shape analysis of odontocete mandibles: functional and

90
SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND EVOLUTIONARY IMPLICATIONS _______________ A Thesis Presented to the Faculty of San Diego State University _______________ In Partial Fulfillment of the Requirements for the Degree Master of Science in Biology _______________ by Celia Barroso Fall 2010

Upload: others

Post on 12-Sep-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

SHAPE ANALYSIS OF ODONTOCETE MANDIBLES:

FUNCTIONAL AND EVOLUTIONARY

IMPLICATIONS

_______________

A Thesis

Presented to the

Faculty of

San Diego State University

_______________

In Partial Fulfillment

of the Requirements for the Degree

Master of Science

in

Biology

_______________

by

Celia Barroso

Fall 2010

Page 2: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND
Page 3: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

iii

Copyright © 2010

by

Celia Barroso

All Rights Reserved

Page 4: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

iv

ABSTRACT OF THE THESIS

Shape Analysis of Odontocete Mandibles: Functional and Evolutionary Implications

by Celia Barroso

Master of Science in Biology San Diego State University, 2010

Odontocete mandibles serve multiple functions, including feeding and hearing. One hypothesis is that sound enters the hearing apparatus via the pan bone of the posterior mandibles (Norris, 1968). Another viewpoint, based on computer models, suggests that sound enters primarily through the gular apparatus and the opening created by the absent medial wall of the posterior mandibles. The posterior region of each mandibular ramus has a large, hollow cavity called the mandibular foramen that contains a bulging mandibular fat body (MFB), which terminates on the bony tympanoperiotic complex. The acoustic properties of these fat bodies suggest that they refract sound. This unambiguous link between form and function has catalyzed this current study of mandibular shape. Previous studies have described odontocete mandibles using linear morphometrics and focused on multiple populations within single genera. Geometric Morphometrics (GM) is used to avoid some limitations of linear morphometric studies, using relative 3-D landmark positions instead of lengths. This technique measures shape only, excluding any scaling, rotational, and positional effects. The primary objective of this study is to use GM to quantify mandibular shape across all major lineages of Odontoceti. Eighty-five mandibles, comprised of 40 species, representing all major lineages were included in the shape analysis. Twelve landmarks found on each mandible represent regions of the symphysis and mandibular foramen. The majority of shape variation was found in Jaw Flare and Symphysis Elongation (85.5%). Shape differences in the mandibular foramen also accounts for a portion the total variation (10.9%). The mandibles are an integral component of the sound reception apparatus in toothed whales and the geometry of the mandibular foramen likely plays a role in hearing. The second objective of this study is to assess correlates of hearing and mandibular foramen geometry derived from the GM shape analysis, as well as from linear morphometrics. Echolocation peak frequency (EPF) was used as a measure of sound reception. Odontocete anatomy may emphasize frequencies they need to hear and suppress others in a returning echo during echolocation. Frequencies can be characterized by corresponding wavelengths. Surface area between the four landmarks that represent the mandibular foramen was calculated to assess the relationship between EPF and dimensions of a receiving structure (i.e. mandibular foramen and corresponding MFB). A significant relationship between size of the foramen and EPF was found, suggesting that size of the foramen may limit lower frequencies (longer wavelengths) from propagating through the mandibular fat body.

Page 5: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

v

The final objective of this study is to examine the amount of phylogenetic signal in the observed morphological variation as well as examine the patterns of mandibular shape evolution. The phylogenetic signal is significant in Jaw Flare and Symphysis Elongation and mandibular foramen shapes. Mandibular foramen shapes, however, may be shaped by other selective pressures. It appears that the shape of the foramen has an optimum shape in the entire toothed whale lineage, perhaps an optimum for sound reception. Phylogenetic relationships may account for mandibular shape on different levels and selective pressures may drive shape variation in less inclusive clades. Convergences among clades is also apparent within Jaw Flare and Symphysis Elongation shapes, indicating a selective force drove two lineages to similar jaw shapes.

Page 6: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

vi

TABLE OF CONTENTS

PAGE

ABSTRACT ............................................................................................................................. iv

LIST OF TABLES ................................................................................................................. viii

LIST OF FIGURES ................................................................................................................. ix

CHAPTERS

1 INTRODUCTION .........................................................................................................1

Mandibular Anatomy ...............................................................................................2

Mandibular Function and Evolution ........................................................................2

Objectives ................................................................................................................4

2 SHAPE ANALYSIS OF ODONTOCETE MANDIBLES ............................................6

Introduction ..............................................................................................................6

Materials and Methods .............................................................................................9

Mandible Collection...........................................................................................9

Landmark Extraction .......................................................................................10

Shape Analysis .................................................................................................13

Results ....................................................................................................................17

Discussion ..............................................................................................................21

3 IMPLICATIONS OF MANDIBULAR MORPHOLOGY ON SOUND RECEPTION................................................................................................................28

Introduction ............................................................................................................28

Echolocation and Sound Sources .....................................................................28

Sound Reception ..............................................................................................30

Materials and Methods ...........................................................................................31

Results ....................................................................................................................34

Discussion ..............................................................................................................35

4 PATTERNS OF MANDIBULAR SHAPE AND PHYLOGENETIC SIGNAL .........39

Introduction ............................................................................................................39

Materials and Methods ...........................................................................................40

Page 7: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

vii

Results ....................................................................................................................43

Discussion ..............................................................................................................45

5 CONCLUSIONS..........................................................................................................51

ACKNOWLEDGEMENTS .....................................................................................................54

REFERENCES ........................................................................................................................56

APPENDIX

A SPECIMENS IN ANALYSIS......................................................................................63

B SPECIMENS IN ANALYSIS FROM CETACEAN DATA LIBRARY ....................68

C SUMMARY OF SPECIES IN ANALYSIS ................................................................70

D X-RAY COMPUTED TOMOGRAPHY (CT) SCANNING PARAMETERS ...........72

E EIGENVALUES AND PERCENT VARIANCE ........................................................74

F PRINCIPAL COMPONENT COEFFICIENTS ..........................................................76

G SPECIMEN PRINCIPAL COMPONENT SCORES – PC1 TO PC4 SPECIES MEANS........................................................................................................................78

Page 8: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

viii

LIST OF TABLES

PAGE

Table 1. Name and Description of Landmarks Corresponding to Numbered Landmarks in Figure 2 .................................................................................................12

Table 2. Echolocation Peak Frequency ....................................................................................32

Table 3. Specimens in Analysis ...............................................................................................64

Table 4. Specimens in Analysis from Cetacean Data Library .................................................69

Table 5. Summary of Specimens in Analysis ..........................................................................71

Table 6. X-ray Computed Tomography (CT) Scanning Parameters .......................................73

Table 7. Eigenvalues and Percent Variance .............................................................................75

Table 8. Principal Component Coefficients .............................................................................77

Table 9. Specimen Principal Component Scores – PC 1 to PC 4 Species Means ...................79

Page 9: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

ix

LIST OF FIGURES

PAGE

Figure 1. Odontocete mandibles depicting some prominent anatomical features. The red dotted line outlines the alveolar groove………………………………………..….3

Figure 2. Odontocete mandibles depicting the twelve bilateral, homologous landmarks of anatomical features extracted from each specimen; a) a lateral view, b) a posterior view. Red filled circles represent landmarks in the feeding component and green represent landmarks in the hearing component…….…………………………………………………………………..….11

Figure 3. Deformation grid modified from D’Arcy Thompson’s (1917) On Growth and Form………………………….…………………………………….……15

Figure 4. Percent variance plotted against ordinal number of principal components. The first four components account for 91.4% of the total variance……………….....17

Figure 5. Shape change associated with PC 1 – Jaw Flare. a) Inset provides mandible orientation in ventral view. b) The filled circles on each color line represent the twelve landmarks. Green arrows indicate direction of change in shape from red (the mean) to yellow, corresponding to an increase of 0.1 PC score in the positive direction………………………………………………………………..…....18

Figure 6. Shape change diagram of PC 1 (Jaw Flare) depicting an estimate of change in shape corresponding to a 0.5 difference in PC 1 score, encompassing approximately entire range in PC 1 shape. a) Inset provides mandible orientation in ventral view. b) Filled circles on each color line represent the twelve landmarks, green represents the low end of the range and blue the high end, orange arrows indicate directions of shape change.…………………..…..…….19

Figure 7. Shape change associated with PC 2 – Symphysis Elongation. a) Inset provides mandible orientation in ventral view. b) The filled circles on each color line represent the twelve landmarks. Green arrows indicate direction of change in shape from red (the mean) to yellow, corresponding to an increase of 0.1 PC score in the positive direction…………………………………………..…20

Figure 8. Shape change associated with PC 3 – Foramen Elongation. a) Inset provides mandible orientation in medial view. b) The filled circles on each color line represent landmarks on one side of the mandibles. Green arrows indicate direction of change in shape from red (the mean) to yellow, corresponding to an increase of 0.1 PC score in the positive direction………..…….20

Page 10: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

x

Figure 9. Shape change associated with PC 4 – Foramen Expansion. a) Inset provides mandible orientation; in medial view. b) The filled circles on each color line represent landmarks on one side of the mandibles. Green arrows indicate direction of change in shape from red (the mean) to yellow corresponding to an increase of 0.1 PC score in the positive direction…………..….21

Figure 10. Principal component plot of Jaw Flare (PC 1) vs. Symphysis Elongation (PC 2). Colors indicate different clades. River dolphins are comprised of Iniidae, Pontoporiidae, and Platanistidae and are enclosed by the olive lines. Lines enclosing groups are for visual purposes only………………………….….….22

Figure 11. Principal component plot of Foramen Elongation (PC 3) vs. Foramen Expansion (PC 4). Colors indicate different clades. River dolphins are comprised of Iniidae, Pontoporiidae, and Platanistidae and are enclosed by the olive lines. Lines enclosing groups are for visual purposes only………………...23

Figure 12. A waveform of a false killer whale (Pseudorca crassidens) echolocation click with a bimodal (two peak) frequency distribution modified from Au et al. (1999)……………………………………………………………………………...…29

Figure 13. Diagram of triangles formed using four landmarks around the mandibular foramen on a right mandible. Surface area of the mandibular foramen opening is estimated by adding area of top triangle to area of bottom triangle…………...…..34

Figure 14. Plots of echolocation mean peak frequency and morphometric parameters. Filled circles represent different species and solid line represents regression line. a) mean peak frequency with high frequency component vs. PC 4 and b) mean peak frequency with low frequency component vs. mandibular foramen surface area………………………………………………...……………….36

Figure 15. Odontocete phylogeny modified from McGowen et al. (2009). Black asterisks (*) indicate species in the study sample. Filled circles indicate weakly supported nodes.……………………...…………………….………41

Figure 16. Ancestral character state reconstruction of PC 1 (Jaw Flare). Larger filled circles indicate less flared mandibles and smaller filled circles indicate more flared mandibles. Node values were raised to a power of four and divided by 11 to emphasize differences in circle size………………………………………..…..44

Figure 17. Ancestral character state reconstruction of PC 2 (Symphysis Elongation). Larger filled circles indicate less elongate symphyses and smaller filled circles indicate more elongate symphyses. Node values were raised to a power of four and divided by 11 to emphasize differences in circle size………………..….45

Figure 18. Ancestral character state reconstruction of PC 3 (Foramen Elongation). Larger filled circles indicate more elongate foramens and smaller filled circles indicate less elongate foramens. Node values were raised to a power of seven and divided by 85 to emphasize differences in circle size…………………………...46

Page 11: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

xi

Figure 19. Ancestral character state reconstruction of PC 4 (Foramen Expansion). Larger filled circles indicate more expanded and smaller filled circles indicate less expanded foramens. Node values were raised to a power of seven and divided by 85 to emphasize differences in circle size……….…….……...………….47

Page 12: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

1

CHAPTER 1

INTRODUCTION

Cetacea is comprised of two main lineages: mysticetes (baleen whales) and

odontocetes (toothed whales). The toothed whales are, with exceptions, smaller than baleen

whales and also differ in feeding strategy, mechanics, and bioacoustic behavior. Mysticetes

utilize baleen to filter feed on large schools of small fish or crustaceans. Odontocetes,

however, capture and/or suction feed on individual prey items, usually fish or cephalopods.

Odontoceti is a diverse lineage comprised of approximately seventy-three described extant

species (Committee on Taxonomy, 2009).

Odontoceti is divided into nine major lineages (McGowen et al., 2009). Delphinidae

consists of oceanic dolphins. The closely related Phocoenidae is composed of seven porpoise

species. The beluga and narwhals comprise the Monodontidae. The non-monophyletic group

of “river dolphins” includes four lineages: Platanistidae, Iniidae, Pontoporiidae, and

Lipotidae. Ziphiidae, beaked whales, have not been extensively studied, and new species

have recently been described (Dalebout et al., 2002). The sixth major lineage is

Physeteroidea, sperm whales and their allies. Although diverse, odontocetes are united by

many characteristics, including their use of sound as a primary sensory modality.

Cetaceans use their acoustic senses for foraging and communication in order to

exploit life in an aquatic environment with low light penetration. Mysticetes and odontocetes

differ in their sound production and reception capabilities. Generally, mysticetes produce

lower frequency sounds, which can travel long distances; whereas odontocetes produce

higher frequency sounds, often with ultrasonic components.

At least 70 published studies have reported echolocation-type clicks in odontocetes

(e.g. Norris, 1961; review in Au, 1993; Miller et al., 1995; Johnson et al., 2004; Li et al.,

2005; Madsen et al., 2005a,b), suggesting that the entire lineage utilizes echolocation. During

echolocation, the animal produces a sound and then listens for a returning echo from the

target. The received echo can provide identifying information about the target, including

distance to and movement of the target. For many years, the supposition has been that the

mandibles play a major role in sound reception by a mechanism known as “jaw hearing”

(e.g., Norris 1968; Brill et al., 1988; Brill and Harder, 1991; Møhl et al. 1999). As this field

Page 13: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

2

of study advances, the evidence for how and where sounds are received suggests that this

process is more complex than originally conceived (Møhl et al. 1999; Popov et al., 2008;

Cranford et al., 2008a,b). Regardless of the exact mechanism(s), it is clear that the hollow

posterior portion of the mandibles is an integral part of sound reception in odontocetes.

The structure of odontocete mandibles is unlike any other mammalian group and

stimulates questions about its evolution. This study uses quantitative techniques to describe

mandibular shape with an aim toward understanding the function and evolution of this

unique structural complex.

MANDIBULAR ANATOMY

Anatomic geometry of the odontocete mandibles (Figure 1) is described in this study.

The mandibular symphysis is in the anterior region where the two mandibles join. The two

mandibles are joined with a bony suture and/or connective tissue that holds the two

mandibles together. The tooth row, if present, is represented by the alveolar groove. Beaked

whales, which may only have one large toothed erupting on each mandible, still have an

alveolar groove. Just posterior to the end of the alveolar groove is the posterior one-third of

the mandible, which marks the opening of the mandibular foramen and the thinning of the

lateral wall. Finally, the condyles are located at the posterior terminus of the mandibles and

are sites for articulation of the jaws with the skull.

MANDIBULAR FUNCTION AND EVOLUTION

Odontocete mandibles serve a variety of functions, such as feeding and hearing.

Toothed whales often prey on single fish or cephalopods and their mandibles play the role of

manipulating their prey. In a few cases, such as some groups of killer whales, the mandibles

are used in shearing and tearing of prey. In addition to predation, the mandibles are also

implicated in male-male competition (MacLeod, 1998; Scott et al., 2005) and sound

reception (e.g., Norris 1968; Brill et al., 1988; Brill and Harder, 1991; Mohl et al. 1999).

Odontocetes are well adapted to high-frequency hearing in the aquatic environment,

in part due to changes in the hearing apparatus, of which the mandibles are only one

component. Over the course of evolution, changes in the hearing apparatus include the loss or

reduction of the outer ear pinna and external auditory meatus and extensive restructuring of

the mandible. The mandibular foramen has expanded and contains a large mandibular

acoustic fat body with uncommon acoustic properties (Norris, 1968; Nummela et al., 2007).

Page 14: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

3

Figure 1. Odontocete mandibles depicting some prominent anatomical features. The red dotted line outlines the alveolar groove. In contrast to terrestrial mammals that receive sound through bilaterally paired external

pinnae and auditory canals, odontocetes are thought to receive vibrations through multiple

pathways. The predominant hypothesis for the primary sound reception pathway is through

the thin lateral wall (“pan bone”) of the mandible (Norris, 1968). Sound vibrations travel into

the internal mandibular fat body and then to the bony ear complex (Norris, 1968). The term

“acoustic window” is used when describing the thin region in the posterior portion of the

jaws through which sound vibrations were thought to penetrate (Norris, 1968). The

mandibular fat body fills the mandibular foramen and extends somewhat medially beyond the

bounds of the mandible. The lipid components of this structure have properties that may

refract sound and assist in transmitting, or channeling, sound to the ear complex (Varanasi

and Malins, 1972; Koopman et al., 2006). Mandibular fat bodies have a consistent

arrangement of lipids based on specimens representing four main odontocete lineages

(Koopman et al., 2006). It is possible that the shape of the foramen, housing the fat body,

limits the geometry and, therefore, the acoustic function of the fat body.

The evolution of the mandibular foramen and thinning pan bone are evident in the

fossil record. The mandibular foramen was present in pakicetids (oldest known whales) and

became enlarged in later whales (Thewissen et al., 1996). For example, the foramen is larger

in ambulocetids than in the earlier pakicetids (Thewissen et al. 1996; Nummela et al., 2007).

Page 15: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

4

The development of a larger foramen may indicate a development of internal mandibular fat

bodies (Nummela et al., 2007). In addition, changes in thickness of the pan bones have

occurred. The pan bone was relatively thicker in archaeocetes, the ancient whales which gave

rise to modern cetaceans, and thinner in extant whales (Thewissen et al., 1996; Nummela et

al., 2007). The thinnest portion of the lateral wall, or pan bone, ranges from 1.5mm – 3.8mm

in thickness in archaeocetes (Nummela et al., 2007). In modern odontocetes, the thinnest part

of pan bone ranges from 0.3mm – 2.8mm with one exception: the killer whale (Orcinus orca)

measuring 5.5mm (Nummela et al., 2007). The shape of the mandibular foramen and

thickness of the pan bone likely played a role in the development of biosonar and high

frequency hearing in toothed whales.

Recently, Cranford et al. (2008a) discovered an alternative hearing pathway to

Norris’ (1968) “jaw hearing” hypothesis through finite element modeling of Cuvier’s beaked

whale (Ziphius cavirostris). According to their work, sound enters the head through the gular

anatomy (e.g., soft tissues between the lower jaws, tongue, and throat) and enters the internal

mandibular fat body, which attaches to the bony ear (tympanoperiotic) complex, through the

missing medial wall of the mandibular foramen. The model also suggested that the pan bone

of the mandibles may fluctuate to transmit sound only at specific frequencies and from

specific angles. The thick pan bones characteristic of early whales would have impeded

sound from transmitting into the mandibular fat bodies through the pan bone. If the primary

pathway of sound was through the pan bone, archaeocetes may have had a low sensitivity to

sound. The thicker pan bone, in essence, would have acted like shutters over the future

“acoustic window.” Perhaps a more likely explanation is that sound was received via a

different pathway, one in which the pan bone does not inhibit sound reception, such as the

“gular pathway.” Models by Cranford and colleagues (2008a) suggests that the “gular

pathway” may be the ancestral and primary mode of hearing in odontocetes.

