal wing element (carpometacarpus) and wing bone diameters, all of which were found to scale isometrically when examined in an explicit phylogeny context.

The carpometacarpus plays an important role in characterizing different flight modes. This element, the attachment site of primary flight feathers, is that portion of the forelimb skeleton from which finely-tuned flight control originates. Its morphology, more than any other individual element, reflects differences in flight mode within pelecaniforms. Additional research is required to assess whether such morphological patterns characterize other groups of birds that also exhibit similar variation in flight mode (e.g., procellariiforms, falconiforms), and importantly, to investigate further the effects that different in-vivo loading patterns may have on element-specific morphology

(e.g., CMC shape as a function of oblique versus normally oriented primary feather attachment). An examination of pelecaniform birds reveals unique examples of both morphological and behavioral convergence and divergence as exemplified by differences in skeletal morphology.

Literature Cited

Abouheif, E., 1999. A method for testing the assumption of phylogenetic independence in

comparative data. Evolutionary Ecology Research 1, 895-909.

Ashmole, N.P. 1971. Seabird ecology and the marine environment. In: D.S. Farner, J.R.

King, K.C. Parkes (Eds.), Avian biology 1 pp. 223-286. New York, NY:

Academic Press.

Baumel, J.J., Witmer, L.M. 1993. Osteologia. In: J.J. Baumel, A.S. King, J.E. Breazile,

H.E. Evans, J.C. Vanden Berge (Eds.), Handbook of Avian Anatomy: Nomina

Anatomica Avium, 2nd Edition pp. 45-132. Cambridge, MA: Publications of the

Nuttall Ornithological Club.

Blomberg, S.P., Garland Jr., T, Ives, A.R., 2003. Testing for phylogenetic signal in

comparative data: behavioral traits are more labile. Evolution 57(4), 717-745.

Bochenski, Z., Bochenski, Z.J., 1993. Correlation between the wing lengths of birds and

the measurements of their bones. Belgian Journal of Zoology 122, 123-132.

Bohonak, A.J., 2002. RMA: Software for Reduced Major Axis Regression.

Bourne, W.R.P., 1976. On subfossil bones of Abbott's booby Sula abbotti from the

Mascarene Islands, with a note on the proportions and distribution of the Sulidae.

Ibis 118, 119-123.

Brewer, M.L., Hertel, F., 2007. Wing morphology and flight behavior of pelecaniform

seabirds. J Morphol 268, 866-877.

Dunning Jr., J.B., 1993. CRC Handbook of avian body masses: CRC Press.

Dyreson, E., Maddison, W.P., 2001. Rhetenor package for morphometrics for the

Mesquite system.

Fain, M.G., Houde, P., 2004. Parallel radiations in the primary clades of birds. Evolution

58, 2558-2573.

Felsenstein, J., 1985. Phylogenies and the comparative method. The American Naturalist

125, 1-15.

Friesen, V.L., Anderson, D.J., 1997. Phylogeny and evolution of the Sulidae (Aves:

Pelecaniformes): A test of alternative modes of speciation. Molecular

Phylogenetics & Evolution 7, 252-260.

Garland Jr., T.D., Harvey, P.H., Ives, A.R., 1992. Procedures for the analysis of

comparative data using phylogenetically independent contrasts. Systematic

Biology 41(1), 18-32.

Garthe, S., Benvenuti, S., Montevecchi, W.A., 2000. Pursuit plunging by northern

gannets (Sula bassana) feeding on capelin (Mallotus villosus). Proceedings of the

Royal Society B 267, 1717-1722.

Gatesy, S.M., Middleton, K.M. 2007. Skeltal adaptations for flight. In: B.K. Hall (Ed.),

Fins into Limbs: Evolution, Development, and Transformation, pp. 269-283.

Chicago, IL: The University of Chicago Press.

Gill, F.B., 1995. Ornithology. New York, W. H. Freeman and Company.

Hackett, S.J., Kimball, R.T., Reddy, S., Bowie, R.C., Braun, E.L., Braun, M.J.,

Chojnowski, J.L., Cox, W.A., Han, K.L., Harshman, J. et al., 2008. A

phylogenomic study of birds reveals their evolutionary history. Science 320,

1763-1768.

Harvey, P.H., Pagel, M.D., 1991. The comparative method in evolutionary biology.

Oxford, Oxford University Press.

Hertel, F., Ballance, L.T., 1999. Wing ecomophology of seabirds from Johnston Atoll.

Condor 101, 549-556.

Johnsgard, P.A., 1993. Cormorants, Darters, and Pelicans of the world.

Washington, Smithsonian Institution Press.

Kennedy, M., Spencer, H.G., 2004. Phylogenies of the Frigatebirds (Fregatidae) and

Tropicbirds (Phaethonidae), two divergent groups of the traditional order

Pelecaniformes, inferred from mitochondrial DNA sequences. Molecular

Phylogenetics and Evolution 31.

Kriloff, A., Germain, D., Canoville, A., Vincent, P., Sache, M., Laurin, M., 2008.

Evolution of bone microanatomy of the tetrapod tibia and its use in

palaeobiological inference. Journal of Evolutionary Biology 21, 807-826.

Laurin, M., Girondot, M., Loth, M., 2004. The evolution of long bone microstructure and

lifestyle in lissamphibians. Paleobiology 30(4), 589-613.

Lavin, S.R., Karasov, W.H., Ives, A.R., Middleton, K.M., Garland Jr., T. 2008.

Morphometrics of the avian small intestine compared with that of nonflying

mammals: a phylogenetic approach. Physiological and Biochemical Zoology

81(5), 526-550.

Livezey, B.C., Zusi, R.L., 2007. Higher-order phylogenetics of modern Aves based on

comparative anatomy. Netherlands Journal of Zoology 51, 179-205.

Maddison, W.P., Maddison, D.R., 2007. Mesquite: a modular system for evolutionary

analysis. Version 2.5 http://mesquiteproject.org.

McGowan, A.J., Dyke, G.J., 2007. A morphospace-based test for competitive exclusion

among flying vertebrates: did birds, bats and pterosaurs get in each other's space?

J Evol Biol 20, 1230-1236.

McMahon, T.A., 1975. Using body size to understand the structural design of animals

quadrupedal loco motion. Journal of Applied Physiology 39, 619-627.

Middleton, K.M., Gatesy, S.M., 2000. Theropod forelimb design and evolution.

Zoological Journal of the Linnean Society 128, 149-187.

Midford, P.E., Garland Jr., T., Maddison, W.P., 2005. PDAP Package of Mesquite.

Version 1.07.

Mosimann, J.E., 1970. Size allometry: size and shape variables with characterizations of

the lognormal and generalized gamma distributions. Journal of the American

Statistical Association 65, 930-945.

Mosimann, J.E., James, F.C., 1979. New statistical methods for allometry with

application to Florida red-winged blackbirds. Evolution 33, 444-459.

Nelson, J.B., 1978. The Sulidae: gannets and boobies. Oxford, Oxford University Press.

Nelson, J.B., 2005. Pelicans, Cormorants, and their Relatives: The Pelecaniformes. New

York, Oxford University Press.

Niemi, G.J., 1985. Patterns of morphological evolution in bird genera of new-world and

old-world peatlands. Ecology (Washington D C) 66, 1215-1228.

Norberg, U.M., 1985. Flying, Gliding, and Soaring. In: M. Hildebrand, D.M. Bramble,

K.F. Liem, D.B. Wake (Eds.), Functional Vertebrate Morphology, pp. 129-158.

Cambridge, MA: Harvard University Press.

Norberg, U.M., 1995. Wing design and migratory flight. Israel Journal of Zoology 41,

297-305.

Norberg, U.M., Norberg, R.A., 1986. Ecomorphology of flight and tree-trunk climbing in

birds. Acta XIX Congressus Internationalis Ornithologici 22, 2271-2282.

Nudds, R.L., 2007. Wing-bone length allometry in birds. Journal of Avian Biology 38,

515-519.

Nudds, R.L., Dyke, G., Rayner, J.M.V., 2004. Forelimb proportions and the evolutionary

radiation of Neornithes. Proceedings of the Royal Society B (Suppl.) 271, S324-

S327.

