HARVARD UNIVERSITY

Ernst Mayr Library

of the Museum of

Comparative Zoology

LIBRARY

DEC 4 2006

.HT-

LIBRARY

CONTRIBUTIONS IN

MARINE MAMMAL PALEONTOLOGY

HONORING

FRANK C. WHTTMORE, JR.

Edited by

Annalisa Berta and Thomas A. Demere

Incorporating the Proceedings of the Marine Mammal Symposium

of the Society of Vertebrate Paleontology

5 1 st Annual Meeting

Held at the San Diego Natural History Museum

San Diego, California

26 October 1991

f

No. 29

1 May 1994

Proceedings of the

San Diego Society of Natural History

ISSN 1059-8707

MCZ

LIBRARY

MAY 1 2 1994

HARVARD

UNIVERSITY

CONTRIBUTIONS IN

MARINE MAMMAL PALEONTOLOGY

HONORING

FRANK C. WHITMORE, JR.

Edited by

Annalisa Berta and Thomas A. Demere

Incorporating the Proceedings of the Marine Mammal Symposium

of the Society of Vertebrate Paleontology

51st Annual Meeting

Held at the San Diego Natural History Museum

San Diego, California

26 October 1991

f

No. 29

1 May 1994

Proceedings of the

San Diego Society of Natural History

ISSN 1059-8707

Contents

Preface Annalisa Berta and Thomas A. Demote 1

Tribute to Frank Clifford Whitmore, Jr. Ralph E. Eshelman and Lauck W. Ward 3

I. Marine Mammals: Evolution and Systematics

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematics and Mode of Life

Richard H. Tedford, Lawrence G. Barnes, and Clayton E. Ray 1 1

Pinniped Phylogeny Annalisa Berta and Andre R. Wyss 33

Basicranial Evidence for Ursid Affinity of the Oldest Pinnipeds Robert M. Hunt, Jr. and

Lawrence G. Barnes 57

The Evolution of Body Size in Phocids: Some Ontogenetic and Phylogenetic Observations

Andre R. Wyss 69

Two New Species of Fossil Walruses (Pinnipedia: Odobenidae) from the Upper Pliocene San Diego

Formation. California Thomas A. Demere 77

The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa Thomas A. Demere 99

Phylogenetic Relationships of Platanistoid River Dolphins: Assessing the Significance of Fossil Taxa

Sharon L. Messenger 125

Are the Squalodonts Related to the Platanistoids? Christian de Muizon 135

Waipatia maerewhenua, New Genus and New Species (Waipatiidae, New Family), An Archaic Late

Oligocene Dolphin (Cetacea: Odontoceti: Platanistoidea) from New Zealand R. Ewan Fordyce 147

A Phylogenetic Analysis of the Sirenia Daryl R Damning 177

A New Middle Miocene Sirenian of the Genus Meta.xytheriam from Baja California and California:

Relationships and Paleobiogeographic Implications Francisco J. Aranda-Manteca, Daryl P. Domning,

and Lawrence G. Barnes 191

A New Specimen of Behemotops proteits (Order Desmostylia) from the Marine Oligocene of Washington

Clayton E. Ray, Daryl P. Domning, and Malcolm C. McKenna 205

II. Marine Mammals: Faunas and Biostratigraphy

Neogene Climatic Change and the Emergence of the Modern Whale Fauna of the North Atlantic Ocean

Frank C Whitmore, Jr. 223

Miocene Cetaceans of the Chesapeake Group Michael D. Gottfried, David J. Bohaska, and

Frank C. Whitmore, Jr. 229

Miocene and Pliocene Marine Mammal Faunas from the Bone Valley Formation of Central Florida

Gary S. Morgan 239

PROCEEDINGS

of the

San Diego Society of Natural History

Founded 1S74

Number 29 1 May 1994

Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr.

Preface

On 26 October 1991 a symposium on marine mammal evolution was held at the 51st Annual Meeting of the Society

of Vertebrate Paleontology in San Diego, California. It was the first symposium on this topic since the one in 1975 held

at the American Institute of Biological Sciences meetings in Corvallis, Oregon. The proceedings of that symposium

were published as a special issue of Systematic Zoology (1976, volume 25:301^146) edited by Charles A. Repenning.

Ten years after the Corvallis symposium three of the original participants, Lawrence G. Barnes, Daryl P. Domning, and

Clayton E. Ray, published an article in Marine Mammal Science (1985, volume 1:15-53) that highlighted notable

paleontological discoveries in the intervening years. Since that publication nine years ago, new fosstf discoveries, new

techniques for investigating evolutionary relationships, and an increase in the number of interested researchers have

contributed important new data. The San Diego symposium was organized to bring together many of these researchers

and to present a synthesis of our current knowledge of the evolution, systematica, and biogeography of marine

mammals (pinnipeds, cetaceans, and sirenians) and marine mammal faunas. The symposium benefited from an

international group of participating scientists representing France, Mexico, New Zealand, Russia, and the United

States.

Thirteen of the current articles are based on papers presented in San Diego; two additional reports were solicited

afterward because of their close relationship to the general theme of the symposium. In the volume, we have arranged

the articles into two broad groups, one covering the taxonomy, systematics, and comparative anatomy of specific taxa,

the other dealing with the biogeography and biostratigraphy of marine mammal faunas.

One of the participants at the symposium, Frank C. Whitmore, Jr.. is widely recognized for his contributions to fossil

cetacean systematics and biogeography and for his role as a mentor to many marine mammal paleontologists.

Accordingly, we have chosen to recognize Frank's past achievements and continuing research by dedicating this

volume to him.

The editors gratefully acknowledge the following individuals who, in addition to most contributors, provided critical

reviews of manuscripts: J. David Archibald. Jon A. Baskin, Mario A. Cozzuol, Francis H. Fay, John J. Flynn, John E.

Heyning, Richard Hulbert, Samuel A. McLeod, James G. Mead, Jeheskel Shoshani, Michael J. Novacek, Donald R.

Prothero. and Charles A. Repenning. We are especially indebted to Clayton E. Ray for his advice and encouragement.

Philip Unitt, managing editor of the San Diego Society of Natural History's scientific publications, provided skillful

editing of the volume. For funding of this publication we also express our appreciation to the National Science

Foundation, National Geographic Society. San Diego State University College of Sciences, and the Kellogg fund of the

Smithsonian Institution.

Annalisa Berta and Thomas A. Demere

Tribute to Frank Clifford Whitmore, Jr.

Ralph E. Eshelman

Department of Paleobiology, National Museum of Natural History, Washington, DC. 20560

Lauck W. Ward

Virginia Museum of Natural History, Martinsville, Virginia 24112

With this volume, we pay tribute to our mentor and colleague

the "good doctor" Frank Clifford Whitmore. Jr., whose gracious

humor, interest in people, and curiosity about the world have meant

so much to so many within the field of paleontology and far beyond

it. We are pleased to have this opportunity to share our respect and

admiration for Dr. Whitmore the scientist, the teacher, and the

friend.

Frank was born at home in Cambridge. Massachusetts, on No-

vember 17. 1915. to Marion Gertrude (Mason) and Frank Clifford

Whitmore. a graduate student at Harvard University. When Frank

Jr. was two, the growing family began a series of moves occasioned

by his father's burgeoning career in organic chemistry: to Houston

and the Rice Institute, to Minneapolis and the University of Minne-

sota, to Evanston and Northwestern University, to Washington.

D. C. and the National Research Council.

Frank, his two younger brothers, and younger sister grew up in a

lively family, which included their Irish grandmother and. as often

as not. a variety of students who rented rooms in their house. Ideas,

adventures, and experiments were encouraged. From these grew

their later devotion to science, literature, medicine, dance, business,

theater, and art.

Frank completed his intermediate schooling in Evanston before

the family relocated, one last time, to State College. Pennsylvania,

and Pennsylvania State College. There his life blossomed with

friendship, studies, and sports. His three high-school buddies have

remained close friends throughout his life. The reporting he did on

the high-school paper and Center Daily Times marks his early work

as a writer, one of the important skills of his career.

In 1 933, Frank enrolled as an English major at Amherst College

in Massachusetts. However, Frank eventually changed his major to

something completely different. The change came about in an

unusual way.

"We had to take a science, and somebody said, 'take geology."

Sol took geology and I was kind of bored by it and I got aC. But the

guy who'd advised me to take it said. 'I know the first course isn't

very good, but you ought to take historical geology. That's really

interesting." So I signed up for historical geology and nobody else

did. This was with F. B. Loomis. . . . Although 1 was the only person

who had registered for the course. [Loomis] agreed to give it. So I

had a one-year course in what was not historical geology [but]

vertebrate evolution, because that was what Loomis was interested

in. and we just sat around three days a week for a year talking about

vertebrate evolution. By the end of that year, I decided I wanted to

be a vertebrate paleontologist" (Cain 1989).

Within his first week at Amherst, Frank was rushed by the

fraternity Phi Kappa Psi. At the end of an exhausting day, he sank

into a couch and remarked, quoting a character from the Barney

Google comic strip. "Moosenose hurt in feet!" One of his new

"brothers" immediately nicknamed him Moose, which remains his

name to many friends and relatives to this day.

In January of 1934, in his freshman year, another fateful event

occurred. While taking his date on a sleigh ride with a group of

other students, the Smith girl seated on his other side got something

in her eye. Frank gallantly helped remove it and life for him was

never the same afterwards; he dated Martha Burling Kremers of

Niagara Falls. New York, until their graduation and engagement.

They were married in 1939, after she'd had a year of business

school and he had earned his master's degree at Pennsylvania State

University.

Although Frank's B. S. cum laude with honorable mention in

geology from Amherst centered on vertebrate paleontology, his M.

S. in paleontology and stratigraphy dealt instead with invertebrates.

Frank M. Swartz was the only resident paleontologist at Penn State;

his specialty was ostracodes and so Frank studied ostracodes. That

year also provided a good background in sedimentology from Paul

Dimitri Kryrine.

Frank continued his education at Harvard University, studying

under one of the world's foremost vertebrate paleontologists, Alfred

Sherwood Romer. Frank was Romer's first student from the geol-

ogy department; all previous students had been from biology.

Romer referred to himself as a zoologist. Frank recalls. "I had quite

a battle with Romer to be allowed to take structural geology be-

cause he thought that would be a waste of time. I felt it would be a

way to get a job. which it turned out to be. ... I remember as a

student, I got to know C. B. Schultz pretty early, and Morris Skinner

and Lloyd Tanner and Mylan Stout, the Nebraska folks, and think-

ing how strange and rather dull that they were always worrying

about stratigraphic correlation. They were talking about the Valen-

tine problem, the Marsland Formation, and so on. In a way. they

certainly were more geological than we easterners were" (Cain

1989). Thanks to his training at Amherst and Penn State. Frank

himself proved to be an exception to this generalization.

A particular admonition of Romer's has in turn been heard by

Frank's own children and young colleagues, probably because it fit

so well his own proclivities: "Learn to write while you're a stu-

dent— you can always learn geology later."

Frank's first real field experience in vertebrate paleontology

came during the summer of 1940. The Harvard field crew headed

west toward the Unita Basin of Utah to collect fossil mammals in a

used pie truck purchased for $85. "I can remember . . . when we

heard that France had fallen, sitting around the campfire. all of us

wondering where we'd be a year from then" (Cain 1989). While in

the field. Frank became the father of twins. Geoffrey Mason and

John Kremers. and their birth kept him out of the army long enough

to complete his doctorate. To support his family. Frank served as a

teaching fellow and university fellow in paleontology.

Harvard was the right place and 1940 the right time to witness

an exciting event, the founding meeting of the Society of Vertebrate

Paleontology. Frank recalled, "of the starving graduate .students,

only those on the premises, like me. could afford to attend. And we

couldn't exactly go overboard socially: the cocktail party at the

founding meeting cost $1.00. My wife Marty and I took stock of our

finances and decided I really should go but that she couldn't afford

to (we would have had to get a sitter for our 6-month-old twins)"

(Whitmore 1989).

Frank's doctoral study, suggested by Romer, was the cranial

morphology of three Oligocene artiodactyls. Frank wrote. "It is the

purpose of this study to examine in detail the cranial anatomy of

some of these extinct genera, because endocranial characteristics

are probably nonadaptive. that is. unlikely to be influenced by the

In A. Berta and T. A. Demere (eds.)

Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore. Jr.

Proc. San Diego Soc. Nat. Hist. 29:3-10, 1994

R. E. Eshelman and L. W. Ward

Figure 1 . Frank C. Whitmore. Jr.. while professor al Rhode Island State

College. Kingston. Rhode Island (now the University of Rhode Island).

ca. 1943.

environment, and therefore useful in determining the taxonomic

position of groups of animals" (Whitmore 1953:117). For this

study, a serial sectioning apparatus was designed and built by F.

Russell Olsen of Harvard's Museum of Comparative Zoology. This

sectioning technique, perfected for paleobotany with cellulose ac-

etate peels, was adapted for vertebrates and described in a paper by

Olsen and Whitmore (1944). It was a pioneering achievement in its

approach and formed the basis of later work by others, in which

details of cranial anatomy such as blood circulation, ear morphol-

ogy, and brain conformation have been used in phylogenetic studies

of fossil mammals.

Frank's first postgraduate job was a teaching position at Rhode

Island State College (now the University of Rhode Island) from

1 942 to 1 944. "I was the entire geology department. I got to teach a

lot of things I'd never studied before, such as specimen petrology

and engineering geology, which I made up myself" (Paleo News

1990). When the Army Specialized Training Program came to the

college, he also taught economic and political geography. Prepara-

tion for these courses aided him most in the next phase of his career.

During this time, Frank Sr. asked his son, "If you could talk to

any paleontologist you wanted to, whom would you choose?"

Frank named W. B. Scott, and his father offered to pay his way to

Princeton for a visit. Frank remembers, "With some qualms, I wrote

to Scott and asked if I could come and see him. and he said yes. So

I got on the train and went down ... to Princeton, and Dr. Scott met

me. He showed me around the lab. I remember he was then about

80. . . . He had me to lunch at the faculty club. . . . Then Scott

showed me the Eocene mammals in the Princeton collection. I had

spent the previous summer in Utah ... so I was interested. At the

end of the day Scott asked me if I would like to collaborate with him

in the study of Eocene mammals in the Princeton collection" (Cain

1989).