The odontocete mandibles serve a variety of functions that drive shape variation

across Odontoceti. The selective pressures that drive changes in shape of the lower jaw may

divide the mandibles into at least two components: generally, an anterior portion driven by

feeding and a posterior component driven by hearing (Perrin, 1975).

OBJECTIVES

The primary objective of this study is to quantify the shape of extant odontocete

mandibles using Geometric Morphometrics (GM) across higher levels of taxa, more inclusive

than the species or generic level. The variance in shape between species may be due to one or

Page 16: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

5

more factors, for example, prey type, feeding mechanics, biosonar behavior, or phylogenetic

relationships. The GM shape analysis is presented in Chapter 2.

Another objective is to assess correlates of hearing and jaw morphology in

odontocetes and examine the evolution of mandibular shape. Norris (1968) hypothesized that

sound travels to the ear complex via the fat bodies that lay internal to the pan bone.

However, Cranford et al. (2008a), produced sound propagation models of a Cuvier’s beaked

whale (Ziphiidae), and suggests that the “gular pathway” of sound reception applies to all

odontocete heads. Understanding correlated functions of mandibular shape may provide

insight into current and ancestral hearing pathways of odontocetes. An assessment of

correlations between mandibular foramen geometry and sound reception is presented in

Chapter 3. Finally, Chapter 4 examines the evolution of mandibular shape and assesses the

amount of phylogenetic signal in the observed morphological variation.

Page 17: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

6

CHAPTER 2

SHAPE ANALYSIS OF ODONTOCETE

MANDIBLES

INTRODUCTION

Morphometric studies of odontocete mandibles have focused on a single species or

genus. These studies have examined sexual dimorphism or compared and contrasted features

of the same species that inhabit different localities, questioning whether they are different

ecotypes (Wang et al., 2000; Murphy et al., 2006; Westgate, 2007). Odontocete mandibular

studies examining higher taxonomic levels have used traditional morphometrics (Werth,

2006; Nummela et al., 2007). Werth (2006) described differences in 34 species of toothed

whales in terms of mandibular bluntness (the ratio of mandibular width to length). This

measure of mandibular bluntness is a simplified description of mandibular shape variation,

rather than providing a detailed depiction of morphological variation. Nummela et al. (2007)

also provided a morphometric study of odontocete mandibles comparing more inclusive

levels of taxa than the species or genus level, but used traditional morphometrics. That study

was also a simplified description of mandibular shape, only examining two morphometric

parameters of the mandibular foramen: greatest height and minimum thickness of the lateral

wall (pan bone). Traditional, or linear, morphometrics consists of comparing distances or

measurements of length between two points across all individuals in a sample. Traditional

morphometrics do not require that these points be homologous landmarks. Homologous

landmarks are points that can be found in all specimens due to a shared ancestry. The

traditional metrics of linear distances sometimes correspond to maxima (e.g., maximum

width) or minima, which may not be defined by homologous points (Adams et al., 2004).

Non-homologous points do not describe the same position on each mandible. These

measurements between landmarks need to be corrected for size in order to remove its effects

on shape variation. Nummela et al. (2007) employed standardized measurements of

mandibular pan bone thickness and foramen height using bicondylar width as an indicator of

size, which was found to be correlated with body mass (Marino et al., 2004). However, body

Page 18: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

7

mass may not be correlated with mandible size. Geometric Morphometrics (GM) uses a

superimposition technique to better assess shape differences while examining a structure as a

whole, not individual distances between points on a structure. Furthermore, the

superimposition technique removes any effects of size such that shape is the only variable

examined.

A study by Allen (2003) used a technique within Geometric Morphometrics to

compare mandibular morphology between two genera of beaked whales (Mesoplodon and

Ziphius), accounting for mandibular shape variation unlike those using standard, linear

morphometrics. Allen’s morphometric study performed a truss analysis in which all pairwise

distances between landmarks were compared. In addition, she corrected for size using a

configuration size, similar to methods used in the present odontocete mandibular shape study.

While a larger number of distances may produce more detailed results about the variation in

mandibular geometry, the relative positions of landmarks are not easily conveyed in the

results with visual graphs. However, descriptions of which distance measurements accounted

for the majority of variation were still possible. For example, a distance between the two

condyloid processes could be gleaned from the results as a measurement which varies greatly

within Mesoplodon and Ziphius.

Geometric Morphometrics (GM) is a relatively new technique for studying shape that

provides tools to visualize shape change. In GM, an entire configuration, or matrix, of

landmarks is considered a single variable, rather than individual distances in between

landmarks. In addition, GM resolves some of the problems that confound linear

morphometrics.

Linear measurements describe size, and, although ratios of lengths are utilized, it may

be difficult to separate size from shape by the ratios (Zelditch et al., 2004). Linear

morphometric studies that use principal components analysis (PCA) create shape variables

from linear combinations of distances/measurements (Zelditch et al., 2004). This analysis

results in components, of which the first is often considered a measure of size, and the rest

measures of shape. However, shape and size information is likely contained in all

components, including the first (Zelditch et al., 2004).

Another limitation of linear morphometrics is that it does not convey relative

positions of the landmarks (Zelditch et al., 2004). For example, if certain landmarks are

closer to each other than to other landmarks, or that a set of three distances might correspond

to a triangle of landmarks on an organism (Rohlf and Marcus, 1993; Rohlf, 1999). A good

Page 19: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

8

example was given by Adams et al. (2004): if maximum length and width were measured for

both an oval and tear drop, the two may have the same results but, clearly their shape

information was not captured.

Furthermore, Geometric Morphometrics allows researchers to visualize how some

landmarks move relative to others to change shape. Visualization tools, such as those

provided by GM, are generally not available using traditional morphometrics. However, there

are still advantages to using traditional morphometrics. One advantage is that a linear

analysis can be used even when data is missing from the analysis.

Geometric Morphometrics is based on multivariate statistics, which allows

researchers to take advantage of tools like principal components analysis (PCA) and

discriminant function analysis. Analyses, such as PCA, used in conjunction with the

superimposition technique within the GM framework, are also useful in answering questions

regarding patterns that bear on ecology and evolution. Potential research questions include,

but are not limited to, morphological similarity/differences between ecological or functional

groups in an effort to explain factors resulting in the observed morphological differences

(e.g., Depecker et al., 2006; Christiansen, 2008; Pierce et al., 2008; Bots et al., 2009).

Furthermore, GM can be used in phylogenetic analyses to determine the extent of

“phylogenetic signal” in the observed shape variation (e.g., Cardini, 2003; Meloro et al.,

2008; Young, 2008; Monteiro and Nogueira, 2009).

Geometric Morphometrics may be used to perform a shape analysis in two

dimensions (2-D) or three dimensions (3-D). This current odontocete mandibular shape

analysis utilizes the powerful tools GM offers using 3-D data. Two-dimensional data is most

often collected and used in GM studies (e.g., Caumul and Polly, 2005). However, shape

analysis in 3-D has gained favor with the advent of new computer-based technology and

software (e.g., Marcus et al., 2000; Terhune, et al., 2007; Young, 2008). The current

odontocete mandibular shape study takes advantage of x-ray Computed Tomography (CT) to

produce 3-D digital images of the mandibles. These images were used to identify landmarks

in 3-D.

Understanding the shape of odontocete mandibles and how they differ, or remain

similar, across higher taxonomic levels may provide clues into understanding how this

structure is influenced by selection, phylogenetic inertia, and the functional implications of

geometry of structures. Selection to hear certain sounds, and/or feed on specific prey items

may drive the shape differences observed in Odontoceti. Conversely, the shape of the bone

Page 20: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

9

might limit the growth of soft tissue structures (e.g., mandibular fat bodies), resulting in

changes to the potential acoustic function of those soft tissues.

MATERIALS AND METHODS

Eighty-five mandibles (56% of described odontocetes), representing major

odontocete lineages were included in this study: 53 delphinids (23 species), 13 phocoenids

(six species), two monodontids (two species), four iniids (one species), five pontoporiids

(one species), one platanistid, one kogiid, and five ziphiids (five species). Adult specimens

were the primary target, but in most instances the age information was not available. Skulls

of corresponding mandibles were examined to determine if cranial sutures had fused; fused

sutures indicated adult age. When available, total length of the stranded animal was used to

estimate age. (Appendices A-C)

Mandible Collection

Mandibles were borrowed from collections at three museums: San Diego Natural

History Museum (SDNHM), National Museum of Natural History (NMNH) at the

Smithsonian Institution, and the Los Angeles County Museum of Natural History (LACM).

These samples were placed in drums or cardboard boxes lined with gelatin packs and then

scanned using x-ray Computed Tomography (CT). A list of specimens used in this project

can be found in Appendix A.

The first mandible scanning events were conducted on July 3 and 5, 2008. Mandibles

borrowed from SDMNH were placed in a drum measuring 45.7 cm diameter x 66.0 cm

height in between layers of therapy hot/cold gel packs. These gel packs not only protected the

mandibles from damage and separated them from one another in layers, but also produced

CT images with fewer artifacts. If the bones are scanned in air, artifacts can be generated due

to the large density difference between the bones and air. Additionally, cotton was used to

prevent the mandibles within a single layer from touching one another and pillow cases were

placed between mandibles and gel packs to prevent the teeth from puncturing the gel packs.

The drum, containing 16 mandibles (Appendix A), was scanned at the University of

California San Diego (UCSD) Medical Center, Hillcrest in a GE LightSpeed CT scanner. A

table of the scanners, protocols, and locations can be found in Appendix D.

Two more scanning events were conducted on July 14 and 15, 2008 at the NMNH.

Mandibles were placed in two boxes (36cm x 36cm x 92cm), which were lined with

Page 21: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

10

homemade gel packs. These gel packs were made using unflavored gelatin in a ratio of four

0.25oz packets of powder to 1cup of hot water, and allowed to cool to a solid at room

temperature in one gallon Zip-Lock bags. These gel packs were placed between layers of

mandibles, with pillow cases preventing the teeth from puncturing the gel packs, and pieces

of foam preventing the mandibles within a layer from touching each other. A total of 36

mandibles (Appendix A) were scanned in these boxes in a Siemens Somatom Emotion CT

scanner. See Appendix D for scanning parameters.

The last set of CT scans were collected on December 29 and 30, 2008. Mandibles

were borrowed from LACM and scanned at University of California, Los Angeles Medical

School in a Siemens Somatom Definition CT scanner (Appendix D). Twenty-three mandibles

(Appendix A) were prepared for scanning using the same method and supplies as the

September 2008 SDMNH event.

In addition to the jaws borrowed from museums, Dr. Ted Cranford at San Diego State

University has a cetacean data library of CT scans of entire odontocete heads (see Appendix

B). I also extracted mandible data from specimens in that library.

Landmark Extraction

The x-ray CT scan images are processed using Analyze 9.0 (Robb, 2001). A

subvolume containing each mandible was extracted from the scanned volumes. Each

mandible was then re-oriented so that all mandibles have a common orientation. The

mandibles were then segmented from the subvolume, effectively isolating the jaws.

Segmenting is the process of defining boundaries of the mandibles in each image. The result

is a 3-D representation of the defined region, which in this case is each mandible. Finally,

homologous landmark locations, those found on each mandible because they are derived

from a common ancestor, are defined by the corresponding (x, y, and z) coordinates.

Twelve homologous landmarks (10 bilateral and two along the midline) were chosen.

See Figure 2 and Table 1 for descriptions and definitions of the landmarks chosen from the

anterior and posterior portions of the mandibles for GM analysis. The landmarks include

points on the symphysis, alveolar groove, mandibular foramen, and condyles.

Page 22: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

11

a)

b)

Figure 2. Odontocete mandibles depicting the twelve bilateral, homologous landmarks of anatomical features extracted from each specimen; a) a lateral view, b) a posterior view. Red filled circles represent landmarks in the feeding component and green represent landmarks in the hearing component.

Six mandibles contained deformities, either in a condyle or the symphysis did not

appear completely fused. The single Ganges river dolphin (Platanista gangetica) mandible

collected and CT scanned was missing part of the right condyle. The condyle was

reconstructed using the mirror image of the left condyle for the missing parts. Similarly, the

only Baird’s beaked whale (Berardius bairdii) mandible also has a deformed condyle. The

right, intact, condyle was used to approximate the shape of the left, deformed, condyle. One

of the three Risso’s dolphins (Grampus griseus) mandibles (Specimen ID# 95836) had a very

loose symphysis, indicating it was degraded, but the jaws were still intact. For this specimen,

an estimate based on external features, such as rough edges, was made to determine the

location of where the two jaws would have initially fused. A similar process was used to

determining the posterior and anterior symphyseal landmarks for three other jaws: vaquita,

Phocena sinus, (ID Vaquita 2) and both narwals (Monodon monoceros). In some species,

Page 23: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

12

Table 1. Name and Description of Landmarks Corresponding to Numbered Landmarks in Figure 2

LANDMARK DESCRIPTION 1 Right condyle centroid (RCC) Centroid of right condyle 2 Left condyle centroid (LCC) Centroid of right condyle 3 Right ventral keel (RVK) Point where medial wall intersects

lateral wall on ventral right side 4 Left ventral keel (LVK) Point where medial wall intersects

lateral wall on ventral left side 5 Right dorsal keel (RDK) Point where medial wall intersects

lateral wall on dorsal right side 6 Left dorsal keel (LDK) Point where medial wall intersects

lateral wall on dorsal left side 7 Right anterior foramen (RAF) Anteriormost point on medial wall

of mandibular foramen 8 Left anterior foramen (LAF) Anteriormost point on medial wall

of mandibular foramen 9 Right posterior end of alveolar groove (RPAG) Posteriormost point of alveolar groove

on right side 10 Left posterior end of alveolar groove (LPAG) Posteriormost point of alveolar groove

on left side 11 Posterior symphysis centroid (PSC) Centroid of the posterior-most point of

fusion between the two mandibles 12 Anterior symphysis centroid (ASC) Centroid of the posterior-most point of

fusion between the two mandibles

complete fusion of the mandibular symphysis occurs at old age (Rommel, et al., 2009). In

addition, photographs of mandibles collected from museums were used to assist in

determining the most appropriate location of these landmarks.

Landmark extractions were repeated to assess sampling variation. Landmark

extraction for eleven mandibles was repeated twice and an analysis of variation (ANOVA)

was performed in R (R Development Core Team, 2009) to determine if the two sampling

periods were significantly different. In addition, landmarks were extracted for a single

mandible five times. An ANOVA was also performed (in R) to determine if the landmark

extraction varied for each of the sampling periods.

The accuracy of estimated landmark extraction was also assessed. Three specimens,

of which the symphyseal landmarks were estimated, are of the same species of at least one

other specimen (not estimated) in the analysis. Each landmark has associated Procrustes

Page 24: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

13

coordinates (see next section for a detailed description of these coordinates), which represent

morphological variation of each specimen. For the symphysis landmarks, the full range of

Procrustes coordinates encompassing the entire range in morphological variation of the entire

sample was calculated. The estimated specimens were then compared to the non-estimated

specimens of the same species. The difference in Procrustes coordinates from the estimated

specimen to non-estimated specimen was calculated. The Procrustes coordinates of the

estimated specimen was subtracted from the coordinates of the non-estimated specimens.

Their percent similarity was then calculated by dividing the absolute value of the difference

by the full range of Procrustes coordinates. The goal is to determine how close estimated

specimens are to non-estimated specimens relative to the entire range in morphological

variation associated with symphyseal landmarks.

Shape Analysis

The landmark configurations are converted into shape variables during the initial step

in Geometric Morphometric (GM) shape analysis. Shape variables represent the differences

in landmark configurations, between individual jaws and a “consensus specimen.” The

consensus specimen is an average of all specimens in the analysis. This shape difference

calculation is made using a technique known as Generalized Procrustes Analysis (GPA),

named after a Greek mythological character, Procrustes. Procrustes, meaning “he who

stretches,” would lure victims into his home with the illusion of a magical bed on which

anyone could fit perfectly. He failed to mention that his methods of making everyone fit

perfectly onto the iron bed included either cutting or stretching limbs. Now, Procrustes is

used in reference to methods which take varying measurements and set them to an arbitrary

standard.

Shape is the remaining geometric information, once location (or position),

orientation, and scale effects are removed from an object (Kendall, 1977). A partial

Procrustes superimposition (Dryden and Mardia, 1998) removes the effects of position, scale,

and orientation. The landmark configuration for each specimen has a centroid, the average of

all x, y, and z coordinates (of all 12 landmarks for that specimen). The centroid size of the

configuration of each specimen is the square root of the sum of squared distances from the

centroid to landmarks. Positional effects are removed by moving the centroid of the landmark

matrix to zero for each specimen. Scaling effects are removed by setting the centroid size to

one for each specimen. Rotational effects are removed by rotating the matrices such that the

partial Procrustes distance between the matrices is minimal. During the rotation step, the

Page 25: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

14

homologous landmarks of each specimen are to match in position as closely as possible to

the corresponding landmarks of all other specimens (Rohlf and Marcus, 1993), minimizing

the partial Procrustes distance between corresponding landmarks. This process removes all

non-shape variation before the rest of the analysis is completed. The process generates an

“average specimen” that is also used to calculate the shape variables. The average, or

consensus, specimen is the mean configuration of all superimposed individuals in the

analysis. Deviations (or Procrustes distances) of each specimen from the mean, or average

specimen, comprise the shape variables which are analyzed using multivariate statistical

tools. The goal is to minimize the Procrustes distance, the sum of squares of residual

distances (or deviations) between matched landmarks (Bookstein, 1997). In contrast to the

partial Procrustes superimposition described above, a full Procrustes fit, or superimposition,

requires an additional step by scaling the target object a second time, to a centroid size of

cos(ρ), where ρ is the Procrustes distance (Rohlf, 1999; Zelditch et al., 2004). This second

round of scaling makes a full Procrustes fit more robust against the influence of outliers by

putting less weight on the observations that are far from the average specimen.

An alternative to analyzing the Procrustes distances directly is a thin-plate spline,

which can be used to map local deformations in shape from one object to another (Bookstein,

1991). This technique stems from D’Arcy Thompson’s (1917) view of shape change as

deformation of a Cartesian plane or grid (Figure 3). This can be thought of as having a sheet

of metal with the landmark points situated in a plane. When the landmarks shift, they “pull”

or deform the sheet of metal accordingly. This deformation is calculated and termed ‘partial

warps.’ The differences in partial warps as one object is deformed into another represents a

shape difference between two objects. Those differences are the shape variables used in

multivariate analysis.

A comprehensive GM software package, MorphoJ version 1.02d (Klingenberg,

2011), was used to analyze mandibular shape. It provides a variety of statistical tools,

including PCA, in addition to a graphical interface that depicts shape variation for 3-D data.

MorphoJ implements a full Procrustes fit. The Procrustes fit in MorphoJ includes steps to

align specimens for the purpose of graphic displays and pair landmarks in datasets with

bilateral symmetry, such as odontocete mandibles or a human face. All specimens were

aligned relative to the principal axes of the mean configuration (default option), in which the

first axis (e.g., x-axis) is oriented parallel to the long axis of the configuration. All specimens

Page 26: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

15

Figure 3. Deformation grid modified from D’Arcy Thompson’s (1917) On Growth and Form.

now share a common alignment. The aligned coordinates are then projected on a linear

tangent space to utilize multivariate analyses. The linear distances between specimens at each

landmark are approximations of the Procrustes distances, which are optimally minimized in

the Procrustes fit. Bilateral landmarks were correctly paired by the program.