Nudds, R.L., Dyke, G., Rayner, J.M., 2007. Avian brachial index and wing kinematics:

putting movement back into the bones. Journal of Zoology (London) 272, 218-

226.

Olmos, M., Casinos, A., Cubo, J., 1996. Limb allometry in birds. Annales des Sciences

Naturelles Zoologie, Paris 17, 39-49.

Pagel M. 1992. A method for the analysis of comparative data. Journal of Theoretical

Biology 156: 431-442.

Pennycuick, C.J., 1972. Soaring behavior and performance of some East African birds,

observed from a motor-glider. Ibis 114, 178-218.

Pennycuick, C.J. 1983. Thermal soaring compared in three dissimilar tropical bird

species, Fregata magnificens, Pelecanus occidentalis, and Coragyps atratus.

Journal of Experimental Biology 102, 307-325.

Purvis, A., Garland Jr., T. 1993. Polytomies in comparative analysis of continuous

characters. Systematic Biology 42(4), 569-575.

Rayner, J.M.V., 1988. Form and function in avian flight. In: R.F. Johnston (Ed.), Current

Ornithology, vol. 5, pp. 1-66. New York: Plenum Press.

Rayner, J.M.V., Dyke, G., 2003. Origins and evolution of diversity in the avian wing. In:

V.L. Bels, J.-P. Gasc, A. Casinos (Eds.), Vertebrate Biomechanics and Evolution,

pp. 297-317. Oxford: BIOS Scientific Publishers Limited.

Rayner, J.M.V., Viscardi, P.W., Ward, S., Speakman, J.R., 2001. Aerodynamics and

energetics of intermittent flight in birds. American Zoologist 41, 188-204.

Savile, D.B.O., 1957. Adaptive evolution in the avian wing. Evolution 11, 212-224.

Schmidt-Nielsen, K., 1984. Scaling: Why is animal size so important? New York,

Cambridge University Press.

Schnell, G.D., 1974. Flight speeds and wingbeat frequencies of the magnificent

frigatebird. The Auk 91, 564-570.

Sibley, C.G., Ahlquist, J.E., 1990. Phylogeny and classification of birds: a study in

molecular evolution. New Haven, Yale University Press.

Siegel-Causey, D., 1988. Phylogeny of the Phalacrocoracidae. Condor 90, 885-905.

Van Tuinen, M., Butvill, D.B., Kirsch, J.A.W., Hedges, S.B., 2001. Convergence and

divergence in the evolution of aquatic birds. Proceedings of the Royal Society B

268, 1345-1350.

Warham, J., 1977. Wing loadings, wing shapes, and flight capabilities of

Procellariiformes. New Zealand Journal of Zoology 4, 73-83.

Weimerskirch, H., Chastel, O., Barbraud, C., Tostain, O., 2003. Frigatebirds ride high on

thermals. Nature 421, 333-334.

Zar, J.H., 1999. Biostatistical Analysis. New Jersey, Prentice Hall, Inc.

Table 1–1. Pelecaniform taxa examined in this study. The range of body mass means (Dunning Jr. 1993), wingspan ranges (Perrins and Middleton 1985), primary flight mode(s), and foraging techniques are listed for each genus. Body mass Wing span Genera (g) (m) Flight mode Foraging mode Fregata 927 - 754 1.76 - 2.3 Static soar on the wing Pelecanus 3702 - 9000 2 - 2.8 Static soar surface feed Static soar, Morus 2350 - 2932 1.7 - 1.9 Dynamic soar plunge dive Anhinga 1235 - 1436 1.2 - 1.27 Static soar foot-propelled dive Sula 1000 - 1550 1.4 - 1.8 Flap-glide plunge dive Phaethon 334 - 750 0.9 - 1.1 Flap-glide plunge dive Phalacrocorax 617 - 3390 0.8 - 1.6 Flap foot-propelled dive

Table 1–2. Summary of reduced major axis (RMA) regression results of skeletal metrics on geometric mean (GMb – body size) for pelecaniform birds. Outgroup species (Balaeniceps rex, Scopus umbretta, Cochlearius cochlearius) are included. Isometry indicated by b = 1. Significant positive allometry (b > 1) at the p ≤ 0.05 level indicated by *. DV = dorsoventral, CC = craniocaudal, CMC = carpometacarpus. Ahistorical RMA Independent Contrasts Measurement (y) n Intercept (a) Slope (b) SEb r p-value n Slope (b) SEb r p-value Humerus length 53 0.04 1.211 0.074 0.90 0.007* 52 1.125 0.067 0.91 0.068 DV Hum. diameter 53 -1.24 1.224 0.076 0.90 0.005* 52 1.109 0.053 0.94 0.051 CC Hum. diameter 53 -1.24 1.196 0.074 0.90 0.011* 52 1.044 0.052 0.94 0.402 Ulna length 53 -0.08 1.295 0.099 0.84 0.002* 52 1.275 0.110 0.79 0.016* DV Ulna diameter 53 -1.26 1.174 0.090 0.84 0.059 52 1.036 0.065 0.90 0.585 CC Ulna diameter 53 -1.16 1.109 0.085 0.84 0.206 52 1.002 0.061 0.90 0.974 CMC length 53 -0.23 1.178 0.097 0.81 0.072 52 1.096 0.084 0.84 0.924 DV CMC diameter 53 -1.32 1.143 0.090 0.83 0.118 52 0.964 0.060 0.90 0.551 CC CMC diameter 53 -1.59 1.270 0.115 0.77 0.022* 52 1.119 0.085 0.84 0.168

Figure 1–1. Composite phylogeny of taxa included in study. Familial level structure based on Livezey and Zusi (2007), inter-familial relationships from Sibley and Alquist (1990), Siegel-Causey (1988), Friesen and Anderson (1997), Kennedy and Spencer (2004). Outgroups included for comparative purposes (Balaeniceps rex, Cochlearius cochlearius, and Scopus umbretta) are based on Livezey and Zusi (2007). 60

Figure 1–2. Representative log-log reduced major axis (RMA) regression plots for pelecaniform birds: △ = Pelecanus, ◇ = Fregata, ○ = Anhinga, ● = Morus, ■ = Sula, □ = Phaethon, ★ = Phalacrocorax, × = Phalacrocorax harrisi (Flightless cormorant). See Table 1–1 for flight modes and Table 1–2 for regression statistics.

Figure 1–3. Plots of first and second canonical discriminant functions from ahistorical and historical multivariate analyses. Flight style groups: Soar1 = Pelecanus, Soar2 = Anhinga + Morus, Soar3 = Fregata, Flap-Glide = Phaethon + Sula, Flap = Phalacrocorax. A) Discriminant Function Analysis (DFA) of wing size-corrected individuals. The model significantly separated the five flight style groups (Wilk’s λ = 0.002, p < 0.00001). Six individuals (from total n = 284) were misclassified. The first function is positively correlated with ulnar length and negatively correlated with humeral length and the second function is positively correlated with dorsoventral humeral diameter and negatively correlated with craniocaudal carpometacarpal diameter. B) Historical canonical variates analysis of wing-size corrected species means. Flight style group locations in canonical space are nearly identical to ahistorical DFA. However, rather than proximal humeral and ulnar variables, distal element (CMC) variables are highly loaded. Dorsoventral CMC diameter is highly loaded (~0.8) on the first function, and total CMC length is highly loaded on the second function (~0.7). Numbers associated with points indicate species as given in Appendix Table 1–1. Scale bar provided for wing profiles. Wing profiles of Sula and Phaethon modified from Brewer (2007).

