Vertebral osteology in Delphinidae (Cetacea)academics.wellesley.edu/Biology/Faculty/Emily/vertebral ost.pdfVertebral osteology in Delphinidae ... The dolphin vertebral column has been

Zoological Journal of the Linnean Society, 2004, 140, 383–401. With 10 figures

© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 383–401 383

Blackwell Science, LtdOxford, UKZOJZoological Journal of the Linnean Society0024-4082The Lin-nean Society of London, 2004? 20041403383401Original Article

E. A. BUCHHOLTZ and S. A. SCHURDELPHINID VERTEBRAL OSTEOLOGY

*Corresponding author. E-mail: [email protected]†Current address: Department of Organismic and Evolutionary Biology, Museum of Comparative Zoology, Harvard University, Cambridge, MA 02138, USA

Vertebral osteology in Delphinidae (Cetacea)

EMILY A. BUCHHOLTZ* and STEPHANIE A. SCHUR†

Wellesley College, Wellesley, MA 02481, USA

Received March 2002; accepted for publication October 2003

Vertebral anatomy in delphinid cetaceans exhibits marked heterogeneity. Description and functional interpretationof this variability is facilitated by the recognition of structural units along the column whose boundaries transgressthose of the classical mammalian series. Vertebral anatomy of the killer whale (Orcinus orca) and the Atlantic white-sided dolphin (Lagenorhynchus acutus) lie near the ends of an anatomical continuum. Primitive columns resemblethose of living delphinapterid delphinoids in exhibiting minimal intervertebral variation, low counts and spool-shaped vertebrae. Derived columns are more regionalized, displaying traits that limit mobility in the anterior torso,enhance flexibility at the point of neural spine syncliny and increase dorsoventral displacement of prefluke verte-brae. Reconstruction of the historical sequence of anatomical innovations identifies syncliny as an early and criticalstep in delphinid column evolution. Trait distribution supports evolutionary isolation of Pseudorca and Orcinus fromremaining delphinids, inclusion of Feresa and Peponocephala among delphinine delphinids, and subdivision of del-phinines on the basis of centrum dimensions, neural spine inclination and count. Details of vertebral anatomy canalso be used to place fragmentary postcranial material, particularly that of fossils, in functional and evolutionarycontext. © 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 383–401.

ADDITIONAL KEYWORDS: anatomy – evolution – postcranium – vertebrae.

INTRODUCTION

Cetaceans are obligate marine mammals with anEocene ancestry among terrestrial ungulates. Taxo-nomic diagnosis is made primarily on the basis oftympanic and dental specializations (Thewissen,1994; Luo, 1998; Berta & Sumich, 1999), but amongliving mammals, whales are easily recognized by theradical transformations of the skull, limbs and axialskeleton required for life in the water.

Monophyly of the odontocete superfamily Delphi-noidea has been recognized on the basis of both molec-ular (Milinkovich, Orti & Meyer, 1993, 1994;Hasegawa, Adachi & Milinkovitch, 1997) and morpho-logical (Heyning, 1989; Fordyce & Barnes, 1994) evi-dence. Relationships of its three extant families(Monodontidae, Phocoenidae, Delphinidae), however,are disputed (Mead, 1975; Heyning, 1989; Arnold &Heinsohn, 1996; Waddell et al., 2000). Delphinids are

characterized by lack of the posterior sac of the nasalpassage (Fordyce, 1994) and by a reduced posteriorend of the left premaxilla (Heyning, 1989; Fordyce &Barnes, 1994). Since its Miocene origin, this speciose(17–19 genera, 34–36 species) and geographicallywidespread family has undergone rapid radiation.

Within Delphinidae, relationships are poorlyresolved (LeDuc, 2002); genera are commonly yet vari-ably assigned to five or six subfamilies. Neither mor-phological (Kasuya, 1973; Mead, 1975; de Muizon,1988; Perrin, 1989; Barnes, 1990) nor molecular(LeDuc, Perrin & Dizon, 1999; LeDuc, 2002) data setsfully resolve subfamilial placement of extant genera.The consensus phylogeny presented in Figure 1 drawsprimarily on the morphological work of Mead (1975),de Muizon (1988) and Perrin (1989), but differs insome respects from each. All three authors isolate thegenus Cephalorhynchus at the subfamilial level, butadditional monotypic subfamilies are erected forOrcaella (by Perrin) and for Lissodelphis (by both Per-rin and Mead). Both de Muizon and Perrin placePeponocephala in the Orcininae instead of the Del-phininae, the affiliation suggested by Mead. deMuizon also recognizes two delphinine subgroups

384 E. A. BUCHHOLTZ and S. A. SCHUR

© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 383–401

(Lagenorhynchus, Lagenodelphis and Lissodelphis;Grampus, Tursiops, Delphinus and Stenella). Thehierarchical placement of the subfamilies in the con-sensus phylogeny follows de Muizon, as both Meadand Perrin present their proposals as non-hierarchicalclassifications.

Despite the physical constraints of an obligateaquatic environment (see e.g., Motani, 2002), ceta-ceans display a considerable range of behaviour andinternal morphology. This variability extends to thevertebral column of delphinids, where it is most evi-dent in extremes of count (47–91 vertebrae) and cen-

trum length (spool vs. disc shape). We initiated thisstudy with the hypothesis that vertebral variation ishierarchically rather than randomly distributed, andthat this distribution pattern can be used both to inferthe history of trait acquisition and to contribute to theresolution of phylogenetic relationships among del-phinids. Because early archeocete whales (Slijper,1936; Thewissen, Hussain & Arif, 1994; Thewissen,Madar & Hussain, 1996; Buchholtz, 1998, 2001; Gin-gerich, 1998, 2003; Thewissen & Bajpai, 2001) and thedelphinapterid odontocetes used as the outgroup inthis project are characterized by low vertebral counts,spool-shaped centra and nearshore habitats, we iden-tified these traits as primitive, and predicted that highcounts would be associated with discoidal centra andpelagic habits.

MECHANICAL EFFECTS OF VERTEBRAL VARIATIONS

In this study we describe key aspects of vertebral oste-ology that differ either along the column of a singledelphinid and/or among delphinid taxa. In the func-tional interpretation of this anatomy, we draw on pre-vious work of general application to all vertebrates(e.g., Slijper, 1946; Filler, 1986; Gál, 1993; Walker &Liem, 1994; McGowan, 1999; Hildebrand & Goslow,2001) as well as more specifically on previous descrip-tions of the cetacean vertebral column and/or muscu-lature (e.g., Slijper, 1946, 1961; Smith, Browne &Gaskin, 1976; DeSmet, 1977; Strickler, 1980; Pabst,1990, 1993, 1996, 2000; Rommel, 1990; Long et al.,1997), and especially on the monographic work of E. J.Slijper (1936).

The dolphin vertebral column has been described asa variably flexible beam (Long et al., 1997; Pabst,2000). Among the factors known to affect that flexibil-ity are muscular and ligamentous tissue, interverte-bral disc composition and size, and vertebral structureand interference (Gál, 1993; Long et al., 1997). Weacknowledge and reiterate the warning offered byLong et al. (1997) that skeletal features can by them-selves only partially predict column function.

Tetrapod vertebrae consist of a basal centrum, typ-ically with neural and (in caudal vertebrae) ventral(haemal) midline processes. These units may be elab-orated by accessory outgrowths for muscle attachmentand/or skeletal articulation. Movement between ver-tebrae may be sagittal (about a horizontal axis thatruns across the anterior centrum face at mid height),lateral (about a vertical axis that runs through thecentre of the anterior centrum face) or rotational(around the long axis of the vertebral column). As inmost mammals, the dominant plane of vertebralmovement in Cetacea is sagittal, and rotational move-ment is minimal. Variations in vertebral structure

Figure 1. Consensus phylogeny of the Family Delphinidaebased on the work of Mead (1975), de Muizon (1988) andPerrin (1989). The Delphinapteridae serves as theoutgroup.

Monodon

Pseudorca Orcinus

Orcaella Feresa

Steno Sotalia

Grampus Tursiops

Delphinus Stenella

Peponocephala

Lissodelphis Lagenodelphis Lagenorhynchus

Cephalorhynchinae Cephalorhynchus

Steninae

Delphininae

Delphinapteridae Delphinapterus

Orcininae

Globicephala

DELPHINID VERTEBRAL OSTEOLOGY 385


that constrain or enhance movement between adja-cent vertebrae can be grouped into considerations ofcentrum shape and spacing, process structure and ori-entation, count and accessory structures.

CENTRUM SHAPE AND SPACING

Centrum shape (face curvature, width, height) affectsnot only the angle through which a vertebra can rotatebut also the absolute distance of displacement of a cen-trum’s posterior face relative to its anterior face.

Face curvature. Delphinid centrum faces range fromflat to gently convex in profile. Given constant inter-vertebral spacing, rounded faces enhance rotation byreducing interference of centrum margins on adjacentvertebrae.

Centrum length. Delphinid centra range from spool-like to discoidal in shape. Given constant angular rota-tion, increased centrum length increases the absolutedisplacement of the posterior face of a vertebra fromthe axis of the vertebra anterior to it.

Centrum width and height. Vertical and horizontaldimensions of delphinid centra vary both in absolutesize and in relationship to each other. Interferencebetween adjacent vertebrae separated by a given inter-vertebral space and rotated through a given angleincreases as dimensions of the centrum faces increase.