To account for the redundancy of bilateral landmarks resulting from symmetry of the

mandibles, MorphoJ amends the dimensions of the data per suggestions by Klingenberg et al.

(2002). Redundant data points are those that can be derived from another point. Parts of

objects with symmetry are repeated in a different orientation and position, and are therefore

partly redundant (Klingenberg et al., 2002). The redundancy inflates the number of degrees

of freedom, thus requiring a larger sample size for proper analysis (Klingenberg et al., 2002;

Zelditch et al., 2004). A standard procedure is to reflect one side across the midline and

calculate the average of the original coordinates and reflected coordinates (Zelditch et al.,

2004). The Procrustes fit is then performed on the average. MorphoJ reflects the left side

onto the right, and then the right onto the left (Klingenberg et al., 2002). On each side, an

average of the original and reflected coordinates is computed and subjected to a full

Procrustes fit (Klingenberg et al., 2002). The full Procrustes fit is performed on “reflection

Page 27: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

16

shape” (Dryden and Mardia, 2000). These methods now reduce the number of degrees of

freedom, now correctly representing the number of independent data points.

Concerns regarding the validity of using reflection shape rather than the original

landmark data may arise if the structure in question is asymmetrical. In order to test whether

odontocete mandibles exhibit asymmetry, the distances between the four landmarks

describing the mandibular foramen (#1, 3, 5, 7 on the right, and #2, 4, 6, 8 on the left in

Figure 2) were used to calculate surface area (mm2) of the mandibular foramen on each side.

The foramen surface areas of the two sides were then compared using methods outlined in

Cranford et al. (1996). A similarity quotient was calculated by dividing the area of the left

side by the area of the right side. A disparity index (DI) was calculated by subtracting the

similarity quotient from 1. A value of zero indicates that the two sides are of equal surface

areas. A value of 0.5 indicates that one side is twice as large as the other. The mean DI of all

specimens in the analysis was close to zero (x̄ = 0.028±0.032) indicating there is little to no

asymmetry in odontocete mandibles. In addition, the foramen landmarks used to calculate DI

constitute eight of the ten bilateral landmarks, representing the majority of the symmetrical

shape included in this study’s GM analysis. As a consequence of the absence of asymmetry,

using reflection shape would not significantly alter the data.

Principal components analysis of shape variables was used to explore shape variation

of odontocete mandibles. This analysis also reduces the number of variables used for later

analyses by combining highly correlated variables, and focusing only on significant

descriptors of shape variation, those that produce the most variation in the sample. The first

principal component (PC) accounts for the greatest amount of variation per PC in a sample

and the following PCs are subsequently smaller than the previous PC.

The PCA was performed on the symmetrical component of the Procrustes fitted

dataset in MorphoJ. In other words, the PCA was performed on a set of shape variables

meant to compare whole structures, not for studies of asymmetry (Klingenberg, 2011). The

most significant PCs were selected by both plotting the percent variance by the ordinal

number of principal components (Figure 4) and also determining how much of the total

variation was described with the PCs chosen. Although the inflection point of the bar graph

(Figure 4) suggests only the first two PCs are most significant, the third and fourth PCs are

included in this analysis because the total variation explained by the first four PCs (91.4%) is

Page 28: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

17

Figure 4. Percent variance plotted against ordinal number of principal components. The first four components account for 91.4% of the total variance.

more inclusive (Appendix E). In addition, important shape information may be contained in

the third and fourth PCs. The individual specimens and their PC scores were then plotted in a

morphospace, which describes aspects of morphological variation (Wills and Fortey, 2000).

A morphospace is a theoretical or empirical construct with reference to any quantifiable

parameter of form (Wills and Fortey, 2000). The parameters of form used in this analysis are

the principal component axes and their plots, which constitute the morphospace.

RESULTS

The first four principal components explain 91.4% of variation in the shape of

odontocete mandibles. Principal component coefficients may be found in Appendix F.

Principal component 1 (45.0%) describes a change in flare (Norris, 1968), or splay, of the

mandibles at the condyles, as well as elongation of the symphysis (Figure 5). The relative

movement of the condylar landmarks (#1 and #2 in Figure 3) in opposite directions

associated with PC 1 cause a narrowing of the jaw. In addition, as the mandibles narrow at

the condyles, the symphysis (#12 in Figure 3) elongates anteriorly. This shape change will be

referred to as ‘Jaw Flare’ and the shape change diagram describes the direction of shape

Page 29: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

18

Figure 5. Shape change associated with PC 1 – Jaw Flare. a) Inset provides mandible orientation in ventral view. b) The filled circles on each color line represent the twelve landmarks. Green arrows indicate direction of change in shape from red (the mean) to yellow, corresponding to an increase of 0.1 PC score in the positive direction.

change associated with a 0.1 increase in PC score (Figure 5). Each mandible will have a

corresponding PC 1 score, as well as scores that correspond to subsequent PCs. The

individual PC scores represent a particular shape within the full range of shape variation

described by that PC. The shape change diagram (Figure 5) does not show the full range of

shape variation, but instead only the shape change and direction which corresponds to a

change in 0.1 PC score in the positive direction. These shape change diagrams may be

thought of as describing direction, not necessarily magnitude. However, a shape change

diagram may be estimated to show a larger range of shape variation (Figure 6). The range of

Jaw Flare shape variation corresponding to approximately a 0.5 difference in PC score is

depicted. This difference of 0.5 encompasses approximately the entire range in shape

variation associated with PC 1. The remaining three shape change diagrams (Figures 7, 8,

and 9) correspond to a 0.1 increase in PC score and ultimately describe the direction of shape

change.

The second PC (35.5%) describes variation in the length of the symphysis (Figure 7)

and is referred to as ‘Symphysis Elongation.’ The shape change associated with PC 2 is the

Page 30: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

19

Figure 6. Shape change diagram of PC 1 (Jaw Flare) depicting an estimate of change in shape corresponding to a 0.5 difference in PC 1 score, encompassing approximately entire range in PC 1 shape. a) Inset provides mandible orientation in ventral view. b) Filled circles on each color line represent the twelve landmarks, green represents the low end of the range and blue the high end, orange arrows indicate directions of shape change. relative movement of the two symphyseal landmarks (#11 and #12 in Figure 3) in opposite

directions either shortening or elongating the symphysis.

The third (Figure 8) and fourth (Figure 9) PCs describe changes in shape of the

mandibular foramen, as well as a shift in position of the posterior limit of the alveolar

groove. Principal component 3 (6.3%) is represented by the relative movement of two

bilateral landmarks: ventral keel (#3 and #4 in Figure 3) and posterior alveolar groove (#9

and #10 in Figure 3). In a medial view (Figure 8), the foramen shape changes due to a shift in

the ventral keel landmarks and appears to narrow, while the posterior limit of the alveolar

groove shifts to a more anterior position. This third PC is referred to as ‘Foramen

Elongation.’ Principal component 4 is represented by relative movement of two different,

bilateral foramen landmarks: dorsal keel (#5 and #6 in Figure 3) and anterior foramen (#7

and #8 in Figure 3). The change in shape appears be an expansion or a contraction of the

foramen and so, PC 4 is referred to as ‘Foramen Expansion.’

The PC scores comparing clades illustrated that Iniidae, Platanistidae, and

Pontoporiidae, as expected, have relatively elongate symphyses, as opposed to phocoenids

(Figure 10). One might have expected river dolphins and sperm whales to occupy the same

Page 31: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

20

Figure 7. Shape change associated with PC 2 – Symphysis Elongation. a) Inset provides mandible orientation in ventral view. b) The filled circles on each color line represent the twelve landmarks. Green arrows indicate direction of change in shape from red (the mean) to yellow, corresponding to an increase of 0.1 PC score in the positive direction.

Figure 8. Shape change associated with PC 3 – Foramen Elongation. a) Inset provides mandible orientation in medial view. b) The filled circles on each color line represent landmarks on one side of the mandibles. Green arrows indicate direction of change in shape from red (the mean) to yellow, corresponding to an increase of 0.1 PC score in the positive direction.

morphospace because they both appear to have elongate symphyses. However, they occupy

different quadrants (Figure 10). The difference is seen in Jaw Flare – the kogiid has a more

elongate symphysis as well as flared mandibles, whereas Iniidae, Platanistidae, and

Pontoporiidae have elongate symphyses and less flared mandibles (Figure 10). The

delphinids encompass much of the shape variation observed in other lineages (Figure 10),

except the kogiid and river dolphins. River dolphins are similar to kogiids in shape variation

Page 32: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

21

Figure 9. Shape change associated with PC 4 – Foramen Expansion. a) Inset provides mandible orientation in medial view. b) The filled circles on each color line represent landmarks on one side of the mandibles. Green arrows indicate direction of change in shape from red (the mean) to yellow corresponding to an increase of 0.1 PC score in the positive direction. described by PC 2, Symphysis Elongation, whereas the ziphiids and some delphinid

mandibles have comparatively shorter symphyses (Figure 10).

Plots of specimens onto morphospaces describing the shape of the mandibular

foramen do not exhibit separation of major lineages. There is much overlap in the

morphospace occupied by delphinids, phocoenids, monodontids, and river dolphins (Figure

10). Beaked whales occupy a smaller range of space on the foramen expansion (PC 4) axis

(Figure 11), but the sample size of this group is low compared to delphinids, porpoises, or

river dolphins.

The variation in landmark extraction was minimal between different sampling

periods. There was no difference between two landmark extraction events in the eleven

mandibles repeated twice (P > 0.05). There was also no difference in the landmark

extractions for the single mandible repeated five times (P > 0.05).

Mandibles that contained estimated symphyseal landmark coordinates were similar to

other specimens of the same species. The estimated and non-estimated specimens fell within

10% of the full range of Procrustes coordinates. Any errors likely did not greatly affect the

final results. Furthermore, the estimated specimens clustered with non-estimated specimens

of the same species in the morphospace (Figures 10 and 11) and only represent 7.1% of the

total sample.

DISCUSSION

Geometric Morphometrics has provided tools to analyze geometry and shape of

biological structures. There are many questions to which answers can be elucidated using

Page 33: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

22

Figure 10. Principal component plot of Jaw Flare (PC 1) vs. Symphysis Elongation (PC 2). Colors indicate different clades. River dolphins are comprised of Iniidae, Pontoporiidae, and Platanistidae and are enclosed by olive lines. Lines enclosing groups are for visualization purposes only.

GM, some of which are addressed in this chapter. This study has attempted to describe the

variation in mandibular shape across all of Odontoceti. A principal component analysis has

allowed visualization of the position of different specimens in a morphospace, effectively

showing how each compare morphologically.

The majority of shape variation appears to be in the anterior “feeding” region of the

mandibles. Perrin (1975) conducted a morphometrics study on spinner dolphins and

concluded that odontocete mandibles could potentially be split into feeding and hearing

components. The anterior region, which includes the tooth row if present, is the feeding

region and the posterior region, containing the mandibular foramen, is the hearing

component. The first two PCs appear to describe changes to the feeding apparatus, changes

Page 34: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

23

Figure 11. Principal component plot of Foramen Elongation (PC 3) vs. Foramen Expansion (PC 4). Colors indicate different clades. River dolphins are comprised of Iniidae, Pontoporiidae, and Platanistidae and are enclosed by olive lines. Lines enclosing groups are for visualization purposes only.

in Symphysis Elongation and Jaw Flare. This may reflect a spectrum of feeding strategies

employed by odontocetes.

Odontocete feeding strategies were most recently reviewed and classified into three

categories by Werth (2000b). Suction feeding, in which the hyolingual apparatus creates

negative pressure in the intraoral cavity, allows some whales (i.e. strap-toothed whale,

Mesoplodon layardii), with a very restricted oral cavity opening, to capture prey. A

secondary method, raptorial feeding, exhibited by river dolphins, includes snapping or biting

Page 35: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

24

of prey before ingestion. The third strategy described is ram feeding and has been observed

in bottlenose dolphins, Tursiops truncatus (Bloodworth and Marshall, 2005). During ram

feeding, the whale engulfs its prey with the power of forward propulsion. All odontocetes

appear to utilize one or more of these strategies. The abundance of variation across toothed

whale species suggests that they do not strictly use one of the strategies mentioned above,

making assigning a feeding mode to each species a daunting task. However, further

examination of feeding strategies and prey preferences in relation to mandibular shape is

warranted.

Suction feeding has recently been studied in greater detail. Werth (2006) suggested

that mandibular bluntness (length:width) is correlated to suction feeding – suction feeders

have wider jaws for a given length. Principal component 1, Jaw Flare, describes a similar

shape (Figure 5). Extreme suction feeders include sperm whales and beaked whales, which

are deep divers and primarily feed on cephalopods (Santos et al., 2006; Walker et al., 2002).

The only sperm whale in this analysis (Kogiidae), does have more flared, or blunt mandibles.

Beaked whales (Ziphiidae), however, appear to have less flared mandibles. They do not share

the morphospace with pygmy sperm whales (Kogiidae), or another known suction feeder,

beluga whales (Monodontidae) (Figure 10). This may confirm that jaw flare, or mandibular

bluntness, is only a part of the morphology associated with suction feeding. Jaw and lingual

muscle morphology (Heyning and Mead, 1996; Werth, 2000a), in conjunction with bone

morphology, may be driven by the advantages of suction feeding strategies.

The variation in jaw flare, due to movement of the condyles is also confirmed by

Allen’s (2003) study of beaked whales. Although the beaked whale mandibles study utilized

a truss analysis, the PCA also found that the distance between the condyles accounted for a

large part of the first PC. The distance between the condyloid processes and alveoli were also

a large contributing factor in PC 1 for the beaked whale analysis. This may be interpreted as

a change in position of posterior alveolar groove, which is a contributing shape change in the

third PC in my GM mandibular shape study across all Odontoceti.

Although feeding strategies vary greatly within Odontoceti, PC 1 (Jaw Flare) and PC

2 (Symphysis Elongation) appear to have phylogenetic signal. In particular, the specimens

cluster at the family level along PC 1 axis (Figure 10). It is possible that both feeding strategy

Page 36: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

25

and evolutionary history contribute to shaping the symphysis and mandibular flare. This

arrangement is not observed along the axes of PC 3 and PC 4 (Figure 11).

Shape of the mandibular foramen (PC 3 and PC 4), while less prominent than the first

two PCs, accounts for a combined 10.9% of total shape variation. The mandibular foramen is

a key component in the sound reception apparatus (Norris, 1968; Cranford et al., 2008a,b)

and changes in shape may reflect differences in sound reception characteristics. Principal

component plots of PC 3 vs. PC 4 (Figure 11) indicate the absence of a strong phylogenetic

signal. The shapes associated with PC 3 and PC 4 may be driven more by environmental, or

ecological, factors as opposed to those associated with PC 1 (Jaw Flare) and PC 2

(Symphysis Elongation). The effects of these shape differences on sound reception are

further explored in Chapter 3 and the amount of phylogenetic signal in observed

morphological variation is assessed in Chapter 4.

Interestingly, Commerson’s dolphins (Cephalorynchus commersonii) and harbor

porpoises (Phocoena phocoena) overlap in the morphospace of PC 3, but not PC 1 and PC 2.

The two species inhabit very similar habitats and are thought to be convergent in many

aspects of their morphology, biology, and ecology (Read, 2002; Galatius, 2010). In addition,

C. commersonii and P. phocoena produce very similar sounds (Madsen et al., 2005a; Kyhn et

al., 2010), perhaps indicating that the two species hear similar sounds, or share prey capture

or predator avoidance strategies. This might confirm that shapes associated with PC 3 are

subject to environmental selective pressures more so than Jaw Flare and Symphysis

Elongation.

Delphinids appear to encompass the morphospace of most odontocete lineages. This

is a very diverse group of odontocetes, as is evident by their shape distributions in the PC

plots. However, this large amount of variation may be a reflection of the radiation event

leading to the species diversity observed in Delphinidae. There are 36 extant species within

Delphinidae, comprising over one-third of the total species within the Odontoceti

(Committee on Taxonomy, 2009).

Odontocetes differ greatly in size. Species that overlap in the morphospace span a

large range of sizes. Killer whales (Orcinus orca), which can reach 9m as adults (Reeves et

al., 2002), appear to have similar Jaw Flare and symphysis shapes to the much smaller

porpoises, which reach 1-3m as adults (Reeves et al., 2002). Similarly, the large false killer

whales (Pseudorca crassidens), is also morphologically similar in mandibular shape to

Page 37: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

26

porpoises. Recall that Geometric Morphometrics provides the means to analyze the shape of

structures which vary greatly in size.

Geometric Morphometrics, as with many analytical techniques, is not without

limitations. For example, one problem encountered in this study was that all landmarks must

be included for all specimens to use in the analysis (Adams et al., 2004). Some mandibles in

museum collections that are rare species had mandibles that were not intact. Therefore, I

could not include them in this GM shape analysis. In addition, because many fossil whale

mandibles are incomplete, they were not included in the current analysis.

Due to the requirement that data sets be complete, studies of very diverse groups of

taxa are difficult. Marcus et al. (2000) conducted a study across many orders of mammals

and found that variation was not as large as anticipated. The authors recognized that the

number of landmarks included in the analysis so that all individuals could be included was

limited, therefore limiting the range of variability the analysis revealed. This presents

difficulties when trying to include extinct taxa in the analysis. Fossils are often incomplete

and extracting all of the landmarks used in specimens of extant species is often not possible.

Another limitation of GM may result from incomplete coverage of shape from

landmarks chosen for a particular study. Landmark-based data, which is also commonly used

in traditional morphometrics studies, does not completely describe a specimen. The

landmarks do not contain information about the curves between them (Richtsmeier et al.,

2002). However, there are methods of selecting ‘semi-landmarks’ representing points

between landmarks (Zelditch et al., 2004). The inclusion of semi-landmarks in the GM

analysis of toothed whale mandibles would likely provide greater information on the true

shape and potentially, the functional differences associated with shape variation.

Geometric Morphometrics may also be used for studies of asymmetry. Klingenberg

and McIntyre (1998) used Procrustes coordinates to measure asymmetry. Many delphinids

have very asymmetrical skulls and associated soft tissues whereas other odontocetes, such as

porpoises, exhibit less extreme cranial asymmetry (Cranford et al., 1996). A GM study of

odontocete cranial asymmetry could supplement existing studies of odontocete skull

asymmetry.

This study presents the first morphometric analysis of Odontoceti that examines

entire mandibles. Nummela et al. (2007) closely examined the linear measurements of the

mandibular foramen, but not the whole jaw. Beaked whale mandibles were also closely

examined (Allen, 2003), but other toothed whales were not incorporated into that study. By

Page 38: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

27

incorporating Geometric Morphometrics, mandibular shape was explored. There are potential

sources of error in a few specimens from this study. Six mandibles were either missing a

piece of the condyle, or had a loose symphysis. For these specimens, the locations of

condylar and symphyseal centroids were estimated by the techniques outlined in the

Materials and Methods section. The resulting errors are unlikely to be significant because

estimated specimens clustered with the non-estimated specimens of the same species. Also,

the number of estimated mandibles was not a large portion (7.1%) of the total sample.

Furthermore, the sample is taxonomically diverse and which represents the spectrum of

shape variation in toothed whales.

Page 39: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

28

CHAPTER 3

IMPLICATIONS OF MANDIBULAR

MORPHOLOGY ON SOUND

RECEPTION

INTRODUCTION

This section first discusses echolocation signals in odontocetes and sources of the

echolocation sounds. Next, sound reception mechanisms are introduced.