CHAPTER 2: CROSS-SECTIONAL GEOMETRY OF THE FORELIMB SKELETON

AND FLIGHT MODE IN PELECANIFORM BIRDS

Abstract

Avian wing elements have been shown to experience both dorsoventral bending and torsional loads during flapping flight. However, not all birds use continuous flapping as a primary flight strategy. The pelecaniforms exhibit extraordinary diversity in flight mode, utilizing flapping, flap-gliding, and soaring. Here I (1) characterize the cross- sectional geometry of the three main wing bone (humerus, ulna, carpometacarpus) and

(2) use elements of beam theory to estimate resistance to loading in fourteen species of pelecaniform birds. Patterns emerge that are common to all species, as well as some characteristics that are flight mode specific. In all birds examined, the distal most wing segment (carpometacarpus) is the most elliptical (relatively high Imax/Imin) at mid-shaft,

suggesting a shape optimized to resist bending loads in a dorsoventral direction. As

primary flight feathers attach at an oblique angle relative to the long axis of the

carpometacarpus, they are likely responsible for inducing bending of this element during

flight. Moreover, among flight modes examined the flapping group (cormorants) exhibits

elements that are more elliptical than other flight modes, perhaps pertaining to the higher

frequency loading in these elements. The soaring birds (pelicans and gannets) exhibit wing elements with near-circular cross sections and higher polar moments of area than in the flap and flap-gliding birds, suggesting shapes optimized to offer increased resistance to torsional loads. Thus, an analysis of the cross-sectional geometry has enhanced our 64 interpretation of how the wing elements may be loaded and ultimately how they are being used during normal activities.

Introduction

The cross-sectional geometry of long bones has been used as a proxy for estimating resistance to the biomechanical loading encountered by that bone during the life of an organism. More specifically, the shape of the cross section determines the ability of a given bone to withstand stresses and helps reduce stress and strain under loading conditions. Parameters derived from beam theory estimate the amount and distribution of cortical bone and have been used to functionally interpret cross-sectional geometry (Roark and Young, 1975). Three commonly used parameters are cortical area

(CA), second moment of area (I) and polar moment of area (J). Cortical area represents the amount of cortical bone in a cross section and has typically been used to estimate resistance to compressional loading. The second moment of area has been used to infer resistance to bending loads, and the polar moment of area has been used to estimate resistance to torsion. Finally, a ratio of two orthogonal second moments provides a measure of relative circularity of the bone shaft. This is functionally relevant as a more elliptical section is typically interpreted to represent a higher resistance to bending in a preferred direction (e.g., Jungers and Minns, 1979; Ruff and Hayes, 1983; Demes et al.,

1991; Ruff, 2002; Carlson, 2005).

Cross-sectional geometric approaches have been used extensively to characterize the limb skeleton in terrestrial mammals. However, most birds predominantly occupy the aerial environment where the forelimb skeleton necessarily experiences different types of

65 loads than those encountered by animals on the ground. Generally speaking, it has been argued that the avian postcranial skeleton is optimized for mass reduction (e.g., Bühler,

1992). Moreover, birds are functionally distinct from most non-chiropteran amniotes and it can be assumed that forelimb and hind limb elements are loaded in vastly different ways. In vivo strain-gauge studies of the humerus in both bats and birds indicate that the most proximal limb element is primarily loaded in torsion and dorsoventral bending during flapping flight due to the production of lift forces acting on the wing distal to the humerus (Swartz et al., 1992; Biewener and Dial, 1995). However, not all birds use continuous flapping flight as their primary locomotor mode. As continuous flapping is energetically expensive, many groups of birds have developed novel energy-saving techniques such as various types of gliding and soaring (Norberg, 1985; Rayner, 1988;

Rayner et al., 2001). At present it is unclear how different habitual flight modes may affect the loading environment, and thus, the cross-sectional geometry, of individual forelimb elements.

The pelecaniforms represent a clade of neognath birds that exhibit an extraordinary diversity in flight modes (Brewer and Hertel, 2007). Flight characteristics exhibited by birds in this group range from continuous flapping (cormorants and shags) to various types of gliding or flap-gliding (e.g., tropicbirds and boobies) and soaring strategies. During soaring, birds exploit moving air currents to gain potential energy, making this the least energetically expensive mode of flight (Norberg, 1985). Static soaring birds, such as the pelican, frigatebird, and anhinga/darter, use rising columns of

66 air, or thermals. By contrast, dynamic soarers such as the gannet utilize velocity differences in stratified currents over the ocean to generate lift.

Ongoing research has demonstrated that the external morphology of forelimb bones (humerus, ulna, and carpometacarpus) varies in a predictable way among flight styles, and specifically, that the external mid-shaft diameter of these elements is particularly important for distinguishing among flight modes (Chapter 1; Simons, In

Press). As such, additional information characterizing the internal morphology of long bones (e.g., cortical thickness, distribution) may be critical for better elucidating specific relationships among whole bone shape, bone cross-sectional properties, whole wing shape and loading regimes associated with different modes of flight in birds.

Two studies to date have investigated long bone cross-sectional geometry in birds across a varied sample of neoganths. Cubo and Casinos (1998) examined the scaling relationship of CA, I in the maximum direction (Imax), and J of the humerus, radius, ulna, as well as hind limb elements. In general, wing bones were found to exhibit isometry under the geometric similarity model (Cubo and Casinos 1998). In addition, the authors provided data on the orientation of Imax for all elements. For proximal elements (humerus and femur) the maximum I was found to be in the PM-AL (posteromedial-anterolateral) orientation, whereas in the ulna and tibiotarsus the maximum I was in the PL-AM

(posterolateral-anteromedial) orientation. More recently, Habib and Ruff (2008) examined section moduli (a metric of structural strength) ratios of the femur and humerus

as a way to differentiate among locomotor categories, such as obligate terrestrial runners,

perchers, hind limb and forelimb-propelled divers, and dynamic soaring birds. Although

67 these studies have revealed important patterns of bone shape and/or inferred strength related to locomotor potential and bone loading, neither examined the entire forelimb skeleton in their analysis. Nonetheless, both studies provide important perspectives on avian long bone morphology in general and provide the impetus to further examine cross- sectional geometry of avian wing bones in a phylogenetically restricted, yet flight-mode diverse group such as the pelecaniforms.

The objective of this study is to characterize the cross-sectional geometry of the forelimb elements in pelecaniform birds and to examine the relationship between the cross-sectional geometry and flight mode, and in particular, of those elements to which the primary and secondary flight feathers are attached. I investigate the cortical area

(CA), ratio of the maximum to minimum second moments of area (Imax/Imin), and the polar moment of area (J) of the humerus, ulna, and carpometacarpus in the flight mode- diverse pelecaniforms.

Materials and Methods

Fourteen species of pelecaniform birds represent the focal sample used in this study (Appendix B). Pelecaniforms include Pelecanidae (pelicans), Sulidae (gannets and boobies), Phaethonidae (tropicbirds), Phalacrocoracidae (cormorants and shags),

Fregatidae (frigatebirds), and Anhingidae (anhingas and darters). A composite phylogeny

(Fig. 2–1) of taxa used in this analysis was assembled based on the following studies:

Siegel-Causey (1988), Friesen and Anderson (1997), Kennedy and Spencer (2004), and

Livezey and Zusi (2007). Species were assigned to a flight mode category based on

68 behavioral data collected from the literature (Table 2–1) (also, see Chapter1; Simons, In

Press).

The humerus, ulna, and carpometacarpus (CMC, major metacarpal only) of 14 species (n = 94, mean of 6 specimens per species; Table 2–1) were scanned at the mid- shaft of each element on a GE eXplore Locus MicroCT (µCT) Scanner housed at Ohio

University. Although cross sections have been shown to vary throughout the length of the bone (Ruff and Hayes, 1983), the maximum stress is predicted to occur at mid-shaft

(Beer et al., 2006). Wing bones were sampled from the right side of skeletally mature specimens and included both sexes depending on availability of specimens. Within the pelecaniforms, differences in body size between the sexes tend to be small, ranging from

0-20% in adults (Johnsgard, 1993). Each element was positioned in the µCT scanner with the long axis of the bone parallel to the scanner bed. A 1 mm diameter circle of barium sulfate was placed at mid-shaft on the three wing elements to ensure a readily identifiable radio-dense marker for use in orienting scan data during subsequent quantitative analyses.

A scan resolution of 44-46 µm was used to acquire cross sections (x-ray tube voltage =

80 kV, x-ray tube current = 450 μm). The total length (L) of each element for use in standardizing J was measured using digital calipers (Mitutoyo Digimatic calipers).