The combined effects of extremes of centrum facecurvature and centrum length are shown diagram-matically in Figure 2. Vertebrae with low curvatureand short centrum length allow little intervertebralrotation or displacement, but column elasticity can bepredicted as the result of the high ratio of interverte-bral disc to bone. Vertebrae with high curvature andshort centrum length allow intervertebral rotationwith minimal absolute displacement. Vertebrae withlow curvature and long centrum length produce rigidcolumns with minimal rotation at a restricted numberof intervertebral sites, whereas those with high cur-vature and long centrum length enhance both rotationand absolute displacement.

The mechanical effects of variable intervertebraldisc length in Delphinus delphis were evaluatedexperimentally by Long et al. (1997). They found thatdisc length was greater in caudal than in precaudalsites, and was negatively correlated with initial bend-ing stiffness in both extension and flexion. An X-ray ofan adult specimen of the dolphin Stenella longirostris(Crovetto, 1991: 140) indicates that intervertebraldiscs are noticeably longer in vertebrae near the flukebase than at adjacent sites. Unfortunately neitherfresh tissue nor X-rays are consistently available forinterspecific comparisons, and intervertebral disclength remains an undocumented variable in thisstudy.

Figure 2. Diagrammatic presentation of vertebral centra in lateral view to demonstrate the variations in face curvatureand dimensions that affect intervertebral movement.

Low curvature, short centrum length High curvature, short centrum lengthElastic stability Rotation with low displacement

Low curvature, long centrum length High curvature, long centrum lengthRigidity Rotation with high displacement



PROCESS STRUCTURE AND ORIENTATION

Neural, haemal and transverse processes act as sitesof muscle origin and insertion. Changes in attachmentsites alter the resulting torque or moment of a givenmuscle system. Effects of variations in process struc-ture and orientation on sagittal rotation of a vertebraare summarized below.

Process height. Given constant muscle attachmentsites and process orientations, elevation of the mus-cle’s point of origin will increase the distance betweenforce application and the centrum, through which theaxis of vertebral rotation passes. As a result, both thelength of the effective in-lever and the mechanicaladvantage of the associated muscle system increase(Walker, 1965; McGowan, 1999). This increase inmechanical advantage comes at the cost of a decreasein angular rotation about the axis and of the velocityratio. Taller neural processes also reduce the amountof angular rotation possible before interferencebetween adjacent processes occurs. Several authors(Smith et al., 1976; Fish & Hui, 1991; McGowan, 1999)have noted the association of tall neural processes,typical of the torso, with increased mechanical advan-tage in cetaceans. Long et al. (1997) experimentallydocumented the association of tall neural processeswith column stability in Delphinus delphis. Theshorter processes immediately anterior to the flukeare conversely associated with low mechanical advan-tage but a greater angular rotation and velocity ratio.

Process orientation. Given constant muscle attach-ment sites and lengths, the inclination of processeschanges the line of muscle action and the effectivelever arm of the system. Any change that brings theangle between force and lever arm closer to the per-pendicular will increase the torque of the system(Hildebrand & Goslow, 2001).

Muscle length. Muscle length varies as the result ofdifferences in vertebral process geometry. Given a con-stant proportion of length contraction and constantcontraction time, longer muscles have greater contrac-tion velocity (Kardong, 1998).

Process length. Processes also have length in theantero-posterior direction. For a given centrumlength, increases in process length will minimize rota-tion by increasing interference between adjacent ver-tebrae during rotation in the plane of the process.

VERTEBRAL COUNT

Given constant vertebral length and intervertebralspacing, increased count must increase not only thetotal length of the column but also the number of sitesfor rotation and the total absolute displacement of the

posterior column from the axis of the first vertebra.Somewhat counterintuitively, however, columns ofhigh count are typical of smaller species and consis-tently display multiple traits associated with limitedintervertebral mobility, especially in the anterior partof the column (see below).

ACCESSORY STRUCTURES

Accessory structures (metapophyses, zygapophyses,ribs) may limit or enhance muscle action and rotationbetween vertebrae.

Metapophyses. Metapophyses are muscle attachmentsites located on the neural processes, typically at thejunction of neural arch and spine. They provide midspine insertion sites for the multifidus and longissi-mus muscle systems, which are the main effectors ofcolumn extension. Elevation of metapophysesincreases in-lever length of attached muscles, increas-ing mechanical advantage. Both Lütken (1888) andHowell (1930) described metapophysis location anddistribution, and Slijper (1936) used metapophysealtraits as major characters in his allocation of cetaceanspecies to locomotor groups (‘Stufen’). Absence ofmetapophyses indicates the absence of mid spine mus-cle attachment sites. Regionally, metapophyses of onevertebra may also overlap neural spines of the nextcranial vertebra, constraining intervertebral move-ment (Long et al., 1997).

Zygapophyses. Zygapophyses are intervertebral artic-ulation sites that restrict vertebral rotation in axesthat intersect the plane of the zygapophyseal surface.Zygapophyses are present on only cervical and ante-rior thoracic vertebrae in most delphinids.

Ribs. Ribs, especially those with sternal connections,stabilize the thorax and limit rotation between thevertebrae with which they articulate (Filler, 1986).

SUMMARY

Structural variations of vertebrae may constrain orenhance intervertebral motion. As a general rule, ver-tebrae in flexible column areas display some combina-tion of high centrum face curvature, relatively smallcentrum face dimensions, and neural processes withhigh angular inclination, limited height and/or shortanterior–posterior length. Such vertebrae bearmetapophyses, typically located close to the centrum;they lack zygapophyses and ribs. Vertebrae in morehighly stabilized column areas typically have flat cen-trum faces with relatively large dimensions, and tall,long and erect neural processes. Metapophyses may beregionally absent; where present they are often ele-vated. Zygapophyses and ribs may be present.



MATERIAL AND METHODS

We have examined and measured postcranial skele-tons of members of the following delphinid genera:Cephalorhynchus, Delphinus, Feresa, Globicephala,Grampus, Lagenodelphis, Lagenorhynchus, Lissodel-phis, Orcaella, Orcinus, Peponocephala, Pseudorca,Sotalia, Stenella, Steno and Tursiops, as well as of thetwo delphinapterid genera, Monodon and Delphi-napterus. For comparison, we have also measuredspecimens of the phocoenids Phocoena, Phocoenoidesand Neophocoena, of the river dolphins Inia, Pontopo-ria, Lipotes and Platanista, and the partial skeletonsof a broad sampling of fossil species. Specimen num-bers and museum sources of the delphinid specimensare listed in Table 1. Complete or nearly completeindividuals were measured whenever possible,although terminal caudal vertebrae were frequentlymissing. Individuals were adult or subadult as evalu-ated by the degree of epiphyseal fusion. All fossil spec-imens were fragmentary.

A suite of measurements was used to describe theindividual vertebrae of each specimen (Fig. 3). Digitalcalipers were used to measure centrum length (CL,ventrally), centrum width (CW, anteriorly) and cen-trum height (CH, anteriorly). Dimensions and inclina-tions of neural processes and their component archesand spines were measured from scaled images of eachvertebra. Neural process height (NPH) is the verticaldistance from the tip of the neural process to the(extended) horizontal line along the dorsal centrumsurface. Neural arch height (NAH) is the running dis-tance from the metapophysis through the centre of thearch to this same centrum surface. Neural arch incli-nation (NAI) is the angle between this line and thehorizontal. Neural spine height (NSH) is the runningdistance from the dorsal tip of the spine through thecentre of the spine to the horizontal line at the level ofthe metapophysis. Neural spine inclination (NSI) isthe angle between this line and the horizontal. Ante-rior inclinations are represented by values >90∞ andposterior inclinations by values <90∞. A process is saidto have an elevated metapophysis when NAH > NSH.Total and series counts from examined specimens arereported, and are supplemented by total counts fromthe literature to reflect variation due to intraspecificdifferences and/or specimen incompleteness.

Vertebral measurements of each specimen wereplotted by column location. Morphological and/ordimensional discontinuities were used to identifystructural subunits of the column. Functional impli-cations of these structural series were then proposed,and variations among delphinid genera noted. Osteo-logical traits were evaluated for polarity using the del-phinapterids Delphinapterus leucas and Monodonmonoceros as primary outgroups. These polarized

traits were then mapped on to the consensus phylog-eny to predict the historical sequence of trait acquisi-tion and to identify possible instances of homoplasy.Implications of trait distribution for delphinid phylog-eny and for the evolutionary interpretation of theincomplete columns of extinct taxa were alsoaddressed.

RESULTS

As the work of E. J. Slijper (1936, 1946) attests, detailsof dolphin postcranial anatomy vary complexly andgradationally. We have chosen to present detaileddimensional data and photographic images of the oste-ology of two delphinid species, Orcinus orca andLagenorhynchus acutus, species that lie near theextremes of the morphological continuum and areknown from multiple specimens. We extend this pre-sentation with short descriptions of individual traitsand then chart their distribution among the extantspecies studied.

DESCRIPTION OF MORPHOLOGICAL EXTREMES

Orcinus orca, the killer whale, is a large dolphin witha worldwide distribution. The vertebral count for the

Figure 3. Left lateral view of an anterior caudal vertebrawith measurements used in this study indicated.CL = centrum length; CH = centrum height;CW = centrum width; NAH = neural arch height;NAI = neural arch inclination; NPH = neural processheight; NSH = neural spine height; NSI = neural spineinclination.