Unfortunately, the war intervened, and Frank was unable to take

up the opportunity. His observations about Dr. Scott that day,

however, presaged Frank's own support of young people in his

field. "I probably never would have done it on my own. . . . But it

does give you an idea of what a really decent person Scott was and

how he would give up a whole day to an unknown" young instruc-

tor (Cain 1989).

Frank joined the honorary scientific fraternities. Gamma Alpha

and Phi Sigma, and he conducted short field trips while in Rhode

Island. But his life was not all science; the war was on. and he had to

take his turn up in the college tower as an airplane spotter, mini-

mally trained in the silhouettes of domestic and enemy aircraft.

In March of 1944. Frank was hired by the U. S. Geological

Survey to edit classified reports in the one-year-old Military Geol-

ogy Unit. By September of 1945, he had become chief editor,

supervising four geologists and 15 typists and draftsmen.

Not long after the birth of their first daughter. Katherine Burling,

Frank and Marty prepared for his assignment to the Engineer Intel-

ligence Division, Southwest Pacific Area, by moving the family to a

small house near Niagara Falls.

Frank's position as scientific consultant on terrain intelligence

took him first to Manila, where he organized the Natural Resources

Section of the General Headquarters of the Supreme Commander

for the Allied Powers, in preparation for the occupation of Japan.

After two months, he relocated to Tokyo, where he served as chief

of the Engineering Geology Unit, Natural Resources Section. He

supervised the field checking of terrain intelligence reports and

consulted with the U. S. Army on foundation conditions, location of

construction materials, and selection of airfield and port sites.

Frank became a commodity specialist in precious metals, com-

piling data on gold and silver production in Japan. As Frank tells it,

"Since I was the paleontologist and didn't know much, they looked

around for the least harmful thing for me to do. That's how I was put

in charge of the precious metals. My job was mainly to hold

audience with Japanese gold and silver mine operators and tell them

'no,' they could not mine gold. It was the perfect bureaucrat's job,

sitting there all day saying 'no.' I was also in charge of the vaults of

Japan, where I saw more money than I will ever see again, with

piles of sheet gold one meter on a side" (Paleo News 1990). There

were also stacks of platinum crucibles and "buckets of diamonds."

The temptations might have been great, except that the military

officer from whom Frank took over the vaults was later arrested for

trying to re-enter the United States with a pocket full of diamonds.

When asked to join the Geological Survey, Frank knew it was

likely to mean an overseas assignment, possibly in a war zone. But

he could not know that his work, and that of his colleagues, would

help directly in saving the lives of Allied troops and hastening the

conclusion both of World War II and the Korean War. In 1946, the

U. S. Army recognized his work with the highest civilian award the

United States bestows, the Medal of Freedom. One would think a

tribute of this magnitude would involve a personal presentation if

not a public ceremony, but the medal was simply sent to his home

with a two-sentence cover letter.

One event during Frank's duty in Japan created enough contro-

versy that its echoes can still be heard, and so we cannot exclude it

from this account. The story of the Peking Man fossils involved

Tribute in Frank Clifford Whilmore. Jr.

bones of Sinanthropus pekingensis and other relics, which the

Japanese had earlier stolen from China. Frank, as part of his duties,

was to take custody of these objects, pending their return to the

National Geological Survey of China. But there has always been

disagreement over the actual location of the objects and the se-

quence of events prior to Frank's arrival on the scene. According to

one story, in 1941 the specimens were packed in three cases marked

"secret" and turned over to U. S. marines who were evacuating

Chinwangtao, China, aboard the Dollar Liner President Harrison.

The liner ran aground in the Yangtze River near Shanghai on

December 8. and the marines were captured. There is documentary

evidence that scientists from the Tokyo Imperial University visited

Peking in August, 1942, at the request of the Japanese North China

Army and took the collection to Tokyo. After the surrender of Japan

to the Allied powers, a letter from the Central Liaison Office of

Japan alerted the Allies to the collection's existence. The Natural

Resources Section was directed to take action to return the speci-

mens, and the section chief dispatched Frank to examine the collec-

tion. In his memo. Frank stated he could "find no traces of Sinan-

thropus" (Lamp and Huang 1990). Many, including Frank, believe

the fossils rest at the bottom of the Yangtze River, but he has been

accused of keeping secret information on their whereabouts. To this

day, the original fossils have never been recovered, but good casts

exist and excavations at the Peking Man site have turned up addi-

tional specimens.

Frank's overseas work did not end in Tokyo. He spent the spring

of 1946 assigned to the 24th Corps, U. S. Army, in Korea to survey

and map railroads, major highways, landing beaches and ports,

including Inchon, which played an important part in the later U. S.

invasion. While in Korea, Frank was promoted to chief of the

Military Geology Unit.

His return to Washington, D. C, did not mean an immediate end

to the family's separation. The government had mushroomed dur-

ing the war, with little or no increase in housing, and Frank spent

months trying to buy or rent a house in the area. Eventually he

found one in West Hyattsville. Maryland, and moved the family

south from Niagara Falls. Frank and Marty's last child. Susan Hale,

was born in 1948, during their ten years in that house.

With the end of war-related work, it was assumed that the

Military Geology Unit would be shut down, and Frank was looking

forward to joining the Paleontology and Stratigraphy Branch and

returning to his fossil vertebrates. But the war had demonstrated to

the U. S. military how little it knew about foreign geology, and so

the unit was transformed into a regular branch of the Geological

Survey. Frank stayed on as chief until 1959.

His management skills and abilities, in part acquired in Asia,

enabled Frank to develop and direct the worldwide activities of the

branch, which employed about 120 scientists and support person-

nel, with headquarters in Washington and field offices in Tokyo,

Heidelberg, and Salzburg. Frank organized interdisciplinary field-

mapping programs involving the study of geology, soils, vegeta-

tion, hydrology, and topography.

Frank's leadership was increasingly recognized and put to use.

He chaired numerous groups including the U. S. Geological Survey's

Geologic Division Staffing Committee and the committee to com-

pile permafrost terms for the first and second editions of the Ameri-

can Geological Institute's Glossary- of Geology. He also served as

security officer for the Geologic Division between 1948 and 1956.

In recognition of the International Geophysical Year in 1958.

the Lake Peters Research Station (renamed the G. William Holmes

Research Station in 1970) was established in the northeastern part

of the Brooks Range of Alaska. Frank was on the team that con-

ducted the initial reconnaissance at this offshoot of the Arctic

Research Laboratory at Point Barrow and formulated plans for

continuing research in the area.

Finally, after 15 years of administration, Frank joined the Pale-

ontology and Stratigraphy Branch of the Geological Survey as a

senior specialist in vertebrate paleontology. He was assigned an

office at the National Museum of Natural History. "When I came

back to paleontology ... I was for all intents and purposes, fresh

out of graduate school, although I'd had my Ph.D. for seventeen

years" (Cain 1989). Frank became the informal chief of the survey's

vertebrate paleontology staff, which included Charles Repenning at

the Menlo Park, California, office and Ed Lewis at the Denver

office.

He launched a series of diverse investigations. In 1959 and

1960, he collected and studied Miocene and Pleistocene vertebrates

from Martha's Vineyard, Massachusetts, as part of the work done

by the Engineering Geology Branch. His biostratigraphy of that

complexly deformed area helped determine the history of Pleisto-

cene deformation on the island.

From 1959 through 1965, Frank conducted biostratigraphic

studies of Paleozoic and Mesozoic fish and Tertiary mammals from

Wyoming and Montana, to aid ongoing geologic mapping there. He

was principal investigator for field and laboratory studies of Mio-

cene mammals from Panama between 1962 and 1965. This resulted

in a biostratigraphic correlation with faunas in Texas and Florida,

established that the Miocene mammal fauna of Panama was en-

tirely of North American affinity, and helped to define a circum-

Caribbean Miocene zoogeographic province and to delineate the

southern extent of the North American land mass. These results

were published in Science.

At about the same time, Frank began collaborating with

Bertrand Schultz and Lloyd Tanner of the University of Nebraska

on work at Big Bone Lick, Kentucky. This important Pleistocene

site is the type locality of Mammut americanum, the American

mastodon, and Bootherium botnhifrons. an extinct musk ox. It is

also the site where, on Thomas Jefferson's orders, explorers Lewis

and Clark collected bones for shipment back to the amateur scien-

tist and president of the United States.

The team's field and lab studies, from 1962 through 1970. of the

late Pleistocene mammals and stratigraphy of Kentucky contrib-

uted to the geomorphic history and paleoclimatology of the Ohio

valley. After five summers of field work, their results helped con-

vince the state of Kentucky to create Big Bone Lick State Park,

ensuring preservation of the site. For his efforts, Frank was anointed

an Honorable Kentucky Colonel by the state.

Meanwhile, Frank was also being exposed to fossil marine

mammals, thanks to his close association with Remington Kellogg,

world-renowned expert, who worked in the Paleobiology Depart-

ment of the museum. It was Dr. Kellogg who made the west side of

the Chesapeake Bay a permanent fixture in Frank's life as Frank

increasingly helped the elder paleontologist and, after his death in

1 969, took over some of his work.

During the late 1960s and early 1970s. Frank was principal

investigator for the Calvert Cliffs Paleontology Project on Chesa-

peake Bay. This project entailed detailed interdisciplinary paleo-

ecological and stratigraphic studies during excavation for the

Calvert Cliffs nuclear power plant. Funding for this work came

from the Ford Foundation and National Geographic Society.

The Calvert Cliffs Paleontology Project opened the door to

Frank's association with the National Geographic Society. In 1971,

he was asked to join the prestigious Committee for Research and

Exploration, serving as Dr. Kellogg's replacement. The committee

grants millions of dollars each year for research projects throughout

the world, some of which are later described in the society's maga-

zine.

In 1 972, Frank returned to Alaska, this time to Amchitka Island,

where he collected fossils of the historically extinct Steller's sea

cow (Hydrodamalis gigas). He and others worked on the rate and

R. E. Eshelman and L. W. Ward

Figure 2. Frank C. Whitmore, Jr., 1965, holding skull of fossil musk ox

collected in 1 807 by William Clark (of Lewis and Clark fame) and exhibited

in the White House in the same year.

mode of Pleistocene uplift of the island, as indicated by beach

deposits, which were critical to the prediction of the effects of

nuclear testing.

Work on Oligocene whales from South Carolina resulted in two

publications in 1975 and 1976. Since 1976, Frank has been princi-

pal investigator on the study of Paleocene vertebrates from Saudi

Arabia. Paleoecologic studies of this estuarine fauna established the

geographical position of part of the ancient Tethys Sea, and contrib-

uted to the delineation of lime deposits needed for cement manufac-

ture.

Like that of the other research scientists at the Geological

Survey, part of Frank's job involved handling "examination and

report" (E & R) requests. Some were submitted by colleagues in

other disciplines whose investigations turned up pieces of what

might be bone. Others came in via USGS public-relations people

from citizens who wanted to know something about their backyard

digs or vacation treasures.

One E & R stands out above all others, both for its unusual

nature and because it landed on the desk of a man whose knowledge

of historical time is as acute as his understanding of geologic time.

This was "The Case of the Papal Proboscidean."

Sylvio Bedini, then deputy director of the National Museum of

History and Technology (now the National Museum of American

History) asked Frank to identify, from photographs, some bones

dug up during the air-conditioning of the papal apartments in the

Vatican. Everyone was puzzled when Frank identified at least one

bone as that of an elephant. Further research revealed that in 1514,

when Pope Leo X had a stranglehold on the spice trade to the Far

Fast. King Emmanuel the Great of Portugal wanted a share of the

action. To get on the good side of the pope. Emmanuel presented

him with a young elephant. No elephant had been seen in Rome

since the time of Hannibal, and it proved to be a great curiosity —

especially as it had been trained to genuflect whenever the pope

appeared. It also held water in its trunk and squirted designated

victims on the command of its trainer. One day. the elephant's

keepers decided they would gild the elephant from head to toe as a

surprise for the pope. The unexpected surprise was that the gilding

killed the elephant. The devastated pope directed the papal painter,

who happened to be Raphael, to paint a life-size mural of the

elephant; Raphael felt this was beneath him and ordered an appren-

tice to complete the mural on the palace wall. The elephant was

subsequently buried beneath the painting. The mural is now gone

but the bones remain (Whitmore 1978).

This and many others stories have been shared gleefully over

lunch at the museum, whose collegiality is perfectly suited to

Frank's temperament. For years, he took a bag lunch to the office of

his longtime friend Harry Ladd. As the other members of the lunch

mess would appear, they would array themselves around the desk

with their sandwiches and the talk would begin. Here, one might

say, is an example of the cross-fertilization that takes place in a

great museum — scholars engaged in the exchange of ideas. Ideas

certainly were exchanged — on politics, reminiscences of fieldwork,

stories of questionable taste, the Washington Redskins; innumer-

able small bets were made, usually on sports or politics, and duly

recorded on Harry's desk calendar (Whitmore and Tracey 1984).

A story Frank tells on himself concerns a day. decades past,

when he was collecting a fossil whale along the shores of Chesa-

peake Bay. Typically, there were a few helpers along that day —

amateur collectors and aspiring young scientists. In the course of

the long day, one of these youths said to Frank — surely out of the

greatest respect — "You must be one of the last of the old-time

collectors!" And so he is.

Perhaps it is Frank's memory of W. B. Scott and that day at

Princeton, or perhaps it is simply Frank's openness and interest in

people that cause him to be so generous with his time and his

knowledge. He frequently gives tours behind the scenes at the

museum to school and college groups, out-of-town visitors, and

amateur collectors. In the 1950s, Washington-area school children

watched him on the early WETA television science series Time for

Science. And school children nationwide now see him on one

segment of PBS's 26-segment story The Voyage of the Mimi when

the young protagonist visits Frank at the museum to learn about

whales and paleontologists.

Frank retired from the United States Geological Survey in 1984,

but he continues his work as a research associate and curator

emeritus of the Smithsonian Institution in his old office there. His

current studies include the taxonomy and description of fossil Plio-

cene whales and terrestrial mammals from the Lee Creek phosphate

mine at Aurora. North Carolina, and description of Miocene marine

mammals from the Pisco Formation of Peru. Many of his papers

and publications have been deposited in the Smithsonian Institution

archives.