Skeletal specimens were borrowed from the Carnegie Museum of Natural History (CM),

National Museum of Natural History (NMNH) and Ohio University Vertebrate

Collections (OUVC).

The following parameters were calculated for each mid-shaft slice using ImageJ version 1.36b (NIH) with MomentMacroJ version 1.3

(www.hopkinsmedicine.org/fae/mmacro.htm): cortical area (CA), total cross-sectional area (TA), second moment of area in the maximum direction (Imax), and second moment

of area in the minimum direction (Imin). The polar moment of area (J) was calculated as

the sum of Imax and Imin. Analysis of Variance (ANOVA) with post-hoc Tukey-Kramer

multiple comparisons were used to test for differences among flight style groups using

the following biomechanical variables calculated for each element: CA/TA, the amount

of cortical bone relative to total bone cross-sectional area; Imax/Imin, a shape ratio

indicating resistance to bending; and J/L, length-standardized resistance to torsion. Non-

parametric Kruskal-Wallis tests were used if groups did not meet the assumptions of

ANOVA.

Results

Results of this study reveal that some common cross-sectional geometric relationships exist among the three forelimb elements in all pelecaniforms examined

(Table 2–2, Fig. 2–2). For example in all species the carpometacarpus exhibited a significantly more (p < 0.0001) elliptical (higher Imax/Imin ratio) cross section than either

the humerus or ulna (Fig. 2–3A), with the major axis of the carpometacarpus oriented

dorsoventrally. Moreover, the humerus exhibited significantly (nonparametric Kruskal-

Wallis test, p = 0.0021) higher J/L than either the ulna or carpometacarpus (Fig. 2–3B).

Relative cortical area (CA/TA) did not differ significantly (p = 0.6157) among the three

elements (Fig. 2–3C).

Additionally, notable differences were detected in cross-sectional geometry

among the flight mode groups. The flapping group exhibited significantly more (p <

0.0001) elliptical (higher Imax/Imin ratio) carpometacarpi and humeri than all other flight modes (Fig. 2–4A). The humeri of the flap-glide and static soaring birds were significantly more (p < 0.0001) elliptical than the dynamic soaring birds (Fig. 2–4C). By contrast, ulnae were significantly more elliptical (p < 0.0001) in the static soaring group than the other flight mode groups (Fig. 2–4B).

For all three elements the soaring birds (dynamic + static) exhibited significantly higher (p < 0.0001) J/L than the flap and flap-gliding flight modes (Fig. 2–5). Of the taxa included in the static soar group, Pelecanus exhibited the largest variation in J/L (Fig. 2–

5B).

Significant differences (p < 0.0001) in relative cortical area (CA/TA) were identified among the flight mode groups (Fig. 2–6). The flapping birds exhibited significantly higher CA/TA than all other flight modes (with exception of anhingas/darters). In addition, the flap-gliding and dynamic soaring birds exhibited

significantly higher CA/TA than the static soaring birds. However, the genus Anhinga is

notable relative to all other birds in that it exhibits extremely high CA/TA values. The

CA/TA values for the humerus in Anhinga were significantly higher than all other

functional groups (see below).

Discussion

Patterns Common to All Pelecaniforms

A more elliptical bone cross section is interpreted to represent higher resistance to bending in a preferred direction. Of the three elements examined, the CMC is the most elliptical at mid-shaft in roughly the dorsoventral direction (Fig. 2–2). This suggests that

71 for all flight modes, the CMC may be experiencing primarily bending loads. This may pertain to the way in which feathers transmit aerodynamic forces to the bone. The primary flight feathers (i.e., those attached to the dorsal margin of the CMC) are attached obliquely to the long axis of the CMC diaphysis (Fig. 2–7). Moreover, primary flight feathers that attach to the phalanges of the major digit are oriented even more obliquely to the long axis of these elements, and these in turn would induce a bending load on the distal end of the carpometacarpus. This contrasts markedly with the nearly perpendicular manner in which secondary flight feathers attach to the ulna. It is suggested that when lift is generated on the primary flight feathers, this orientation of feather attachment to the

CMC (and major digit phalanges) would impart bending loads through a roughly dorsoventral plane (around the craniocaudal axis). Note: analysis was restricted to the major metacarpal (i.e., the main component of the CMC and the attachment site of the primary flight feathers) and did not characterize the cross section of the minor metacarpal

(Fig. 2–7). It is clear that the compound nature of the element would provide additional support and reduces the amount of bending possible along a craniocaudal plane (around the dorsoventral axis). Thus, the bone may be more likely to experience dorsoventral bending and structural (shape) adaptation of the major metacarpal reflects a response over evolutionary time.

Within limb analyses of J/L reveal relatively high resistance to torsional loads

(increased J/L) in the humeri of all pelecaniforms (Fig. 2–3). This is consistent with the concept that torsional loads applied to distal elements are additive from distal to proximal through the wing. In addition, this interpretation is supported by the concept that lift

72 generated by secondary flight feathers, distal to the humeral axis, creates torsional loads on the humerus, as indicated by in vivo strain gauge analyses (Swartz et al., 1992;

Biewener and Dial, 1995). It is important to note that the interpretation of J as a measure of resistance to torsion is most robust when the shape ratio of the section (Imax/Imin) is less than or equal to 1.5 (i.e., it is relatively circular; Daegling, 2002). The mean Imax/Imin of the humerus and ulna for all flight modes, with the exception of the humerus of the flapping group, is less than 1.5.

The proportion of cortical bone per total bone area (CA/TA) is not significantly different among the three elements for pelecaniform taxa (Fig. 2–3C). Whereas there are differences among taxa (and indeed among flight modes, see below), within each species

CA/TA is consistent for the three elements. For example, Pelecanus occidentalis exhibits the lowest values for CA/TA, ranging from 0.22 for the humerus to 0.28 for the CMC

(Table 2–2). This differs greatly from Anhinga melanogaster, which exhibits CA/TA values of 0.66-0.77. Within each of these species, however, the results for the three elements are similar. Such differences likely pertain to skeletal variation in the locomotor apparatus between these two species, and specifically, variation that marks buoyancy requirements for different types of foraging behavior (see below). This suggests that for cortical area, the bones of the entire wing (and perhaps the entire skeleton; see O’Connor,

In Press) are responding to the similar pressures, whether environmental, behavioral, or phylogenetic.

Flight Mode Patterns

There are also clear differences in cross-sectional characteristics when examined as a function of flight mode category. For example, the shape ratio (Imax/Imin) is

significantly different among flight modes. Flapping flight taxa possess more elliptical carpometacarpi and humeri than other flight modes. This suggests that carpometacarpi and humeri of flapping birds may be experiencing predominantly bending loads, at least more so than the corresponding bones of birds that use other (i.e. non-flapping) primary flight modes such as soaring. Specifically, the CMC exhibits a dorsoventrally elliptical cross section, a shape that is consistent with resistance to bending loads from the obliquely oriented primary flight feathers (Fig. 2–7). In addition, the relatively elliptical humerus (also with a major axis oriented dorsoventrally) in the flapping flight category suggests that during continuous flapping, even the proximal-most element may be experiencing predominantly bending loads. This may pertain to the frequency of loading experienced by the elements of the continuous-flapping category. For example, the flapping birds examined in this study (cormorants, shags) exhibit wing beat frequencies exceeding five beats per second (Meinertzhagen, 1955; Pennycuick, 1983; Pennycuick,

1990). By contrast, the non-flapping specialists, such as the pelican, exhibit average wing beat frequencies of approximately two beats per second (Johnsgard, 1993). Thus, flapping flight would necessarily result in a higher frequency of wing loading in birds.

Taken together, the relatively high wing beat frequency, combined with the inferred orientation of the applied load, is here hypothesized to influence structural adaptation of the avian wing skeleton. This interpretation is consistent with the results of Biewener and

Dial (1995), who used in vivo strain gauges on the pigeon humerus and documented both dorsoventral bending and torsion during high frequency flapping flight. In sum, it is clear that additional experimental work is required to examine potential loading differences in birds utilizing different flight modes. Importantly, the cross-sectional geometric differences identified herein will allow the development of specific hypotheses that take into account both shape and flight mode variation.