Tab

le 1

.D

elph

inid

an

d de

lph

inap

teri

d ce

tace

an s

peci

men

s ex

amin

ed. S

erie

s co

un

ts o

f ex

amin

ed s

peci

men

s ar

e su

pple

men

ted

wit

h s

peci

es r

ange

s ci

ted

in t

he

lit-

erat

ure

. Su

bfam

ilia

l m

embe

rsh

ip r

eflec

ts t

hat

of

the

con

sen

sus

phyl

ogen

y. T

C =

tot

al c

oun

t; M

axC

= m

axim

um

cou

nt

ran

ge f

or s

peci

es;

ME

= m

etap

oph

ysis

ele

-va

tion

; S

yn =

neu

ral

spin

e sy

ncl

iny;

2R

= s

econ

dary

ris

e in

CL

/CH

; TC

L =

ave

rage

th

orac

ic C

L/C

H;

LC

L =

ave

rage

lu

mba

r C

L/C

H;

ML

= r

egio

nal

met

apop

hys

islo

ss;

NP

H =

ave

rage

lu

mba

r n

eura

l pr

oces

s h

eigh

t. I

nst

itu

tion

al a

bbre

viat

ion

s: A

MN

H =

Am

eric

an M

use

um

of

Nat

ura

l H

isto

ry;

LA

CM

= L

os A

nge

les

Cou

nty

Mu

seu

m;

MC

Z =

Mu

seu

m o

f C

ompa

rati

ve Z

oolo

gy,

Har

vard

Un

iver

sity

; U

SN

M =

Nat

ion

al M

use

um

of

Nat

ura

l H

isto

ry.

‘+’ =

ter

min

al c

auda

l ve

rteb

rae

abse

nt.

Mis

sin

g da

ta w

ere

eith

er i

nac

cess

ible

or

not

col

lect

ed

Spe

cies

Spe

cim

enS

erie

s co

un

tT

CM

axC

Cit

atio

n f

or m

axim

um

cou

nt

ME

Syn

2RT

CL

LC

LM

LN

PH

Del

phin

apte

rida

eD

elph

inap

teru

s le

uca

sU

SN

M 5

7102

1C

v7T

11L

7Cd2

5+50

+49

–54

Bro

die

(198

9)n

on

on

o1.

101.

24n

o2.

67M

onod

on m

onoc

eros

AM

NH

141

56C

v7T

11L

9Cd2

3+50

+50

–55

Hay

& M

ansfi

eld

(198

9)n

on

on

o0.

981.

17n

o1.

57O

rcin

inae

Pse

ud

orca

cra

ssid

ens

AM

NH

996

81C

v7T

10L

9Cd2

1+47

+47

–52

Ode

ll &

McC

lun

e (1

999)

no

no

no

1.05

1.36

no

1.75

Pse

ud

orca

cra

ssid

ens

MC

Z 3

0330

Cv7

T9L

11C

d18+

45+

no

no

no

0.98

1.26

no

2.10

Orc

inu

s or

caU

SN

M 2

3004

Cv7

T11

L10

Cd2

6+54

+50

–54

Dah

lhei

m &

Hey

nin

g (1

999)

yes

no

no

0.75

0.94

no

2.17

Orc

inu

s or

caA

MN

H 2

1527

0C

v7T

12L

9Cd1

1+37

+ye

sn

on

o0.

730.

91n

oO

rcin

us

orca

AM

NH

342

76C

v7T

11L

11C

d24

53ye

sn

on

o0.

97n

o2.

00G

lobi

ceph

ala

mac

rorh

ynch

aU

SN

M 2

2571

Cv7

T8L

13C

d26

5458

–61

Ber

nar

d &

Rei

lly

(199

9)ye

sye

sn

o0.

691.

05n

o2.

46O

rcae

lla

brev

iros

tris

MC

Z 2

1929

C7T

12L

15C

d17+

51+

58–6

0A

rnol

d &

Hei

nso

hn

(19

96)

yes

yes

yes

0.91

0.84

no

1.92

Fer

esa

atte

nu

ata

MC

Z 5

1458

Cv7

T11

L17

Cd3

469

67–7

0R

oss

& L

eath

erw

ood

(199

4)ye

sye

sye

s0.

960.

86ye

s3.

11C

eph

alor

hyn

chin

aeC

eph

alor

hyn

chu

s co

mm

erso

nii

US

NM

550

156

Cv7

T12

L13

Cd3

162

61–6

6G

ooda

ll (

1994

)ye

sye

sye

s0.

830.

74n

o2.

68S

ten

inae

Ste

no

bred

anen

sis

US

NM

550

221

Cv7

T12

L15

Cd2

862

65–6

7M

iyaz

aki

& P

erri

n (

1994

)ye

sye

sye

s0.

930.

86n

o3.

71S

otal

ia fl

uvi

atil

isM

CZ

709

7C

v7T

12L

12C

d23

5653

–55

da S

ilva

& B

est

(199

4)ye

sye

sye

s0.

980.

95n

o2.

82D

elph

inae

Gra

mpu

s gr

iseu

sU

SN

M 5

0432

8C

v7T

12L

16C

d31

6668

–69

Kru

se, C

aldw

ell

& C

aldw

ell

(199

9)ye

sye

sye

s0.

850.

78ye

s2.

67T

urs

iops

tru

nca

tus

MC

Z 7

899

Cv7

T12

L16

Cd2

7+62

+59

–67

Wel

ls &

Sco

tt (

1999

)ye

sye

sye

s0.

930.

82ye

s3.

18T

urs

iops

tru

nca

tus

AM

NH

120

920

Cv7

T12

L19

Cd2

9+67

+ye

sye

sye

s1.

000.

83ye

sT

urs

iops

tru

nca

tus

US

NM

550

422

Cv7

T10

L16

Cd2

4+57

+ye

sye

sye

s1.

000.

95ye

s2.

91P

epon

ocep

hal

a el

ectr

aU

SN

M 5

0494

8C

v7T

12L

19C

d40

7881

–82

Per

rym

an e

t al

. (19

94)

yes

yes

yes

0.92

0.75

yes

3.15

Del

phin

us

del

phis

AM

NH

130

119

Cv7

T13

L23

Cd3

073

73–7

4E

van

s (1

994)

yes

yes

yes

0.82

0.62

yes

Del

phin

us

del

phis

US

NM

550

868

Cv7

T13

L22

Cd3

1+73

+ye

sye

sye

s0.

800.

61ye

s2.

72S

ten

ella

fro

nta

lis

US

NM

571

139

Cv7

T12

L20

Cd3

372

67–7

2P

erri

n, C

aldw

ell

& C

aldw

ell

(199

4)ye

sye

sye

s0.

810.

68ye

sS

ten

ella

lon

giro

stri

sU

SN

M 5

0001

7C

v7T

14L

18C

d34

7369

–77

Per

rin

& G

ilpa

tric

k (1

994)

yes

yes

yes

0.99

0.76

yes

2.82

Ste

nel

la c

oeru

leoa

lba

US

NM

504

350

Cv7

T14

L20

Cd3

879

71–8

0P

erri

n, W

ilso

n &

Arc

her

(19

94)

yes

yes

yes

0.75

0.59

yes

2.96

Lis

sod

elph

is b

orea

lis

LA

CM

434

72C

v7T

15L

33C

d34

8988

–92

Jeff

erso

n e

t al

. (19

94)

no

no

yes

0.82

0.44

no

1.49

Lis

sod

elph

is b

orea

lis

US

NM

484

929

Cv7

T15

L33

Cd3

1+86

+n

on

oye

s1.

080.

59n

o1.

86L

agen

odel

phis

hos

eiM

CZ

543

79C

v7T

14L

24C

d33+

78+

78–8

1P

erri

n, L

eath

erw

ood

& C

olle

t (1

994)

yes

yes

yes

0.96

0.72

yes

2.86

Lag

enor

hyn

chu

s ac

utu

sM

CZ

623

82C

v7T

14L

23C

d35+

79+

77–8

2R

eeve

s et

al.

(199

9)ye

sye

sye

s0.

700.

52ye

s2.

50L

agen

orh

ynch

us

acu

tus

US

NM

504

754

Cv7

T13

L22

Cd3

777

yes

yes

yes

0.77

0.59

yes

2.66

Lag

enor

hyn

chu

s ac

utu

sM

CZ

609

39C

v7T

14L

25C

d36

82ye

sye

sye

s0.

750.

55ye

s2.

39L

agen

orh

ynch

us

acu

tus

MC

Z 6

2380

Cv7

T14

L22

Cd4

083

yes

yes

yes

0.75

0.58

yes

2.51

Lag

enor

hyn

chu

s al

biro

stri

sM

CZ

532

2C

v7T

14L

24C

d46

9188

–94

Ree

ves

et a

l. (1

999)

yes

yes

yes

0.61

0.45

yes

2.85

Lag

enor

hyn

chu

s al

biro

stri

sU

SN

M 5

5020

8C

v7T

13L

24C

d47

91ye

sye

sye

s0.

660.

49ye

s2.

72L

agen

orh

ynch

us

obli

quid

ens

US

NM

504

851

Cv7

T12

L24

Cd3

275

72–7

6B

row

nel

l, W

alke

r &

For

ney

(19

99)

yes

yes

yes

0.72

0.52

yes



species, Cv7,T11–13,L10–12,Cd20–24 = 50–54 (Dahl-heim & Heyning, 1999), is among the lowest for anydolphin. AMNH 34276 (Fig. 4A) is a large adult ani-mal, with a total CL of 5349 mm. All seven cervicalvertebrae are fused (Fig. 5A). Thoracic, lumbar(Fig. 5B) and anterior caudal (Fig. 5C) vertebrae arerelatively uniform in shape, with elongate centra andnearly circular cross-section CH ª CW (Fig. 4A). Thelongest vertebrae occur in the mid column, with CL/CH ª 0.92. CL decreases in the prefluke vertebrae(Fig. 5D), which are laterally compressed (CH > CW).Neural processes average 2.0 times centrum height,and all neural spines incline posteriorly – there is noreversal of spine inclination. Metapophyses arepresent on all lumbar vertebrae, and are elevated

(NAL > NSL) in the posterior lumbos and anterior tail.The last nine caudal vertebrae are dorsoventrallycompressed, lack processes and have very short cen-trum length; in life they supported the fluke (Fig. 5E).