While Frank came late to his life's research, his colleagues and

employers valued highly the management and leadership he brought

with him to the museum. His ability to listen, to draw others out.

and to mediate discussion of touchy subjects has often been tested.

Frank was appointed chair of the joint U. S. Geological Survey/

Smithsonian Institution committees for the design of new labs and

for decisions regarding the paleontology collections. In 1971, he

was general chair of the Geological Society of America meetings in

Washington, with 4300 people attending. For the American Asso-

ciation for the Advancement of Science and the Mexican Science

Council, he chaired a symposium on land connections between

North and South America.

Tribute to Frank Clifford Whitmore, Jr.

<

Figure 3. Frank C. Whitmore, Jr.. studying fossil whales at the Univer-

sity of Otago. Dunedin, New Zealand. 1988.

The list of institutions on whose panels and committees he has

served is long and varied. It includes the Department of Defense,

American Geological Institute, National Research Council, and the

Center for the Study of the First Americans. He has provided

scientific guidance to the Schoelkopf Geological Museum in

Niagara Falls, New York, and to exhibit specialists at the National

Museum of Natural History and the state of South Carolina.

In 1979, he served as general chair of the International Centen-

nial Symposium of the USGS on "Resources for the 21st Century."

Brought together were some 500 scientists, corporate executives,

and government officials from 48 countries. Frank spent five years

planning the agenda, booking speakers, and editing and rewriting

many of the foreign papers to make them publishable in English.

Professional societies have provided Frank with a continuous

thread throughout his varied career. And they have all benefited by

his membership. He's at home both at small, convivial monthly

meetings and at huge national conventions. His wit at the dinner

table is as valuable to many as his chairing of committees and

symposia.

Frank was a founding member of the Society of Vertebrate

Paleontology in 1940: he later served on its executive board and is

an honorary life member. He's been a member of the Paleontologi-

cal Society of America since 1942. In 1944. he joined the Paleonto-

logical Society of Washington, later serving as vice-president and

president. That same year, he joined the Geological Society of

Washington, becoming councillor, secretary, first vice-president,

and then president. In 1945, perhaps to show that he couldn't live

without meetings wherever he might be stationed, he helped found

the Geological Society of the Philippines.

He became a fellow of the Geological Society of America in

1947, and has served on its Penrose Medal Committee and as its

Penrose citationist. Three years later, he became a fellow of the

American Association for the Advancement of Science, eventually

serving as section secretary and chairman, councillor, and chair of

the Newcomb Cleveland Prize Committee.

Although not of the highest scientific significance, one mem-

bership gave Frank a wider forum for his well-known humorous

talents. In 1944. he joined other young U. S. Geological Survey

types as a member of the Pick and Hammer Club. Begun in 1 894 as

the Association of Aspiring Assistants, it was originally an offshoot

of the Geological Society of Washington. By the mid-twentieth

century, this platform for talks by unestablished geologists was best

known for its annual show, which spoofs the stuffed shirts of the

geologic bureaucracy. Frank participated fully in writing songs and

dialogue, acting in minor and major roles, singing lustily, and

dancing, among other things, the academic gavotte in cap and

gown.

In 1967. he was invited to be a guest on the television program

To Tell the Truth. He was, in fact, the "true" vertebrate paleontolo-

gist, but got himself in hot water by misunderstanding one of the

questions. Instead of answering "dinosaur" to the query. "What was

the largest carnivore in the world?" he replied, "bear." Conse-

quently, the panel was led astray and guessed the wrong contestant.

Immediately following the show. Kitty Carlisle lit into him, "Imag-

ine, a bear! How ridiculous!" And when Frank returned to his office

at the museum, there in his chair sat a huge femur of a meat-eating

dinosaur.

In 1981, Frank was traveling in China when he was awarded the

Meritorious Service Award by the Interior Department. This got

him out of shaking hands, on stage, with then-secretary James G.

Watt, not one of his heroes. When it was Frank's time to receive his

40-year service scroll and pin, a representative came to his office,

stated something to the effect that he knew Frank wouldn't want a

lot of pomp and ceremony, discovered he didn't have the award in

his pocket, and left saying it would be sent in the mail.

There is a type of honor frequently bestowed by his colleagues

that has been harder to receive. Because of his friendship, respect

for biography, and skill as a writer, over the years Frank has been

asked to prepare memorials to several fellow scientists. As a result,

his bibliography includes memorials to Alfred Romer, Remington

Kellogg. Harry Ladd, Willy Postel. Louis Ray, John Huddle, and

Charley Johnson. They are graceful tributes to men whose work and

character he admired.

In addition to his own writing. Frank has reviewed many books

and papers. One stands out from all the rest, and we quote it here as

an example of the good doctor's breadth of knowledge, attention to

detail, and mellow humor. From his review of the sixth edition of

"Eoornis pterovelox gobiensis" Whitmore stated, "Far ahead of his

time. Fotheringham thoughtfully melded together every aspect of

the natural history of his subject: the occurrence of its fossil ances-

tors, references to it in Egyptian hieroglyphics, its appearance in

Cro-Magnon cave paintings, and at the other extreme of scientific

inquiry, the most painstaking physiologic studies of wing-beat fre-

quency and of the pH cycle of the bird's beak fluid, observed under

the most difficult conditions over an entire year. The author's

discovery that each dumbbell-shaped egg contains a male and a

female, and his analysis of the part played by parthenogenesis in the

evolution of the genus, makes the mind boggle" (Whitmore 1967).

But a truer indication of a person's worldliness and erudition

R E. Eshelman and L. W. Ward

may be the letter to the editor, of which Frank has written many. Here

we quote from a joint response by Whitmore and Hotton to a 1972

article in Smithsonian Magazine entitled "Fantastic Animals Proved

Tall Timber of our Mythology." "The taxonomy of the sidehill

gouger {Membriinequales declivitous) is more complicated than

Carson suspects. There are subspecies, not yet formally named but

recognized by local inhabitants, and there is more than a scintilla of

evidence that the morphology and habits of these taxa can be

correlated. Among many examples we can cite is the sidehill dodger,

which inhabits the Driftless Area of Wisconsin; the dextrosinistral

limb ratio approaches unity although the metapodials on the down-

hill side are noticeably stouter. The sidehill gouger is the common

Pennsylvanian species, but it has been speculated (Hotton, in lilt.

1972) that all of the different varieties spring from the little-known

strike-runner (Crestophilus ambiguus) in which the right front and

the left rear limbs are short and the left front and right rear are long,

or vice versa. These animals ran the long Tennessee ridges in late

Pleistocene time but became extinct during the Pleistocene epoch,

perhaps because of trauma resulting from attempts to run ridges of

newly formed glacial cirques" (Hotton and Whitmore 1972).

A professional life as busy as Frank*s might leave the impres-

sion that he has little activity outside his career, but his nonscientific

interests are many. When their family was young, he and Marty

took their children on many day trips, to Rock Creek Park and the

zoo, on picnics, to Civil War battlefields, and the like (mysteriously,

after he began working at the museum, Saturday trips to the

Smithsonian decreased markedly). Drives to the country, and longer

trips to visit relatives, invariably involved a stop along the road so

he could give a brief lesson on an especially noteworthy outcrop.

Despite the financial strains of a growing family, the Whitmores

made sure to attend an occasional play and took the children along

at a young age. This meant Saturday matinees at the National

Theater, where dramas and musicals tried out on the way to Broad-

way, and a brand-new local group. Arena Stage, put on productions

at the Hippodrome and the Old Vat. The Whitmores have had

season tickets to Arena Stage since the 1960s.

Certain aspects of Frank's work have intersected with his and

Marty's other interests, especially their love of travel. Field trips to

Martha's Vineyard took them back to the New England of their

student years. Frank's membership on the Committee for Research

and Exploration means their joining the National Geographic

Society's triennial tours of current research projects; hence their

trips to South America. Africa, Europe, Asia, and Australia.

At home, alongside mementoes of their long life and many

travels together, the Whitmores display tokens of affection from

friends and family. Half of the pictures, statuettes, and trinkets

depict the moose, Frank's long-time alter ego. The other half are

whales, in honor of the marine-mammal paleontologist. In addition.

he has accumulated a veritable wardrobe of whale neckties, in

colors suitable for every occasion — from Christmas to St. Patrick's

Day to a meeting at the National Geographic Society.

Frank continues to serve as vice-chair of the National Geo-

graphic Society's Committee for Research and Exploration. He

attends to his mail and his research at the museum. When in town.

Frank and Marty still frequent the Tuesday Lunch Group of mu-

seum denizens, at the Beeffeeders on Tenth Street. They continue to

travel extensively, in part to keep up with the four children, now

scattered across the country. There are five grandsons, three step-

grandchildren, and great-grands may soon be on the way.

Many of us owe some measure of our professional progress to

this man who took lime to listen, teach, and put up with the ignorance

of budding scientists and up-and-coming collectors. While working

with Frank in the late 1960s, we used to refer to him as "the good

doc," a nickname we still use with reverence and respect for a man

who means that and more. To us. he will always be "the good doc."

ACKNOWLEDGMENTS

We are first and foremost thankful to Martha Kremers

Whitmore. who was generous in her assistance. When Ralph

Eshelman called Marty at home (knowing Frank was safely down-

town in his office) to tell her of this tribute, he told her he had

something secret to discuss concerning her husband. In typical

Marty fashion she responded in a happy, friendly voice. '•How-

mysterious and exciting! What is it?" She immediately offered a file

she has kept on Frank from early in their marriage, including

newspaper clippings, programs, personal notes, letters, photo-

graphs, certificates, and more. Without this material, our tribute to

Frank would be woefully incomplete. But just as revealing were the

little notes she has frequently written him. supporting his work and

commending his achievements. They illustrate her admiration for,

and devotion to. her husband. For this love he is most fortunate.

Many of Frank's friends and co-workers added to this tribute.

Everyone approached was enthusiastic and supportive, making our

job all the easier. It is obvious from these contacts that Frank has

many friends who wish him well. Among them are Edwin Snider.

Joshua Tracey. Ellis Yochelson, Barry Bishop, Tom Dutro. Richard

and Mary Ellen Williams, Pamela Henson, Clayton Ray, David

Bohaska, Nicholas Hotton, Warren Blow, and Joseph Cain.

LITERATURE CITED

Cain. J. 1989. Oral history interview of Frank C. Whitmore. Jr.. 8 August

1989. Smithsonian Institution Archives.

Hotton. N.. and Whitmore. F. C, Jr. 1972. Letters to the editor: Fantastic

Animals. Smithsonian Magazine 3 (7): 13.

Lamp. J., and Weiwen. H. 1990. The story of Peking Man: From

archaeology to mystery. Foreign Language Press. Beijing. China.

and Oxford University Press, Hong Kong.

Olsen. F. R.. and Whitmore. F. C Jr. 1 944. Machine for serial sectioning

of fossils. Journal of Paleontology 18: 210-215.

Paleo News. 1990. Department of Paleobiology newsletter. 1(1): 5.

Whitmore, F. C, Jr. 1953. Cranial morphology of some Oligocene

Artiodactyla. United States Geological Survey Professional Paper

243-H: 117-159.

. 1967. Review: "Eoornis pterovelox gobiensis" by Augustus G.

Fotheringham. Journal of Paleontology 41: 1302-1303.

— . 1978. The papal proboscidean. The Cross Section 4(3): 12.

1989. Letter to John A. Wilson. 17 March 1989. Smithsonian

Institution Archives

— , and J. I. Tracey, Jr. 1984. Memorial to Harry Stephen Ladd.

1899-1982: Geological Society of America Memorials 14: 1-7.

SELECTED BIBLIOGRAPHY OF FRANK C. WHITMORE. JR.

1938. (with F. B. Phleger. Jr.). Two young merycoidodonts. American

Journal of Science, 5th ser., 36 (215): 377-388.

1942. Endocranial anatomy of some Oligocene Artiodactyla. Geological

Society of America Bulletin 53: 1842-1843.

1944. (with F. R. Olsen). Machine for serial sectioning of fossils. Journal

of Paleontology 18: 210-215.

1946. Planned peacetime work in military geology. Geological Society

of America Bulletin 57: 1242.

1947. Cranial morphology of some early Tertiary Artiodactyla. Harvard

University, Summary of Theses 1943-1945, pp. 198-200.

1948. Military geology. Professional Geographer 7: 7-16.

— . Digest on problems of military geology in the event of a

national emergency. Research and Development Board. National

Military Establishment, Digest Series 10.

1040. Review: "Geology and Paleontology (Fiat Review of German

Science, 1939-1946)." Science 109 (2834): 425-426.

1951. Translation: Kleinschmidt. A. 1951. Uber ein skelet und eine

Rekonstruktion des ausseren Habitus der Reisenseekuh, Rhytina

gigas Zimmerman 1780. Zoologischer Anzeiger 146, heft 9-10'

292-314.

1953. Cranial morphology of some Oligocene Artiodactyla. United

States Geological Survey Professional Paper 243-H: 1 17-159.

Tribute to Frank Clifford Whitmore, Jr.

. Currenl Japanese studies of Desmostylus. Society of Vertebrate

Paleontology News Bulletin 37: 9.

1956. (with F. M. Swart/). Ostracoda of the Silurian Decker and Manlius

limestones in New Jersey and eastern New York. Journal of Paleon-

tology 30: 1029-1091.

1957. (with R. Kellogg). Marine mammals — annotated bibliography.

Pp. 1223-1226 in J. W. Hedgpeth (ed.). Treatise of marine ecology

and paleoecology: Ecology. Geological Society of America Mem-

oir 67. vol. I

— . (with R. Kellogg). Mammals — annotated bibliography. Pp.

1021-1024 in H. S. Ladd (ed.). Treatise on marine ecology and

paleoecology: Paleoecology. Geological Society of America

Memoir 67. vol. 2.

1958. Further information on archeological and fossil material from

Choukoutien. China. Asian Perspectives 2(1): 54-56.

— . Geologic writing for the nongeologist. Geological Society of

America Bulletin 69: 1662.

1960. Fossil mammals from Ishigaki-shima. Ryukyu-retto. United States

Geological Survey Professional Paper 400-B. art. 171: B372-

B374.