Echolocation and Sound Sources

Cetaceans have adapted the use of sound to forage, navigate, and communicate in an

aquatic environment which has low visibility. Odontocetes, specifically, have evolved the use

of echolocation, in which a sound is produced by the whale and then that whale listens for a

returning echo bouncing off of a target. Echolocation type signals have been reported in over

70 studies in a variety of odontocete species, indicating that the entire clade utilizes

echolocation (e.g., review in Au, 1993; Miller et al., 1995; Johnson et al., 2004; Li et al.,

2005; Madsen et al., 2005a,b). Evidence for echolocation in dolphins was described early,

but not until a study by Norris et al. (1961) was the discrimination ability of dolphins using

echolocation demonstrated. This was the first study to blindfold a toothed whale (bottlenose

dolphin, Tursiops truncatus) and conduct experiments testing the animal’s ability to

discriminate between inanimate objects and fish. In addition, it was noted that the

echolocation beam appeared to be directional (Norris et al., 1961).

Echolocation signals generally consist of pulses with relatively high frequency

spectra, often containing ultrasonic components. Furthermore, these signals often show a

bimodal distribution (Figure 12) in their frequency spectrum (Au et al., 1995). A bimodal

spectrum is defined as amplitude in a secondary peak that is greater than 1/3 amplitude of

primary peak (Au and Wursig, 2004). In the case of unimodal and bimodal spectral

distributions, odontocetes emphasize certain frequencies, perhaps those most useful for

Page 40: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

29

Figure 12. A waveform of a false killer whale (Pseudorca crassidens) echolocation click with a bimodal (two peak) frequency distribution modified from Au et al. (1995).

echolocation. Because structure and function are connected, it has been suggested that the

structure of sound generators is causing the bimodal distribution (Cranford et al., 1996).

Early studies hypothesized that sound in odontocetes was generated in the larynx, much like

their terrestrial relatives (Schenkkan, 1973). Later, it was shown that most soundsare

produced in the nasal region (Diercks et al., 1971; Hollien et al., 1976; Dormer, 1979;

Mackay and Liaw, 1981; Amundin and Andersen, 1983), but the exact location and

mechanism remained unknown. In 1997, Cranford and colleagues described results of using

an endoscopy to examine sound generation in a bottlenose dolphin (Tursiops truncatus).

Studies also described the anatomy of two potential sources of sound, possibly accounting for

the bimodal frequency distribution observed (Cranford et al., 1996, 2008a, 2008b).

Through x-ray Computer Tomography images (CT), Cranford et al. (1996) described

the anatomy of sound generators in odontocetes. They proposed that the site of sound

generation is in the dorsal bursae in the monkey lips, leading to the term monkey lip/dorsal

bursae (MLDB) complex, which is homologous in Odontoceti. The monkey lips were later

referred to as phonic lips (Cranford, 2000). The location of sound generation was confirmed

using endoscopic video placed in the nasal region of a captive dolphin (Cranford et al.,

1997). The dorsal bursae, made of fatty tissue, are in pairs, each pair located bilaterally in the

nasal complex. The bursae are imbedded within the phonic lips. During experiments with a

Page 41: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

30

live animal, an endoscope was inserted into the nasal complex via the blowhole of a

bottlenose dolphin (Cranford et al., 1997). They found that the pulse of the phonic lips were

synchronous with acoustic pulses and confirmed that the phonic lips were at the source of

pulsed sound production in odontocetes. These phonic lips and bursae, in one form or

another, are present in all odontocetes.

Cranford et al. (1996) described the asymmetry of sound generators in modern

odontocetes and made some inferences about the potential function of the asymmetry. Those

odontocetes which exhibit a bimodal frequency distribution also have highly asymmetrical

craniofacial features, including the dorsal bursae. In the case of bottlenose dolphins, the right

sound generators are approximately twice the size of the sound generators on the left,

perhaps providing a structural explanation for the presence of a bimodal frequency

distribution. On the other hand, Commerson’s dolphins (Cephalorhynchus commersonii) and

harbor porpoises (Phocoena phocoena) both produce unimodal frequency distributions and

have more symmetrical bursae.

The biosonar apparatus produces a sound that leaves the head of an odontocete in a

particular trajectory. The directionality of echolocation signals was suggested by Norris et al.

(1961), but experiments by Au et al. (1995) provided evidence for this hypothesis.

Recordings of echolocation clicks that were off axis were more distorted (Au et al., 1995).

Sound Reception

There are inadvertent mechanical “noises” that arise in the process of underwater

sound generation (Cranford, 2000). In odontocetes, the mandibles are a part of the sound

reception apparatus that may play a role in modifying received sounds (Norris, 1968;

Cranford, 2008a; Cranford et al., 2010).

Echolocation in toothed whales is at least used in foraging, navigation, predatory

avoidance, and social facilitation. Echolocation is a complex behavior which includes sound

generation, reception, and central nervous system processing. These diverse functions likely

make hearing echolocation sounds as important as producing them. Consequently, the shape

of the mandibles, and the associated fat bodies that fill the hollow mandibular foramen,

probably play a major role in the sounds heard by the animal.

Page 42: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

31

The mandibular fat bodies are not impedance matched to the bony mandibles.

Therefore, the mandibles may act as a sound barrier in some circumstances. Sounds entering

the mandibular fat bodies are propagated to the ear complex. The shape of the mandibular

foramen, housing the fat body, may affect the types or characteristics of sounds that reach ear

complex. The dimensions of the mandibular foramen may also affect the sounds propagated

to the ear complex by the fat bodies.

The dimensions of the foramen provide limits to the size and shape of the mandibular

fat bodies, which in turn may influence the length of a sound wave which propagate through

the fat body. A small opening may not allow a long wavelength sound to pass. Therefore, it is

reasonable to suggest that the dimensions of the foramen may act as a band-pass filter for

lower frequencies and longer wavelengths.

This chapter tests two relationships. The first addresses whether a relationship

between mandibular shape and received sound exists. The second examines a potential

relationship between size of the mandibular foramen and sounds received. In both tests, the

spectral characteristics of the echolocation pulses produced by some species are used as an

estimate of the sounds received by those species.

MATERIALS AND METHODS

To assess relationships between received sound and shape of the foramen, principal

component shape scores for PC 3 (Foramen Elongation) and PC 4 (Foramen Expansion)

were regressed against sound production characteristics for eleven odontocete species. In

species for which multiple specimens were used in this project (Appendix A), a mean PC

score for each PC 3 and PC 4 (Appendix G) was used in regression analyses. For example,

five Dall’s porpoise (Phocoenoides dalli) mandibles in the sample have five individual PC 3

scores. The mean of the five PC 3 scores was used in the regression analysis as a measure of

Foramen Elongation. Mean peak frequency, the average of the frequency with the highest

amplitude in a set of echolocation clicks, was used as an estimate of sound production

capabilities, and therefore an estimate of sound reception. These are the frequencies in which

the most energy is produced by the species, as reported in the literature. Some delphinids

produce bimodal frequency spectra, containing a low frequency component peak and a high

frequency component peak. In these cases, two mean peak frequencies were included in the

Page 43: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

32

analysis. Mean peak frequency values were gathered from published literature for as many

species as possible (Table 2). Au et al. (1995) determined that echolocation beams are highly

directional and their characteristics degrade as the recording moves off-axis of the sound

beam. To ensure that the most accurate information was used, only studies that took

measures to assure that on-axis echolocation pulses were analyzed were included in this

regression.

Table 2. Echolocation Peak Frequency

Family Species Mean Peak Frequency

Mean Peak Low

Frequency Component

Mean Peak High

Frequency Component

Delphinidae Cephalorhynchus commersonii 132 --- --- Feresa attenuata --- 45 117 Grampus griseus --- 40 90 Lagenorhynchus obscurus --- 55 105 Orcinus orca --- 24 108 Psedorca crassidens --- 48.6 99.6

Phocoenidae Neophocaena phocaenoides 123.5 --- --- Phocoena phocoena 135 --- ---

Ziphiidae Mesoplodon densirostris 31.5 --- --- Mesoplodon europaeus 37 --- --- Ziphius cavirostris 40 --- ---

Note: For species producing unimodal frequency distribution, only mean peak frequency is reported. For species producing bimodal frequency distribution, mean peak frequencies for both the lower and higher frequency components is reported. Dashed lines indicate category of frequency data is not applicable for that species. Numbers in superscript indicate sources of data. Table adapted from: Kyhn et al., 2010 Simon et al., 2007 Madsen et al., 2005b Madsen et al., 2004 Au et al., 1995 Johnson et al., 2004 Philips et al., 2002 Li et al., 2005 Gillespie et al., 2009 Au and Wursig, 2004 Li et al., 2007 Zimmer et al., 2005

Some delphinids produce bimodal spectra, and so the regression analysis was run

twice for each PC against: (a) mean peak frequency, with low frequency component peaks

(MPFL) if applicable, and (b) mean peak frequency, with high frequency component peaks

(MPFH) if applicable.

Page 44: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

33

To test for a potential relationship between foramen size and received sound, surface

area between the landmarks describing the foramen was regressed against mean peak

frequency. The area across four landmarks (#1, #3, #5, and #7 in Figure 13) served as a

measurement of foramen size. The 3-D coordinates of these mandibular foramen landmarks

on the right mandible extracted for the shape analysis were used to determine an approximate

surface area between those four landmarks (in mm2). Euclidean distances between each pair

of landmarks were calculated using the following formula to find the distance between point

1 and point 2:

Distance12 =√(X1-X2)2(Y1-Y2)

2(Z1-Z2)2

The digital distances calculated with the formula above represent real distances (in mm)

corresponding to scan and voxel sizes of the 3-D volume renderings (Appendix D). The

Euclidean distances were multiplied by the corresponding voxel size to determine the true

distance in millimeters. Next, the four landmarks were divided into two triangles (Figure 13),

with the distance between right condyle centroid (#1 in Figure 13) and right anterior foramen

(#7 in Figure 13) as the common side. The area of each triangle, with sides a, b, and c, was

estimated using Heron’s formula:

If p = a + b + c ,

2

then area = √p(p-a)(p-b)(p-c)

The areas of the two triangles were added to get a total surface area estimate of the

mandibular foramen opening. As with the first hypothesis tested, because some delphinids

produce bimodal spectra, the regression analysis against surface area was run twice: a) mean

peak frequency, with low frequency component peaks, if applicable, and b) mean peak

frequency, with high frequency component peaks, if applicable. All linear regression

analyses were performed in R (R Development Core Team, 2010).

The same relationships tested above using a regression analysis were tested again

using independent contrasts in order to account for evolutionary history while testing for

correlated character evolution. The following analysis is performed to supplement the

regression analyses above. Because all species are related to each other to various degrees,

they cannot be regarded as independent individuals sampled from the same distribution

Page 45: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

34

Figure 13. Diagram of triangles formed using four landmarks around the mandibular foramen on a right mandible. Surface area of the mandibular foramen opening is estimated by adding area of top triangle to area of bottom triangle.

(Felsenstein, 1985). This non-independence can cause an exaggeration of the significance in

hypothesis tests (Felsenstein, 1985). Independent contrasts are estimates of the character, or

variable (e.g. PC 3 shape score, mean peak frequency), with the evolutionary history and

relationships removed (Felsenstein, 1985). The branch lengths, which are estimates of change

in a phylogeny, are taken into account to determine the new values of the characters which

are now independent (i.e. independent contrasts). Then, a regression of the independent

contrasts tests for a relationship between two characters. All linear regressions of

independent contrasts were performed in R, using the package Geiger (Harmon et al., 2009).

The independent contrasts assume Brownian motion, and so assumes no selection is

occurring on a trait, or character. Successive changes to a character (i.e. changes in character

state) along one branch are independent of changes in that character on another branch, as

well as the previous character state on the same branch.

A molecular phylogeny and associated branch lengths (McGowen et al., 2009) was

incorporated to calculate the independent contrasts. This odontocete phylogeny is the most

recent and comprehensive of which the consensus phylogeny was determined from both

mitochondrial DNA and nuclear DNA. McGowen et al. (2009) estimated the branch lengths

from cytochrome b sequences and constrained them to the consensus phylogeny.

RESULTS

Mean peak frequency with high frequency component (MPFH) shows a significant

linear relationship to PC 4 (Foramen Expansion) (R2 = 0.6478, P = 0.00281) (Figure 14a).

Mean peak frequency with low frequency components (MPFL) shows a significant linear

relationship to surface area (R2 = 0.4047, P = 0.03537) (Figure 14b). However, the

Page 46: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

35

relationship between independent contrasts of PC 4 and MPFH was insignificant (R2 = 0.24,

P = 0.1261). Surface area and MPFL remained significantly correlated after performing a

regression of independent contrasts (R2 = 0.5053, P = 0.01420).

DISCUSSION

The results of the regression analyses were somewhat expected. The idea that sound

waves may not fit through a reception apparatus with specific dimensional restrictions is

reasonable. Thus, the significant results from testing the relationship between sound

wavelength and foramen size dimensions were to be expected. The finding of a significant

relationship between the area of the opening in the foramen and the wavelength of the low

frequency components (MPFL) supports the notion that size of a mandibular foramen may

act as a filter for the sounds that reach the ears by way of the mandibular fat bodies.

Cranford et al. (2008a) described a pathway for sound to reach the ear complex in a

beaked whale model. This sound reception pathway passes through the opening in the

mandibular foramen, described by four landmarks (1, 3, 5 and 7) in the current GM

mandibular shape analysis. Those four landmarks provide not only shape information, but

also information on the dimensions of the opening outside of the GM analysis. The size of

the opening may be capable of “tuning” the sounds received. Lower frequency sounds, those

with longer wavelengths, may be filtered out or adjusted by the anatomy of the lower jaws

and associated structures (muscles, connective tissue sheets, bony skull components, and air

spaces).

The initial regression analyses suggested a significant relationship between foramen

shape and mean peak frequency. However, the independent contrast analysis found the

relationship to be insignificant. It appears that the beaked whales (Ziphiidae), which are those

points below 55kHz (Figure 14a), are driving the significance in the initial regression

analysis. Once those points are corrected to show their closer relationships relative to other

data points, the relationship between Foramen Expansion (PC 4) and mean peak frequency

with high frequency components is dissolved.

The small sample size is a problem which may provide inaccurate results for

regression analyses described in this study. Unfortunately, methods of recording on-axis

Page 47: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

36

a)

b)

Figure 14. Plots of echolocation mean peak frequency and morphometric parameters. Filled circles represent different species and solid line represents regression line. a) mean peak frequency with high frequency component vs. PC 4 and b) mean peak frequency with low frequency component vs. mandibular foramen surface area.

Page 48: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

37

echolocation beams are still in development and difficult to implement in the wild. Only a

few studies in the passed fifteen years have succeeded in ensuring on-axis recordings (e.g.

Au and Würsig, 2004).

There may be additional variation that is explained by other factors not addressed in

this current study. The intra-species variation in the high frequency components of species

which exhibit a bimodal frequency distribution may be due to prey preferences or availability

For example, an animal will need to use a higher frequency signal in order to capture a

smaller target in the signal and receive the echo bouncing off of the small target. Killer

whales which feed on large Chinook salmon (Oncorhynchus tshawytscha) of up to

approximately 85cm in length emphasize different frequencies in their echolocation than

killer whales which feed on large groups of schooling herring (Clupea spp). Herring

schooling behavior was studied in Norwegian waters when threatened by killer whales. The

size of their densely packed schools averaged 180m2 in area (Nøttestad and Axelsen, 1999),

a much larger target than a single salmon. Salmon eating killer whales from the northeastern

Pacific Ocean produce higher frequencies than Norwegian herring eaters (Au et al., 2004;

Simon et al., 2007). Both studies reported center frequency, which is the frequency at which

the energy in a click spectrum is split into two equal halves. The peak frequencies (both high

and low because killer whales produce bimodal spectra) are likely not included in the center

frequency. However, the range of center frequencies does reflect differences between the

herring- and salmon-eating populations (Au et al., 2004; Simon et al., 2007). Salmon-eater

center frequency range is 45 – 80kHz (Au et al., 2004), whereas herring-eater center

frequency range is 22 – 49kHz. Simon et al. (2007) suggested it may be a consequence of the

larger target size in a large school of herring as compared to a single salmon. The lower

frequencies used by herring eaters may be a reflection of the target size. Such variation in

prey preference or availability may be a source of variation in posterior, sound reception this

toothed whale mandibular shape study.

The mandibles are likely an important component in sound reception in odontocetes.

The entire sound reception apparatus is complicated and complex. There are many structures

(e.g. mandible, mandibular fat bodies, air sinuses) and their exact interactions with one

another are not yet worked out. The complex system of reflective and refractive surfaces may

Page 49: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

38

assist in filtering or amplifying sounds. It is possible that toothed whales have an internal

acoustic pinna, which acts to emphasize or suppress different sounds. The mandibles are only

a single component of the sound reception apparatus which may provide sound barriers and

reflective surfaces to the associated structures involved in sound reception. However, this

shape analysis provides preliminary data regarding the role of the mandibles. It appears that

mandibular foramen size and shape influences the characteristic frequencies that enter and

propagate through mandibular fat bodies to the ear complexes. The mandibles may, in fact,

act as a sound filtering device to reduce the amount of low frequency noise the whales

receive. This might, for example, allow the whales to focus on the echoes that are most

beneficial to them, such as the echoes bouncing off of targets.

Understanding the mechanics of sound reception and potential filtering of sounds by

the anatomy of the sound reception apparatus has implications in the conservation of toothed

whales. Anthropogenic, man-made, noise in the ocean has recently become a major concern

for the conservation of many whale species (e.g. Erbe, 2002). There are many sources of

anthropogenic sound in the oceans, such as seismic exploration, shipping traffic, and naval

sonar. If we understand which sounds are filtered out by the sound reception apparatus for

various whale species, appropriate mitigation measures could be implemented to protect both

whales and the interests of the sources of anthropogenic noise (e.g., oil drilling, seismic

exploration, commercial shipping).

Page 50: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

39

CHAPTER 4

PATTERNS OF MANDIBULAR SHAPE

AND PHYLOGENETIC SIGNAL

INTRODUCTION

The morphological variation observed in organisms is a result of both selective

pressures and phylogenetic history. Similar selective pressures, such as niche exploitation or

prey preferences, may cause species to resemble one another. On the other hand, organisms

may resemble each other because they share a recent common ancestor. Phylogenetic signal

is defined as a “tendency for related species to resemble each other more than they resemble

species drawn at random from the tree” (Blomberg and Garland, 2002). Testing for

phylogenetic signal is useful in estimating how much of an observed trait is due to selection

(i.e. adaptation) or phylogeny. It is important to keep in mind that both evolutionary history

and natural selection of the recent and current environment play a role in producing a trait

(Blomberg and Garland, 2002).

The shape analysis in the second chapter presented patterns of shape variation in

Odontoceti. The majority of shape variation is due to changes in the width of the mandibles

at the condyles (Principal Component [PC] 1 – Jaw Flare) and symphysis (PC 2 –Symphysis

Elongation). Some variation was also attributed to changes in the shape of the mandibular

foramen (PC3 – Foramen Elongation, PC 4 – Foramen Expansion). It was noted that there

appeared to be more phylogenetic signal in the first two principal components; the specimens

clustered at the family level in the PC plots (Figure 10 on p. 21 and Figure 11 on p. 22). An

assessment of phylogenetic signal will provide insight regarding which parts of the

mandibles are possibly more shaped by selection than others.