Lagenorhynchus acutus, the Atlantic white-sideddolphin, is a mid size pelagic delphinid. Reported ver-tebral counts are Cv7,T14–15,L18–22,Cd38–41 = 77–82 (Reeves et al., 1999), among the highest in theDelphinidae. MCZ 60939 (Fig. 4B) has 82 vertebraeand a total postcranial length of 1705 mm. The cervi-cal vertebrae (Fig. 5F) are highly discoidal and thefirst four cervicals are fused. Thoracic vertebrae haveposteriorly inclined spines and nearly circular cross-sections (CH ª CW, Fig. 4B). Anterior lumbars haveerect neural arches and spines; metapophyses areabsent or effectively absent on L9–19 (Fig. 5G). Poste-rior lumbar and anterior caudal vertebrae have ante-riorly inclined neural arches and spines; they alsobear elevated metapophyses (Fig. 5H). Reversal ofspine orientation (syncliny) occurs in the mid cau-dals, at vertebra 60 (Fig. 5I). The shortest vertebraeoccur in the midcolumn, immediately before the syn-clinal point, with CL/CH ª 0.40. Tail stock vertebraeare laterally compressed and longer than torso orfluke vertebrae (Fig. 5J). The last 12 caudal vertebraeare dorsoventrally compressed and supported thefluke.

TRAIT ANALYSIS

The traits used for functional and evolutionary anal-ysis are discussed below. In each case, trait expressionis described in delphinids and compared with that indelphinapterids. Delphinapterids are broadly recog-nized (Heyning, 1989; Arnold & Heinsohn, 1996;Waddell et al., 2000) as lying outside the Family Del-phinidae but inside the Superfamily Delphinoidea,making them an appropriate outgroup.

Vertebral count. Total count in Delphinapterus leucasis 49–54 (Cv7,T11–12,L9,Cd22) (Brodie, 1989). Likeliving delphinapterids and almost all fossil odonto-cetes, early delphinids must have had very low totalcounts (ª 50). Within the family, cervical count is fixed(at seven), and thoracic count has a very small range(8–14). Lumbar count equals or exceeds thoracic countin all genera except Pseudorca and Orcinus. Almost allincrements to total count are to the lumbar and caudalseries (Fig. 6A). At low and intermediate total counts(ª 50–70) lumbar and caudal counts increase inapproximately a 1 : 1 ratio. At high total counts (>70),lumbar count appears to be ‘capped’ (at ª24) and alladditional increase is in the caudal series. The soleexception to this pattern is Lissodelphis, in which thevery high lumbar count may reach or even exceed thecaudal count.

Figure 4. Vertebral dimensions in Orcinus orca (AMNH34276) and Lagenorhynchus acutus (MCZ 60939), withclassic divisions of the column. Cervical and anterior tho-racic vertebrae of the Orcinus column are mounted, makingCH and CW measurements impossible. Cv = cervical.

0

50

100

150

200

250

0 10 20 30 40 50 60

Vertebra number

Dim

ensi

on, m

m

Centrum lengthCentrum widthCentrum height

A cervical thoracic lumbar caudal

0

10

20

30

40

50

60

70

0 10 20 30 40 50 60 70 80 90

Vertebra number

Dim

ensi

on, m

m

Centrum lengthCentrum widthCentrum height

Cv thoracic lumbar caudal B



A

B

C

F

G

H

D

E J

I



Relative centrum length. Lumbar vertebral centra ofdelphinapterids are elongate (spool-shaped), with CL/CH ≥ 1.0. All delphinids except Pseudorca have lum-bar vertebrae with average CL/CH < 1.0, although the

extent of CL reduction varies. Shape and count areinversely associated, with more extreme reduction inlumbar CL in genera with higher counts (Fig. 6B,C).Peponocephala, Delphinus, Stenella, Lagenodelphis

Figure 5. Vertebral structure in Orcinus orca AMNH 34276 (A–E, scale bars = 10 cm) and Lagenorhynchus acutus MCZ60939 (F–J, scale bars = 2 cm). A, left lateral view of fused neck (Cv1–7) vertebrae and first chest (T1) vertebra (truncated);B, left lateral view of mid lumbar (L5) vertebra; C, left lateral view of anterior caudal vertebrae (Cd1–4); D, left lateral viewof immediate prefluke caudal vertebrae (Cd10–13); E, dorsal view of fluke vertebrae (Cd14–24). F, left lateral view of cer-vical vertebrae; G, left lateral view of anterior torso vertebra (L9); H, left lateral view of mid torso vertebrae (L24–25, Cd1–9) showing anterior neural arch inclination; I, left lateral view of mid torso/posterior torso transition (synclinal point)(Cd12–16); J, dorsal view of elongate tail stock vertebrae (Cd19–24) and first three fluke vertebrae (Cd25–27).

Figure 6. Vertebral count and shape in delphinids. A, relationship between total counts and series counts. B, relationshipbetween total count and lumbar CL/CH. C, CL/CH traces for four species, each with syncliny and a secondary rise in CL(arrows). Note the progressive reduction in CL/CH with increasing total count.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 10 20 30 40 50 60 70 80 90 100

Vertebra number

CL

/ C

H

Sotalia fluviatilisTursiops truncatusDelphinus delphisLagenorhynchus albirostris

C

0

10

20

30

40

50

60

70

80

40 60 80 100 120 140Total count

Seri

es c

ount

Alumbar + caudal seriesy = 0.901x–12.528

R2 = 0.99

caudal series y = 0.431x + 1.872 R2 = 0.81

R2 = 0.86

R2 = 0.66

y = -0.018x + 2.0283

R2 = 0.83

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

40 60 80 100Total count

Ave

rage

lum

bar

CL

/ C

H

B

lumbar seriesy = 0.472x–14.547

caudal series, y = 7

thoracic seriesy = 0.108x +5.083

�



and Lagenorhynchus all have discoidal lumbar verte-brae with average lumbar CL/CH £ 0.75. Species withhigher counts also show extension of discoidal shapeanteriorly into the thorax and posteriorly into theanterior tail. In species with very high counts (Lageno-rhynchus spp.), as well as a few other isolated species(Orcinus, Globicephala), the average CL/CH of all tho-rax vertebrae is £ 0.75.

Orientation of neural arches and spines. Orientationof neural arches and spines is uniformly posterior indelphinapterids. At least some vertebrae bear anteri-orly inclined neural arches in all delphinids, as notedby Slijper (1936). Extreme anterior inclination of thearch (here termed a ‘reclining arch’) displaces the tipof the neural spine anterior to the plane of the cen-trum face in caudal vertebrae in Lagenorhynchus(Fig. 5H) and other delphinids with extremely highcounts (Peponocephala, Delphinus, Stenella, Lageno-delphis). In most delphinids, a region of anteriorlyinclined neural spines introduces a point of divergence(syncliny) in the anterior caudal series (Fig. 5I). Thesynclinal point is located more posteriorly (% totalpostcranial length) in species with higher counts(Fig. 7). Neither Pseudorca nor Orcinus displayssyncliny.

Elongation of prefluke (tailstock) vertebrae. In delphi-napterids, CL peaks in the middle of the column,decreasing both anteriorly and posteriorly from apoint near the lumbar/caudal transition. Orcinus(Fig. 4A) exhibits a similar pattern. In most delphin-ids, the mid column is instead the point of shortest CL(Lagenorhynchus, Fig. 4B). CL increases posteriorly ina ‘secondary rise’ to a maximum anterior to the fluke(Fig. 6C). This rise is continuous and gentle in del-phinids with low counts, but is abrupt and postponeduntil after the synclinal point in species with highcounts.

Metapophysis presence and location. Metapophysesare uniformly present on lumbar and prefluke cau-dals in delphinapterids, but are regionally lacking(Fig. 5G) from mid and posterior lumbars in dol-phins with counts ≥ 65. Located at the transitionbetween neural arch and neural spine, metapophy-ses are ‘elevated’ when the running distance of thearch is greater than that of the spine (NAH > NSH).Elevated metapophyses occur in posterior lumbarand anterior caudal vertebrae in all delphinidsexcept Pseudorca and the highly unusual genusLissodelphis.

DISCUSSION

Variations in vertebral osteology are used below toidentify structural units of the delphinid column, to

predict functions of these structural units, to predictthe historical sequence of trait origination and to con-tribute to the resolution of phylogenetic relationshipswithin the Family Delphinidae.

STRUCTURAL UNITS IN THE DELPHINID COLUMN

Axial sites of marked morphological discontinuity aretraditionally used to define vertebral series (e.g.Gadow, 1933). The mismatch between terrestrialmammals and cetaceans in the locations of these sitescomplicates interpretation of delphinid vertebralanatomy. Although cervical and thoracic series may bedefined similarly in whales and in terrestrial mam-mals, cetacean lumbar and caudal vertebrae havebeen dramatically reconfigured during evolution. As aresult, cetacean morphological units transgress classicseries borders. This is most marked for the large lum-bar series that must include, in addition to ‘true’ lum-bars, vertebrae homologous with the sacral vertebraeof terrestrial mammals (Fig. 8A,B). Slijper (1936) ten-tatively identified these ‘sacral lumbars’ on the basisof pudendal nerve roots. He also argued for the exist-ence of a variable number of apparent lumbars thatwere in fact post-sacral (‘caudal lumbars’) but unrec-ognizable as caudals because they lack haemal archscars.