1961. Edited: Discoveries in Ancient Life, by V. Brown. Science Materi-

als Center. New York.

1962. Paleontology, evolution, findings on 50-foot fossil whale. United

States Geological Survey Professional Paper 450-A: 74.

— . Review: "Mestonakhozhdeniya Tretichnykh Nazemnykh

Mlekopitayushchikh na Territorii SSSR." by A. A. Borisiak and E.

I. Behaeva (Tertiary terrestrial mammalian localities in the territory

of the USSR). International Geology Review 4: 863-864.

1963. (with C. B. Schultz, L. G. Tanner. L. L. Ray. and E. C. Crawford).

Paleontologic investigations at Big Bone Lick State Park, Ken-

tucky: A preliminary report. Science 142 (3596): 1 167-1 169.

1965. (with R. H. Stewart). Miocene mammals and Central American

seaways. Science 148 (3667): 180-185.

— . Presentation of the Paleontological Society Medal to G. Arthur

Cooper. Journal of Paleontology 39: 520-522.

1966. (with 1. L. Ray. C. B. Schultz, L. G. Tanner, and E. C. Crawford).

Kentucky. Pp. 53-63 in Guidebook for Field Conference G, Great

Lakes-Ohio River Valley: International Association for Quaternary

Research. 7th Congress, U.S.A., 1965. Nebraska Academy of Sci-

ence, Lincoln, Nebraska.

— . Review: "The Age of Reptiles," by E. H. Colbert. Geotimes 10

(6): 32.

— . Review: "Dinosaur Hunt," by G. O. Whitaker. Geotimes 10(6):

36-37.

— (withC. B. Schultz and L. G. Tanner). Pleistocene mammals and

stratigraphy of Big Bone Lick State Park. Kentucky. Geological

Society of America Special Paper 87: 262-263.

1967. (with H. L. Foster). Pamhera atrox (Mammalia: Felidae) from

central Alaska. Journal of Paleontology 4 1 : 247-25 1 .

. (with C. B. Schultz, L. G. Tanner, L. L. Ray. and E. C.

Crawford). Big Bone Lick, Kentucky: A pictorial story of the

paleontological excavations at this famous fossil locality from

1962 to 1966. University of Nebraska State Museum. Museum

Notes 33.

— . (with K. O. Emery, H. B. S. Cooke, and D. J. P. Swift). Elephant

teeth from the Atlantic continental shelf. Science 1 56(378 1 ): 1 477-

1480.

— . Elephants under the sea. Science Digest 61 (4): 15-16.

— . Edited: "The Changing Earth." by J. Viorst. Bantam. New York.

— . Presentation of the Paleontological Society Medal to Alfred S.

Romer. Journal of Paleontology 41: 817-819.

— . Review: "Marsh's Dinosaurs." by J. H. Ostrom and J. S. Mcin-

tosh. Geotimes 12(4): 34-36.

— . Review: "Fossil Lake. Oregon: Its Geology and Fossil Faunas."

by I. S.Allison. Journal of Paleontology 41: 814-815.

— . Review: "A Review of the Macropodid Genus Slhenurus," by

R. H. Tedtord. Journal of Paleontology 41: 815.

— . Review: "Eoornis pterovelox gobiensis," by A. G.

Fotheringham. Journal of Paleontology 41: 1302-1303.

— . Review : "The Fossil Macropodidae from Lake Menmdee. New

South Wales." by R. H.Tedford. Journal of Paleontology 41: 1303-

1 304.

1468. Review: "The Bering Land Bridge," edited by D. M. Hopkins

American Scientist 56: I50A-152A.

— . Memorial to Albert Williams Postel. Geological Society of

America. Proceedings for 1966. pp. 341-346.

1%9. Adaptive radiation of Cetacea. United States Geological Survey

Professional Paper 650-A: Alt)

— . (with C. B. Schultz. L. G. Tanner, and L. L. Ray). Geologic and

faunal evidence of the Quaternary deposits at Big Bone Lick.

Kentucky. Geological Society of America. South Central Section

Meeting. Lawrence. Kansas. Abstracts with Programs for 1969. pt.

2, pp. 24-25.

— . Review: "Fossil Vertebrates of Southern California," by T

Downs. Journal of Paleontology 43: 1307-1308.

. Review: "Notes and Comments on Vertebrate Paleontology,"

by Alfred S. Romer. Geotimes 14 (9): 33, 36.

-. Ecologic and stratigraphic implications of some Cenozoic ver-

tebrates in the Atlantic coastal plain. Geological Society of

America. Abstracts with Programs for 1969. pt. 7, p. 236.

1970. (with A. E. Sanders). Extinct porpoises collected near Charleston.

South Carolina. Marine Newsletter I (6): 1.

— . Cenozoic of the U. S.. New toothed whale from the Yorktown

Formation in Virginia. United States Geological Survey Profes-

sional Paper 700-A: A 145.

1 97 1 . Calvert Cliffs project. Science 173(3993): 192-193.

. Vertebrate biofacies and paleoenvironmental history of Maryland

Miocene. Guidebook 3. Maryland Geological Survey, pp. 31-36.

— . (with R. E. Gemant and T G. Gibson). Description of selected

Miocene exposures. Guidebook 3. Maryland Geological Survey,

pp. 49-58.

1972. (with L. M. Gard. Jr., and G. E. Lewis). Steller's sea cow in

Pleistocene interglacial beach deposits on Amchitka, Aleutian Is-

lands. Geological Society of America Bulletin 83: 867-870.

— . (with W. I. Finch and J. D. Sims). Stratigraphy, morphology,

and paleoecology of a fossil peccary herd from western Kentucky.

United States Geological Survey Professional Paper 790.

— . Remington Kellogg ( 1892-1969). American Philosophical So-

ciety Yearbook 1972. pp. 205-210.

1973. (with C. E. Ray). Paleontology. Pp. 67-68 in T Simkin, W. G.

Reeder. and C. MacFarland (eds.). Galapagos Science: 1972 Status

and Needs. Smithsonian Institution, Washington. D. C.

1974. Collections of South African mammal-like reptiles in the United

States: A correction. Journal of Paleontology 48: 607.

— . Memorial to Remington Kellogg, 1892-1969. Geological Soci-

ety of America Memorials 4: 117-129.

— . (with J. Knapp). Remington Kellogg. October 5. 1892-May 8.

1969. Biographical Memoirs, National Academy of Sciences 46:

159-189.

— . (with A. E. Sanders). The Cetacea 30 million years ago. Ameri-

can Zoologist 15: 812.

. A thousand nights' entertainment. The Geological Society of

Washington. 1893-1975. Geotimes 20 (7): 14-15.

. Presentation of the Penrose Medal to Maurice Ewing. citation

by Frank C. Whitmore. Jr. Geological Society of America Bulletin

86: 1161-1168.

1976. (with A. E. Sanders). Review of the Oligocene Cetacea. System-

atic Zoology 25: 304-320.

1977. Memorial to Alfred Sherwood Romer. 1894-1973. Geological

Society of America Memorials 5: 1-10.

— . Review: "Origin and Evolution of the Elephantidae." by V J.

Maglio. Journal of Mammalogy 58: 457—159.

— . (with S. H. Whitmore). Review: "Dinosaurs," by N. Sullivan.

AAAS Science Books and Films 13(2): 94-95.

— . (with L. M. Gard. Jr.) Steller's sea cow (Hydrodamalis giga.s)

of late Pleistocene age from Amchitka. Alaska. United States Geo-

logical Survey Professional Paper 1036.

. The papal proboscidean. The Cross Section 9(3): 12.

— . (with S. H. Whitmore). Review: "What Really Happened to the

Dinosaurs'?" by D. Cohen. AAAS Science Books and Films 14(3):

181.

— . Review: "Athlon: Essays on Palaeontology in Honour of Loris

Shano Russell." edited by C. S. Churcher. Journal of Paleontology

52: 1401-1403.

10

R. E. Eshelman and L. W. Ward

1979. Alexander Wetmore. 1886-1978. Society of Vertebrate Paleontol-

ogy News Bulletin 1 16: 64-65.

. Review: "The Ecology of Fossils: Illustrated Guide," edited by

W. S. McKerrow. AAAS Science Books and Films 15(3): 147.

. (with C. T. Madden. 1. M. Nagvi. D. L. Schmidt. W. Langston,

— . Review: "A New Look at the Dinosaurs." by A. Charig. AAAS

Science Books and Films 19 (2): 78.

Hemphillian vertebrate fauna from Mobile County. Alabama.

Jr.. and R. C. Woodl. Paleocene vertebrates from coastal deposits in

the Harrat Hadan area. At Taif region. Kingdom of Saudi Arabia.

United States Geological Survey. Saudi Arabian Mission. Project

Report 269, Open-file Report OF-80-227.

1980. Memorial to Louis Lamy Ray. 1909-1975. Geological Society of

America Memorials 10: 1-8.

. Review: "Dinosaurs and People: Fossils. Facts, and Fantasies."

by L. Pringle. AAAS Science Books and Films 15 (4): 226.

— Review: "Mesozoic Mammals: The First Two-thirds of Mamma-

lian History." edited by J. A. Lillegraven, Z. Kielan-Jaworowska,

and W. A. Clemens. AAAS Science Books and Films 16(2): 75.

. Resources for the 21st Century: Summary and conclusions of

the International Centennial Symposium of the United States Geo-

logical Survey. United States Geological Survey Circular 857.

1982. Review: "Dinosaurs in Your Backyard," by W. Manetti. AAAS

Science Books and Films 18 (2): 76.

— . (with M. E. Williams). Edited: Resources for the Twenty-first

Century — Proceedings of the International Centennial Symposium

of the United States Geological Survey. United States Geological

Survey Professional Paper 1 193.

— . The International Centennial Symposium: Background and

objectives. P. 2 in F C. Whitmore, Jr.. and M. E. Williams (eds.).

Resources for the Twenty-first Century — Proceedings of the Inter-

national Centennial Symposium of the United States Geological

Survey. United States Geological Survey Professional Paper 1193.

. Remains of Delphimdae from Sahabi Formation. Garyounis

Scientific Bulletin. University of Garyounis, Libya. Special Issue

4: 27-28.

. (with C. T. Madden. K. W. Glennie. R. Dhem. D. L. Schmidt, R.

J. Ferfaglia, and P. J. Whybrow). Stegotatra belodont (Proboscidea.

Gomphothenidae) from the Miocene of Abu Dhabi. United States

Geological Survey, Saudi Arabian Project Report, Jiddah, pp. 1-22.

1983. (with J. E. Repetski). Memorial to John Warfield Huddle, 1907-

1975. Geological Society of America Memorials 13: 1-7.

— . (with C. T. Madden and D. L. Schmidt). Masritherium

(Artiodactyla, Anthracotheriidae) from Wadi Sabya, southwestern

Saudi Arabia: An earliest Miocene age for continental rift-valley

volcanic deposits of the Red Sea margin. United States Geological

Survey Open-file Report OF-83-61 .

— . Review: "First Look at Dinosaurs." by M. E. Selsam and J.

Hunt. AAAS Science Books and Films 18 (5): 273-274.

. Remington Kellogg, 1892-1969. Pp. 15-24 in C. E. Ray (ed.).

Geology and Paleontology of the Lee Creek Mine, North Carolina,

vol. 1. Smithsonian Contributions to Paleobiology 53.

. (with C. T. Madden). Tertiary vertebrate faunas of Arabian

Peninsula. Geological Society of America, Abstracts with Pro-

grams 15 (6),: 633.

Pp. 71-72 in W. C. Isphording and G. C. Flowers (eds). Differen-

tiation of unfossiliferous clastic sediments: Solutions from the

southern portion of the Alabama-Mississippi coastal plain. Tulane

Studies in Geology and Paleontology 17(3).

1984. Cetaceans from the Calvert and Eastover Formations. Pamunkey

River. Virginia. Pp. 227-231 in L. W. Ward and K. Krafft (eds.).

Stratigraphy and paleontology of the outcropping Tertiary beds in

the Pamunkey River region, central Virginia coastal plain. Guide-

book for Atlantic Coastal Plain Geological Association 1984 Field

Trip. Atlantic Coastal Plain Geological Association.

. Land mammals from the Calvert Formation. Pamunkey River,

Virginia. Pp. 236-239 in L. W. Ward and K. Krafft (eds.). Stratigra-

phy and paleontology of the outcropping Tertiary beds in the

Pamunkey River region, central Virginia coastal plain. Guidebook

for Atlantic Coastal Plain Geological Association 1984 Field Trip.

Atlantic Coastal Plain Geological Association.

. (with J. I. Tracey, Jr.). Memorial to Harry Stephen Ladd, 1899-

1982. Geological Society of America Memorials 14: 1-7.

— . Review: "American Science in the Age of Jefferson," by J. C.

Green. Earth Science History 3: 188-190.

1986. (with G. V Morejohn and H. T Mullins). Fossil beaked whales —

Mesoplodon longirostris dredged from the ocean bottom. National

Geographic Research 2(1): 47-56.

. Whale worries confirmed. Geotimes 31 (4): 2.

. Review: "General Features of the Paleobiological Evolution of

Cetacea," by G. A. Mchedlidze. Quarterly Review of Biology 61:

249-250.

1987. Cetacea from the Sahabi Formation. Libya. Pp. 145-152 in N. T.

Boaz et al. (eds.). Neogene Paleontology and Geology of Sahabi.

Liss, New York, New York.

— . Review: "Mammal Evolution: An Illustrated Guide." by R. J.

G. Savage. AAAS Science Books and Films 22 (4): 232-233.

— . (with R. N. Oldale and J. R. Grimes). Elephant teeth from the

western Gulf of Maine, and their implications. National Geographic

Research 3 (4): 439-146.

. A delphinoid ear bone from the Dam Formation (Miocene) of

Saudi Arabia. Pp. 447—450 in P. J. Wybrow (ed.). Miocene geology

and paleontology of Ad Dabtiyah, Saudi Arabia: Bulletin of the

British Museum (Natural History) (Geology) 41: 447-450.

. (with R. N. Oldale and J. R. Grimes). Late-Wisconsinan el-

ephant teeth from the western Gulf of Maine. Geological Society of

America, Abstracts with Programs, p. 794.

-. Edited: "Fossil Cetacea of the Caucasus," by G. A. Mchedlidze.