Selection may result in organisms resembling others less related to them. Such

convergent evolution is present on a higher level, such as unrelated fish and marine

mammals. Both groups are characterized by paddle-like limbs (fins and flippers), an

adaptation to living in an aquatic environment. Convergent evolution has also occurred

within Odontoceti. For example, river dolphins are a non-monophyletic group which are

morphologically similar to one another. Morphology-based phylogenies have placed all

monotypic river dolphin lineages (Platanistidae, Pontoporiidae, Iniidae, and Lipotidae) into a

Page 51: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

40

monophyletic group (Geisler and Sanders, 2003). However, recent molecular-based

phylogenies place Platanistidae outside of the clade consisting of other river dolphins (May-

Collado and Agnarsson, 2006; McGowen et al., 2009). In fact, the most recent, and

comprehensive, molecular phylogeny places the three remaining river dolphin lineages

(Pontoporiidae, Iniidae, and Lipotidae) sister to beaked whales (Ziphiidae), not Platanistidae

(McGowen et al., 2009). This suggests that odontocetes dispersed from or took refuge in

riverine systems twice and evolved similar morphologies to adapt to fresh water ecological

niches. Platanistids and other river dolphins demonstrate convergent evolution.

The goal of this chapter is to begin to understand components of the observed

morphological variation by assessing how much of the variation is due to evolutionary

history, as well as examine the patterns of mandibular shape evolution. This is accomplished

using three methods. First, the total observed shape variation in odontocetes is compared to

cetacean phylogeny to estimate phylogenetic signal in total shape variation. Second, the

amount of phylogenetic signal in specific traits, or characters, which account for the majority

of mandibular shape variation are assessed. Finally, an ancestral character state

reconstruction reveals patterns of mandibular shape evolution.

MATERIALS AND METHODS

The Mantel test (Mantel, 1967) is used to test the association between total

mandibular shape variation and phylogeny. This test examines the correlation between two

distance matrices. A Procrustes coordinates distance matrix, an estimate of morphological

difference, was compared to a phylogenetic distance matrix. Procrustes coordinates for each

specimen in the shape analysis were exported from MorphoJ (Klingenberg, 2011) after the

Procrustes superimposition was performed. The Procrustes coordinates represent the amount

of deviation of each specimen from the average, consensus specimen. If multiple specimens

of a single species were in the dataset, the means per species were used when calculating the

morphology-based distance matrix. The morphology-based distance matrix were created

using R (R Development Core Team, 2010). The distance matrix created in R incorporates a

Euclidean distance criteria. This only refers to the program’s (R) method of calculating the

pairwise distances between each coordinate. The original data file containing Procrustes

coordinates consists of 36 coordinates, three (one in each plane) for 12 landmarks. The three

coordinates per landmark are not treated as a single point in 3-D space, but as 36 individual

coordinates compared to the individual coordinates of the average specimen. So, the

Euclidean criteria implemented in R does not imply that Euclidean distance between 12

Page 52: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

41

landmarks in 3-D space are calculated. Instead, it is a method to calculate the distance

between each of the 36 coordinates.

The phylogenetic distance matrix was created using branch lengths as an estimate of

phylogenetic distance. A molecular cetacean phylogeny (McGowen et al., 2009) was used for

all phylogenetic comparative methods described in this odontocete mandibular shape study

(Figure 15). The phylogenetic distance matrix was created using the package ape (Paradis et

al., 2004) for R. The Mantel test was performed using the package ade4 (Dray et al., 2007)

for R with 9999 permutations.

Figure 15. Odontocete phylogeny modified from McGowen et al. (2009). Asterisks (*) indicate species in the current study. Filled circles indicate weakly supported nodes.

To assess the phylogenetic signal in shape variables, or characters, the package

picante (Kembel et al., 2010) for R was used to calculate Blomberg’s K-statistic and p-value.

Page 53: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

42

The K-statistic represents a measure of how much phylogenetic signal is present in the

character and the p-value is a measure of phylogenetic signal significance. A value of K > 1

indicates that more than expected phylogenetic signal is present in a character if the character

was evolving under Brownian motion (Blomberg et al., 2003). A character evolving under

Brownian motion is randomly evolving. Under Brownian motion, characters are expected to

evolve randomly following a random walk. A successive character state is independent of the

previous state. In other words, selective forces are not acting on a character. A value of K < 1

suggests that relatives resemble each other less than would be expected if the character was

evolving under Brownian motion and another factor (e.g. adaptation) affected character

evolution (Blomberg et al., 2003). For example, scale color in lizards varies with the

environment the lizard lives in to be best camouflaged. This particular character might

exhibit low phylogenetic signal and result in a low K value, providing environment

differences do not correspond closely with phylogenetic relationships. The p-value is a

measure of statistical significance of phylogenetic signal. The shape variables, or characters,

tested are the first four principal components (PC) resulting from the principal components

analysis in Chapter 2. The characters are: jaw flare (PC 1), symphysis elongation (PC 2),

foramen elongation (PC 3), and foramen expansion (PC 4).

To supplement the abovementioned assessment of phylogenetic signal, model

goodness-of-fit tests were also performed on the four characters (PC1 to PC 4). The Akaike

Information Criterion (AIC) was used to compare three models using the package geiger

(Harmon et al., 2008) for R. The first model is Brownian motion, the second is Orstein-

Uhlenbeck (OU) with one optimum (Butler and King, 2004), and the third is called “white

noise.” Brownian motion was described above, and a character follows a random walk. The

OU model with a single optimum implies a character is driven to an optimal state by

selection in the entire lineage (i.e. Odontoceti). The Orstein-Uhlenbeck model incorporates

an optimal trait value and strength of selection parameter. This model is also a generalization

of the BM model. The selection parameters simulate stabilizing selection in which a trait

evolves to become the most optimal state. The farther away a trait is from the optimum, the

stronger the selection to pull a trait to that optimum. So, selection stabilizes the trait. The

white noise model is a completely random model, in which all species are considered

independent; phylogeny is excluded in this model. In the lizard scale color example from

above, the best fit model tests would potentially confirm that Brownian motion is does not

Page 54: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

43

explain the evolution of scale color. If the OU model is best fit, this would suggest that the

entire lineage of lizards is under selection for a specific color of scale.

The patterns observed in odontocete mandibular shape evolution were examined

using ancestral character state reconstructions. The PC scores of the first four PCs

highlighted in Chapter 2 were continuous characters in a maximum likelihood ancestral

character state reconstruction using the R package ape version 2.5-3 (Paradis et al., 2004).

Maximum likelihood methods optimize the likelihood of each character state transition. In

ape, when using continuous characters, a Brownian motion model (the default) of evolution

is incorporated. The mean PC score for each species is used to provide values for the

terminal taxa.

RESULTS

The phylogenetic and morphological distance matrices were significantly correlated

(R = 0.47, R2 = 0.22, P = 0.001). The R2 value suggests that phylogeny accounts for

approximately 22% of the observed morphological shape variation.

Phylogenetic signal was significant in all four characters (PC 1, PC 2, PC 3, PC 4).

The greatest phylogenetic signal is present in PC 2, or Symphysis Elongation (K = 1.825,

P = 0.001). The second largest amount of phylogenetic signal is in PC 1, Jaw Flare

(K = 0.627, P = 0.001). Principal component 3, Foramen Elongation, (K = 0.435, P = 0.004)

and PC 4, Foramen Expansion (K = 0.383, P = 0.003) contain the least amount of

phylogenetic signal.

The best fit models were not the same for each character, or PC. A Brownian Motion

model of evolution is the best fit for Jaw Flare (PC 1) and Symphysis Elongation (PC 2). In

contrast, the best fit model of evolution is the Orstein-Uhlenbeck (OU) with one optimum for

Foramen Elongation (PC 3) and Foramen Expansion (PC 4).

The ancestral character state reconstructions reveal interesting trends in the evolution

of lower jaw shape in odontocetes. There are convergences between some delphinids,

monodontids, and porpoises in shapes associated with PC 1, or Jaw Flare (Figure 16). Those

delphinids, the killer whale (Orcinus orca) and globicephalinids, porpoise, and monodontid

mandibles are more flared than others. Globicephalinae includes Risso’s dolphins (Grampus

griseus), false killer whales (Pseudorca crassidens), melon-headed whales (Peponocephala

electra), short-finned pilot whales (Globicephala macrorhynchus), and pygmy killer whales

(Feresa attenuata). In addition, the kogiid (pygmy sperm whale in the current sample)

appears to share a similar Jaw Flare shape. This particular trait, or mandibular shape, may

Page 55: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

44

Figure 16. Ancestral character state reconstruction of PC 1 (Jaw Flare). Larger filled circles indicate less flared mandibles and smaller filled circles indicate more flared mandibles. Node values were raised to a power of four and divided by 11 to emphasize differences in circle size. have evolved at least two times in Odontoceti: within phocoenids and monodontids,

Globicephalinae, and possibly sperm whales.

River dolphins have extremely elongate symphyses. The most recent, molecular

phylogenies defines river dolphins as non-monophyletic. The presence of very elongate

symphyses may have evolved twice in Odontoceti (Figure 17), with river dolphins

(Platanista gangetica, Pontoporia blainvillei, and Inia geoffrensis). In addition, the Kogia

simus also has a relatively elongate symphysis (Figure 17).

The shapes of the mandibular foramen (PC 3 and PC 4) and position of posterior

alveolar groove (PC 3) do not exhibit clear defining patterns (Figures 18 and 19). However,

Page 56: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

45

Figure 17. Ancestral character state reconstruction of PC 2 (Symphysis Elongation). Larger filled circles indicate less elongate symphyses and smaller filled circles indicate more elongate symphyses. Node values were raised to a power of four and divided by 11 to emphasize differences in circle size.

the beaked whales and kogiid appear to have more elongate (PC 3) and expanded (PC 4)

foramina.

DISCUSSION

Some phylogenetic signal is present in the shape variation observed in odontocete

mandibles. The Mantel test found that the phylogenetic and morphological distance matrices

were significantly correlated. Phylogeny accounts for 22% of the observed shape variation.

This explains why some of the shape variables (PC 1 and PC 2) exhibit phylogenetic signal.

Recall, it was suggested that there was more phylogenetic signal in the first two PCs than the

Page 57: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

46

Figure 18. Ancestral character state reconstruction of PC 3 (Foramen Elongation). Larger filled circles indicate more elongate foramens and smaller filled circles indicate less elongate foramens. Node values were raised to a power of seven and divided by 85 to emphasize differences in circle size.

third and fourth PC. Blomberg’s K-statistic for each PC confirms that there is more

phylogenetic signal in PC 1 and PC 2 than PC 3 and PC 4.

The correspondence between phylogeny and morphology of approximately 22%

might indicate that phylogenetic inertia provides broad structure to morphological variation.

But at the species level, that structure breaks down and individual variation is more

influenced by environmental factors. It is more likely that the evolutionary process leading to

shape change is one of selective pressures, not Brownian motion (BM). If the process

followed Brownian motion, then we would expect morphologic distance to be proportional to

phylogenetic distance. Furthermore, Blomberg’s K statistic of less than one implies that there

Page 58: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

47

Figure 19. Ancestral character state reconstruction of PC 4 (Foramen Expansion). Larger filled circles indicate more expanded and smaller filled circles indicate less expanded foramens. Node values were raised to a power of seven and divided by 85 to emphasize differences in circle size.

is less resemblance in relatives than expected under Brownian motion evolution (Blomberg et

al, 2003). The small values of K from PC 3, and PC4 suggest these characters evolved under

a different mode of evolution, such as adaptation (Blomberg et al., 2003).

The test of best fit models supports the notion that Foramen Elongation (PC 3) and

Foramen Expansion (PC 4) shapes may have evolved under a selective regime and specific

shapes are structurally important for sound reception. The foramen shapes appear conserved,

only accounting for less than 11% of the total shape variation observed in odontocete

mandibles. Mandibular foramens, an integral component of the sound reception apparatus,

may require specific shapes to function at an optimum level. The foramina appear to contain

optimal shapes that have been shaped by selection. In addition, selective pressures for sound

Page 59: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

48

reception may drive shape variation observed in the foramen. The shapes do vary, albeit a

relatively small amount, and may reflect individual differences in environment and optimal

sound reception abilities for that environment.

Blomberg’s K-statistic from Jaw Flare, PC 1, is also less than one (K = 0.627)

implying Brownian motion is not the best fit model of evolution of the trait. However, the K

value is close to 1 and the test for best fit model suggests BM is more likely than OU with

one optimum. Together, these results indicate that BM is the best fit model for Jaw Flare

evolution, as opposed to OU with a single optimum. It is possible that the selective forces

(e.g. feeding mechanics) are driving shape variation on a shorter time scale which appears as

BM. Selecting multiple optima, not just one, based on feeding strategies or prey preference

may provide a more accurate model of evolution, but that is beyond the scope of this current

study. The current results show that the Jaw Flare evolves randomly. This pattern may be

attributed to selective forces (e.g. feeding mechanisms) that change on a short time scale, but

when taken together, produce what appears to be random variation.

Principal component 2 (Symphysis Elongation) was the only shape variable tested

that had a K > 1, indicating that more than expected phylogenetic signal is present in that

character. This may be a reflection of the river dolphin lineages (Platanistidae, Iniidae, and

Pontoporiidae) exhibiting extremely elongate symphyses, whereas the other lineages are less

variable in this shape. Two lineages that possess very elongate symphyses, or extreme

difference in shape from the rest of the lineages, may be the cause of high phylogenetic

signal observed in this trait. In addition, the best fit model for Symphysis Elongation (PC 2)

is BM. Brownian motion might explain the evolution of Symphysis Elongation in lineages

other than the river dolphins. As with Jaw Flare (PC 1), a detailed analysis of prey preference

included in a multiple optima model may be more appropriate, potentially indicating that

multiple selective forces drive symphysis shape variation.

The Mantel test has been shown to be less effective than Blomberg’s K-statistic in

detecting phylogenetic signal (Harmon and Glor, 2010). Harmon and Glor (2010) suggest

that tests requiring data to be converted to a matrix should only be used when the data can

only be expressed as pairwise distances. In the case of the current study, it would be more

appropriate to use the Procrustes coordinates to create a morphological distance matrix than

the characters chosen for further analysis (PC 1 to PC 4). The principal components are linear

combinations of the Procrustes variables, which describing morphological differences among

Page 60: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

49

specimens. It was best to convert the original Procrustes shape coordinates to a matrix to

compare to a phylogenetic distance matrix when testing for phylogenetic signal in overall

shape variation. However, because phylogenetic signal of certain characters was also of

interest, phylogenetic signal was assessed using Blomberg’s K-statistic as well.

The ancestral character state reconstructions reveal potential convergent evolution in

jaw shape of different clades. Globicephalinid Jaw Flare morphology closely resembles that

of less closely related phocoenids (porpoises) and monodontids. It is possible that these

animals utilize similar feeding strategies (e.g. suction feeding, prey preference) which result

in similar jaw morphology. The two clades of river dolphins appear similar in their

symphysis elongation shape. Perhaps elongate symphyses assist in capturing fish in riverine

ecosystems. The results of ancestral character state reconstructions are tenuous at deeper

nodes. The lack of an outgroup affects the accuracy of deeper nodes, affecting the

interpretations observed patterns.

Ancestral character state reconstructions for continuous characters were traditionally

performed using maximum parsimony criteria in which the preferred explanation is one of

the least evolutionary changes in a character. However, new methods to incorporate models

of evolution using maximum likelihood methods have been developed. In ape, the Brownian

motion model of evolution is assumed, but it is possible that these characters are not evolving

following a random walk. The Ornstein-Uhlenbeck (OU) model of evolution in estimating

ancestral states is a potentially more appropriate model of evolution (Hansen, 1997), as is

possibly the case for Foramen Elongation (PC 3) and Foramen Expansion (PC 4). Butler and

King (2004), however, found that unless the selective model of evolution is accurate, then the

BM model of evolution performs better. It is best to assume BM unless the accurate model is

incorporated. This current study assumes a BM model of evolution because determining

selective parameters is beyond the scope of the study. More research into parameters to best

estimate a selective model of evolution for shape characters might provide more accurate

ancestral character state reconstructions. In addition, the inclusion of an outgroup, such as a

stem odontocete, will provide an estimate of the mandibular shape of the common ancestor to

all odontocetes.

It is apparent that mandibular shape is driven by both environmental selective

pressures and phylogenetic inertia. Approximately 78% of variation in jaw shape that is not

Page 61: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

50

influenced by phylogenetic history is likely driven by selective pressures such as feeding

strategy and hearing. This current study provides some understanding regarding which

shapes may be more driven by selection than others. Jaw Flare and Symphysis Elongation are

influenced more by phylogeny than mandibular foramen shapes. However, environmental

selective pressures still play a role in driving the majority of shape variation observed in

extant odontocetes. Symphysis Elongation and Jaw Flare may be subject to pressures from

prey preference and variation in foramen shape may be due to selective pressures of

ecological environment. Future studies closely examining prey preference, and perhaps

sounds best used to echolocate on those prey items, with respect to mandibular morphology

could shed light into what shapes odontocete lower jaws.

The enlarged foramen is an important component in the sound reception apparatus,

which is likely shaped, in part, by the types produced and received. A broad spectrum of

selective pressures on acoustic function should be considered. Evolution of biosonar capacity

may also serve as predator avoidance. For example, porpoises may produce sounds so high in

frequency that predators, such as killer whales, may not hear. Regardless of the pressures

driving shape variation in odontocetes, the mandibles are only one component of a complex

sound reception apparatus (Norris, 1968; Cranford et al., 2008). Each component in the

complex apparatus may respond differently to selective pressures and influence one another.

The current study provides the first look at the mandibular component by describing shape

variation and assessing factors driving that variation in all odontocetes.

Page 62: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

51

CHAPTER 5

CONCLUSIONS

The odontocete mandibles serve a variety of functions, which include hearing and

feeding. These complex functions have likely driven the shape of mandibles. The primary

goal of this current project is to describe mandibular shape variation. Other goals include

increased understanding the influence of mandibular shape on function, as well as beginning

to identify the selective pressures that may drive mandibular shape variation in toothed

whales. Geometric morphometrics (GM) is used to analyze shape because it involves a

superimposition technique to remove size from the analysis, allowing a wide range of

mandible sizes to be included. In addition to removing size, GM resolves some of the other

problems confounding traditional, linear morphometrics. In part, GM does this by providing

visual graphs which depict relative movement of landmarks (Zelditch et al., 2004).