Figure 7. Neural spine inclination (NSI) at different loca-tions along the column in four delphinid species: Orcinusorca, AMNH 34276, total count = 53; Sotalia fluviatilis(MCZ 7097, vertebral count = 54); Tursiops truncatus(MCZ 7899, vertebral count = 62+); Lagenorhynchus acu-tus (MCZ 60939; vertebral count = 82). Note that the posi-tion of the synclinal point (arrows) is more posterior inspecies with higher counts, and that Orcinus does notexhibit syncliny.

30

40

50

60

70

80

90

100

110

120

0 10 20 30 40 50 60 70 80 90 100

Per cent postcranial length

NSI

Lagenorhynchus acutus Sotalia fluviatalis Tursiops truncatus Orcinus orca

ANTERIOR

POSTERIOR



Alternatively, the cetacean column may be subdi-vided on the basis of its own structural discontinuities,which may be predicted to correspond to units of dis-tinctive function in whales. Unambiguous subdivisionof the column is possible on the basis of centrumdimensions and the presence of ribs into neck (N),chest (C), torso (T), tail stock (TS) and fluke (F) series(Fig. 8). The torso may be further subdivided in mostdelphinid genera (those with neural spine syncliny)into anterior, mid and posterior units on the basis ofneural spine orientation. We propose recognition ofthe following vertebral series for delphinids, andanticipate that they may be broadly applied to othercetaceans as well:

Neck. The neck is foreshortened and supported byseven vertebrae of exceptionally short centrum length.

Anterior vertebrae are typically fused, although thenumber included in the fused unit varies among spe-cies, among individuals of the same species and withontogeny in the same individual. Neural spines arevery short and posteriorly inclined.

Chest. Chest vertebrae have posteriorly inclined neu-ral spines and relatively long centra with nearly roundcross-section (CW ª CH). Each vertebra is associatedwith an unfused rib that surrounds and protects thethoracic viscera. Double-headed ribs are succeededposteriorly by single-headed and/or floating ribs.

Torso. The torso comprises numerous highly uniform,discoidal vertebrae of circular cross-section and (inmost delphinids) short centrum length. Neural pro-cesses are a major site of origin for the epaxial musclesof locomotion that insert posteriorly on tail stock and

Figure 8. Two representations of the allocation of vertebrae to series in delphinid cetaceans. A, comparison of column sub-divisions in terrestrial mammals, in cetaceans using terrestrial nomenclature and in cetaceans using cetacean nomencla-ture. Subdivision of the torso is applicable only to delphinids with syncliny. S = synclinal point. B, regional morphology ofLagenorhynchus acutus (MCZ 61008) by classic and cetacean column series in dorsal (above) and lateral (below) views.N = neck, TS = tail stock, F = fluke.

A Classical series in terrestrial mammals

Classical series in cetaceans

Cetacean series

cervical cervical neck thoracic thoracic chest

anterior lumbar

sacral

lumbar true lumbar sacral lumbar postsacral lumbar mid

torso S

posterior

tail stock

caudal

caudal

fluke

B



fluke vertebrae. In taxa with syncliny, torso vertebraemay be subdivided into an anterior region (with effec-tively erect neural arches and spines), a middle region(with anteriorly inclined arches and spines) and a pos-terior region (with ‘bent’ processes composed of ante-riorly inclined arches and posteriorly inclined spines).The angular transition between spines of middle andposterior torso vertebrae marks the synclinal point(Slijper, 1936). The lumbar/caudal transition occurswithin the middle torso with minimal morphologicaldiscontinuity.

Tail stock. Tail stock vertebrae are laterally com-pressed (CH > CW), and support the laterally com-pressed peduncle anterior to the fluke. Neural archesare erect or incline anteriorly, and the very short neu-ral spines incline posteriorly. In most delphinids, tailstock vertebrae have longer centrum lengths (‘second-ary rise’) than do adjacent torso or fluke vertebrae.The last tail stock vertebra has highly convex faces.

Fluke. The fluke is characterized by vertebrae that aredorsoventrally compressed (CW > CH), have short CLand lack processes.

FUNCTIONAL INTERPRETATION OF VERTEBRAL SERIES

Neck (cervical) and chest (thoracic) vertebral series indelphinids are defined as in terrestrial mammals andparallel them in their structural and supportive func-tions. The remainder of the postcranium generates themovements that propel the animal during swimming.During swimming, the post-thoracic column effectsboth vertical displacement and oscillation of the fluke,which is the propulsive surface (Slijper, 1961; Fish &Hui, 1991; Curren, Bose & Lien, 1994). The site offluke oscillation is uniform and discrete; it is easilyidentified in osteological specimens by the convex ver-tebra at the end of the tail stock (Slijper, 1936; Watson& Fordyce, 1993). By contrast, vertebral morphologyindicates that the axial extent of the displacing unitvaries among delphinid taxa. The laterally com-pressed tail stock is located immediately anterior tothe fluke, and is certainly part of the displacing unit inall dolphins. The torso, however, may be either unimo-dally (without morphological discontinuity) or bimod-ally (with morphological discontinuity) constructed.We propose that unimodal and bimodal torsos achievefluke displacement differently. Vertebrae of unimodaltorsos all contribute to fluke displacement, but dis-placement in bimodal torsos is more localized, occur-ring preferentially at the synclinal point.

Vertebral morphology in unimodal torsos (e.g. Orci-nus, Figs 4, 5) is highly uniform. Count is low, centraare spool-shaped, and anterior and posterior centrumfaces are gently rounded. Neural spines are relativelyshort and posteriorly inclined. Short spines limit the

mechanical advantage of the epaxial musculature, butenhance angular rotation. The neural spines are sep-arated from each other by the long centra that bearthem, minimizing interference between adjacent ver-tebrae during rotation. Torso vertebrae are longerthan chest or tail stock vertebrae; there is no ‘second-ary rise’ anterior to the fluke. Similarity of structurethroughout the torso signals similarity of function. Weinfer that all torso vertebrae are involved to at leastsome extent in the vertical displacement of the flukecaused by the contraction of the epaxial and hypaxialmusculature.

Bimodal torsos (e.g. Lagenorhynchus, Figs 4, 5) havehigh vertebral counts and discoidal centra with flatanterior and posterior faces. Torso vertebrae areshorter than either chest or tail stock vertebrae. Ante-riorly, neural spines are very tall and almost as longaxially as the centra that bear them, threateningmechanical interference with minimal centrum rota-tion. The tall neural spines, many with elevatedmetapophyses, increase the distance between forceapplication and the axis of rotation, enhancingmechanical advantage but limiting angular rotation.Metapophyses are regionally lacking in the anteriortorso, inferring absence of the shorter (mid-spine) mus-cle fascicles that would produce greater rotation. In themid-torso, metapophyses reappear, and neural spinesshorten progressively and also incline anteriorly. Atthe mid-torso/posterior torso transition, neural processorientation reverses, creating a synclinal point atwhich angular divergence and distance betweenadjacent neural spines is maximized. Posterior to thesynclinal point, neural spine height decreases dramat-ically, signalling the possibility of greater angular rota-tion. Centrum length increases in a ‘secondary rise’that encompasses the posterior torso and tail stock. Weinfer from the morphology above that during swim-ming of bimodal animals, the anterior torso is highlystabilized. Flexibility in the posterior torso is localizedat the synclinal point and displacement of the fluke isdue primarily to dorsoventral movement of vertebraeposterior to that point. Elongation of these vertebraemaximizes the vertical displacement possible.

EVOLUTIONARY IMPLICATIONS OF TRAIT DISTRIBUTION

The unimodal and bimodal torsos described above rep-resent points near the extremes of a morphologicalcontinuum. Their differences are bridged by living andfossil taxa that display some, but not all, of the traitsthat limit intervertebral mobility in the anterior torsoand enhance it in the posterior torso. This hierarchicaldistribution of traits strongly suggests the stepwiseacquisition of separate aspects of this anatomy in thehistory of the family. Used in conjunction with anestablished phylogeny, evolutionary transitions may



be identified by derived traits limited to progressivelysmaller subgroups (Fig. 9).

We use the following derived traits in reconstructionof the evolutionary history of delphinid vertebral anat-omy. All delphinids display anteriorly inclined neuralarches (trait 1, Fig. 9), a character that separatesthem from other delphinoids. With the exception ofPseudorca (and Lissodelphis, see below), all delphin-ids have elevated torso metapophyses (trait 2), identi-fying this as a very early historical innovation. Thelack of syncliny (trait 5) isolates Pseudorca and Orci-nus from the remaining delphinids and implies theearly origin of this key step in delphinid locomotorevolution. Elongation of posterior torso and tail stockvertebrae (trait 6) occurs in most but not all animalswith syncliny, signalling its subsequent origin. Asmaller generic subset displays regional loss of torsometapophyses (trait 8), identifying this as a still morerecent innovation.