Translation published by Smithsonian Institution Press. Washing-

ton. D. C.

1990. Review: "Frozen Fauna of the Mammoth Steppe: The Story of

Blue Babe." by R. D. Guthrie. AAAS Science Books and Films 26

(2): 111).

The Early Miocene Littoral Ursoid Carnivoran Kolponomos:

Systematics and Mode of Life

Richard H. Tedford

Department of Vertebrate Paleontology, American Museum of Natural History, Central Park West at 79th Street,

New York, New York 10024

Lawrence G. Barnes

Vertebrate Paleontology Section, Natural History Museum of Los Angeles County, WO Exposition Boulevard,

Los Angeles, California 90007

Clayton E. Ray

Department of Paleobiology, United States National Museum of Natural History, Smithsonian Institution.

Washington. D.C. 20560

ABSTRACT. — Species of the large extinct early Miocene carnivoran Kolponomos Stirton. 1960. are known from a few fossils found in marine

rocks along the northeastern margin of the Pacific Ocean in Oregon and Washington, U.S.A. These animals are notable for their massive skulls with

markedly deflected rostra and broad, crushing cheek teeth like those of a sea otter. Originally based on an incompletely preserved snout from the

marine lower Miocene Clallam Formation at Clallam Bay. Clallam County. Washington, and questionably assigned by Stirton to the Procyonidae,

the taxon has until recently remained enigmatic and not certainly assigned to any particular carnivoran family. Additional specimens from the type

locality, including a nearly complete cranium with some teeth, provide new data on the cranial morphology of the species. Another specimen.

consisting of a nearly complete cranium, mandible with dentition, and some postcranial bones, from the lower Miocene Nye Mudstone on the

Oregon Coast, represents a new species. Kolponomos newportensis.

The new material demonstrates that Kolponomos is an ursoid most closely related to members of the paraphyletic family Amphicynodontidae.

Similar phylogenetic roots have been postulated for the pinnipeds as a whole, and cladistic analysis implies a sister-taxon relationship of

Kolponomos with the Pinnipedimorpha. The few postcranial bones available demonstrate that Kolponomos was amphibious but not a strong

swimmer.

Kolponomos was probably littoral in distribution, all specimens having been discovered in nearshore marine rocks. The crushing cheek teeth

would have been suited to a diet of hard-shelled marine invertebrates. The anteriorly directed eyes and narrow snout indicate that Kolponomos could

view objects directly in front of its head, of benefit to an animal that would selectively eat epifaunal marine invertebrates. The elongated upper canine

and third incisor teeth clustered in thickened bone at the anterior end of the down-turned snout and the posteriorly retracted nasal opening are

adaptations that would allow the animal to pry organisms from rocks while keeping its nostrils away from the substrate. Large paroccipital and

mastoid processes indicate strong neck muscles that could provide powerful downward movements of the head. These features indicate that

Kolponomos probably fed on marine invertebrates living on rocky substrates, prying them off with the incisors and canines, crushing their shells, and

extracting the soft parts, as do sea otters.

Kolponomos represents an unique aquatic adaptation for marine carnivorans, whose mode of living and ecological niche are approached only by

modem sea otters.

INTRODUCTION

all known specimens of Kolponomos, to redescribe and rediagnose

The early Miocene carnivoran genus Kolponomos Stirton, 1960, K. clallamensis. to describe a new species from Oregon, to corn-

is known from a few fossils found in marine rocks along the ment on the relationships and taxonomy of the genus, and to discuss

northeastern margin of the Pacific Ocean in Oregon and Washing- implications for its functional morphology and behavior,

ton. Kolponomos was originally based on an incompletely pre-

served snout from the marine lower Miocene Clallam Formation at

Clallam Bay, Clallam County. Washington. The relationships and

morphology of the type species of the genus. K. clallamensis All specimens were in hard concretionary sandstone matrix.

Stirton. 1960, have remained problematic, and for many years the Those from Washington were prepared by use of pneumatic chisels

animal was not assigned with certainty to any particular carnivoran and formic acid; those from Oregon were prepared by mechanical

family. Stirton (I960) questionably assigned it to the Procyonidae, and air abrasive methods.

and was followed by Piveteau ( 1961 ), Romer ( 1966). and Thenius Geologic ages cited herein are modified according to the re-

(1969). Carroll (1988) classified the taxon as Carnivora, incertae vised radiometric scale of Dalrymple (1979) and the correlations

sedis, and Ray (in Barnes et al. 1985:43) regarded it as an proposed by Armentroutet al. ( 1983). The acronyms for institutions

enaliarctine pinniped. are as follows: AMNH, American Museum of Natural History, New

Additional specimens, including a nearly complete cranium York, New York; BM(NH). British Museum (Natural History),

with some teeth, have now been collected from the Clallam Forma- London, England; LACM. Natural History Museum of Los Angeles

tion near the type locality of Kolponomos clallamensis at Clallam County. Los Angeles, California; UCMP. University of California

Bay. These provide additional data on the cranial morphology of the Museum of Paleontology, Berkeley, California; USNM. National

type species. Barnes et al. (1985) announced the discovery of a Museum of Natural History, Smithsonian Institution. Washington,

nearly complete cranium and mandible with some postcranial bones D.C.

of Kolponomos from the lower Miocene Nye Mudstone on the Casts of the crania have been placed in AMNH, USNM, UCMP,

Oregon coast. This specimen represents a new species of LACM, and the University of Nebraska State Museum. Measure -

Kolponomos. The purpose of this study is to describe and illustrate ments, in millimeters, of the crania, dentitions, and mandible of the

In A. Berta and T. A. Demere (eds.)

Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr.

Proc. San Diego Soc. Nat. Hist. 29:1 1-32, 1994

METHODS AND MATERIALS

12

R. H. Tedford. L. G. Barnes, and C. 1 Raj

species of Kolponomos have been provided in Tables 1 and 2.

Cranial restorations of Kolponomos clallamensis are based on all

available specimens.

SYSTEMATICS

Class Mammalia Linnaeus. 1758

Order Carnivora Bowdich, 1821

Suborder Caniformia Kretzoi, 1943

lnfraorder Arctoidea Flower. 1869

Parvorder Ursida Tedford, 1 976

Superfamily Ursoidea (Gray). 1825

Family Amphicynodontidae (Simpson). 1945

Included genera. — This paraphyletic family includes Amphi-

cynodon Filhol. 1882. Pachycynodon Schlosser. 1887 (including

Paracynodon Schlosser, 1899). Allocyon Merriam, 1930. and

Kolponomos Stirton, 1960.

Kolponomos Stirton. 1960

Kolponomos Stirton, 1960:346.

Kolponomus Carroll. 1988:635. 672.

Emended diagnosis of genus. — Stirton's diagnosis was based

solely on the holotype of K. clallamensis and amounted to a de-

scription of that specimen. New material allows a refinement of that

diagnosis. The characters noted are all derived with respect to other

Carnivora: I3 with large root; I, vestigial or missing; cheek teeth

with strongly inflated principal cusps; P1"3 and P, , with anterior

and posterior cingular cusps, P4 also with prominent posterolingual

cingular cusp; P4 molariform with large protoconc; M' with large

conules. lingual cingulum only between principal cusps; M: mark-

edly smaller than and lying posterolingual to M1 with postero-

lingually placed metaconule; M, quadrate in occlusal outline; M,

triangular with reduced talonid; M, absent.

Facial region of skull markedly flexed downward relative to

basicranial plane; muzzle deep; nasal retracted to above P1 or P:, its

sutural contact with frontal wide, slightly wider anteriorly than

posteriorly; long process of premaxilla nearly meeting correspond-

ing processes from frontal along nasal suture; palate highly vaulted

anteriorly; infraorbital foramen greatly enlarged and opening into

shallow fossa in maxilla; infraorbital canal very short; orbit facing

forward and relatively small; zygomatic arch widely flaring with

strong postorbital process and variably developed masseteric pro-

cess; postorbital process of frontal lacking; lacrimal foramen small,

variably present; sphenopalatine foramen large and closely associ-

ated with posterior palatine foramen; optic foramen small, nearly

same size as ethmoid foramen; anterior process of alisphenoid

forming strut bracing palate against braincase; mastoid process

hypertrophied into long column extending laterally and ventrally;

posterior carotid foramen well anterior to posterior lacerate fora-

men; foramina for venous occipital sinus in foramen magnum;

lambdoidal crest strongly extended posteriorly on either side of

midline.

Type species. — By original designation. Kolponomos clalla-

mensis Stirton, I960.

Included species. — Kolponomos clallamensis Stirton, I960, late

Early Miocene, Washington; Kolponomos newportensis, new spe-

cies. Early Miocene. Oregon.

Kolponomos clallamensis Stirton, 1960

Figures 1-7, 13-14

Kolponomos clallamensis Stirton, 1960:347, figs. 1— t.

Diagnosis of species. — A species of Kolponomos differing from

K newportensis, new species, by the following derived features:

cranium with anterior part of palate more highly vaulted;

infraorbital foramen larger, approximately twice the diameter, open-

ing into prominent fossa; large masseteric process on ventral sur-

face of zygomatic arch at jugal/maxillary suture: maxilla rather

than jugal forming anterior rim of orbit; paroccipital process larger,

oriented vertically rather than posteroventrally; basioccipital nar-

rower, especially posteriorly between posterior lacerate foramina;

narrow, prominent vertical crest on occiput dorsal to foramen mag-

num lacking; zygomatic arch more strongly arched dorsally. In

addition, K. clallamensis is distinguished by the following primi-

tive features: rostrum narrower, with anterolateral margin of snout

around canine and I' not flaring laterally, nearly vertical; hamular

process of pterygoid straighter, smaller, not extending so far ven-

trally: mastoid process shorter, straighter. and oriented vertically

rather than being twisted and projecting anteroventrally beneath the

external auditory meatus; intercondylar notch present and deep.

Holotype. — UCMP 50056. anterior part of cranium with roots

of left I3 and both M"s, collected in 1957 by Mrs. Betty Willison.

Type locality.— UCMP V5761, 250 yards east of Slip Point

lighthouse near section line. NW 1/4 NE~ 1/4. Sec. 21, T. 32N. R.

12W. and LACM 5933, Slip Point, Clallam Bay, Clallam County,

Washington.

Referred specimens. — LACM 1 3 1 148. from the type locality, a

nearly complete cranium with parts of the left I2 and I' and right and

left P: and M1, collected by Albin Zukofsky, II, February 1988;

LACM 123547. from Merrick's Bay. Clallam Co., Washington, a

fragment of tooth (M,7) collected by William Buchanan. May

1983.

Formation and age. — Extensive marine sedimentary deposits

are exposed on the north side of the Olympic Peninsula in Washing-

ton. The lower Miocene Clallam Formation and underlying Eocene

and Oligocene deposits are exposed in wave-cut cliffs and terraces

and in man-made excavations for more than 70 miles (112 km)

along the south shore of the Strait of Juan de Fuca. The stratigraphy

and invertebrate paleontology of this thick sedimentary sequence

are well known (e.g., Addicott 1976a,b.c; Armentrout et al. 1983:

Feldmanetal. 1991; Rau 1964;Snavely 1983; Snavely et al. 1980;

Tabor and Cady 1978). but only a very few fossil vertebrates have

been recorded [Stirton I960 (carnivore); Olson 1980 (bird):

Domning et al. 1986. and Ray et al. 1994, this volume (desmo-

stylian); Barnes 1987, 1989 (whale)]. Specimens of Kolponomos

clallamensis are from the Clallam Formation, deposited during the

Pillarian Molluscan Stage (Addicott 1976a; Moore and Addicott

1987) of the early Miocene. A measured section at Slip Point

(Addicott 1976b: tig. 4) shows that the rocks exposed there belong

to the lower part of the Clallam Formation. The lower part of the

Pillarian Molluscan Stage containing K. clallamensis includes an

interval of time correlative with the late Arikareean North Ameri-

can Mammal Age.

Skull.— The nearly complete referred skull (LACM 131148)

has parts of the left i:and I3 and both P:'s and M''s, but is missing

the right zygoma and the anterolateral margin of the right premax-

illa and maxilla, much of the ascending ramus of the right maxilla,

the right nasal, the dorsal surface of the interorbital region, the

sagittal crest, much of the roof of the braincase on the right side, and

the lambdoidal crest (Figs. 1-4). Structures of the right orbit are

fully exposed.

The Early Miocene Littoral Cirsoid Carnivoran Kolponomos: Systematics and Mode of Life

13

The skull of Kolponomos clallamensis is roughly triangular in

dorsal aspect, with a broad occipital region and a narrow snout

(Figs. 5-7). The dorsal profile is arched, and the zygomatic arches

are prominent. At its anterior extremity, the rostrum is thick and

ventrally deflected. In anterior view, the narial opening is elongated

dorsoventrally and tapered ventrally. A considerable thickness of

the premaxillae anterior to the narial opening separates the incisors

from the anterior margin of the naris. There is a rather prominent

vertically oriented premaxillary protuberance. On either side of it.

the anterior surface of the premaxilla slopes abruptly antero-

ventrally. On either side in this area, the distal part of the root for I3

forms a bulge in the premaxilla. Otherwise, the bone surface in this

area is depressed. The vertical premaxillary eminence extends

posterodorsally and is continuous with the relatively narrow and

sharp margins of the naris. Immediately anterior to the naris, the

bone surface is rugose and punctured by many small foramina.

The maxilla— premaxilla suture is fused and obliterated anteri-

orly on both the holotype and the referred cranium, but on both

specimens the sutures bounding the premaxilla lateral to the naris

and the nasal bone are discernable. The ascending process of the

premaxilla extends posteriorly to about mid-length on the nasal. It

does not meet the anterior process of the frontal as in living bears

but stops approximately 2 to 3 mm from the frontal. Along the

lateral margin of the naris. the premaxilla forms a nearly vertical

lateral surface. It is almost horizontal adjacent to the nasal bone,

however. The lateral surface of the snout is dorsoventrally high and

flat. Between the canine and the zygomatic arch it is concave.

There is a nasolabials fossa on the dorsal part of the ascending

ramus of the maxilla immediately anterior to the orbital margin.