The majority of shape variation may be attributed to differences in feeding strategy;

however, some shape variation may be associated with sound reception. Four principal

components depicting shape variation are described in detail in the present study. The

majority of shape variation is explained by Jaw Flare (50.0%) and Symphysis Elongation

(35.5%), which can be attributed to feeding differences. Variation in Foramen Elongation

(6.3%) and Foramen Expansion (4.6%) may be due to differences in sound reception,

The sound reception apparatus in odontocetes has undergone changes through the

course of evolution to incorporate the mandibles. The mandibular foramen houses a

mandibular fat body, through which sound travels to the ear complex (Norris, 1968; Cranford

et al., 2008a,b). The predominant toothed whale sound reception hypothesis over the

previous four decades is that sound travels through the thin lateral wall of the mandibles, also

known as the pan bone (Norris, 1968). In the “jaw hearing” hypothesis, sound travels

through the pan bone and into the mandibular fat body, where it is propagated toward the ear

complex (Norris, 1968). The lipid properties of the mandibular fat body suggest that it

functions to propagate sound to the ear complex and may serve to filter or otherwise

manipulate acoustic properties. The most recent hypothesis of sound reception, the gular

Page 63: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

52

pathway, suggests that sound enters the mandibular fat bodies primarily through the soft

tissues of the throat (Cranford et al., 2008a). These computer models indicate that sound is

propagated along the gular pathway and that the mandibles may cause specific frequencies to

be excluded or filtered out. The conspicuous absence of the medial wall of the posterior

mandibles in all extant odontocetes can be traced back into the fossil record, suggesting that

the gular pathway may be the primary and ancestral pathway of sound reception. Clearly, the

mandibles are an integral component of the sound reception apparatus (e.g. Norris, 1968;

Brill et al., 1988; Brill and Harder, 1991; Mohl et al. 1999; Cranford et al., 2008a,b) and its

shape is conserved; much more variability exists in the anterior two-thirds of the mandibles.

The size of the mandibles appears to influence the sounds received by toothed whales.

Results presented in Chapter 3 suggest that the geometry of the mandibles limits the lower

frequencies received by the animal. The bony mandibles provide boundaries to the

mandibular fat bodies, which may limit the geometry and influence the acoustic function of

the fat bodies. The lower frequencies have longer corresponding wavelengths, which may not

propagate through smaller openings in the receiving structure (i.e. mandibular boundaries on

the fat bodies). Further understanding of the potential for sound reception anatomy to either

amplify or diminish certain sounds may provide useful information on the effects of

anthropogenic sound on toothed whales.

Sound reception is not the only influence on odontocete mandibular shape. Results

presented in Chapter 4 suggest that phylogenetic relationships, as well as environmental

factors probably affect the shape of the mandibles. The exact selective pressures driving the

shapes are beyond the scope of this study, but the variation in shape observed in Jaw Flare

and Symphysis Elongation may be a reflection of the various feeding strategies employed by

toothed whales. The variation in foramen shapes exhibit less phylogenetic signal than Jaw

Flare and Symphysis Elongation, implying that the relatively small amount of foramen shape

variation is more driven by environmental factors. This is supported by the correlates of

foramen geometry and sound reception described in this mandibular shape study. All shapes

described in detail in this study are subject to environmental pressures, as is evidenced by

convergent evolution of mandibular shape in less closely related lineages observed in

ancestral character state reconstructions presented in Chapter 4. This current study provides a

Page 64: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

53

base for future work identifying the pressures driving shape variation, as well as being the

first to describe odontocete mandibles in detail.

Page 65: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

54

ACKNOWLEDGEMENTS

I’d like to acknowledge my committee for all of their hard work and assistance. Dr.

Berta, thank you for your encouragement and expert advice. Dr. Cranford, thank you for your

patience, guidance, and expertise. Dr. Bailey, your ability to help me understand aspects of

this project are greatly appreciated. Dr. Burns, I am grateful for your wonderful input and

counseling.

Special thanks to our funding source. This research was funded in part by a grant

from the Chief of Naval Operations Environmental Readiness Division (CNO45) to Dr. Ted

W. Cranford.

This project would not have been brought to fruition without the assistance from

personnel at hospitals and museums during mandible collection and scanning and I offer

them all my sincere appreciation. Dr. Robert Mattrey, Dr. David Vera, Ms. Jacqueline

Corbeil, Mr. Thomas Ellerbrock, Mr. Steven Robitaille, Mr. Matt Konishi , provided time

and assistance on many occasions to scan specimens at the Hillcrest Medical Center and

UCSD Medical School. Dr. Claudia Kirsch, Dr. Robert Lufkin, and Shiva Borgheian

arranged the CT scan appointment at UCLA Medical Center and Ms. Emilia Patterson

assisted with scanning. Phillip Unitt and Jim Wilson from San Diego Natural History

Museum, helped gather mandibles from their collection for scanning. Dr. James G. Mead,

Charley Potter, John Ososky, provided access to mandible specimens from the National

Museum of Natural History. Kellyn Goler and Janine Hinton conducted the CT scans at the

National Museum of Natural History. In particular, I appreciate the efforts of Janine Hinton,

she committed to doing a complete and superb job as a volunteer and assisted for many,

tireless hours during CT scanning. James P. Dines and David Janiger loaned specimens from

their collection at the Los Angeles County Museum of Natural History.

The fellow Evolutionary Biology graduate students were indispensable during this

project. To all of you, thank you for your suggestions, and providing relief during stressful

moments. To my lab mates, Cassie Johnston, Sarah Kienle, Jessica Martin and Samantha

Young, thank you for your endless support and project advice. Allison Shultz, your

assistance has helped bring this project to completion.

Page 66: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

55

I thank the staff at San Diego State University for their help in many aspects of this

study ranging from project advice to administrative help. In particular, the Evolutionary

Biology program area staff was always available to answer project questions and for this, I

am grateful.

Michael R. McGowen, thank you for your time in providing the phylogeny and

branch lengths to incorporate into this study. These files were incredibly valuable.

To the support staff for Analyze software at Mayo Clinic, your time and assistance

while developing project methods were incredibly beneficial.

This project could not have been completed without the patience and support of my

family and friends. To John Ishigami, your help and patience has been unbelievable. Thank

you very much for all of your wonderful encouragement. To my sister, Analia, whose has a

particular understanding of graduate school pressures, you have been an invaluable sounding

board. I greatly appreciate your sisterly and professional advice. To my parents, Julio and

Norma, this project and graduate school could not have started without your encouragement.

I am grateful for your love and support every step of the way. I’d like to thank Laura Madden

for her support as a friend and colleague. Laura, you have been a remarkable friend who

seems to understand every aspect of my life and so, I thank you. Jessica Bond, I am grateful

for your support and time spent assisting me in this project. And finally, to all of my friends

who have given their time to help me throughout this project, thank you for giving me

amazing feedback.

Page 67: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

56

REFERENCES

Adams, DC, FJ Rohlf, and DE Slice. 2004. Geometric morphometrics: ten years of progress following the ‘revolution.’ Ital J Zool 71:5-16.

Allen, BM. 2003. Morphometric analysis of mandibular variation in two genera of beaked whales, Mesoplodon and Ziphius: using measurements to predict species and gender of specimens. Masters Thesis, George Washington University. 178pp.

Amundin M, Andersen SH. 1983. Bony nares air pressure and nasal plug muscle activity during click production in the harbour porpoise, Phocoena phocoena, and the bottlenosed dolphin, Tursiops truncatus. J Exp Biol 105:275-282.

Au WWL. 1993. The sonar of dolphins. Springer-Verlag, New York.

Au WWL, Ford JKB, Horne JK, Allman KAN. 2004. Echolocation signals of free-ranging killer whales (Orcinus orca) and modeling of foraging for chinook salmon (Oncorhynchus tshawytscha). J Acoust Soc Am 115:901-909.

Au WWL, Pawloski JL, Nauchitgall PE. 1995. Echolocation signals and transmission beam

pattern of a false killer whale (Pseudorca crassidens). J Acoust Soc Am 98:51-59. Au WWL, Würsig B. 2004. Echolocation signals of dusky dolphins (Lagenorhynchus

obscurus) in Kaikoura, New Zealand. J Acoust Soc Am 115:2307-2313.

Blomberg SP, Garland Jr. T. 2002. Tempo and mode in evolution: phylogenetic inertia, adaptation and comparative methods. J Evol Biol 15:899-910.

Blomberg SP, Garland T Jr., Ives AR. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57: 717-745.

Bloodworth, B and CD Marshall. 2005. Feeding kinematics of Kogia and Tursiops (Odontoceti: Cetacea): characterization of suction and ram feeding. J Exp Biol 208:3721-3730.

Bookstein, FL. 1991. Morphometric tools for landmarks data: geometry and biology. Cambridge University Press, Cambridge.

Bookstein, FL. 1997. Shape and information in medical images: a decade of the morphometric synthesis. Comput Vis Image Und 66:97-118.

Bots J, Breuker CJ, Kerkhove AV, Dongen SV, Bruyn LD, Gossum HV. 2009. Variation in flight morphology in a female polymorphic damselfly: intraspecific, intrasexual, and seasonal differences. Can J Zool 87:86-94.

Brill RL, Harder PJ. 1991. The effects of attenuating returning echolocation signals at the lower jaw of a dolphin (Tursiops truncatus). J Acoust Soc Am 89:2851-2857.

Page 68: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

57

Brill RL, Sevenich ML, Sullivan TJ, Sustman JD, Witt RE. 1988. Behavioral evidence for hearing through the lower jaw by an echolocating dolphin (Tursiops truncatus). Mar Mamm Sci 4:223-230.

Butler MA, King AA. 2004. Phylogenetic comparative analysis: a modeling approach for adaptive evolution. Am Nat 164:683-695.

Cardini A. 2003. The geometry of the marmot (Rodentia: Sciuridae) mandible: phylogeny and patterns of morphological evolution. Syst Biol 52:186-205.

Caumul R, Polly PD. 2005. Phylogenetic and environmental components of morphological variation: skull, mandible, and molar shape in marmots (Marmota, Rodentia). Evolution 59:2460-2472.

Christiansen P. 2008. Evolution of skull and mandible shape in cats (Carnivora: Felidae). PLosONE 3:e2807.

Committee on Taxonomy. 2009. List of marine mammal species and subspecies. Society for Marine Mammalogy. Available at: www.marinemammalscience.org, consulted on [September 1, 2010].

Cranford TW. 2000. In search of impulse sound sources in odontocetes. In: Au WWL, Popper AN, Fay RR, editors. Hearing by Whales and Dolphins. New York: Springer- Verlag. p 109-156.

Cranford TW, Amundin M, Norris KS. 1996. Functional morphology and homology in the odontocete nasal complex: implications for sound generation. J Morphol 228:223- 285.

Cranford TW, Krysl P, Amundin M. 2010. A new acoustic portal into the odontocete ear and vibrational analysis of the tympanoperiotic complex. PLosONE 5(8):e11927.

Cranford TW, Krysl P, Hildebrand JA. 2008a. Acoustic pathways revealed: simulated sound transmission and reception in Cuvier’s beaked whale (Ziphius cavirostris). Bioinspiration Biomimetics 3:1-10.

Cranford TW, McKenna MF, Soldevilla MS, Wiggins SM, Goldbogen JA, Shadwick RE, Krysl P, St. Leger JA, Hildebrand JA. 2008b. Anatomic geometry of sound transmission and reception in Cuvier’s beaked whale (Ziphius cavirostris). Anat Rec 291:353-378.

Cranford TW, Van Bonn WG, Chaplin MS, Carr JA, Kamolnick TA, Carder DA, Ridgway SH. 1997. Visualizing dolphin sonar signal using high-speed video endoscopy. J Acoust Soc Am 102:3123-3123.

Dalebout ML, Mead JG, Baker CS, Baker AN, van Helden AL. 2002. A new species of beaked whale Mesoplodon perrini sp. N. (Cetacea: Ziphiidae) discovered through phylogenetic analyses of mitochondrial DNA sequences. Mar Mamm Sci 18:577-608.

Depecker M, Berge C, Penin C, Renous S. 2006. Geometric morphometrics of the shoulder girdle in extant turtles (Chelonii). J Anat 208:35-45.

Page 69: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

58

Diercks KJ, Trochta RT, Greenlaw CF, Evans WE. 1971. Recording and analysis of dolphin echolocation signals. J Acoust Soc Am 49:1729-1732.

Dormer KJ. 1979. Mechanism of sound production and air recycling in delphinids: cineradiographic evidence. J Acoust Soc Am 65: 229-239.

Dray S, Dufour AB, Chessel D. (2007). The ade4 package-II: Two-table and K-table methods. R News, 7:47-52.

Dryden IL, Mardia KV. 1998. Statistical Shape Analysis. Jon Wiley and Sons, Chichester.

Erbe C. 2002. Underwater noise of whale-watching boats and potential effects on killer whales (Orcinus orca), based on an acoustic impact model. Mar Mamm Sci 18:394- 418.

Felsenstein J. 1985. Phylogenies and the comparative method. Amer Nat 125:1-15. Galatius, A. 2010. Paedomorphosis in two small species of toothed whales (Odontoceti): how and why? Biol J Linn Soc 99:278-295.

Geisler JH, Sanders AE. 2003. Morphological evidence for the phylogeny of Cetacea. J Mamm Evol 10:23-129.

Gillespie D, Dunn C, Gordon J, Claridge D, Embling C, Boyd I. 2009. Field recordings of Gervais' beaked whales Mesoplodon europaeus from the Bahamas. J Acoust Soc Am 125:3428-3433.

Hansen TF. 1997. Stabilizing selection and the comparative analysis of adaptation. Evolution 51:1341-1351.

Harmon LJ, Glor RE. 2010. Poor statistical performance of the Mantel test in phylogenetic comparative analyses. Evolution 64:2173-2178.

Harmon LJ, Weir J, Brock C, Glor RE, Challenger W, Hunt G. 2008. geiger: Analysis of evolutionary diversification. R package version 1.3-1. Available at: http://CRAN.R- project.org/package=geiger.

Heyning JE, Mead JG. 1996. Suction feeding in beaked whales: morphological and observational evidence. Contrib Sci 464:1-12.

Hollien H, Hollien P, Caldwell DK, Caldwell MC. 1976. Sound production by the Atlantic bottlenosed dolphin Tursiops truncatus. Cetology 26:1-8.

Johnson M, Madsen PT, Zimmer WMX, Aguilar de Soto N, Tyack PL. 2004. Beaked whales echolocate on prey. Proc R Soc Lond B Biol 271:S383-S386.

Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, Blomberg SP, Webb CO. 2010. Picante: R tools for integrating phylogenies and ecology. Bioinformatics 26:1463-1464.

Kendall DG. 1977. The diffusion of shape. Adv Appl Prob 9:428-430.

Klingenberg CP. 2011. MorphoJ: an integrated software package for geometric morphometrics. Mol Eco Res, advance online publication. doi: 10.1111/j.1755-0998.2010.02924.x

Page 70: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

59

Klingenberg CP, Barluenga M, Meyer A. 2002. Shape analysis of symmetric structures: quantifying variation among individuals and asymmetry. Evolution 56:1909-1920.

Klingenberg CP, McIntyre GS. 1998. Geometric morphometrics of developmental instability: analyzing patterns of fluctuating asymmetry with Procrustes methods. Evolution 52:1363-1375.

Koopman HN, Budge SM, Ketten DR, Iverson SJ. 2006. Topographical distribution of lipids inside the mandibular fat bodies of odontocetes: remarkable complexity and consistency. IEEE J Oceanic Eng 31:95-106.

Kyhn LA, Jensen FH, Beedholm K, Tougaard J, Hansen M, Madsen PT. 2010. Echolocation in sympatric Peale's dolphins (Lagenorhynchus australis) and Commerson's dolphins (Cephalorhynchus commersonii) producing narrow-band high-frequency clicks. J Exp Biol 213:1940-1949.

Li S, Wang D, Tomonari A. 2005. Echolocation signals of the free-ranging Yangtze finless porpoise (Neophocaena phocaenoides asiaorientialis). J Acoust Soc Am 117:3288- 3296.

Li S, Wang D, Wang K, Akamatsu T. 2007. Echolocation click sounds from wild inshore finless porpoise (Neophocaena phocaenoides sunameri) with comparisons to the sonar of riverine N. p. asiaeorientalis. J Acoust Soc Am 121:3938-3946.

Mackay RS, Liaw HM. 1981. Dolphin vocalization mechanisms. Science 212:676-678.

MacLeod CD. 1998. Intraspecific scarring in odontocete cetaceans: an indicator of male ‘quality’ in aggressive social interactions? J Zool 244:71-77.

Madsen PT, Carder DA, Bedholm K, Ridgway SH. 2005a. Porpoise clicks from a sperm whale nose – convergent evolution of 130 kHz pulses in toothed whale sonars? Bioacoustics 15:195-206.

Madsen PT, Johnson M, Aguilar de Soto N, Zimmer WMX, Tyack P. 2005b. Biosonar performance of foraging beaked whales (Mesoplodon densirostris). J Exp Biol 208:181-194.

Madsen PT, Kerr I, Payne R. 2004. Source parameter estimates of echolocation clicks from wild pygmy killer whales (Feresa attenuata) (L). J Acoust Soc Am 116:1909-1912. Mantel N. 1967. The detection of disease clustering and a generalized regression approach. Cancer Res 27:209-220.

Marcus LF, Hingst-Zaher E, Zaher H. 2000. Application of landmark morphometrics to skulls representing the orders of living mammals. Hystrix 11:27-47.

Marino L, McShea DW, Uhen MD. 2004. Origin and evolution of large brains in toothed whales. Anat Rec 281:1247-1255.

May-Collado L, Agnarrson I. 2006. Cytochrome b and Bayesian inference of whale phylogeny. Mol Phylogenet Evol 38:344-354.

McGowen MR, Spaulding M, Gatesy J. 2009. Divergence date estimation and a comprehensive molecular tree of extant cetaceans. Mol Phylogenet Evol 53:891-906.

Page 71: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

60

Meloro C, Raia P, Piras P, Barbera C, O’Higgins P. 2008. The shape of mandibular corpus in large fissiped carnivores: allometry, function and phylogeny. Zool J Linn Soc, 154:832-845.

Miller LA, Pristed J, Møhl B, Surlykke A. 1995. The click-sounds of narwhals (Monodon monoceros) in Inglefield Bay, northwest Greenland. Mar Mamm Sci 11:491-502.

Møhl B, Au WWL, Pawloski J, Nachtigall PE. 1999. Dolphin hearing: relative sensitivity as a function of point of application of a contact sound source in the jaw and head region. J Acoust Soc Am 105:3421-3424.

Monteiro LR, Nogueira MR. 2009. Adaptive radiations, ecological specialization, and the evolutionary integration of complex morphological structures. Evolution 64:724- 744.

Murphy S, Herman JS, Pierce GJ, Rogan E, Kitchener A. 2006. Taxonomic status and geographical cranial variation of common dolphins (Delphinus) in the eastern north Atlantic. Mar Mamm Sci 22:573-599.

Norris KS. 1968. The evolution of acoustic mechanisms in odontocete cetaceans. In: Drake ET, editor. Evolution and Environment. New Haven: Yale University Press. p 297- 324.

Norris KS, Prescott JH, Asa-Dorian PV, Perkins P. 1961. An experimental demonstration of echolocation behavior in the porpoise, Tursiops truncatus (Montagu). Bio Bull 120:163-176.

Nøttestad L, Axelsen BE. 1999. Herring schooling manoeuvers in response to killer whale attacks. Can J Zool 77:1540-1546.

Nummela S, Thewissen JGM, Bajpai S, Hussain ST, Kumar K. 2007. Sound transmission in archaic and modern whales: anatomical adaptations for underwater hearing. Anat Rec 290:716-733.

Paradis E, Claude J, Strimmer K. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289–290.

Perrin WF. 1975. Variation of spotted and spinner porpoise (Genus Stenella) in the eastern Pacific and Hawaii. Bull Scripps Inst Oceanogr 21: 211pp.

Philips JD, Nachtigall PE, Au WWL, Pawloski JL, Roitblat HL. 2002. Echolocation in the Risso's dolphin, Grampus griseus. J Acoust Soc Am 113:605-616.