In addition to these discrete traits, others are gra-dational and are therefore more difficult to use in his-torical reconstruction. In some cases, clinal gapsprovide ‘steps’ in these traits, allowing them to be usedlike discrete traits. The most notable gradationalchange is in vertebral count, with small counts at theprimitive end of the continuum. Steps in this cline areabrupt, and may reflect evolutionary changes in HOXgene expression patterns. Steps occur where lumbarcount exceeds thoracic count (trait 4), at total count of60 (trait 7) and at the capping of lumbar count syn-chronous with total count of 70 (trait 9). Other grada-tional trends include the reductions in torso CL/CHbelow 1.0 (trait 3) and below 0.75 (trait 9), the increasein neural arch inclination that carries neural spinesanterior to the centrum face (trait 11), and the ante-rior extension of discoidal vertebrae (RCL £ 0.75) intothe chest (trait 12). Increase in lumbar NPH and theposteriad movement (% of postcranial length) in theposition of the synclinal point (Fig. 7) are generaltrends but not discrete enough to allow unambiguoususe.

Our interpretation of vertebral morphology sug-gests that the ancestors of living dolphins had rela-tively low vertebral counts, spool-like torso vertebrae,and relatively short neural processes without syncliny,traits possessed today by delphinapterids. We inferthat the torsos of these ancestors were less regional-ized and more uniformly flexible than those of mostliving delphinids. The cumulative functional effect ofthe evolutionary innovations noted above is theenhanced regionalization of the column, in particularthe stabilization of the anterior column, the pro-gressive localization of flexibility to anterior (syncli-nal) and posterior (fluke base) sites, and theelongation of the tail stock vertebrae responsible forfluke displacement.

Analysis of trait distribution also reveals the prob-able existence of homoplasy. Homoplasy is stronglyimplicated in the occurrence of low chest CL in taxa(Orcinus, Globicephala, Lagenorhynchus) widely sep-arated by count and syncliny. Another likely occur-rence of homoplasy is suggested by the reducedlumbar CL in Cephalorhynchus, otherwise restrictedto taxa with regional metapophysis loss and countsabove 70. The unusual combination of vertebraltraits in the northern right whale dolphin Lissodel-phis can be added to its anomalous lack of a dorsalfin, elongate body form and leaping style of locomo-tion (Jefferson et al., 1994) in setting it apart fromother dolphins. It combines features typical of primi-tive species (lack of syncliny, low torso metapophyses,short neural spines) with those of highly derived spe-cies (very high vertebral count, highly discoidal lum-bar vertebrae, ‘secondary rise’ in posterior torso CL).These distribution patterns imply multiple origins ofone or more traits, with high count and reduced CLthe most likely candidates. The lability of these traitsis suggested by high variation in segment count invertebrates generally (Raff, 1996; Richardson et al.,1998; Polly, Head & Cohn, 2001) and by the existenceof parallel sequences of count enhancement and CLreduction in phocoenid delphinoids (Buchholtz,2001).

Despite these considerations, we believe that verte-bral anatomy provides a previously little-used sourceof traits that can contribute to the resolution of phy-logenetic relationships within Delphinidae. Limita-tion of our data to a single morphological systemargues against its use as a basis for the definition ofnew taxonomic categories. Rather, we propose the fol-lowing modifications to the consensus phylogeny(Figs 1, 9):1. Removal of genera possessing syncliny from the

Orcininae into a separate (‘globocephaline’) sub-family, emphasizing the morphological isolation ofPseudorca and Orcinus.

2. Addition of Sotalia, which exhibits syncliny, alumbar count ª thoracic count and metapophysesthroughout the torso, to this globocephaline sub-family. Consistent with the conclusions of Arnold &Heinsohn (1996), this same combination of traitssupports the inclusion of Orcaella in this subfamilyas well.

3. Removal of Feresa, which lacks metapophysesregionally and has a total count of 67–70, from theOrcininae and placement of it in the Delphininae.These same traits also support Mead’s (1975) place-ment of Peponocephala in the Delphininae.

4. Subdivision of the Delphininae into two subgroupson the basis of total count, lumbar count, neuralarch inclination and chest CL. These subgroupsinclude: (a) Feresa, Tursiops, Grampus; (b) Pepono-



1 2 3 4 5 6 7 8 9 10 11 12

Monodon

Pseudorca

Orcinus

Orcaella

Feresa

Steno

Sotalia

Grampus

Tursiops

Delphinus

Stenella

Trait Primitive state Derived state

1 neural arches inclined posteriorly regional neural arch anterior inclination

2 all metapophyses low regional elevated metapophyses

3 lumbar vertebrae with CL / CH > 1.0 lumbar vertebrae with CL / CH ≤ 1.0

4 lumbar count < thoracic count lumbar count ≥ thoracic count

5 neural spines inclined posteriorly neural spine syncliny

6 torso CL > tail stock CL secondary rise in tail stock CL

7 total count < 60 total count > 60

8 torso metapophyses uniformly present regional absence of torso metapophyses

9 total count < 70 lumbar count capped, total count > 70,

10 lumbar vertebrae with CL / CH > 0.75 lumbar vertebrae with CL / CH ≤ 0.75

11 reclining neural arches absent in mid torso reclining neural arches in mid torso

12 chest vertebrae with Cl / CH > 0.75 chest vertebrae with CL / CH ≤ 0.75

Delphinapteridae

Delphinapterus

Orcininae

Globicephala

Cephalorhynchinae

Cephalorhynchus

Steninae

Delphininae

Peponocephala

Lissodelphis

Lagenodelphis

Lagenorhynchus



cephala, Lagenodelphis, Delphinus, Stenella,Lagenorhynchus.

5. Isolation of Lissodelphis, in agreement with Mead(1975) and Perrin (1989). The vertebral morphologyof this genus is enigmatic. Its lack of syncliny andretention of metapophyses argue against its place-ment in Delphininae, but its high count, discoidallumbar vertebrae and ‘secondary rise’ prevent itsplacement in other, less derived, subfamilies.

INFERENCES FROM VERTEBRAL ANATOMY OF FOSSIL DELPHINIDS

The vertebral traits discussed here offer the possibil-ity of placing partial columns or even isolated fossilvertebrae into functional and evolutionary contexts.Among the most useful traits for such evaluation isthe presence of anteriorly inclined neural spines or‘bent’ neural processes that indicate the presence ofsyncliny. Loss of metapophyses further constrainsplacement to a small and derived group of delphinids.Ratio of CL to CH of isolated torso vertebrae may beused as an indicator of total count (Fig. 6B) and allowsat least generalized inference of phylogenetic place-ment. Examples of evaluation of fossil taxa are pre-sented below.

‘Odontocete indet.’, CMM-V-1694, 21 vertebrae fromBed 13 (Mid Miocene) of the Calvert Formation, Cal-vert Co., MD, USA. Vertebrae of the torso, tail stockand even of the fluke have CL/CH ≥ 1.0 (Fig. 10A). Noknown delphinid or delphinoid has fluke vertebraewith length > width, placing this specimen outside theDelphinoidea.

Hadrodelphis calvertense, CMM-V-11, a nearlycomplete column of a kentriodontid from the Cal-vert Formation of Charles County, MD, USA,described by Dawson (1996). Neural arches lackanterior inclination. Neural spines are short, broadand lack syncliny; metapophyses are universallypresent and low (Fig. 10B). This vertebral profile isalmost identical to that of living monodontids, whichcan be grouped with kentriodontids as ‘non-del-phinid delphinoids.’

Albireo whistleri, UCR 14589, seven neck, 13 chestand 25 torso vertebrae from a specimen described byBarnes (1984) from the Late Miocene Almejas Forma-tion of Baja California, Mexico. Barnes erected a newfamily (Albireonidae) for this specimen on the basis ofits unique suite of cranial characters. Mid torso verte-brae of this specimen have anteriorly inclined neural

spines, implying syncliny (Fig. 10C); metapophysesare universally present and lack elevation. This suiteof characters suggests placement among basal ‘syncli-nal’ delphinids.

‘Tursiops sp.’ USNM 15727, three chest, three ante-rior torso and two mid torso vertebrae (previouslynine in total: see Blake, 1939, 1953) from the Pleis-tocene Talbot Formation at Wailes Bluff, St Marys Co.,MD, USA (Fig. 10D). The three most anterior torsovertebrae lack metapophyses, placing this specimenamong the delphinines. Preserved chest vertebraehave CL/CH ª 1.0, probably placing the specimen out-side the most derived taxa. Average CL/CH of pre-served torso vertebrae = 0.72, predicting a totalvertebral count of 73 using the regression equation ofFigure 6B. The dimensions of the fossil specimen sug-gest that it falls just outside of and on the derived sideof the range of total count for living Tursiops (Wells &Scott, 1999).

CONCLUSIONS

The delphinid vertebral column has been radicallyreorganized for axial locomotion in the aquatic envi-ronment. Interpretation of its structure and func-tion is facilitated by recognition of structuraldiscontinuities that mark the boundaries of theneck, chest, torso, tail stock and fluke. Delphinid col-umns show significant intrafamilial variation, withdifferences in vertebral count, shape and neuralspine orientation being most prominent. Low counts,spool-shaped vertebrae and posterior inclination ofneural spines are traits shared with living non-del-phinid odontocetes and also with almost all fossilodontocete taxa; we identify them as primitive. Gen-era with these traits have unimodal torsos; all post-thoracic vertebrae are involved to lesser or greaterextent in fluke displacement. Syncliny of neuralspine orientation in the mid tail was a key del-phinid innovation, and signals the evolution of thebimodal torso in which prefluke flexibility is local-ized. Localization of caudal flexibility at the syncli-nal point is accompanied by traits that limitmobility in the anterior torso (increase in neuralprocess height; regional loss of metapophyses) andenhance displacement of the tail stock (tail stock CLelongation). Bimodal torsos are typical of generawith high counts. The hierarchical distribution ofvertebral traits suggests that evolution of the bimo-dal torso occurred step-wise, with syncliny a histori-

Figure 9. Distribution of vertebral traits in delphinid cetaceans mapped on to the consensus phylogeny. Note both the gen-eral agreement of vertebral traits with the existing phylogeny and the discordant distribution of traits in the genus Lis-sodelphis. Shading indicates primitive (light) or derived (dark) trait states.