The carpometacarpus has also received recent focus in the context of a whole- bone metric analysis of the wing skeleton in pelecaniforms. Simons (In Press; Chapter 1) documented that total length and dorsoventral diameter of the CMC was useful for distinguishing among most flight modes. Interestingly, the flappers in the whole-bone study (cormorants, shags) were not completely separated from all other flight mode groups based on these specific external measures, and instead, overlapped in morphospace with the static soaring pelicans. By incorporating the internal structure of the bones, as in the calculation of Imax/Imin here, increased resolution is gained for

exploring the flapping/static soaring interface. Whereas the two groups have similar

relative dorsoventral CMC diameter (relative to total wing size), the cross-sectional shape

is significantly different and thereby useful for more fine-tuned separation of the flight

modes.

The static and dynamic soaring birds exhibit higher J/L values than the flap and

flap-glide modes. Relatively large polar moment values suggest that a bone is shaped to

resist one of several loading environments: bending moments that are not large or

frequent enough to require an elliptical section, predominant/frequent bending that is

75 occurring but in multiple directions (Carlson, 2005), or torsion is the predominant load encountered. As element cross sections in soaring birds clearly exhibit a circular section with material distributed distant from the neutral axis, it is suggested that soaring is indeed placing higher torsional loads on the wing skeleton, and in particular, the humerus. Other characteristics that support this interpretation include the fact that soaring birds typically have very large, broad wings. Soaring birds exhibit wings with a relatively high aspect ratio, and in static soaring birds in particular, a significantly longer chord length (Fig. 2–7). Such long secondary feathers would act as longer lever arms on the ulna as lift is generated, ultimately transferring relatively larger torsional loads up through the humerus. Experimental studies have shown that, in general, avian humeri experience torsional loads (Biewener and Dial, 1995), but this is one of the first studies to postulate how the shape of the whole wing (including feathers) may differentially impact the amount of torsional loading experienced by the forelimb skeleton (see also de

Margerie et al., 2005). As predicted, the birds in this study with large broad wings

(pelicans) do indeed have skeletal elements that exhibit higher resistance to torsional loads than birds with smaller, more slender wings. Interestingly, the dynamic soaring bird in the sample (the gannet) possesses a high aspect ratio, but relatively slender wings

(shorter chord length). The gannet also exhibits forelimb elements with cross sections optimized to resistant to torsion, suggesting that the wing shape and soaring behavior both affect the loading environment acting on the wings.

Cortical bone area is generally considered in association with resistance to compression. Whereas the forelimb bones of most birds are likely not experiencing

76 significant and/or sustained axial compression, major differences in relative cortical area were identified among different flight mode categories. Most notably, darters and anhingas (Anhinga) exhibit extremely high CA/TA values, not only among their flight mode partners (soarers), but also among pelecaniforms generally. Anhingas and darters are highly specialized for sustained underwater foraging. Unlike any other bird examined, anhingas and darters rarely swim at the surface of the water and instead often swim submerged with only their head and neck above the surface. In addition, during foraging they remain totally submerged for up to a minute to stealthily stalk their prey (Owre,

1967; Johnsgard, 1993; Nelson, 2005). High CA/TA values may work to impart a reduction in whole-body buoyancy that would be beneficial in this type of foraging strategy. It is also notable in that Anhinga is the only member of pelecaniforms with a completely apneumatic postcranial skeleton, whereas other members of the clade variably exhibit air-filled bones (O’Connor, In Press). In addition to Anhinga, Phalacrocorax

(cormorants), members of the flapping flight mode, also exhibit relatively high CA/TA values and a general reduction in whole-body skeletal pneumaticity (although not to the extreme level of apneumaticity observed in Anhinga). Cormorants also use an underwater foot-propelled pursuit foraging mode (Owre, 1967). Thus, in pelecaniforms, a relatively high CA/TA may not reflect increased resistance to axial loading (or even flight mode in general), but more likely represents a skeletal modification related to buoyancy reduction in underwater foragers. Similar adaptations have been suggested in other diving specialists among both birds (e.g., auks and penguins) and mammals (Taylor, 1994;

Habib and Ruff, 2008; Kriloff et al., 2008).

In conclusion, some general patterns emerge when considering the cross-sectional geometry of the wing skeleton in pelecaniform birds. Importantly, some trends appear independent of flight mode, whereas others partition along flight modes categories.

Regardless of flight mode, the carpometacarpus is the most elliptical of all limb elements, likely reflecting the manner in which primary flight feathers attach to the long axis of the bone and transmit aerodynamic loads to the skeleton. In contrast, the humerus exhibits the highest polar moment of area (a metric of resistance to torsional loading) of the three limb segments. This is interpreted to reflect generation of lift distal to the humeral axis during any flight style and the additive nature of loads from proximal to distal through the wing. High relative cortical area was identified in underwater foragers such as

Anhinga and Phalacrocorax, suggesting a role in buoyancy modulation. Among flight modes, flapping birds exhibit the most elliptically-shaped bones, whereas soaring birds, especially static soarers, exhibit circular cross sections with bone distributed relatively distant from the hypothesized neutral axis. This analysis of long-bone cross-sectional anatomy has enhanced our interpretation of how avian wing elements relate to hypothesized loading regimes, and generally, how the postcranial skeleton reflects locomotor and foraging activities in birds. Future in vivo studies, especially of distal wing elements, are necessary to test and further refine the hypotheses developed herein related to wing-bone cross-sectional shape, whole-wing morphology and flight mode variation in birds.

Literature Cited

Beer FB, Johnston ER, DeWolf JT. 2006. Mechanics of Materials. New York: McGraw

Hill Co.

Biewener AA, Dial KP. 1995. In vivo strain in the humerus of pigeons (Columba livia)

during flight. Journal of Morphology 225: 61-75.

Brewer ML, Hertel F. 2007. Wing morphology and flight behavior of pelecaniform

seabirds. Journal of Morphology 268: 866-877.

Bühler P. 1992. Light bones in birds. In: Campbell KEJ, editor. Papers in avian

paleontology honoring Pierce Brodkorb. Los Angeles: Natural History Museum

of Los Angeles County. p 385–393.

Carlson KJ. 2005. Investigating the form-function interface in African apes: relationships

between principal moments of area and positional behaviors in femoral and

humeral diaphyses. American Journal of Physical Anthropology 127: 312-334.

Cubo J, Casinos A. 1998. Biomechanical significance of cross-sectional geometry of

avian long bones. European Journal of Morphology 36(1): 19-28.

Daegling DJ. 2002. Estimation of torsional rigidity in primate long bones. Journal of

Human Evolution 43: 229-239.

Demes B, Jungers WL, Selpien K. 1991. Body size, locomotion, and long bone cross-

sectional geometry in indriid primates. American Journal of Physical

Anthropology 86: 537-547.

Friesen VL, Anderson DJ. 1997. Phylogeny and evolution of the Sulidae (Aves:

Pelecaniformes): A test of alternative modes of speciation. Molecular

Phylogenetics & Evolution 7: 252-260.

Habib MB, Ruff CB. 2008. The effects of locomotion of the structural characteristics of

avian long bones. Zoological Journal of the Linnean Society 153: 601-624.

Johnsgard PA. 1993. Cormorants, Darters, and Pelicans of the World. Washington:

Smithsonian Institution Press. 445 p.

Jungers WL, Minns RJ. 1979. Computed tomography and biomechanical analysis of

fossil long bones. American Journal of Physical Anthropology 50: 285-290.

Kennedy M, Spencer HG. 2004. Phylogenies of the Frigatebirds (Fregatidae) and

Tropicbirds (Phaethonidae), two divergent groups of the traditional order

Pelecaniformes, inferred from mitochondrial DNA sequences. Molecular

Phylogenetics and Evolution 31.

Kriloff A, Germain D, Canoville A, Vincent P, Sache M, Laurin M. 2008. Evolution of

bone microanatomy of the tetrapod tibia and its use in palaeobiological inference.

Journal of Evolutionary Biology 21: 807-826.