Pierce SE, Angielczyk KD, Rayfield EJ. 2008. Patterns of morphospace occupation and mechanical performance in extant crocodilian skulls: a combined geometric morphometric and finite element modeling approach. J Morphol 269:840-864.

Popov VV, Supin AY, Klishin VO, Tarakanov MB, Pletenko MG. 2008. Evidence for double acoustic windows in the dolphin, Tursiops truncatus. J Acoust Soc Am 123:552-560.

Page 72: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

61

R Development Core Team. 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Australia. Available at: http://www.r-project.org.

Read AJ. 2002. Porpoises, overview. In: Perrin WF, Würsig B, Thewissen JGM, editors. Encyclopedia of marine mammals. Academic Press, San Diego. p 982-985.

Reeves RR, Stewart BS, Clapham PJ, Powell JA. 2002. National Audubon Society Guide to marine mammals of the world. Alfred A. Knopf, Inc., New York.

Richtsmeier JT, Deleon VB, Lele SR. 2002. The promise of geometric morphometrics. Yearb Phys Anthropol 45:63-91.

Robb RA. 2001. The Biomedical Imaging Resource at Mayo Clinic. IEEE Trans Med Imagin 20:854-67.

Rohlf FJ. 1999. Shape Statistics: Procrustes superimpositions and tangent spaces. J Classif 16:197-223.

Rohlf FJ, Marcus LF. 1993. A revolution in morphometrics. Trends Ecol Evol 8:129-132.

Rommel SA, Pabst DA, McLellan WA. 2009. Skull anatomy. In: Perrin WF, Würsig B, Thewissen JGM, editors. Encyclopedia of Marine Mammals, Second Edition. Academic Press, San Diego. p 1033-1047.

Santos, MB, Pierce GJ, Lopez A, Reid RJ, Ridoux V, Mente E. 2006. Pygme sperm whales Kogia breviceps in the northeast Atlantic: new information on stomach contents and strandings. Mar Mamm Sci 22:600-616.

Schenkkan EJ. 1973. On the comparative anatomy and function of the nasal tract in odontocetes (Mammalia, Cetacea). Bijdr Dierk 43:127-159.

Scott EM, Mann J, Watson-Capps JJ, Sargeant BL, Connor RC. 2005. Aggression in bottlenose dolphins: evidence for sexual coercion, male-male competition, and female tolerance through analysis of tooth-rake marks and behaviour. Behaviour 142:21-44.

Simon M, Wahlberg M, Miller LA. 2007. Echolocation clicks from killer whales (Orcinus orca) feeding on herring (Clupea harengus) (L). J Acoust Soc Am 121:749-752.

Terhune CE, Kimbel WH, Lockwood CA. 2007. Variation and diversity in Homo erectus: a 3D geometric morphometric analysis of the temporal bone. J Hum Evol 53:41-60.

Thewissen JGM, Madar SI, Hussain ST. 1996. Ambulocetus natans, an Eocene cetacean (Mammalia) from Pakistan. Cour Forsch-Inst Senckenberg 191:1-86.

Thompson DW. 1917. On growth and form. Cambridge University Press, Cambridge.

Varanasi U, Malins DC. 1972. Triaclyglycerols characteristic of porpoise acoustic tissues: molecular structures of diisovaleroylglycerolides. Science 176:926-928.

Walker WA, Mead JG, Brownell RL Jr. 2002. Diets of Baird’s beaked whales, Berardius bairdii, in the southern sea of Okhotsk and off the Pacific coast of Honshu, Japan. Mar Mamm Sci 18:902-919.

Page 73: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

62

Wang JY, Chou LS, White BN. 2000. Osteological differences between two sympatric forms of bottlenose dolphins (genus Tursiops) in Chinese waters. J Zool 252:147-162.

Werth A. 2000a. A kinematic study of suction feeding and associated behavior in the long- finned pilot whale, Globicephala melas (Traill). Mar Mamm Sci 16:299-314.

Werth A. 2000b. Feeding in marine mammals. In: Schenk K., editor. Feeding form, function and evolution in tetrapod vertebrates. Academic Press, San Diego. p 487-526.

Werth A. 2006. Mandibular and dental variation and the evolution of suction feeding in Odontoceti. J Mamm 87:579-588.

Westgate AJ. 2007. Geographic variation in cranial morphology of short-beaked common dolphins (Delphinus delphis) from the north Atlantic. J Mamm 88:678-688.

Wills MA, Fortey RA. 2000. The shape of life: how much is written in stone? BioEssays 22:1142-1152.

Young NM. 2008. A comparison of the ontogeny of shape variation in the anthropoid scapula: functional and phylogenetic signal. Am J Phys Anthropol 136:247-264.

Zelditch ML, Swiderski DL, Sheets HD, Fink WL. 2004. Geometric morphometrics for biologists: A primer. Elesevier Academic Press, San Diego.

Zimmer WMX, Johnson MP, Madsen PT, Tyack PL. 2005. Echolocation clicks of free- ranging Cuvier's beaked whales (Ziphius cavirostris). J Acoust Soc Am 117:3919- 3927.

Page 74: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

63

APPENDIX A

SPECIMENS IN ANALYSIS

Page 75: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

64

Tab

le 3

. Sp

ecim

ens

in A

nal

ysis

ID

No.

M

useu

m

Gen

us

Spe

cies

D

ate

Col

.S

can

Loc

. D

ate

Sca

n.S

ex

TL

A

ge

Est

. Age

23

811

SD

MN

H

Del

phin

us

delp

his

3-Ju

l-08

U

CS

D

5-Ju

l-08

--

- --

- --

- --

- 23

024

SD

MN

H

Del

phin

us

delp

his

3-Ju

l-08

U

CS

D

5-Ju

l-08

--

- --

- --

- --

- 23

741

SD

MN

H

Del

phin

us

delp

his

3-Ju

l-08

U

CS

D

5-Ju

l-08

--

- --

- --

- --

- 20

146

SD

MN

H

Lag

enor

hync

hus

obli

quid

ens

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 21

225

SD

MN

H

Lag

enor

hync

hus

obli

quid

ens

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 23

567

SD

MN

H

Ste

nell

a at

tenu

ata

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 23

564

SD

MN

H

Ste

nell

a at

tenu

ata

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 23

563

SD

MN

H

Ste

nell

a at

tenu

ata

3-Ju

l-08

U

CS

D

5-Ju

l-08

M

--

- --

- --

- 23

539

SD

MN

H

Ste

nell

a lo

ngir

ostr

is

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 23

541

SD

MN

H

Ste

nell

a lo

ngir

ostr

is

3-Ju

l-08

U

CS

D

5-Ju

l-08

M

--

- --

- --

- 23

538

SD

MN

H

Ste

nell

a lo

ngir

ostr

is

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 23

540

SD

MN

H

Ste

nell

a lo

ngir

ostr

is

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 23

529

SD

MN

H

Ste

nell

a lo

ngir

ostr

is

3-Ju

l-08

U

CS

D

5-Ju

l-08

M

--

- --

- --

- 23

798

SD

MN

H

Tur

siop

s tr

unca

tus

3-Ju

l-08

U

CS

D

5-Ju

l-08

--

- --

- --

- --

- 21

213

SD

MN

H

Tur

siop

s tr

unca

tus

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 20

145

SD

MN

H

Tur

siop

s tr

unca

tus

3-Ju

l-08

U

CS

D

5-Ju

l-08

F

--

- --

- --

- 39

5375

N

MN

H

Cep

halo

rhyn

chus

eutr

opia

14

-Jul

-08

NM

NH

15

-Jul

-08

F

150

---

Adu

lt

3953

74

NM

NH

C

epha

lorh

ynch

useu

trop

ia

14-J

ul-0

8N

MN

H

15-J

ul-0

8 M

15

2 --

- A

dult

21

167

NM

NH

C

epha

lorh

ynch

useu

trop

ia

14-J

ul-0

8N

MN

H

15-J

ul-0

8 --

- --

- --

- --

- 55

0067

N

MN

H

Cep

halo

rhyn

chus

heav

isid

ii

14-J

ul-0

8N

MN

H

15-J

ul-0

8 --

- 16

9 --

- A

dult

55

0389

N

MN

H

Fer

esa

atte

nuat

a 15

-Jul

-08

NM

NH

15

-Jul

-08

M

208

---

Adu

lt

4849

95

NM

NH

F

eres

a at

tenu

ata

15-J

ul-0

8N

MN

H

15-J

ul-0

8 M

20

7 --

- A

dult

49

582

NM

NH

In

ia

geof

fren

sis

15-J

ul-0

8N

MN

H

15-J

ul-0

8 M

--

- --

- --

- 23

9667

N

MN

H

Inia

ge

offr

ensi

s 15

-Jul

-08

NM

NH

15

-Jul

-08

M

255

---

Adu

lt

3956

14

NM

NH

In

ia

geof

fren

sis

15-J

ul-0

8N

MN

H

15-J

ul-0

8 F

13

6 --

- Ju

veni

le

3956

02

NM

NH

In

ia

geof

fren

sis

15-J

ul-0

8N

MN

H

15-J

ul-0

8 F

20

0 --

- A

dult

(

tabl

e co

ntin

ues)

Page 76: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

65

Tab

le 3

(co

ntin

ued)

ID

No.

M

useu

m

Gen

us

Spe

cies

D

ate

Col

. S

can

Loc

. D

ate

Sca

n.S

ex

TL

A

ge

Est

. Age

57

1619

N

MN

H

Lag

enod

elph

is

hose

i 14

-Jul

-08

NM

NH

15

-Jul

-08

F

227

---

Adu

lt

5713

90

NM

NH

L

agen

orhy

nchu

s ac

utus

15

-Jul

-08

NM

NH

15

-Jul

-08

M

242

---

Adu

lt

5713

42

NM

NH

L

agen

orhy

nchu

s ac

utus

15

-Jul

-08

NM

NH

15

-Jul

-08

M

210

---

Juve

nile

50

4628

N

MN

H

Lag

enor

hync

hus

albi

rost

ris

14-J

ul-0

8 N

MN

H

15-J

ul-0

8 F

23

9 --

- A

dult

27

0418

N

MN

H

Lag

enor

hync

hus

obsc

urus

14

-Jul

-08

NM

NH

15

-Jul

-08

M

---

---

---

5507

63

NM

NH

L

agen

orhy

nchu

s ob

scur

us

14-J

ul-0

8 N

MN

H

15-J

ul-0

8 M

--

- --

- --

- 55

0105

N

MN

H

Mes

oplo

don

euro

paeu

s 15

-Jul

-08

NM

NH

15

-Jul

-08

M

---

Mat

ure

---

5713

57

NM

NH

M

esop

lodo

n m

irus

15

-Jul

-08

NM

NH

15

-Jul

-08

M

---

Mat

ure

---

2400

02

NM

NH

N

eoph

ocae

na

phoc

aeno

ides

15

-Jul

-08

NM

NH

15

-Jul

-08

M

---

---

---

5045

03

NM

NH

P

epon

ocep

hala

el

ectr

a 14

-Jul

-08

NM

NH

15

-Jul

-08

M

---

---

---

5042

50

NM

NH

P

epon

ocep

hala

el

ectr

a 14

-Jul

-08

NM

NH

15

-Jul

-08

F

219

---

Juve

nile

57

1486

N

MN

H

Pho

coen

a di

optr

ica

14-J

ul-0

8 N

MN

H

15-J

ul-0

8 --

- --

- --

- --

- 39

5746

N

MN

H

Pho

coen

a sp

inip

inni

s 15

-Jul

-08

NM

NH

15

-Jul

-08

---

---

---

---

3957

53

NM

NH

P

hoco

ena

spin

ipin

nis

15-J

ul-0

8 N

MN

H

15-J

ul-0

8 --

- --

- --

- --

- 27

6062

N

MN

H

Pho

coen

oide

s

dall

i 15

-Jul

-08

NM

NH

15

-Jul

-08

M

173

---

Adu

lt

2760

63

NM

NH

P

hoco

enoi

des

da

lli

15-J

ul-0

8 N

MN

H

15-J

ul-0

8 M

16

7 --

- A

dult

27

6394

N

MN

H

Pho

coen

oide

s

dall

i 15

-Jul

-08

NM

NH

15

-Jul

-08

M

---

---

---

2345

6 N

MN

H

Pla

tani

sta

gang

etic

a 15

-Jul

-08

NM

NH

15

-Jul

-08

---

---

---

---

5506

65

NM

NH

P

onto

pori

a bl

ainv

ille

i 15

-Jul

-08

NM

NH

15

-Jul

-08

F

---

---

---

5507

31

NM

NH

P

onto

pori

a bl

ainv

ille

i 15

-Jul

-08

NM

NH

15

-Jul

-08

M

---

---

---

5507

30

NM

NH

P

onto

pori

a bl

ainv

ille

i 15

-Jul

-08

NM

NH

15

-Jul

-08

M

---

---

---

5506

57

NM

NH

P

onto

pori

a bl

ainv

ille

i 15

-Jul

-08

NM

NH

15

-Jul

-08

---

160

---

Adu

lt

4827

17

NM

NH

P

onto

pori

a bl

ainv

ille

i 15

-Jul

-08

NM

NH

15

-Jul

-08

M

147

---

Adu

lt

2534

76

NM

NH

S

otal

ia

fluv

iati

lis

14-J

ul-0

8 N

MN

H

15-J

ul-0

8 M

--

- --

- --

- 21

499

NM

NH

S

otal

ia

fluv

iati

lis

14-J

ul-0

8 N

MN

H

15-J

ul-0

8 --

- --

- --

- --

- 25

8859

N

MN

H

Sou

sa

chin

ensi

s 14

-Jul

-08

NM

NH

15

-Jul

-08

---

195

---

Juve

nile

(

tabl

e co

ntin

ues)

Page 77: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

66

Tab

le 3

(co

ntin

ued)

ID N

o.

Mus

eum

G

enus

S

peci

es

Dat

e C

ol.

Sca

n L

oc.

Dat

e S

can.

Sex

T

L

Age

E

st. A

ge

8401

3 L

AC

MN

H

Cep

halo

rhyn

chus

com

mer

soni

i 29

-Dec

-08

UC

LA

30

-Dec

-08

F

---

---

---

8408

1 L

AC

MN

H

Del

phin

apte

rus

leuc

as

29-D

ec-0

8U

CL

A

30-D

ec-0

8F

--

- --

- --

- 95

716

LA

CM

NH

D

elph

inus

ca

pens

is

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

18

8.5

---

Juve

nile

95

741

LA

CM

NH

D

elph

inus

ca

pens

is

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

20

4.5

---

Adu

lt

9575

3 L

AC

MN

H

Del

phin

us

cape

nsis

29

-Dec

-08

UC

LA

30

-Dec

-08

M

228

---

Adu

lt

5412

3 L

AC

MN

H

Glo

bice

phal

a m

acro

rhyn

chus

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

43

1 --

- A

dult

72

546

LA

CM

NH

G

ram

pus

gris

eus

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

36

0.7

---

Adu

lt

9583

6 L

AC

MN

H

Gra

mpu

s gr

iseu

s 29

-Dec

-08

UC

LA

30

-Dec

-08

M

307

---

Juve

nile

86

055

LA

CM

NH

G

ram

pus

gris

eus

29-D

ec-0

8U

CL

A

30-D

ec-0

8F

29

8 --

- Ju

veni

le

5460

0 L

AC

MN

H

Kog

ia

sim

a 29

-Dec

-08

UC

LA

30

-Dec

-08

---

---

---

---

8602

1 L

AC

MN

H

Lag

enor

hync

hus

albi

rost

ris

29-D

ec-0

8U

CL

A

30-D

ec-0

8F

22

0 --

- Ju

veni

le

4347

0 L

AC

MN

H

Lag

enor

hync

hus

obli

quid

ens

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

--

- --

- --

- 51

395

LA

CM

NH

L

agen

orhy

nchu

s ob

liqu

iden

s 29

-Dec

-08

UC

LA

30

-Dec

-08

M

---

---

---

9192

7 L

AC

MN

H

Lis

sode

lphi

s bo

real

is

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

20

1 --

- A

dult

91

950

LA

CM

NH

L

isso

delp

his

bore

alis

29

-Dec

-08

UC

LA

30

-Dec

-08

M

201

---

Adu

lt

9189

8 L

AC

MN

H

Lis

sode

lphi

s bo

real

is

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

--

- --

- --

- 72

539

LA

CM

NH

P

hoco

ena

phoc

oena

29

-Dec

-08

UC

LA

30

-Dec

-08

M

121

---

Juve

nile

84

214

LA

CM

NH

P

hoco

ena

phoc

oena

29

-Dec

-08

UC

LA

30

-Dec

-08

M

131

---

Adu

lt

7258

0 L

AC

MN

H

Pho

coen

oide

s

dall

i 29

-Dec

-08

UC

LA

30

-Dec

-08

M

203.

2--

- A

dult

84

048

LA

CM

NH

P

hoco

enoi

des

da

lli

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

21

3 --

- A

dult

84

047

LA

CM

NH

P

seud

orca

cr

assi

dens

29

-Dec

-08

UC

LA

30

-Dec

-08

M

480

---

Juve

nile

72

154

LA

CM

NH

S

teno

br

edan

ensi

s 29

-Dec

-08

UC

LA

30

-Dec

-08

F

212

---

Adu

lt

7255

8 L

AC

MN

H

Ste

no

bred

anen

sis

29-D

ec-0

8U

CL

A

30-D

ec-0

8M

19

2 --

- Ju

veni

le

ID

No.

= I

dent

ific

atio

n nu

mbe

r

TL

= T

otal

leng

th li

sted

in m

useu

m d

atab

ase

Dat

e C

ol. =

Dat

e C

olle

cted

Age

= A

ge li

sted

in m

useu

m d

atab

ase

Sca

n L

oc. =

Sca

n L

ocat

ion

E

st. A

ge =

Age

est

imat

ed f

rom

tota

l len

gth;

ada

pted

fro

m

Dat

e S

can.

= D

ate

Sca

nned

re

fere

nces

of

age-

leng

th o

n ne

xt p

age

Page 78: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

67

Dawsons, SM. 2002. Cephalorhynchus dolphins. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 200-203.

Donahue, MA and WL Perryman. Pygme killer whale. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 1009-1010.

da Silva, VMF. Amazon river dolphin. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 18-20.

Dolar, MLL. Fraser's dolphin. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 485-487.

Ciprano, F. Atlantic white-sided dolphin. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 49-51.

Kinze, CC. White-beaked dolphins. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 1332-1334.

Perryman, WL. Melon-headed whale. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 733-735.

Jefferson, TA. Dall's porpoise. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 308-310.

Crespo, EA. Franciscana. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 482-485.

Ross, GJB. Humpback dolphins. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 585-589.

Perrin, WF. Common dolphins. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 245-248.

Reeves RR, Stewart BS, Clapham PJ, Powell JA. 2002. National Audubon Society guide to marine mammals of the world. New York: Alfred A. Knoff, Inc. 524 p.

Lipsky, J. Right whale dolphins. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 1030-1033.

Bjorge, A and KA Tolley. Harbor porpoise. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 549-551.

Jefferson, TA. Dall's porpoise. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 308-310.

Baird, RW. False killer whale. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 411-412.

Jefferson, TA. Rought-toothed dolphins. In: Perrin WF, Wursig B, Thewissen JGM, editors. Encyclopedia of marine mammals. San Diego: Academic Press. p 1055-1059.