�



cally earlier innovation than metapophysis loss ortail stock elongation. At least some vertebral traitsdisplay homoplasy; increases in count are known inphocoenids and are also implicated in Lissodelphis.Vertebral trait distribution may also have phyloge-netic content. Our data suggest recognition of syn-cliny by subfamilial separation of Pseudorca andOrcinus from remaining ‘globocephaline’ delphinids,placement of both Feresa and Peponocephala in theSubfamily Delphininae, and subdivision of the Del-phininae. We believe that vertebral characters havegreat potential in the elucidation of vertebrate func-tion and history, and encourage their inclusion infuture investigations to achieve a more completeview of vertebrate evolution.

ACKNOWLEDGEMENTS

We thank Jim Mead, Charley Potter, Dave Bohaska,Bob Purdy, Maria Rutzmoser, Judy Chupasko, LarryBarnes, Stephen Godfrey, Robert Randall and JohnAlexander who allowed access to specimens undertheir care. Jim Mead graciously allowed us to use hislarge collection of cetacean necropsy photographs, andwas a generous source of information and discussion.We are grateful to George Dikmak for the constructionof the physical model used to test effects of changes invertebral process placement and inclination, to LouisBuchholtz for mathematical modelling of variations inprocess orientation and centrum curvature, and toKate Webbink for help with computer graphics. X-rays

Figure 10. Isolated vertebrae from fossil cetaceans for which functional and/or evolutionary context may be predicted. A,11 caudal vertebrae of ‘Odontocete indet.’ CMM-V-1694 in dorsal view, with an inferred gap in vertebral sequence. Note thatthe terminal (fluke) vertebrae are elongate. B, two mid torso vertebrae of Hadrodelphis calvertense CMM-V-11 in left lateralview. Metapophyses are present, neural processes are short and vertebrae are spool-shaped. C, six anterior and mid torsovertebrae of Albireo whistleri UCR 14589 in left lateral view. Metapophyses are present and low, but neural spines are verytall and centra are discoidal. D, three chest and five torso vertebrae of USNM 15727 ‘Tursiops sp.’ in left lateral view. Notethe lack of metapophyses on anterior torso vertebrae. All scale bars = 5 cm.

A

C D

B



of dolphin vertebral columns were made at the TuftsUniversity School of Veterinary Medicine, whose staffwe gratefully acknowledge. The thoughtful sugges-tions of an anonymous reviewer are greatly appreci-ated; we believe that they resulted in significantimprovements to the paper. Michelle Gillett, DeniseChing and Cynthia Efremoff previously contributedboth data and ideas to related projects. We also thankWellesley College, which supported our travel withawards from the Jerome Schiff, Katherine Mulhearn,Howard Hughes Medical Institute Biological SciencesEducation Program, Sigma Xi and Faculty Awardsfunds.

REFERENCES

Arnold PW, Heinsohn GE. 1996. Phylogenetic status of theIrrawaddy dolphin Orcaella brevirostris (Owen in Gray): acladistic analysis. Memoirs of the Queensland Museum 39:141–204.

Barnes LG. 1984. Fossil odontocetes (Mammalia: Cetacea)from the Almejas Formation, Isla Cedros, Mexico. Paleobios42: 1–46.

Barnes LG. 1990. The fossil record and evolutionary relation-ships of the genus Tursiops. In: Leatherwood S, Reeves RR,eds. The bottlenose dolphin. San Diego: Academic Press, 3–26.

Bernard HJ, Reilly SB. 1999. Pilot whales – GlobicephalaLesson, 1828. In: Ridgway SH, Harrison R, eds. Handbook ofmarine mammals, Vol. 6. San Diego: Academic Press, 245–280.

Berta A, Sumich JL. 1999. Marine mammals; evolutionarybiology. San Diego: Academic Press.

Blake SF. 1939. A Pleistocene porpoise (Tursiops sp.) fromMaryland. Proceedings of the Biological Society of Washing-ton 52: 99–100.

Blake SF. 1953. The Pleistocene fauna of Wailes Bluff andLangleys Bluff, Maryland. Smithsonian Miscellaneous Col-lections 121: 1–32.

Brodie PF. 1989. The white whale Delphinapterus leucas. In:Ridgway SH, Harrison R, eds. Handbook of marine mam-mals, Vol. 4. London: Academic Press, 119–144.

Brownell RL Jr, Walker WA, Forney KA. 1999. Pacificwhite-sided dolphin – Lagenorhynchus obliquidens Gill,1865. In: Ridgway SH, Harrison R, eds. Handbook ofmarine mammals, Vol. 6. San Diego: Academic Press, 57–84.

Buchholtz EA. 1998. Implications of vertebral morphology forlocomotor evolution in early Cetacea. In: Thewissen JGM,ed. The emergence of whales. New York: Plenum Press, 325–351.

Buchholtz EA. 2001. Vertebral osteology and swimming stylein living and fossil whales (Order: Cetacea). Journal of Zool-ogy, London 253: 175–190.

Crovetto A. 1991. Etude osteometrique et anatomo-func-tionelle de la colonne vertebrale chez les grands cetaces.Investigations on Cetacea 23: 7–189.

Curren K, Bose N, Lien J. 1994. Swimming kinematics of aharbor porpoise (Phocoena phocoena) and an Atlantic white-sided dolphin (Lagenorhynchus acutus). Marine MammalScience 10: 485–492.

Dahlheim ME, Heyning JE. 1999. Killer whale – Orcinusorca (Linnaeus, 1758). In: Ridgway SH, Harrison R, eds.Handbook of marine mammals, Vol. 6. San Diego: AcademicPress, 281–322.

Dawson SD. 1996. A description of the skull and postcrania ofHadrodelphis calvertense Kellogg 1966, and its positionwithin the Kentriodontidae (Cetacea; Delphinoidea). Jour-nal of Vertebrate Paleontology 16: 125–134.

DeSmet WMA. 1977. The regions of the cetacean vertebralcolumn. In: Harrison J, ed. Functional anatomy of marinemammals, no. 3. New York: Academic Press, 58–80.

Evans WE. 1994. Common dolphin, white-bellied porpoise –Delphinus delphis Linnaeus, 1758. In: Ridgway SH, Harri-son R, eds. Handbook of marine mammals, Vol. 5. London:Academic Press, 191–224.

Filler AG. 1986. Axial character seriation in mammals.Unpublished DPhil Thesis, Harvard University.

Fish F, Hui C. 1991. Dolphin swimming – a review. MammalReview 21: 181–195.

Fordyce RE. 1994. Waipatia maerewhenua, new genus andnew species (Waipatiidae, new family), an archaic Late Oli-gocene dolphin (Cetacea: Odontoceti: Platanistoidea) fromNew Zealand. Proceedings of the San Diego Society of Natu-ral History 29: 147–176.

Fordyce RE, Barnes LG. 1994. The evolutionary history ofwhales and dolphins. Annual Review of Earth and PlanetaryScience 22: 419–455.

Gadow HF. 1933. Evolution of the vertebral column. Cam-bridge: Cambridge University Press.

Gál JM. 1993. Mammalian spinal biomechanics. II. Interver-tebral lesion experiments and mechanisms of bending resis-tance. Journal of Experimental Biology 174: 281–297.

Gingerich PD. 1998. Paleobiological perspectives on Mesony-chia, Archaeoceti, and the origin of whales. In: ThewissenJGM, ed. The emergence of whales. New York: Plenum Press,423–449.

Gingerich PD. 2003. Land-to-sea transition in early whales:evolution of Eocene Archeoceti (Cetacea) in relation to skel-etal proportions and locomotions of living semiaquatic mam-mals. Paleobiology 29: 429–454.

Goodall RNP. 1994. Commerson’s dolphin – Cephalorhyn-chus commersonii (Lacépède 1804). In: Ridgway SH, Harri-son R, eds. Handbook of marine mammals, Vol. 5. London:Academic Press, 241–268.

Hasegawa M, Adachi J, Milinkovitch MC. 1997. Novelphylogeny of whales supported by total evidence. Journal ofMolecular Evolution 44 (Suppl. 1): S117–S120.

Hay KA, Mansfield AW. 1989. Narwhal – Monodon mono-ceros Linnaeus, 1758. In: Ridgway SH, Harrison R, eds.Handbook of marine mammals, Vol. 4. London: AcademicPress, 145–177.

Heyning JE. 1989. Comparative facial anatomy of beakedwhales (Ziphiidae) and a systematic revision among thefamilies of extant Odontoceti. Natural History Museum



of Los Angeles County Contributions to Science 405: 1–64.

Hildebrand M, Goslow GE. 2001. Analysis of vertebratestructure, 5th edn. New York: John Wiley and Sons.

Howell AB. 1930. Aquatic mammals. Baltimore: Charles C.Thomas.

Jefferson TA, Newcomer MW, Leatherwood S, Van Waer-ebeek K. 1994. Right whale dolphins, Lissodelphis borealis(Peale, 1848) and Lissodelphis peronii (Lacépède, 1804). In:Ridgway SH, Harrison R, eds. Handbook of marine mam-mals, Vol. 5. San Diego: Academic Press, 335–362.

Kardong KV. 1998. Vertebrates: comparative anatomy, func-tion, evolution. Boston: McGraw-Hill.

Kasuya T. 1973. Systematic consideration of Recent toothedwhales based on the morphology of tympano-periotic bone.Scientific Report of the Whales Research Institute 25: 1–103.