Livezey BC, Zusi RL. 2007. Higher-order phylogenetics of modern Aves based on

comparative anatomy. Netherlands Journal of Zoology 51: 179-205. de Margerie E, Sanchez S, Cubo J, Catanet J. 2005. Torsional resistance as a principal

component of the structural design of long bones: comparative multivariate

evidence in birds. Anatomical Record 282A: 49-66.

Meinertzhagen R. 1955. The speed and altitude of bird flight. Ibis 97: 81-117.

Nelson JB. 2005. Pelicans, Cormorants, and their Relatives. New York: Oxford

University Press. 661 p.

Norberg UM. 1985. Flying, Gliding, and Soaring. In: Hildebrand M, Bramble DM, Liem

KF, Wake DB, editors. Functional Vertebrate Morphology. Cambridge, MA:

Harvard University Press. p 129-158.

O’Connor PM. In press. Evolution of Archosaurian Body Plans: Skeletal Adaptations

of an Air-Sac Based Breathing Apparatus in Birds and other Archosaurs. Journal

of Experimental Zoology Part A: Ecological Genetics and Physiology.

Owre OT. 1967. Adaptations for locomotion and feeding in the anhinga and the double-

crested cormorant. American Ornithological Union, Ornithological Monograph,

No. 6.

Pennycuick CJ. 1983. Thermal soaring compared in two dissimilar tropical bird species.

Journal of Experimental Biology 102: 307-325.

Pennycuick CJ. 1990. Predicting wingbeat frequency and wavelength in birds. Journal of

Experimental Biology 150:171-187.

Rayner JMV, 1988. Form and function in avian flight. In: Johnston RF, editor. Current

Ornithology, vol. 5. New York: Plenum Press. p 1-66.

Rayner JMV, Viscardi PW, Ward S, Speakman JR. 2001. Aerodynamics and energetics

of intermittent flight in birds. American Zoologist 41: 188-204.

Roark RJ, Young WC. 1975. Formulas for Stress and Strain, Fifth Edition. New York:

McGraw Hill Book Company. 624 p.

Ruff CB, Hayes WC. 1983. Cross-sectional geometry of Pecos Pueblo femora and tibiae

– a biomechanical investigation: I. method and general patterns of variation.

American Journal of Physical Anthropology 60: 359-381.

Ruff CB. 2002. Long bone articular and diaphyseal structure in old world monkeys and

apes. I: locomotor effects. American Journal of Physical Anthropology 119: 305-

342.

Siegel-Causey D, 1988. Phylogeny of the Phalacrocoracidae. Condor 90: 885-905.

Simons, ELR. In press. Forelimb skeletal morphology and flight mode evolution in

pelecaniform birds. Zoology.

Swartz SM, Bennett MB, Carrier DR. 1992. Wing bone stresses in free flying bats and

the evolution of skeletal design for flight. Nature 359: 726-729.

Taylor MA. 1994. Stone, bone or blubber? Buoyancy control strategies in aquatic

tetrapods. In: Maddock L, Bone Q, Rayner JMV, editors. Mechanics and

Physiology of Animals Swimming. Cambridge: Cambridge University Press. p

151-161.

Table 2–1. List of taxa scanned and flight mode groups. Species Common name n Flight style Phalacrocorax auritus Double-crested Cormorant 10 Flap Phalacrocorax africanus Long-tailed Cormorant 6 Flap Phalacrocorax bougainvillii Guanay Cormorant 4 Flap Phalacrocorax penicillatus Brandt's Cormorant 7 Flap Sula dactylatra Masked Booby 7 Flap-glide Sula sula Red-footed Booby 6 Flap-glide Phaethon lepturus White-tailed Tropicbird 7 Flap-glide Anhinga melanogaster Darter 3 Static soar Anhinga anhinga Anhinga 6 Static soar Fregata ariel Lesser Frigatebird 7 Static soar Fregata magnificens Magnificent Frigatebird 4 Static soar Pelecanus erythrhorynchos American White Pelican 7 Static soar Pelecanus occidentalis Brown Pelican 10 Static soar Morus bassanus Atlantic Gannet 8 Dynamic soar

Table 2–2. Species means of Imax/Imin, length standardized polar moment of area (J/L), and relative cortical area (CA/TA) for each of the three main forelimb bones in birds; Abbreviations: Hum, humerus; Uln, ulna, CMC, carpometacarpus. Imax/Imin J/L CA/TA Species Hum Uln CMC Hum Uln CMC Hum Uln CMC Phalacrocorax auritus 1.426 1.110 1.854 1.545 0.545 0.374 0.559 0.609 0.701 Phalacrocorax africanus 1.587 1.165 1.948 0.367 0.135 0.105 0.588 0.611 0.708 Phalacrocorax bougainvillii 1.424 1.133 1.671 1.584 0.561 0.387 0.557 0.566 0.628 Phalacrocorax penicillatus 1.626 1.146 1.711 1.638 0.515 0.337 0.655 0.677 0.759 Sula dactylatra 1.284 1.242 1.533 1.831 0.898 0.857 0.465 0.401 0.425 Sula sula 1.387 1.230 1.497 1.207 0.602 0.483 0.430 0.352 0.414 Phaethon lepturus 1.297 1.361 1.507 0.384 0.190 0.181 0.485 0.518 0.556 Anhinga melanogaster 1.359 1.341 1.409 1.163 0.626 0.416 0.772 0.660 0.753 Anhinga anhinga 1.322 1.520 1.676 1.188 0.755 0.425 0.766 0.687 0.733 Fregata ariel 1.354 1.387 1.653 1.593 0.715 0.695 0.341 0.256 0.299 Fregata magnificens 1.408 1.289 1.584 3.290 1.446 1.329 0.302 0.232 0.292 Pelecanus erythrhorynchos 1.451 1.308 1.511 10.817 3.013 2.666 0.240 0.250 0.307 Pelecanus occidentalis 1.392 1.455 1.473 5.272 1.448 1.434 0.223 0.275 0.284 Morus bassanus 1.188 1.248 1.491 3.128 1.702 1.498 0.503 0.425 0.446 84

Figure 2–1. Composite phylogeny of pelecaniform taxa used in study.

Figure 2–2. Example cross sections of the humerus, ulna, and carpometacarpus (CMC) of four species, representing the four main flight mode categories. Static soarer, Pelecanus erythrorhynchos (USNM 13668); Dynamic soarer, Morus bassanus (CM S15516); Flapper, Phalacrocorax auritus (OUVC 9772); Flap-glide, Phaethon lepturus (USNM 490836). Abbreviations: D, indicates dorsal direction; Cr, indicates cranial direction. Scale bar applies to all three rows of cross sections.

Figure 2–3. Pooled species means for the humerus, ulna, and carpometacarpus (CMC): (A) Imax/Imin, (B) Length standardized polar moment of area (J/L), (C) Relative cortical bone area (CA/TA). Significant differences indicated by lowercase letters: a, b.

Figure 2–4. Imax/Imin ratio of the (A) carpometacarpus, (B) ulna, and (C) humerus for the four flight modes (static soar, dynamic soar, flap, flap-glide). Significant differences indicated by lowercase letters: a, b, c.

Figure 2–5. Length standardized polar moment of area (J/L) of the humerus for (A) the four flight modes (static soar, dynamic soar, flap, flap-glide) and (B) with the static soaring flight mode group deconstructed into its constituent groups to illustrate the taxon- specific variance breakdown. Significant differences indicated by lowercase letters: a, b.

Figure 2–6. Relative cortical area (CA/TA) of the humerus for the four main flight mode groups with the Anhinga separated out of the static soaring group.

Figure 2–7. Illustration of wing anatomy of (A) Morus bassanus (OUVC 10587) and (B) Pelecanus occidentalis (OUVC 10586) in ventral view. (C) Schematic of the distal forelimb skeleton and proximal feather attachments. The secondary flight feathers (shaded light gray in A and B) are oriented perpendicular to the axis of the ulna. The primary flight feathers are oriented obliquely to the axis of the carpometacarpus. Note also difference in mean chord length (width of wing) between species. Abbreviations: CMCmaj, major metacarpal; CMCmin, minor metacarpal.