Page 79: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

68

APPENDIX B

SPECIMENS IN ANALYSIS FROM

CETACEAN DATA LIBRARY

Page 80: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

69

Tab

le 4

. Sp

ecim

ens

in A

nal

ysis

fro

m C

etac

ean

Dat

a L

ibra

ry

F

amil

y G

enus

S

peci

es

Fie

ld I

D

Acc

essi

on I

D

Ow

ner

Sex

A

ge

Del

phin

idae

O

rcin

us

orca

C

RC

-480

02

NW

R01

001

NM

ML

F

emal

e A

dult

Del

phin

idae

S

teno

br

edan

ensi

s --

- R

TD

L-0

09

NM

FS

F

emal

e A

dult

Del

phin

idae

T

ursi

ops

trun

catu

s --

- T

t491

M

NO

SC

M

ale

Adu

lt

Mon

odon

tida

e M

onod

on

mon

ocer

os

"fem

ale"

--

- --

- F

emal

e A

dult

Mon

odon

tida

e M

onod

on

mon

ocer

os

"mal

e"

---

---

Mal

e A

dult

Pho

coen

idae

P

hoco

ena

sinu

s "V

aqui

ta 1

" --

- --

- F

emal

e A

dult

Pho

coen

idae

P

hoco

ena

sinu

s "V

aqui

ta 2

" --

- --

- F

emal

e A

dult

Zip

hiid

ae

Ber

ardi

us

bair

dii

CR

C-4

97

03N

WR

0800

37

PS

U

Fem

ale

Adu

lt

Zip

hiid

ae

Mes

oplo

don

dens

iros

tris

--

- 57

2752

N

MN

H

Fem

ale

Adu

lt

Zip

hiid

ae

Zip

hius

ca

viro

stri

s P

SU

3-1

3-02

02

NW

R03

004

PS

U

Mal

e A

dult

N

MM

L =

Nat

iona

l Mar

ine

Mam

mal

Lab

orat

ory

N

MF

S =

NO

AA

/Nat

iona

l Mar

ine

Fis

heri

es S

ervi

ce

N

OS

C =

Nav

al O

cean

Sys

tem

s C

ente

r, H

awai

i Lab

orat

ory

P

SU

= P

ortl

and

Sta

te U

nive

rsit

y

NM

NH

= N

atio

nal M

useu

m o

f N

atur

al H

isto

ry

Page 81: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

70

APPENDIX C

SUMMARY OF SPECIES IN ANALYSIS

Page 82: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

71

Table 5. Summary of Specimens in Analysis

Family Species No. IndividualsDelphinidae Cephalorhynchus commersonii 1

Cephalorhynchus eutropia 3 Cephalorhynchus heavisidii 1 Delphinus capensis 3 Delphinus delphus 3 Feresa attenuata 2 Globicephala macrorhynchus 1 Grampus griseus 3 Lagenodelphis hosei 1 Lagenorhynchus acutus 2 Lagenorhynchus albirostris 2 Lagenorhynchus obliquidens 4 Lagenorhynchus obscurus 2 Lissodelphis borealis 3 Orcinus orca 1 Peponocephala electra 2 Pseudorca crassidens 1 Sotalia fluviatilis 2 Sousa chinensis 1 Stenella attenuata 3 Stenella longirostris 5 Steno bredanensis 3 Tursiops truncatus 4

TOTAL 53 Phocoenidae Neophocoena phocaenoides 1

Phocoena dioptrica 1 Phocoena phocoena 2 Phocoena sinus 2 Phocoena spinipinnis 2 Phocoenoides dalli 5

TOTAL 13 Monodontidae Delphinapterus leucas 1

Monodon monodon 2 TOTAL 3

Iniidae Inia geoffrensis 4 TOTAL 4 Pontoporiidae Pontoporia blainvillei 5 TOTAL 5 Platanistidae Platanista gangetica 1 TOTAL 1 Kogiidae Kogia simus 1 TOTAL 1 Ziphiidae Berardius bairdii 1

Mesoplodon densirostris 1 Mesoplodon europaeus 1 Mesoplodon mirus 1 Ziphius cavirostris 1

TOTAL 5 TOTAL NUMBER OF INDIVIDUALS 85 TOTAL NUMBER OF SPECIES 40

Page 83: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

72

APPENDIX D

X-RAY COMPUTED TOMOGRAPHY (CT)

SCANNING PARAMETERS

Page 84: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

73

Tab

le 6

. X-r

ay C

omp

ute

d T

omog

rap

hy

(CT

) S

can

nin

g P

aram

eter

s

UC

SD

N

MN

H -

Box

1

NM

NH

- B

ox 2

U

CL

A -

Box

1

UC

LA

- B

ox 2

Sca

nner

G

E

Lig

htS

peed

Sie

men

s S

omat

om

Em

otio

n

Sie

men

s S

omat

om

Em

otio

n

Sie

men

s S

omat

om

Def

init

ion

Sie

men

s S

omat

om

Def

init

ion

kV

---

110

110

120

120

mA

--

- 80

80

10

0 10

0

Fie

ld o

f V

iew

(m

m)

---

412

/ -16

/ 0

412

/ 8 /

-5

485

500

Vox

el S

ize

(mm

) 0.

9765

62

0.5

0.5

0.6

0.6

Dat

a T

ype

16-b

it s

igne

d 16

-bit

sig

ned

16-b

it s

igne

d 16

-bit

sig

ned

16-b

it s

igne

d

UC

SD

= U

nive

rsit

y of

Cal

ifor

nia

San

Die

go M

edic

al C

ente

r –

Hil

lcre

st

N

MN

H =

Nat

iona

l Mus

eum

of

Nat

ural

His

tory

UC

LA

= U

nive

rsit

y of

Cal

ifor

nia

Los

Ang

eles

Med

ical

Sch

ool

Page 85: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

74

APPENDIX E

EIGENVALUES AND PERCENT VARIANCE

Page 86: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

75

Table 7. Eigenvalues and Percent Variance

Principal

Component Eigenvalues % Variance Cumulative

% 1 0.0147374 44.956 44.956 2 0.01164473 35.522 80.478 3 0.0020538 6.265 86.743 4 0.00151977 4.636 91.379 5 0.00125058 3.815 95.194 6 0.00058178 1.775 96.969 7 0.00026046 0.795 97.764 8 0.00020037 0.611 98.375 9 0.0001489 0.454 98.829 10 0.00012113 0.37 99.198 11 0.00010381 0.317 99.515 12 0.00006579 0.201 99.716 13 0.00005099 0.156 99.871 14 0.00002342 0.071 99.943 15 0.00001873 0.057 100

Page 87: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

76

APPENDIX F

PRINCIPAL COMPONENT COEFFICIENTS

Page 88: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

77

Tab

le 8

. Pri

nci

pal

Com

pon

ent

Coe

ffic

ien

ts

PC

1

PC

2

PC

3

PC

4

PC

5

PC

6

PC

7

PC

8

PC

9

PC

10

P

C11

PC

12

P

C13

PC

14

P

C15

x1

0.03

50

-0.0

696

-0.0

245

-0.0

086

-0.1

028

0.53

97

0.01

40

-0.0

297

0.05

29

-0.0

934

0.03

63

-0.0

844

-0.0

138

0.07

27

0.07

88

y1

0.

2655

0.

1861

0.

1748

0.

0054

-0

.022

8 0.

0138

-0

.303

8 -0

.209

6 -0

.132

2 -0

.189

0 0.

1257

-0

.225

9 -0

.183

8 0.

0572

0.

1622

z1

-0.0

020

0.00

69

-0.0

298

0.01

90

0.00

95

0.02

76

-0.3

069

0.23

64

-0.1

208

-0.1

089

-0.1

319

0.13

13

0.29

36

0.03

51

0.14

47

x2

0.

0350

-0

.069

6 -0

.024

5 -0

.008

6 -0

.102

8 0.

5397

0.

0140

-0

.029

7 0.

0529

-0

.093

4 0.

0363

-0

.084

4 -0

.013

8 0.

0727

0.

0788

y2

-0.2

655

-0.1

861

-0.1

748

-0.0

054

0.02

28

-0.0

138

0.30

38

0.20

96

0.13

22

0.18

90

-0.1

257

0.22

59

0.18

38

-0.0

572

-0.1

622

z2

-0

.002

0 0.

0069

-0

.029

8 0.

0190

0.

0095

0.

0276

-0

.306

9 0.

2364

-0

.120

8 -0

.108

9 -0

.131

9 0.

1313

0.

2936

0.

0351

0.

1447

x3

-0.0

356

-0.0

063

0.32

91

-0.0

349

-0.3

915

-0.2

818

-0.0

530

0.04

34

0.03

86

0.06

35

0.11

65

0.06

58

-0.0

231

0.03

49

-0.0

696

y3

0.

2086

0.

1517

0.

0317

0.

0445

0.

0462

0.

1273

-0

.073

9 0.

0155

0.

3260

0.

2075

0.

1130

0.

4255

-0

.060

0 0.

1953

0.

0128

z3

0.10

46

0.01

13

-0.0

025

-0.0

105

0.07

53

-0.0

094

0.18

52

0.11

10

0.22

80

0.12

44

0.30

54

-0.3

618

0.09

76

0.00

51

-0.0

857

x4

-0

.035

6 -0

.006

3 0.

3291

-0

.034

9 -0

.391

5 -0

.281

8 -0

.053

0 0.

0434

0.

0386

0.

0635

0.

1165

0.

0658

-0

.023

1 0.

0349

-0

.069

6

y4

-0.2

086

-0.1

517

-0.0

317

-0.0

445

-0.0

462

-0.1

273

0.07

39

-0.0

155

-0.3

260

-0.2

075

-0.1

130

-0.4

255

0.06

00

-0.1

953

-0.0

128

z4

0.

1046

0.

0113

-0

.002

5 -0

.010

5 0.

0753

-0

.009

4 0.

1852

0.

1110

0.

2280

0.

1244

0.

3054

-0

.361

8 0.

0976

0.

0051

-0

.085

7

x5

-0.0

400

0.14

61

0.15

69

0.38

81

0.40

01

-0.0

845

-0.0

539

-0.0

606

0.02

19

0.00

26

-0.0

399

-0.0

193

0.09

20

-0.0

888

-0.0

809

y5

0.

2227

0.

0969

0.

0780

-0

.025

8 -0

.139

6 0.

1896

-0

.007

7 -0

.028

0 -0

.128

0 0.

0317

-0

.096

0 0.

0502

0.

1669

-0

.309

7 -0

.452

9

z5

-0.0

507

0.01

45

-0.0

203

0.04

48

0.07

40

0.02

44

0.05

82

0.11

41

-0.1

082

-0.1

285

-0.0

349

0.12

05

-0.5

280

-0.0

144

-0.2

091

x6

-0

.040

0 0.

1461

0.

1569

0.

3881

0.

4001

-0

.084

5 -0

.053

9 -0

.060

6 0.

0219

0.

0026

-0

.039

9 -0

.019

3 0.

0920

-0

.088

8 -0

.080

9

y6

-0.2

227

-0.0

969

-0.0

780

0.02

58

0.13

96

-0.1

896

0.00

77

0.02

80

0.12

80

-0.0

317

0.09

60

-0.0

502

-0.1

669

0.30

97

0.45

29

z6

-0

.050

7 0.

0145

-0

.020

3 0.

0448

0.

0740

0.

0244

0.

0582

0.

1141

-0

.108

2 -0

.128

5 -0

.034

9 0.

1205

-0

.528

0 -0

.014

4 -0

.209

1

x7

0.12

97

0.07

18

0.01

30

-0.4

962

0.26

04

-0.0

679

0.05

58

0.05

75

-0.0

734

0.13

01

-0.0

991

0.03

01

-0.0

494

-0.1

346

0.12

58

y7

0.

1323

0.

0597

0.

0872

0.

2166

-0

.149

5 0.

0236

0.

2619

0.

1934

0.

0571

-0

.043

7 -0

.096

5 0.

0426

-0

.078

2 -0

.355

0 0.

3707

z7

-0.0

193

-0.0

274

0.06

72

-0.1

046

-0.0

114

-0.0

666

0.18

56

-0.4

312

0.19

31

-0.2

951

-0.1

012

0.14

76

0.12

78

-0.0

335

0.03

57

x8

0.

1297

0.

0718

0.

0130

-0

.496

2 0.

2604

-0

.067

9 0.

0558

0.

0575

-0

.073

4 0.

1301

-0

.099

1 0.

0301

-0

.049

4 -0

.134

6 0.

1258

y8

-0.1

323

-0.0

597

-0.0

872

-0.2

166

0.14

95

-0.0

236

-0.2

619

-0.1

934

-0.0

571

0.04

37

0.09

65

-0.0

426

0.07

82

0.35

50

-0.3

707

z8

-0

.019

3 -0

.027

4 0.

0672

-0

.104

6 -0

.011

4 -0

.066

6 0.

1856

-0

.431

2 0.

1931

-0

.295

1 -0

.101

2 0.

1476

0.

1278

-0

.033

5 0.

0357

x9

0.19

42

0.11

85

-0.5

025

0.11

06

-0.1

556

-0.1

961

0.05

89

-0.0

004

-0.0

305

-0.1

115

-0.0

258

-0.0

071

0.02

09

0.10

81

-0.0

489

y9

0.

1339

0.

0414

0.

2295

0.

0428

0.

0122

0.

0285

0.

3182

0.

0900

-0

.190

6 0.

0019

-0

.268

6 -0

.063

6 0.

0717

0.

4342

-0

.028

7

z9

-0.0

393

0.02

48

-0.0

720

0.10

64

-0.1

051

0.03

98

0.00

16

-0.2

692

-0.2

603

0.46

79

-0.0

237

0.00

54

-0.0

219

-0.0

171

0.13

67

x1

0 0.

1942

0.

1185

-0

.502

5 0.

1106

-0

.155

6 -0

.196

1 0.

0589

-0

.000

4 -0

.030

5 -0

.111

5 -0

.025

8 -0

.007

1 0.

0209

0.

1081

-0

.048

9

y10

-0.1

339

-0.0

414

-0.2

295

-0.0

428

-0.0

122

-0.0

285

-0.3

182

-0.0

900

0.19

06

-0.0

019

0.26

86

0.06

36

-0.0

717

-0.4

342

0.02

87

z1

0 -0

.039

3 0.

0248

-0

.072

0 0.

1064

-0

.105

1 0.

0398

0.

0016

-0

.269

2 -0

.260

3 0.

4679

-0

.023

7 0.

0054

-0

.021

9 -0

.017

1 0.

1367

x11

0.09

96

-0.8

187

0.04

61

0.13

45

0.08

23

-0.0

053

-0.0

833

-0.0

504

-0.0

246

0.04

33

-0.0

049

0.05

12

-0.0

617

-0.0

106

-0.0

116

y1

1 0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

0.

0000

z11

-0.0

046

-0.0

176

0.03

44

-0.0

495

-0.1

225

-0.0

341

-0.3

080

0.20

14

0.41

59

0.08

64

-0.4

937

-0.2

858

-0.1

008

0.02

21

-0.0

648

x1

2 -0

.666

0 0.

2977

0.

0098

-0

.052

5 -0

.103

7 0.

1864

0.

0396

0.

0300

0.

0057

-0

.026

0 0.

0290

-0

.021

5 0.

0086

0.

0261

0.

0012

y12

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

0.00

00

z1

2 0.

0183

-0

.042

7 0.

0805

-0

.060

8 0.

0379

0.

0024

0.

0605

0.

2763

-0

.279

7 -0

.206

0 0.

4664

0.

1996

0.

1624

0.

0274

0.

0204

Page 89: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

78

APPENDIX G

SPECIMEN PRINCIPAL COMPONENT SCORES –

PC1 TO PC4 SPECIES MEANS

Page 90: SHAPE ANALYSIS OF ODONTOCETE MANDIBLES: FUNCTIONAL AND

79

Table 9. Specimen Principal Component Scores – PC 1 to PC 4 Species Means

Species PC1 PC2 PC3 PC4

Cephalorhynchus commersonii -0.013169908 0.069441866 -0.003674553 0.028128135 Cephalorhynchus eutropia 0.000195332 0.078252556 0.006785685 -0.011784029Ceohalorhynchus heavisidii -0.033007317 0.039483903 -0.042662778 -0.023203286Delphinus capensis 0.12306697 0.123514133 0.026724961 -0.034049107Delphinus delphis 0.129555486 0.128871134 0.013940873 -0.01999576 Feresa attenuata -0.133949006 -0.038574378 -0.035381334 -0.031244071Globicephala macrorhynchus -0.196402916 -0.082464432 -0.056582058 0.056412839 Grampus griseus -0.165167779 -0.075890121 -0.007439318 0.059067138 Lagenodelphis hosei -0.000853 0.076016127 0.013123409 -0.007750949Lagenorhynchus acutus -0.026056752 0.05571236 -0.036796162 -0.002822556Lagenorhynchus albirostris -0.027158188 0.046599668 -0.043409729 -0.050922181Lagenorhynchus obliquidens -0.018754486 0.064178057 -0.017594343 0.000662673 Lagenorhynchus obscurus 0.021792755 0.098269736 -0.000232021 -0.02939521 Lissodelphis borealis 0.053127855 0.087601788 -0.005989094 0.04499485 Orcinus orca -0.164413316 -0.05313715 -0.034055374 -0.049955391Peponocephala electra -0.078339103 0.022345983 -0.023152757 0.014030181 Pseudorca crassidens -0.117301989 -0.061721093 -0.052516826 -0.042154394Sotalia fluviatlis 0.086287142 0.065306615 -0.072456268 0.036342445 Sousa chinensis 0.122658307 0.041166478 -0.045029355 0.014235199 Stenella attenuata 0.092484082 0.076711716 0.008335166 0.031856857 Steno bredanensis 0.062205272 -0.002436493 -0.009085479 0.014024025 Stenella longirostris 0.138940329 0.117010492 0.024204419 -0.017164011Tursiops truncatus 0.008868589 0.067200192 -0.046222667 -0.017008041Delphinapterus leucas -0.140142676 -0.063014502 -0.032661474 -0.029879398Monodon monoceros -0.196926508 -0.077319157 0.063501104 -0.036098781Neophocaena phocaenoides -0.218845981 -0.099213503 -0.058227892 -0.060148659Phocoenoides dalli -0.123270713 -0.008820305 0.052990208 0.011829277 Phocoena dioptrica -0.116659712 0.015858263 0.020040445 0.047477307 Phocoena phocoena -0.120789486 -0.020902158 0.026234896 -0.049597217Phocoena sinus -0.187124866 -0.056480646 -0.03139462 -0.065464927Phocoena spinipinnis -0.093815498 0.001044048 0.077239942 -0.026142746Kogia simus -0.205057998 -0.139106853 0.046371481 0.086906061 Inia geoffrensis 0.131767265 -0.236695604 -0.056867344 0.003844501 Platanista gangetica 0.17569364 -0.228330629 0.071472624 -0.08023099 Pontoporia blainvillei 0.209718165 -0.238299112 0.041924631 -0.0030789 Berardius bairdii 0.016967387 0.045534094 -0.044665831 0.062398595 Mesoplodon densirostris 0.079934094 0.012951145 -0.042814437 0.108008462 Mesoplodon europaeus -0.125376057 0.09328583 0.146332539 0.047189546 Mesoplodon mirus -0.00895008 -0.003944819 0.054445554 0.072297454 Ziphius cavirostris -0.04377928 -0.032868962 0.000121 0.092629499