This fossa is broad and shallow, and is bordered anterodorsally by a

slight protuberance, dorsally by a faint horizontal ridge, and poste-

riorly by the orbital margin, which has a vertically elongated

antorbital process. The infraorbital foramen has a large anterior

opening. 18 to 21 mm high by 10 to 11 mm wide. The foramen

opens into a broad fossa, strongly emarginated ventrally. and

extending anteriorly nearly to the P2.

The nasal bones are elongated, nearly parallel-sided, with

rounded posterior borders. Where they join posteriorly they are not

separated by the frontals. They slope anteroventrally and are nearly

flat transversely. On the holotype they are approximately 74 mm

long; on the referred cranium, although incomplete, the left nasal is

69 mm long as preserved. The nasals are narrowest just anterior to

their mid-length and are wider both anteriorly and posteriorly. Their

anterior margin is thick and slightly up-turned, especially near the

sagittal line. They expand laterally at the anterior margin.

The posterior part of the nasals is at the highest part of the

cranium, which has a smooth, domelike profile. The low supraor-

bital ridges of the frontals extend posterodorsomedially from just

lateral to the posterior ends of the nasals. Posteriorly the interorbital

constriction tapers uniformly toward the braincase, to its narrowest

point in the intertemporal region. The top surface of the interorbital

region on the referred cranium is weathered away, but on the holo-

type cranium a low crest extends posteromedially from each su-

praorbital process toward the sagittal plane. These crests lap anteri-

orly onto the frontals, and where they merge posteriorly on the mid-

line they form a slight V-shaped sulcus. Posteriorly from this point

there is a low, broad sagittal ridge on the holotype, but the cranium is

missing posterior to the intertemporal region. The mid-sagittal re-

gion posterior to this area is broken on the referred cranium also.

The braincase is elongated and tapers anteriorly toward the

interorbital constriction. Its dorsal surface slopes laterally toward the

temporal fossa and flares posterodorsolaterally to the nuchal crest,

which is, however, almost entirely lost on the referred cranium. The

surface of the bone is slightly undulating but not strongly rugose or

pitted. The frontal-parietal suture ascends the anterolateral wall of

the braincase from the back of the orbital region, approaches the

midline, then bends posteriorly over the dorsal surface of the brain-

case to extend posteromedially toward the midline. The squamosal

fossa, forming the floor of the temporal fossa over the squamosal, is

broad and shallow, does not slope anteriorly, and extends posteriorly

into a broad sulcus in the lateral part of the nuchal crest.

The occipital shield is in the shape of a broad isosceles triangle,

with the apical part broken away. A broad median crest extends

dorsally from the foramen magnum toward the nuchal crest. This

crest is flanked by a pair of broad fossae. In turn, each fossa is

flanked laterally by a broad eminence, preserved on the left side and

in part on the right of the referred cranium, that extends dorso-

lateral^ to merge with the posterior side of the nuchal crest. From

this point, the nuchal crest becomes narrower ventrolaterally and

curls posteriorly in the area posterior to the temporal fossa. Dorsal

to each condyle is a prominent transversely oriented fossa that is

continuous with a large fossa lying dorsal to the paroccipital pro-

cess and below the nuchal crest. The paroccipital process curves

posteroventrally almost as far ventrally as do the occipital condyles.

It has a vertical posterior swelling that extends dorsally into the

lateral fossa. The foramen magnum is wide and compressed dorso-

ventrally. Its dorsal margin is a broad arch, only slightly peaked

medially.

The occipital condyles are relatively small, canted dorso-

lateral^", with sharp edges around the lateral and ventral margins of

the articular facets. The medial side of each is slightly excavated

and has a small condyloid foramen. The condyles are separated

ventrally by a broad U-shaped intercondylar notch. Their articular

surfaces are not continuous ventrally but separated by a fossa that is

a continuation of the intercondylar notch and aligned antero-

posteriorly with the median ridge on the basioccipital. The condyles

are positioned relatively ventrally in relationship to the basi-

cranium. At the anterior margin of each condyle, immediately

posterior to the hypoglossal foramen, is a recess in the margin of the

articular surface.

The orbit is of rather small diameter, measuring approximately

33 mm transversely in the referred cranium. At the anterior margin of

the orbit, immediately ventral to the antorbital process, is a small (3

mm diameter) lacrimal foramen. The anterior part of the orbit has a

convex medial wall that protrudes into the orbit in the area posterior

to the infraorbital foramen. Immediately posterior to this area is a

large (9 mm diameter), round sphenopalatine foramen that is imme-

diately dorsal to and very close to the slightly smaller (6 mm) orbital

aperture of the posterior palatine foramen. The tract for the optic

nerve is elongated and relatively deep, leading posteroventrally from

the ethmoidal to the optic foramen and into the orbital fissure; these

foramina are approximately equal in size. The anterior lacerate

foramen ( orbital fissure) and foramen rotundum lie in a large fossa as

in pinnipeds, although a partition of bone separates them. As in other

caniform carnivorans, the anterior aperture of the alisphenoid canal

opens into the ventral wall of the anterior lacerate foramen and does

not have a separate opening into the orbit.

The zygomatic arch is stout and curves uniformly outward from

the orbit. The zygomatic process of the squamosal increases only

slightly in thickness posteriorly, and the posterior extremity of the

jugal diminishes equally in diameter as it extends posteriorly. The

zygomatic arch is not straight in its middle but curves both laterally

and dorsally. At its anterior end. it is massive, forming thick dorsal

and ventral margins around the infraorbital foramen. The ventral

root of the zygomatic arch is very stout and descends vertically to

form a buttress dorsal to the M'. This buttress forms a vertical

component to the anterior end of the zygomatic arch that continues

dorsally through the dorsal margin of the infraorbital foramen.

14

R. H. Tedford, L. G. Barnes, and C. E. Ray

B

t \

\

5cm

-

'W

"■m**

Figure 1 . Lateral views of the crania of species of Kolponomos. A, K. clallamensis Stirton, 1 460. referred, LACM 1311 48, left side; B, K. clallamensis,

holotype, UCMP 50056, left and right sides; C, K. newportensis n. sp„ holotype, USNM 2 1 5070, right side reversed to left for comparison. All specimens

to same scale.

From this point, the zygomatic arch flares posterolateral^ to form

the lateral margin of the orbit. The jugal bone extends

anteromedially over the maxilla, forming a partially mortised joint.

The jugal does not form the anterior rim of the orbit. Ventral to the

orbit, the maxilla flares where it meets the jugal, and together they

form a prominent masseteric process. This process projects

ventrolateral^, and a crest extends posteriorly from it along the

ventrolateral border of the jugal. The postorbital process of the

jugal is stout, broad anteroposteriorly, and its apex is located anteri-

orly. The anterior extremity of the zygomatic process of the squa-

mosal abuts the posterior side of the postorbital process and is

dorsoventrally expanded and slightly up-turned. From this point the

zygomatic process curves uniformly posteroventrally to the glenoid

fossa. At the glenoid fossa, the zygomatic process curves medially

to form the dorsal surface of the glenoid fossa.

The glenoid fossa is elongated transversely, narrow anteropos-

teriorly. and has a smoothly curved articular surface. In contrast to

that of typical Ursinae. the glenoid fossa is situated in a plane dorsal

to the basioccipital plane. In ursines. the glenoid fossa is ventral to

the plane of the basioccipital. As is typical of the Ursidae. the

postglenoid process is well developed medially, projecting antero-

ventrally to the glenoid fossa, and diminishes laterally. The

postglenoid process is thin anteroposteriorly and does not form an

anteroposteriorly thickened buttress as in the Ursinae. There is a

low preglenoid process laterally.

On the dorsal surface of the zygomatic process of the squamosal.

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematics and Mode of Life

15

Figure 2. Outline drawings of restored crania of Kolponomos species viewed from the left and oriented so that the hasicranial plane is horizontal. A. K.

clallamensis Stirton. 1 960. referred. LACM 131 148; B. /C. clallamensis. holotype; UCMP 50056; C. K. newponensis n. sp„ holotype, USNM 2 1 5070 with

tooth row restored. All drawings to same scale. Symbols for anatomical features: ac, alisphenoid canal (posterior aperture!; earn, external acoustic meatus;

fio. infraorbital foramen; fl, lacrimal foramen; fla. anterior lacerate foramen; fo, foramen ovale; fop, optic foramen; Fr, frontal; fr, foramen rotundum; fs,

sphenopalatine foramen; Ju. jugal; mp, mastoid process; Mx. maxilla; Na, nasal; nf. nasolabialis fossa; Pa. parietal; Pmx. premaxilla; pp. paroccipital ( =

jugular) process; Sq. squamosal.

16

R. H. Tedford, L. G. Barnes, and C. E Ray

V

I

B

Figure 3. Ventral views of the crania of species of Kolponomos. A, K. clallamensis Stirton. 1 960. referred. LACM 1 3 1 1 48; B. A', clallamensis, holotype.

UCMP 50056; C. K. newportensis n. sp.. holotype, USNM 21 5070. AM specimens to same scale.

where the zygomatic arch meets the squamosal fossa, there is a

prominent tuberosity. This tuberosity also is at the anterolateral edge

of the shelf dorsal to the external auditory meatus. This shelf slopes

posteroventrally and expands dorsoventrally as it merges with the

base of the mastoid process.

The palate is elongated and concave both anteroposteriorly and

transversely. On either side of the midline the palate is subplanar,

essentially flat transversely and gently arched anteroposteriorly.

from the incisive foramina to the palatal notch. Anteriorly and

laterally the palate descends abruptly to the inner margin of the

dental arcade. The anterior end of the palate is deflected ventrally.

so that the alveolar margins of the canines and incisors are posi-

tioned more ventrally than those of the cheek teeth.

The septum separating the incisive foramina is continuous pos-

teriorly with a slight raised ridge extending more than 30 mm

posteriorly along the midline suture of the palate. These foramina

are large and reniform. On either side of the palate, at the posterolat-

eral corner just medial to the M: alveoli, are the posterior palatine

foramina, closely associated with the maxillo-palatine suture.

These foramina are variable in size and number. All are large on the

holotype. On the referred cranium, the anterior foramina are both of

intermediate size, while the posterior foramina are of different

17

5cm

occ

ten

Figure 4. Outline drawings of restored crania of Kolponomos species, viewed ventrally. A, K. clallamensis Stirton, 1960. referred, LACM 131148; B./C.

clallamensis, holotype, UCMP 50056; C, K. newportensis n. sp., holotype, USNM 215070, with tooth row restored. All drawings to same scale. Symbols for

anatomical features: Bo, basioccipital; Bs, basisphenoid; cc. carotid canal; earn, external acoustic meatus; fh. hypoglossal foramen; fi. incisive foramen ( =

palatine fissure); flp. posterior lacerate foramen; fpal, palatine foramen; fsm. stylomastoid foramen; hf, tympanohyal pit (= hyoid fossa); mp, mastoid

process; Mx, maxilla; occ, occipital condyle; Pal, palatine; pp. paroccipital (= jugular) process; Ps, presphenoid; Pt, pterygoid; tec, ectotympanic; ten,

entotympanic.

sizes, intermediate on the left and small on the right. On both crania

the anterior foramen is continuous with a prominent antero-

posteriorly elongated sulcus that extends anteriorly to a point where

it disappears medial to the P\ Posterior to the palatine foramina and

the M2 the posterolateral palatal margin is formed by a narrow

vertically oriented crest pierced by a foramen. This crest is continu-

ous posteriorly with the sharply keeled ventral border of the ptery-

goid hamulus. On the referred cranium, the hamulus is very thin

transversely and concave laterally, as is typical of the Ursinae, but is

bent sharply ventrally. The narrow posterior process of the ptery-

goid hamulus extends posteriorly. The main part of the pterygoid-

palatine strut ascends posteriorly, to join the basicranium around

the posterior aperture of the alisphenoid canal. The lateral surface

of the strut is rounded and convex and continues onto the ventrolat-

18

R. H. Tedford. L. G. Barnes, and C. E. Ray

r:

w

■*»

B

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematica and Mode of Life

19

-

B

B

5cm

5 cm

Figure 6. Anterior views of the crania of Kolponomos species. A, K.

clallamensis Stirton, 1960, referred, LACM 131148; B, K. clallamensis,

holotype, UCMP 50056; C, K. newportensis n. sp., holotype, USNM

215070. All specimens to same scale.

eral surface of the braincase wall. From the pterygoid hamulus, a

fine ridge extends posteriorly along the medial side of the posterior

aperture of the alisphenoid canal and the foramen ovale and contin-

ues into the auditory tube of the bulla.

The internal narial opening is highly arched and wide. The

palatal notch is broadly rounded in the referred specimen, has a

slightly acute apex in the holotype, and extends anteriorly almost to

a point between the centers of the M:'s. On either side of the

internal narial opening, the palatine-pterygoid struts sweep medi-

ally to form a sharp, underhanging border. The roof of the internal

narial opening ascends anteriorly and in its anterior part has a

medial keel formed by the vomer and presphenoid. The pre-

sphenoid-basisphenoid suture is transversely oriented and is not

Figure 7. Posterior views of the crania of Kolponomos species. A, K.

clallamensis Stirton, 1960. referred, LACM 131148; B, K. clallamensis,

holotype, UCMP 50056; C, K. newportensis n. sp., holotype, USNM

215070. All specimens to same scale.

coossified. The basisphenoid is nearly flat where it forms the roof

of the internal naris between the pterygoid hamulae. It expands

posteriorly and at its lateral edge, dorsal to the pterygoid hamulus,

bears a groove that extends posterolaterally into the median lacerate

foramen. The median lacerate foramen is elongated antero-

posteriorly and is situated at the anteromedial corner of the bulla.

The basioccipital-basisphenoid suture is fused, and its precise

location is not visible. The basioccipital has a median crest that

widens posteriorly and spreads toward each condyle. On either side

are a curved fossa and a rugosity that mark the insertion of the

rectus capitis ventralis muscles. Posterolateral to the fossa, between

the condyle and the bulla, is the hypoglossal foramen, which is

transversely oval and approximately 3 mm in diameter.

The tympanic bulla is small and has a rugose ventral surface. It

is fused laterally to the squamosal and the base of the mastoid

Figure 5. Dorsal views of the crania of Kolponomos species. A, K. clallamensis Stirton, 1960, referred, LACM 131 148; B. K. clallamensis. holotype,

UCMP 50056; C, K. newportensis n. sp., holotype. USNM 215070. All specimens to same scale.