CHAPTER 3: WHOLE BONE AND CROSS-SECTIONAL MORPHOLOGY OF THE

WING SKELETON IN PROCELLARIIFORM SEABIRDS: IMPLICATIONS FOR

DIFFERENCES IN FLIGHT BEHAVIOR

Abstract

Procellariiforms are wide-ranging pelagic seabirds that span a large body size range, exhibit high wing size and shape variability, and employ a variety of flight behaviors (i.e., flap-gliding, dynamic soaring, and underwater flapping). The objectives of this study were to (1) characterize the morphology (external whole bone and cross- sectional geometry) of the wing skeleton in procellariiform birds, (2) examine the relationship between morphology and flight mode in the group, and (3) to compare and integrate these results with previous research on other birds. External length and diameter measurements of wing elements were taken on 530 skeletal specimens from 101 species of procellariiforms. In addition, the main wing elements of 17 of these species were microCT scanned to examine the cross-sectional geometry. In general, the lengths of all three main wing elements were found to be positively allometric when phylogenetic relationships were taken into account. In addition, principle components analysis (PCA) and classification and regression tree (CART) analysis found that the diameters of distal elements and lengths of proximal elements were able to successfully partition procellariiform species into primary flight mode groups. Analysis of the cross-sectional geometry indicates that the elements of birds that utilize more high frequency flapping

(small size flap-gliding birds) exhibit more elliptically shaped cross sections of the distal elements, a shape optimized to resist bending loads. The cross sections of elements of dynamic soaring birds exhibit large polar moments, or shapes optimized to resist

92 torsional loads. These results are congruent with results observed in other bird groups, namely pelecaniforms, and therefore may indicate general trends in the structure and function of the avian wing skeleton.

Introduction

Procellariiforms are wide-ranging seabirds that are highly specialized for aerial locomotion and foraging over the open ocean. Most procellariiforms can travel thousands of kilometers in search of food and only return to land to breed (Warham, 1990; Brooke,

2004). Because of their highly marine lifestyle and colonial breeding habit, procellariiforms are a commonly examined group. For example, a number of studies have recently been published regarding the olfactory ability, foraging behavior, physiology, and conservation of the group (e.g., Kabat et al., 2007; Pinaud and Weimerskirch, 2007;

Delord et al., 2008; Navarro et al., 2008; Nevitt, 2008; Van Buskirk and Nevitt, 2008).

Most relevant to this study, procellariiforms span a large body size range, exhibit high wing size and shape variability, and employ a variety of flight behaviors (Figure 3–1,

Table 3–1)(Murphy, 1936; Warham, 1977; Pennycuick, 1982; Warham, 1990; Warham,

1996; Spear and Ainley, 1998; Hertel and Ballance, 1999; Pennycuick, 2002; Brooke,

2004). All procellariiforms utilize flapping, flap-gliding and soaring flight to some degree

(Murphy, 1936; Pennycuick, 1982; Brooke, 2004), but depending on body size, a bird may use more flapping (small birds) or more soaring (large birds). During flapping, continuous wingbeats are required to stay aloft. Gliding and soaring are both energy- saving flight strategies that have been hypothesized to minimize the cost of flight

(Norberg, 1985; Rayner, 1988; Rayner et al., 2001). During gliding, a bird is not actively flapping and is therefore either losing altitude or speed. Gliding is often interspersed with

93 flapping, i.e., flap-gliding. In contrast, soaring birds use moving air currents to maintain or even increase altitude without flapping (Norberg, 1985). There are two main types of soaring that differ based on the type of wind current used. During static soaring, a bird utilizes rising columns of air (thermals) whereas during dynamic soaring, a bird utilizes the gradient of horizontal wind currents over the ocean. When soaring, procellariiforms utilize predominantly dynamic soaring. The largest procellariiform birds almost exclusively use dynamic soaring to minimize the cost of flight (Bevan et al., 1994, 1995) and rarely utilize flapping except at take-off (Brooke, 2004). In addition to this diversity of aerial flapping, flap-gliding, and soaring flight behaviors, several birds (e.g.,

Pelecanoididae) within the group use their wings for underwater propulsion during prey pursuit. This range in flight behavior, body size, and wing size and shape in procellariiforms provides an opportunity to investigate how skeletal morphology relates to presumed biomechanical loading of the wing during different kinds of flight behavior.

The size and shape of the whole wing is traditionally described using two parameters: wing loading (body weight/wing area) and aspect ratio (wing span2/wing

area). Warham (1977) investigated the wing shape and flight behavior of 48 species of

procellariiform and found several trends with body size (Figure 3–1, Table 3–1). As size increases, the wing becomes relatively longer and narrower (higher aspect ratio). This increased length is interpreted to be the result of relatively longer proximal wing bones

(humerus and radius/ulna). Indeed, small procellariiforms, such as storm petrels, have as few as 10 secondary flight feathers (feathers that attach to the ulna) whereas the large albatrosses can have as many as 34 secondaries, a difference that reflects the increase in ulnar length (Brooke, 2004). Differences in aspect ratio relate directly to how well a wing

94 functions during different behaviors. A bird with a low aspect ratio wing generally has more maneuverability whereas a bird with a high aspect ratio wing is more aerially efficient (Savile, 1957; Warham, 1977). However, no study has investigated how the relative size and shape (including cross section) of all three of the main wing elements that support the wing of procellariiforms vary with whole wing shape and flight behavior.

In vivo strain studies on the humerus of birds and bats have shown that these elements experience both dorsoventral bending and torsional loading during flapping flight (Swartz et al., 1992; Biewener and Dial, 1995). Investigation of cross-sectional geometric properties, which take into account both the external and internal morphology of a cross section of a long bone, can help infer resistance to biomechanical loading during the life of the organism. Although cross-sectional geometric parameters have been applied extensively to terrestrial animals, several recent studies have used them to investigate the shape and distribution of bone in the appendicular skeleton in birds (Cubo and Casinos, 1998; Habib and Ruff, 2008; Simons and O’Connor, In Review; Chapter 2).

The parameters commonly investigated, namely cortical area (CA), second moment of area (I), and polar moment of area (J), are derived from beam theory (Roark and Young,

1975). In general, these parameters of wing bones have been found to exhibit isometry under the geometric similarity model for a variety of species of bird (Cubo and Casinos

1998). The amount of cortical bone area relative to the total area (TA) of the section has traditionally been used to estimate resistance to axial loading (compression). Although this model is adequate for terrestrial tetrapods where ground reaction forces are typically transmitted up through the long axis of the limb bones, such a model for birds (and other flying animals) may not be appropriate. However, relative cortical bone measures in birds

95 may convey other biologically relevant information. For example, in one group of birds, the pelecaniforms, relative cortical bone has been found to reflect skeletal modification related to buoyancy reduction in specialized divers (Chapter 2; Simons and O’Connor, In

Review). The second moment of area is used to infer resistance to bending loads and a ratio of two orthogonal second moments of area provides a measure of the relative circularity of a bone section. More elliptical sections are interpreted to better resist bending loads in a preferred direction. The most distal element (carpometacarpus) in pelecaniforms was found to be the most elliptical and was therefore interpreted to exhibit a shape best suited to resist predominantly bending loads, an interpretation that is consistent with the hypothesized loading regime on this element as a function of the orientation of primary feather attachment (Chapter 2; Simons and O’Connor, In Review).

Simons and O’Connor also documented that the wing bones of birds utilizing different primary flight modes exhibit shape differences reflecting differing amounts (cyclicity) of bending, with the flapping birds possessing relatively elliptical elements when compared to the flap-gliding and soaring birds. The polar moment of area (J) is generally used as an estimate of resistance to torsion. Sections that experience high torsional loads tend to be highly circular with the cortical bone distributed distant from the neutral axis. The wing bones, especially the humeri, of soaring pelecaniforms (pelicans and gannets) exhibit cross sections with a shape optimized to resist to torsional loads, particularly more so than in flapping or flap-gliding birds. An examination of the whole wing shape, including the mean cord length of secondary flight feathers, suggests that relatively long secondary feathers in static soaring birds act as large lever arms and create potentially high torsional loads on the humerus. The polar moment standardized to the maximum radius of a

96 section yields the torsional strength of a bone (Z). A recent study of the torsional strength of the femur and humerus in two procellariiforms (Diomedea exulans and Puffinus griseus) suggests that despite the dynamic soaring and underwater flapping behaviors exhibited by each of these species, respectively, the humerus and femur exhibit similar strength (Habib and Ruff, 2008).