Kruse S, Caldwell DK, Caldwell MC. 1999. Risso’s dolphin– Grampus griseus (G. Cuvier, 1812). In: Ridgway SH, Har-rison R, eds. Handbook of marine mammals, Vol. 6. SanDiego: Academic Press, 183–213.

LeDuc RG. 2002. Delphinids, overview. In: Perrin WF, Wür-sig B, Thewissen JGM, eds. Encyclopedia of marine mam-mals. San Diego: Academic Press, 310–314.

LeDuc RG, Perrin WF, Dizon AE. 1999. Phylogeneticrelationships among the delphinid cetaceans based on fullcytochrome b sequences. Marine Mammal Science 15: 619–648.

Long JH, Pabst DA, Shepherd WR, McLellan WA. 1997.Locomotor design of dolphin vertebral columns: bendingmechanics and morphology of Delphinus delphis. Journal ofExperimental Biology 200: 65–81.

Luo Z. 1998. Homology and transformation of cetacean ecto-tympanic structures. In: Thewissen JGM, ed. The emergenceof whales. New York: Plenum Press, 269–301.

Lütken MD. 1888. Critical studies upon some Odontoceti ofthe genera Tursiops, Orca, and Lagenorhynchus. Annals andMagazine of Natural History, 6th Series 2: 179–186.

McGowan C. 1999. A practical guide to vertebrate mechanics.Cambridge: Cambridge University Press.

Mead JG. 1975. Anatomy of the external nasal passages andfacial complex in the Delphinidae (Mammalia: Cetacea).Smithsonian Contributions to Zoology 207: 1–72.

Milinkovich MC, Orti G, Meyer A. 1993. Revised phylogenyof whales suggested by mitochondrial ribosomal DNAsequences. Nature 361: 346–348.

Milinkovitch MC, Orti G, Meyer A. 1994. Phylogeny of allmajor groups of cetaceans based on DNA sequences fromthree mitochondrial genes. Molecular Biology and Evolution11: 939–948.

Miyazaki N, Perrin WF. 1994. Rough-toothed dolphin –Steno bredanensis (Lesson, 1828). In: Ridgway SH, HarrisonR, eds. Handbook of marine mammals, Vol. 5. London: Aca-demic Press, 1–22.

Motani R. 2002. Scaling effects in caudal fin propulsion andthe speed of ichthyosaurs. Nature 415: 309–312.

de Muizon C. 1988. Les relations phylogénétiques des Del-phinida (Cetacea, Mammalia). Annales de Paléontologie 74:159–227.

Odell DK, McClune KM. 1999. False Killer Whale – Pseud-orca crassidens (Owen, 1846). In: Ridgway SH, Harrison R,eds. Handbook of marine mammals, Vol. 6. San Diego: Aca-demic Press, 213–244.

Pabst DA. 1990. Axial muscles and connective tissues of thebottlenose dolphin. In: Leatherwood S, Reeves RR, eds. Thebottlenose dolphin. San Diego: Academic Press, 51–67.

Pabst DA. 1993. Intramuscular morphology and tendon geom-etry of the epaxial swimming muscles of dolphins. Journal ofZoology, London 230: 159–176.

Pabst DA. 1996. Morphology of the subdermal connective tis-sue sheath of dolphins: a new fibre-wound, thin-walled pres-surized cylinder model for swimming vertebrates. Journal ofZoology, London 238: 35–52.

Pabst DA. 2000. To bend a dolphin: convergence of force trans-mission designs in cetaceans and scombrid fishes. AmericanZoologist 40: 146–155.

Perrin WF. 1989. Dolphins, porpoises, and whales. An actionplan for the conservation of biological diversity: 1988–1992,2nd edn. Gland, Switzerland: IUCN.

Perrin WF, Caldwell DK, Caldwell MC. 1994. Atlantic spot-ted dolphin – Stenella frontalis (G. Cuvier, 1829). In: Ridg-way SH, Harrison R, eds. Handbook of marine mammals,Vol. 5. London: Academic Press, 173–190.

Perrin WF, Gilpatrick JW Jr. 1994. Spinner dolphin –Stenella longirostris (Gray, 1828). In: Ridgway SH, HarrisonR, eds. Handbook of marine mammals, Vol. 5. London: Aca-demic Press, 99–128.

Perrin WF, Leatherwood S, Collet A. 1994. Fraser’s Dol-phin – Lagenodelphis hosei Fraser, 1956. In: Ridgway SH,Harrison R, eds. Handbook of marine mammals, Vol. 5.London: Academic Press, 225–240.

Perrin WF, Wilson CE, Archer FI II. 1994. Striped dolphin– Stenella coeruleoalba (Meyen, 1833). In: Ridgway SH, Har-rison R, eds. Handbook of marine mammals, Vol. 5. London:Academic Press, 129–160.

Perryman WL, Au DWK, Leatherwood S, Jefferson TA.1994. Melon-headed whale – Peponocephala electra Gray,1846. In: Ridgway SH, Harrison R, eds. Handbook of marinemammals, Vol. 5. London: Academic Press, 363–386.

Polly PD, Head JJ, Cohn MJ. 2001. Testing modularityand dissociation: the evolution of regional proportions insnakes. In: Zelditch ML, ed. Beyond heterochrony: the evo-lution of development. New York: John Wiley and Sons,305–335.

Raff RA. 1996. The shape of life: genes, development, and theevolution of animal form. Chicago: University of ChicagoPress.

Reeves RR, Smeenk C, Kinze CC, Brownell L Jr, Lien J.1999. White-beaked dolphin – Lagenorhynchus albirostrisGray, 1846. In: Ridgway SH, Harrison R, eds. Handbook ofmarine mammals, Vol. 6. San Diego: Academic Press, 1–30.

Richardson MK, Allen SP, Wright GM, Raynaud A, Han-ken J. 1998. Somite number and vertebrate evolution.Development 125: 151–160.

Rommel S. 1990. Osteology of the bottlenose dolphin. In:Leatherwood S, Reeves RR, eds. The bottlenose dolphin. SanDiego: Academic Press, 29–49.



Ross GJB, Leatherwood S. 1994. Pygmy killer whale – Fer-esa attenuata Gray, 1974. In: Ridgway SH, Harrison R, eds.Handbook of marine mammals, Vol. 5. London: AcademicPress, 387–404.

da Silva VMF, Best RC. 1994. Tucuxi – Sotalia fluviatilis(Gervais, 1853). In: Ridgway SH, Harrison R, eds. Hand-book of marine mammals, Vol. 5. London: Academic Press,43–70.

Slijper EJ. 1936. Die Cetaceen, Vergleichend-Anatomisch undSystematisch. Capita Zoologica 6–7: 1–590.

Slijper EJ. 1946. Comparative biologic–anatomical investiga-tions on the vertebral column and spinal musculature ofmammals. Verhandelingen, Afdeling Natuurkunde Konin-klijke Nederlandse Akademie van Wetenschappen. TweedeReeks 42: 1–128.

Slijper EJ. 1961. Locomotion and locomotory organs inwhales and dolphins (Cetacea). Symposia of the ZoologicalSociety of London 5: 77–94.

Smith GFD, Browne KW, Gaskin DE. 1976. Functionalmyology of the harbour porpoise, Phocoena phocoena (L.).Canadian Journal of Zoology 54: 716–729.

Strickler TL. 1980. The axial musculature of Pontoporia bla-invillei, with comments on the organization of this systemand its effect on fluke-stroke dynamics in the Cetacea. Amer-ican Journal of Anatomy 157: 49–59.

Thewissen JGM. 1994. Phylogenetic aspects of cetacean ori-gins: a morphological perspective. Journal of Mammal Evo-lution 2: 157–183.

Thewissen JGM, Bajpai S. 2001. Whale origins as a posterchild for macroevolution. Bioscience 51: 1037–1049.

Thewissen JGM, Hussain ST, Arif M. 1994. Fossil evidencefor the origin of aquatic locomotion in archaeocete whales.Science 263: 210–212.

Thewissen JGM, Madar SI, Hussain ST. 1996. Ambuloce-tus natans, an Eocene cetacean (Mammalia) from Pakistan.Courier Forschungsinstitut Senckenberg 191: 1–86.

Waddell VG, Milinkovitch MC, Bérubé M, Stanhope MJ.2000. Molecular phylogenetic examination of the Delphi-noidea trichotomy: congruent evidence from three nuclearloci indicates that porpoises (Phocoenidae) share a morerecent common ancestry with white whales (Monodontidae)than they do with true dolphins (Delphinidae). MolecularPhylogenetics and Evolution 15: 314–318.

Walker WF. 1965. Vertebrate dissection, 3rd edn. Philadel-phia: W.B. Saunders Co.

Walker WF, Liem KR. 1994. Functional anatomy of the vert-ebrates: an evolutionary perspective, 2nd edn. Fort Worth:Saunders College Publishing.

Watson AG, Fordyce RE. 1993. Skeleton of two minkewhales, Balaenoptera acutorostrata, stranded on the south-east coast of New Zealand. New Zealand Natural Sciences20: 1–14.

Wells RS, Scott MD. 1999. Bottlenose dolphin – Tursiopstruncatus (Montagu, 1821). In: Ridgway SH, Harrison R,eds. Handbook of marine mammals, Vol. 6. San Diego: Aca-demic Press, 137–182.

Documents

Vertebral osteology in Delphinidae (Cetacea)academics.wellesley.edu/Biology/Faculty/Emily/vertebral ost.pdfVertebral osteology in Delphinidae ... The dolphin vertebral column has been