20

R. H. Tedford. L. G. Barnes, and C. E. Ray

process. Posteromedially it is separated from the basioccipital by a

sulcus. Anterolaterally it expands posleroventral to the medial part

of the glenoid fossa and is broadly appressed to the posteromedial

part of the postglenoid process. An oblique crest on the ventral

surface of the bulla that extends from the stylomastoid foramen to

the anteromedial margin appears to mark the junction between the

entotympanic and the ectotympanic. If this is true, the entotympanic

contribution to the tympanic bulla is approximately equal to that of

the ectotympanic.

There appears to be a small postglenoid foramen located in a

fissure where the medial part of the postglenoid process is over-

lapped by the bulla. The ventral surface of the bulla is retracted

posteriorly at the anteromedial corner ventral to the median lacerate

foramen. The posterior lacerate foramen lies at the posteromedial

corner of the bulla and is semicircular and positioned obliquely. The

external auditory meatus is round, approximately 4 mm in diameter,

and recessed far beneath a wide shelf formed by the squamosal.

The mastoid process is very long, extending outward from the

cranium variably 38 and 44 mm on either side, measured from the

notch where it joins the paroccipital process. It projects

anteroventrolaterally from the basicranium. The process is basi-

cally three-sided; one flattened surface medial, one anterior, and

one posterior. The medial surface is concave in contrast to the other

two, which are slightly convex. At its distal end, the mastoid

process is compressed anteroposteriorly. The concave medial

surface expands proximally toward the basicranium and is confluent

with a large recess surrounding the hyoid fossa. This same recess

extends onto the anterolateral surface of the paroccipital process,

which is at this place deeply excavated. The paroccipital process

projects posteroventrolaterally from the basicranium and is com-

pressed transversely. Its anteroventral margin is a crest that extends

anteromedially toward the bulla, reaching the posterior side of the

bulla between the stylomastoid foramen and the posterior lacerate

foramen.

As in the Ursinae. the hyoid fossa is separated from the poste-

rior lacerate foramen by a ridge of bone. The hyoid fossa sits within

a deep recess. In ursines, the hyoid fossa is widely separated from

the external auditory meatus by a wide expanse of the bulla. In

Kolponomos clallamensis, however, the hyoid fossa is very close to

the external auditory meatus. Also, in K. clallamensis the

stylomastoid foramen lies midway between the hyoid fossa and the

external auditory meatus, whereas in the Ursinae the stylomastoid

foramen is widely separated from the external acoustic meatus and

is within the recess that houses the hyoid fossa.

Dentition. — The upper dentition consists of l'~\ canine, P1 4,

and M1 2. The actual teeth present in the referred cranium are the

roots of the left I2"-' and the complete left and right P:'s and M''s. On

the left side, all alveolar margins are preserved, and it is that side

that forms the basis for the following description.

The incisors and canines are clustered, without significant di-

astemata, in the thickened and downturned anterior end of the

snout. The incisors are aligned transversely anterior to the canines.

I1 and I2 are small and have transversely compressed roots. I1 is

smaller than I2, and both teeth are implanted essentially vertically in

the palate. The I^'s are much larger, being approximately four times

the diameter of I2 at the alveolar rim. Unlike the medial incisors, the

P's are procumbent and deeply rooted in the premaxilla between

the canine and the incisive foramen. The left I1 measures 18.2 mm

anteroposteriorly and 12.4 mm transversely at the alveolar rim. A

diastema of 4 mm separates the alveolus for the left I1 from that of

the upper canine. The canine alveolus is oval and measures 20 mm

anteroposteriorly and 17 mm transversely at the alveolar rim. The

bulge in the lateral side of the rostrum indicates that the root for the

canine is extremely long and extends nearly as far into the rostrum

as the lateral edge of the nasal bone. The root is procumbent.

The cheek-tooth row curves laterally from the canine posteri-

orly to M'. then M2 is positioned more medially. P1 has a single root

that is round in cross-section, procumbent, and closely appressed to

the posterior side of the canine alveolus. The alveolus indicates that

the root was tapered, approximately 7 mm in diameter at the alveo-

lar margin, and extended for approximately 1 5 mm into the maxilla.

P2 is a robust tooth, with two roots and a large smooth crown.

The tooth is oriented obliquely to the axis of the cheek-tooth row.

the anterior root medial to the axis. The posterior root is on the axis

of the tooth row and is approximately twice the diameter of the

anterior root. The tooth tilts medially into the oral cavity. The crown

of this tooth has a flat apical wear facet on the principal cusp. A

smooth cingulum borders the lingual side of the crown from the

anterior to the posterior border of the tooth. The posterior part of the

crown is formed into a talon lying between the principal cusp and

the posterior end of the cingulum.

P1 had roots aligned in the axis of the cheek-tooth row. Judged

by the size of the alveoli, the two roots of this tooth were more

nearly equal in size than those of P2, the posterior one being only

slightly larger in diameter than the anterior one. The maxilla

projects ventrally, forming a crest of bone between the two roots of

this tooth.

P4 was a large tooth, nearly as large as M1. and had three roots.

Of these, the medial (protocone) root is the largest, being more than

twice the depth and diameter of either of the lateral (paracone and

metacone) roots. Between P3 and P4, the lateral margin of the

maxilla begins a strong lateral bend, so that P4 is oriented obliquely

to the cheek-tooth row. A diastema of approximately 3 mm sepa-

rates the anterior (paracone) root of P4 from the posterior root of P\

M1 is a massive tooth. It appears to have been approximately

30% larger than P4 in surface area. It is greatly expanded trans-

versely and is triangular in occlusal view. It has five mammiform

cusps of nearly equal sizes. The lateral two cusps are the paracone

and metacone; the most medial cusp forms the apex of the triangle

and is the protocone; intermediate cusps are the para- and

metaconules. The surface of each M' shows extensive wear that

breaks through the enamel to expose the inner dentinal core. There

is a labial cingulum between the para- and metacones, but other

cingula are lacking.

The M2 alveoli indicate that this tooth was approximately half

the size of the M1. M2 is positioned lingually opposite the talon of

M'. Like both P4 and M1, M2 had three roots. Of these, the

anterolateral (paracone) root was broadly joined with the medial

(protocone) root to form a transversely oriented bilobate root. The

posterior root is the metacone root and is rotated so that it is actually

the most medial root of the tooth.

Kolponomos newportensis, new species

Figures 1-12

Kolponomos clallamensis Stilton, 1960. Barnes etal.. 1985:43, figs. 9a. b.

Diagnosis. — A species of Kolponomos differing from K. clalla-

mensis in the following derived cranial features: broad muzzle

flaring laterally above canines and incisors; mastoid process twisted

clockwise, extending forward beneath external auditory meatus as

far as postglenoid process; intercondylar notch lacking so that the

articular surfaces of the occipital condyles are continuous ventrally.

In addition. K. newportensis is distinguished from K. clallamensis

by the following primitive features: infraorbital foramen smaller

and lacking marked excavation of maxilla anterior to it; prominent

masseteric process of maxilla lacking and masseteric process on

jugal reduced; jugal forming anterior rim of orbit; paroccipital

process smaller and less downwardly pointing; paroccipital process

lacking hyoid fossa.

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematic* and Mode of Lite

21

Holotype. — USNM 215070, cranium lacking parts of dorsal

surface with only right and left P: in situ; mandible lacking tips of

coronoid processes and lacking incisors and right P,; isolated asso-

ciated right I3 and C. right P1. left P1, left P\ right P4. left M'. and

right and left M:. Recovered from the same concretion were the

axis, third cervical vertebra, a broken proximal lumbar vertebra, a

sternebra, a proximal part of an anterior rib, a thyrohyal lacking the

proximal end, a complete '.'ceratohyal, a metapodial lacking the

distal end and half of the proximal end, a median phalanx, and

unidentifiable bone fragments. The original half of the concretion

containing the occipital part of the skull and most postcranial

elements was collected by Douglas R. Emlong, October 1969

(Emlong field no. 603). On 26 January 1976 Emlong found the

remainder of the concretion, containing the balance of the skull,

mandible, and isolated teeth (Emlong field no. E76-B). recognizing

it as associated and pertaining to Kolponomos.

Type locality. — A concretion found on the beach at low tide

line, approximately 300 yards (274 m) south of the mouth of Big

Creek and 100 yards (91 m) seaward of the sea cliff just north of

Newport, Lincoln County. Oregon.

Formation and age. — Lower part of the Nye Mudstone, repre-

senting the early Pillarian Molluscan Stage ( Addicott 1976a; Moore

and Addicott 1987). correlative with the Late Arikareean Land

Mammal Age. early Miocene.

Etymology. — Named for the town near the type locality to

record the occurrence of the type species in a manner similar to that

for the genoholotype.

Skull.— The skull of USNM 215070 lacks most of the dorsal

surface and is toothless except for both P:'s. which are crushed into

their alveoli and forward into those for P1. The concretion enclosing

the specimen was subspherical and aproximately 27 cm in diameter.

Prior to its consolidation, all upper teeth except left and right P: had

fallen out. In the course of gross preparation these were found in a

tight cluster beneath the palate and between the horizontal rami of

the mandible. Also prior to consolidation, the mandible had slipped

out of articulation and moved forward and upward forcibly, coming

to rest in a symmetrical undershot false occlusal position, undoubt-

edly causing the anterior displacement of left and right P2 and

severely crushing and comminuting the thin alveolar walls of left

and right M: and M1 and, to a lesser extent, the alveolar margins of

the upper premolars. The apices of the coronoid processes were

removed by abrasion of the smooth surface of the concretion as was

much of the dorsal surface of the skull. The tightly appressed

mandible was painstakingly separated from the skull and the tightly

clustered isolated teeth were extracted by Gladwyn B. Sullivan in

1976 in the course of gross preparation of the specimen. This

specimen represents an old individual as judged by its heavily worn

teeth and advanced cranial fusion. Despite the latter, it is possible to

make out many sutures, particularly in the orbital region.

A striking major feature of the skull, in common with all re-

mains of the genus, is the flexure of the facial part of the skull

relative to the basicranial plane (Fig. 2). Measured as the angle

between the palate and basisphenoid, the flexure is approximately

155° in USNM 215070. The widely flaring zygomatic arches, the

forward-oriented orbits, and the great hypertrophy of the mastoid

processes are also distinctive features of the remarkable skulls of

Kolponomos,

In USNM 2 1 5070 the muzzle is nearly as broad at the carnassial

as the palate. The large anteroposteriorly elongated incisive fo-

ramina lie in the trough of the strongly arched palate with distinct

grooves extending anteriorly from them nearly to the incisor al-

veoli. The interforamen septum forms a low S-curve anteriorly.

Posteriorly the vaulted palate has strong anteroposteriorly oriented

depressions along either side of the flattened medial part of the

palate, becoming progressively deeper posteriorly and leading into

the anterior palatine foramina at the maxillary-palatine suture adja-

cent to the anterior root of M:. Behind that a series of pits extends

the posterior palatine groove to a foramen that penetrates the thin

rim of the palatine portion of the palate.

The pterygoid hamuli are arcuate in palatal view and terminate

in dorsoventrally flattened processes. Sutures with bones surround-

ing the pterygoid are too coossified for the precise outline of this

element to be determined. The anterior or pterygoid process of the

alisphenoid is defined by its suture with the palatine; with the

palatine it forms a strong strut bracing the back of the palate against

the braincase. Ventral to this strut a depression for the origin of the

pterygoid muscle extends downward toward the hamular process.

The posterior end of the large alisphenoid canal penetrates the base

of the strut. This opening is closely followed by the foramen ovale,

which lies in a common pit with the opening of the canal on the left

side of USNM 215070, but on the right a groove joins these orifices

as in most arctoids. The dorsal and posterior sutures of the

alisphenoid with adjacent bones are closed.

The basisphenoid and basioccipital bones are strongly

coossified and the sutures between them are eliminated. They form

a trapezoidal figure with its base lying across the rectus capitis

insertions just anterior to the posterior lacerate foramina. Thus the

basioccipital is broadest across the rectus insertions where the

winglike lateral processes of this bone overlap the medial edge of

the petrosal and presumably floor the large tract for the inferior

petrosal sinus. The knoblike processes for the rectus are situated

bilaterally at the posterolateral corners of ovoid shallow depres-

sions for muscle insertion that presumably extend forward onto the

basisphenoid and medially to a low crest at the midline. The

exoccipital is solidly fused with surrounding elements, except for

its irregular contact with the posterior end of the bulla (caudal

entotympanic). This bone contains the hypoglossal foramen, which

is situated posteromedial to the posterior lacerate foramen as well

as the posterior rim of the latter opening. Presumably the exoccipital

also forms the medial wall and spine of the paroccipital process.

The occipital condyles protrude slightly posterior to the nuchal

crest. There is no intercondylar notch as the condyles are conjoined

ventrally, uniting their articular surfaces. A shelflike extension of

the floor of the foramen magnum extends posteromedially beyond

these articular surfaces. Inside the foramen magnum the paired

posterior openings of the hypoglossal foramina can be seen on its

floor. An additional pair of foramina lying on the lateral wall of the

foramen magnum at the level of the dorsal part of the condyles

presumably accommodated venous drainage for sinuses within the

occiput.

The auditory region is very small and nestled deeply within the

ventrally projecting elements surrounding it, particularly the greatly

hypertrophied mastoid process. The bulla is uninflated and exten-

sively coossified with surrounding elements; nevertheless, most of

its outline as well as its composition can be determined from

bilaterally symmetrical suture traces and rugose coossification

tracts. These observations indicate that the ectotympanic lacks an

ossified meatal tube; its anterior limb spreads over the postero-

medial surface of the postglenoid. forming the posterior wall of the

slitlike postglenoid foramen. Anteromedially the ectotympanic

overlaps the entotympanic and coossifies laterally with the

alisphenoid behind the foramen ovale. A styloid process of the

ectotympanic lies beneath the opening for the eustachian tube. The

posterior limb of the ectotympanic is fused to the base of the

mastoid process. Behind this union, the stylomastoid foramen

emerges from the mastoid. The facial canal is thus separated from

the large pit for the tympanohyal that opens above a prominent

hyoid process of the entotympanic at the posterolateral corner of the

bulla. There is no large hyoid fossa excavated into the anterior wall

of the paroccipital process as in K. clallamensis.