Ongoing research on the wing elements of this other group of marine birds, the pelecaniforms, has shown that the external morphology (lengths and mid-shaft diameters) and cross-sectional geometry (amount and distribution of cortical bone) varies in a predictable way with differences in both whole wing shape and flight behavior (see

Chapters 1, 2; Simons, In Press; Simons and O’Connor, In Review). The objectives of this study are to (1) characterize the morphology (external whole bone and cross- sectional geometry) of the wing skeleton in procellariiform birds, (2) examine the relationship between morphology and flight mode in the group, and (3) compare and integrate these results with previous research on pelecaniform and other birds to examine the generality of findings underpinning the structure and function of the avian forelimb skeleton.

Materials and Methods

Procellariiforms represent a large clade (~125 species) of marine birds consisting of: the Diomedeidae (albatrosses), Procellariidae (gadfly petrels, fulmars, prions, and shearwaters), Hydrobatidae (storm petrels), and Pelecanoididae (diving petrels).

Procellariiforms range in size from the small storm petrels (~ 17 - 84 g) to the large albatrosses (~ 2460-8400 g) (Dunning, 1993; Table 3–1). For the comparative analysis conducted in this study, the MRP (matrix representation with parsimony) Adams

97 consensus seabird supertree (Kennedy and Page, 2002) was used with additional resolution in the shearwaters (Puffinus) added from Austin et al. (2004), and outgroups from Livezy and Zusi (2007) (Figure 3–2). Eleven species sampled as part of this study were not included in any previous phylogenetic analysis. As such, these species were intercalated as soft polytomies into the composite phylogeny near species inferred to be closely related based on natural history information (Purvis and Garland, 1993; Brooke,

2004).

All procellariiforms are capable of utilizing a range of flight behaviors depending on the environmental conditions and circumstances (Murphy, 1936; Brooke, 2004).

However, small birds in this group use relatively more flapping and large birds use more soaring. Additionally, several taxa use underwater flapping in pursuit of prey. Therefore, for this study, taxa were divided into one of two generalist (small and medium birds that use differing degrees of flapping and gliding) and two specialist (large birds that use primarily soaring or birds that are capable of flapping underwater) categories. These grouping were based on information on flight behavior obtained from the literature and are as follows: Dynamic soar (large birds including Diomedeidae and Macronectes),

Underwater flap (Pelecaniodidae and Puffinus), Flap-glide1 (generalist flap-gliders larger than 100 grams, namely the remaining Procellariidae), and Flap-glide2 (generalist flap- gliders smaller than 100 grams, the Hydrobatidae) (Table 3–1) (e.g., Murphy, 1936;

Warham, 1977; Withers, 1979; Pennycuick, 1982; Pennycuick, 1987; Warham, 1990;

Obst and Nagy, 1992; Tickell, 2000; Brooke, 2004; Sachs, 2005).

Whole-Bone Morphometrics

External length and diameter measurements were taken on 530 skeletal specimens from 101 species (mean of 5 individuals per species) of procellariiforms, representing

~80% of species diversity in the group, and two outgroup species (Gavia immer and

Podiceps auritus) from collections at the American Museum of Natural History

(AMNH), the Carnegie Museum of Natural History (CM), the Field Museum of Natural

History (FMNH), the National Museum of Natural History (NMNH), and the Ohio

University Vertebrate Collections (OUVC). Whole bone measures included nine from the forelimb: total bone length, craniocaudal and dorsoventral mid-shaft diameters of the humerus, ulna, and carpometacarpus (CMC). Anatomical orientation of wing bones

(craniocaudal vs. dorsoventral) was determined based on a fully extended wing position and measurements were taken with digital calipers (Mitutoyo Digimatic calipers).

Anatomical reference points and orientations used standardized nomenclature as outlined in Nomina Anatomica Avium (Baumel and Witmer, 1993). In addition, a geometric mean

(GM) was established as a proxy for body size from five additional measurements: femur length, synsacral length, sternal length, sternal width, and height of sternal keel

(Mosimann, 1970; Mosimann and James, 1979; Niemi, 1985). The species means of the calculated GM were significantly correlated with the general species means of body mass from Dunning (1993) (p < 0.0001, r2 = 0.99); therefore the GM is appropriate for use in

regression analyses. To remove the effects of body size in multivariate analyses, the log10 transformed body size estimate (GM) was subtracted from each log10 transformed

variable.

All variables were tested for phylogenetic signal by randomly re-shuffling the data across the tips of the tree 10,000 times and comparing the results to the measured squared length (number of character steps) in the given tree (Blomberg et al., 2003;

Laurin et al., 2004; Kriloff et al., 2008) in Mesquite 2.0 (Maddison and Maddison, 2007).

Significant phylogenetic autocorrelation was present in all variables (p < 0.0001) and justified the calculation of phylogenetic independent contrasts (Felsenstein, 1985), which were computed using the PDAP package (Midford et al., 2005) in Mesquite 2.0

(Maddison and Maddison, 2007). As the phylogeny used in this study represents a composite of multiple analyses based on both molecular and morphological data, branch lengths were arbitrarily set using Pagel’s (1992) method (e.g., Lavin et al., 2008).

The subsequent analyses conducted on external wing element measurements were completed in both historical and ahistorical contexts. An allometric analysis was conducted using a Reduced Major Axis (RMA) regression model for each log10

transformed external measurement, using log10GM as a proxy for body size. RMA lines

were expressed with equations in the form of:

log(y) = b * log(x) + a

t = (b – 1)/SEb

100

where b = slope and SEb = standard error of slope (Zar, 1999). Ahistorical RMA was

performed using RMA: Software for Reduced Major Axis Regression (Bohonak, 2002).

Phylogenetically-informed RMA analyses were completed using the PDAP package

(Midford et al., 2005) in Mesquite 2.0 (Maddison and Maddison, 2007). For slopes of

contrasts, the standard error of the OLS slope was used as the standard error of the RMA

slope following Nudds (2007).

In addition, multivariate Principle Components Analysis (PCA) and Classification

and Regression Tree Analysis (CART) were performed on size-corrected variables in

JMP version 7.0.1. Historical PCA for comparison was completed using the Rhetenor

package (Dyreson and Maddison, 2001) in Mesquite 2.0 (Maddison and Maddison,

2007).

Cross-Sectional Geometry

The internal structure of the mid-shaft region of the forelimb elements was

assessed in a subsample of the species included in the whole bone analysis (Figure 3–2).

The humerus, ulna, and carpometacarpus (CMC) of 17 species (n = 106, mean of 6

specimens per species) were scanned on a GE eXplore MicroCT (μCT) Scanner housed

at Ohio University. Elements were scanned at mid-shaft where the maximum stress is

predicted to occur (Beer et al., 2006). Wing bones were sampled from the right side of

skeletally mature specimens and included both sexes depending on availability of each.

The degree of sexual dimorphism present is variable within procellariiforms. In general,

males are slightly to significantly larger (up to 34% in large fulmars) than females,

depending on the species. However, sexual dimorphism is absent in many species (e.g.,

species within Pelecanoididae) and even reversed in Hydrobatidae, in which females are

101 larger than males. For this study, in order to obtain a large enough sample size, both sexes and specimens of unknown sex were used. Each element was positioned in the µCT scanner with the long axis of the bone parallel to the scanner bed. A 1 mm diameter circle of barium sulfate was placed at mid-shaft to ensure a reliable radio-dense marker for use in the orienting scan data during subsequent quantitative analyses. A scan resolution of

44-46 µm was used to acquire cross sections (x-ray tube voltage = 80 kV, x-ray tube current = 450 μm) on all but the smallest species (Oceanites oceanicus) for which a scan resolution of 21 µm was used. The total length (L) of each element for use in standardizing J was measured using digital calipers (Mitutoyo Digimatic calipers).

Specimens for scanning were borrowed from the Carnegie Museum of Natural History

(CM), National Museum of Natural History (NMNH) and Ohio University Vertebrate