22

R. H. Tedford, L. G. Barnes, and C. E. Ray

The entotympanic is irregularly exposed along the medial edge

of the bulla because of variable overlap of the ectotympanic; poste-

riorly the caudal element is sutured to a process from the exoccipital

and posterolateral^ to a process from the mastoid. There is a large

posterior opening of the carotid canal well anterior to the posterior

lacerate foramen. This opening is formed medially by the basioc-

cipital wing and laterally by the entotympanic, but anteriorly the

arterial tube is nearly completely surrounded by the entotympanic.

The anterior carotid foramen lies medial to the styloid process of

the ectotympanic and opens into a groove in the basisphenoid

anterior to the median lacerate foramen. From this groove the artery

must loop posteriorly to enter the median lacerate foramen and/or

the presumed channel in the basioccipital that accommodates the

inferior petrosal sinus.

The large mastoid process takes the form of an anteroposteriorly

flattened column bending outward and downward from its base and

curving forward at its tip to pass nearly under the postglenoid

process (Fig. 7C). Its components are totally coossified. but the

nuchal crest extends along the lateral surface of the process, thus

marking the position of the mastoid-squamosal suture and indicat-

ing that the process is composed about equally of the two elements.

The process terminates in a raised ovoid area that lies within the

suture. This element may represent the secondary ossification cen-

ter (epiphysis) frequently observed at the tip of the mastoid process

in adult ursids. A ridge arises from the posteroproximal surface of

the mastoid process and passes upward and posteriorly to join the

paroccipital process. The paroccipital process curves postero-

ventrally and terminates in a sharp point.

The supraoccipital bones are solidly coossified with surround-

ing elements. They are concave and highly rugose beneath the

nuchal crest; a thin ridge is present sagittally. At their lateral ex-

tremities a shallow pit is present dorsal to the base of the

paroccipital processes. The lateral processes of the nuchal crest

extend behind the inion and beyond the occipital condyles; conse-

quently, the crest has a broad inflection at the midline. The parietal

and squamosal bones are coossified. but their junction is probably

marked by the bilaterally symmetrical collapse of the braincase

wall under lithostatic load. Breakage dorsally has removed most of

the sagittal crest of the parietal, but the remaining evidence indi-

cates the presence of at least a low crest. The squamosal apparently

makes a significant contribution to the anterior part of the mastoid

process. Its glenoid fossa forms a cylindrical articulation nearly at

right angles to the basicranial axis. Prominent recurved postglenoid

processes are present, deepest medially, and the anterolateral part of

the articular surface is bordered by a low preglenoid process. Most

of the squamosal-jugal suture is visible; the squamosal contribution

to the zygomatic arch seems to taper out at the base of the jugal

postorbital process.

The anterior ends of the frontals have been removed by erosion,

exposing a natural section of the narial cavity. Frontal-parietal

sutures are coossified and not traceable, but they may have crossed

the midline at about the point where the parasagittal crests appear to

diverge anteriorly. The frontal sinuses seem to have extended back-

ward over the braincase to about this point. Beneath these sinuses

the braincase is sharply constricted anteriorly, probably indicating

the greatly constricted form of the olfactory lobes of the brain as

shown by the holotype of K. clallamensis.

Sutures in the orbital wall can be partially seen and the relative

position of foramina and bones can be determined. In general the

arrangement is like that described by Stirton for the holotype of the

genotypic species. There is a large common opening for the anterior

orifice of the alisphenoid canal, the foramen rotundum, and the

anterior lacerate foramen as in pinnipeds. Anterodorsal to this opening

the slitlike small optic foramen opens into a short groove. Dorsal to the

Table 1. Measurements (in mm) of crania of Kolponomos clallamensis and

A', newportensis, new species.

Total (condylobasal) length

Postpalatal length (palatal notch to basion)

Basion to anterior edge of zygomatic root

Length C alveolus to M: alveolus

Width of rostrum across canines

Width of skull across alveoli of M1

Width of skull at infraorbital foramen

Width of skull across antorbital process

Width of greatest intertemporal constriction

Width of braincase, anterior edge of glenoid fossa

Zygomatic width

Auditory width

Mastoid width

Paroccipital width

Greatest width of occipital condyle

Greatest width, anterior nares

Greatest height, anterior nares

Greatest width of foramen magnum

Greatest height of foramen magnum

Transverse diameter of infraorbital foramen

Height of infraorbital foramen

"For additional measurements see Stirton ( 1960:355).

Bilateral measurements of the referred cranium of K. clallamensis are made on the left

side. Parentheses indicate estimated measurements made hy doubling a half width.

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematics and Mode of Life

23

Table 2. Measurements of cheek teeth and mandible of

Kolponomos. Where available, measurements of the left side en-

tered first.

optic foramen lies the equal-sized ethmoid foramen. The

orbitosphenoid bone containing these foramina seems to pass

anterodorsally along the orbital wall to a greater extent than in K.

clallamensis. Ventral to these foramina, the sphenopalatine and poste-

rior palatine foramina are closely associated, the former on the

orbitosphenoid-palatine suture and the latter penetrating the adjacent

palatine bone. The sphenopalatine foramen is about twice the size of

the posterior palatine foramen; neither is as large as its counterpart in

K. clallamensis. Indistinct sutures suggest that the palatine is attenu-

ated between the frontal and maxilla and fails to reach the anterior end

of the orbital fossa. There is a very small lacrimal foramen but no trace

of the limits of the corresponding lacrimal bone.

The jugal forms the ventral part of the orbit and appears to

extend nearly to the lacrimal foramen (in contrast to K.

clallamensis). With the maxilla it forms the roof of the short

infraorbital canal. There is a strong postorbital process and a low,

elongated, rugose masseteric process formed by the jugal.

The dorsal part of the maxilla is eroded away so that only the

lateral and palatal parts of this bone are preserved. The oval

infraorbital canal penetrates the zygomatic process of the maxilla.

This canal is large (see Table 1) and short (11.0 mm left), as is

characteristic of Kolponomos and pinnipeds. It opens onto the face

into a shallow depression (infraorbital fossa ) that is more extensively

developed in K. clallamensis. There appears to have been a shallow

fossa for the levator labii marked by a dorsal preorbital fossa as in A'.

clallamensis. The suture with the premaxilla is thoroughly

coossified. The rostrum anterior to the canines is laterally expanded

and bears low rises over the roots of the canine and I3.

A natural oblique section through the narial cavity exposes the

dorsal part of this region from the level of the canines to just behind

the orbits, showing that the ethmoturbinals are placed laterally and

are excluded from the narial opening as in other arctoids. The dorsal

frontal sinus is also evident and must extend backward close to the

frontoparietal suture.

Upper dentition. — Most of the upper cheek teeth of USNM

215070 (Fig. 8). with the exception of the in situ P2's, were found

grouped together in the matrix between the rami of the associated

mandible. When prepared and fitted into their respective alveoli the

following teeth were recovered: right l\ C, right P1, left P1, left P\

right P4, left M', and right and left M:. All except I' and C are highly

worn and so furnish little information about their crowns' mor-

phology.

The alveoli for I'~: indicate that these teeth were much smaller

than I3. I3 is procumbent and has a long cylindrical root and a

relatively short crown with a recurved tip. There is no cingulum or

carina. A facet for the lower canine is present, worn through the

enamel at the posterolateral base of the crown. The tip has been

fractured and a transverse groove has been cut across the anterior

face of the crown, both probably representing damage during use of

this tooth as a lever. The upper canine is broken across the root so

that only about half the total length of the tooth remains. It is also

procumbent; its crown is an attenuated cone, slightly recurved at

the tip, and has no cingulum or carina. A wear facet for the lower

canine extends from the tip to within 2 mm of the base of the crown.

The tip is also worn apically. and a short transverse groove cuts the

anterior face near the tip.

The P1 is closely appressed to the canine. Its crown is oriented

transverse to the axis of the tooth row. The crown has an ovoid

outline and a posterior cingulum. Heavy wear has truncated the

apex to a medially sloping wear surface cut to the crown's base. Its

single root is anteroposteriorly flattened and bends posteriorly to

accommodate the roots of C and P2. The P: has two roots; its crown

is an elongated oval in occlusal outline with cingular shelves ante-

rior and posterior to the stout principal cusp. Wear has formed a

medially truncated surface across the principal cusp that has cut

nearly to the crown's base. P3 is only slightly larger than P2 and two-

rooted, the more anterior root passing inside the posterior root of P2

so that the tooth has an orientation oblique to the tooth row. The

crown is ovoid in occlusal outline; wear has removed the principal

cusp and anterior cingulum. but the posterior cingular shelf re-

mains. This cingulum bears a low cuspule laterally and is bounded

medially by a well-developed facet of interdental wear.

P4 bears three roots, the lateral pair about the size of those of the

anterior premolars, although the most posterior is slightly smaller

than the anterior. A large medial root is nearly symmetrically placed

with the lateral roots, but the whole tooth has an oblique orientation

to align with the anterior surface of M', thus removing the embra-

sure pit found in most carnivores where carnassial shear is impor-

tant. At the advanced stage of wear shown by the P4 only an

encircling band of thin enamel remains. The enamel is broken on

the posterior side by abrasion between P4 and the adjacent M1. A

remnant of the metastyle remains on the posterolateral part of the

crown marked by a notch that may represent the base of the carnas-

sial notch. Dentine-filled pulp cavities on the gently concave worn

crown indicate the presence of a principal external cusp or coa-

lesced cusps forming an anteroposteriorly elongated structure and

large internal cusp ("protocone") supported by the strong internal

24

R. H. Tedford. L G Bames. and C. E. Rav

5cm

Figure 8. Upper cheek teeth of holotype of Kolponomos newportensis n.

sp„ USNM 215070. Right side with left P:. P3, and M1"2 reversed to restore

the complete cheek-tooth series.

root. Anterolateral to the latter a filled pulp cavity indicates the

presence of another smaller cusp linked to the "protocone," prob-

ably indicating that the anterior cingulum bore a cusp analogous to

the paraconule of tribosphenic molars. Thus the upper carnassial of

USNM 215070 has been molarized, its sectorial nature changed to

function with the molars as part of the masticatory battery.

The first upper molar has three roots. The labial roots support-

ing the paracone and metacone are anteroposteriorly compressed

structures; the lingual root supporting the protocone is a short

faceted cone. Filled pulp cavities indicate a pattern of cusps similar

to the M1 of A-, clallamensis. A short section of the labial cingulum

bridges the indentation between the paracone and metacone and. as

in K. clallamensis, indicates that the paracone was larger than the

metacone. Thin enamel rims the concave wear surface of this tooth,

broken only at the junction with P4. The M: is triangular in form and

least worn on the left side. It is positioned lingually opposite the M1

talon, in contrast to a labial position opposite the trigon as in most

carnivorans. It bears two short stout roots; the anterior is anteropos-

teriorly flattened, the posterior triangular. The worn crown shows

four inflated cusps that have coalesced, separated only by thin

grooves similar to the condition in the less worn M' of K.

clallamensis. These are interpreted as follows: the most labial is the

paracone with the closely allied metacone immediately postero-

lingual to it; the large protocone occupies the anterolingual border.

Figure 9. Mandible of Kolponomos newportensis n. sp., holotype,

USNM 215070. A, occlusal view; B, left side. At same scale as figures of

crania.

the labial projection of its wear facet representing the paraconule;

the cusp at the posterolingual corner of the tooth is the metaconule.

The enamel covering of these cusps is remarkably thin, as revealed

by apical wear. An interdental wear facet with M1 occurs on the

anterior face of the tooth.

Mandible. — The nearly complete mandible of 215070 lacks

only the tips of both coronoid processes (Figs. 9, 10). The rami are

thoroughly ankylosed at the symphysis; the junction is raised exter-

nally, creating a symphyseal boss ventrally. The horizontal ramus is

deepest at this boss and shallowest posteriorly. Anteroventrally

paired foramina lie on either side of the symphyseal suture about at

mid-depth. Laterally there are three mental foramina; the largest is

the most ventral, lying beneath P, at the posterior end of a shallow

depression. A second foramen lies anterodorsal to the first and

beneath P, on the right side or posteriorly beneath the anterior root

of P4 on the left side. The third and most posterior foramen lies at

mid-depth of the horizontal ramus beneath the posterior root of P4.

The rami are markedly rugose beneath the molars; the right side

shows bone resorption around the protoconid root ofM,, and on the

left side there is a pit in the lateral surface adjacent to the hypoconid

root of the same tooth. The masseteric fossa has a deep anteroposte-

riorly elongated pit in its deepest recess. The masseteric crest does

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematic* and Mode of Life

25

Figure 1 0. Outline drawing of the holotype of Kolponomos newportensis

n. sp., USNM 215070, viewed from the left. Upper dentition restored from

isolated teeth found with the type.

not extend to the anterior end of the fossa but arises just above the

base of the angular process and passes to the articular condyle. In

harmony with the form of the glenoid fossa, the articular condyle is

cylindrical, pointed laterally, and deepest medially; a pit for inser-

tion of the external pterygoid muscle occurs at the anteromedial

base of the condyle. The angular process is relatively small and

markedly inflected medially. Its dorsal surface has a pit and ridge,

and the medial surface is rugose, all for insertion of parts of the

internal pterygoid muscle. There is a large mandibular foramen that

lies below the level of the tooth row and the condyle and just above

the level of the dorsal surface of the angular process, about midway

along the base of the ascending ramus.

Lower dentition. — The central incisors seem to have been lost

in life in USNM 215070 but it is not certain that I, was in fact

present. The position of this tooth is occupied by spongy bone. The

alveolus for the right I2 is filled with spongy bone but the alveolus

for the left 1, is discernable. The roots of both I,'s are present. The

canine has a long root and short crown. It is fully preserved only on

the left side; the right canine appears to have had its apex broken

away in life; the broken surface is polished and a secondary wear

facet was established on the medial side of the broken tip. Occlu-

sion with the upper canine has worn the tip and posterolateral side

of each lower canine.

The premolar row diverges slightly posteriorly along an axis

that lies wholly inside the axis of the molar row. The left P, is badly

broken; the right P, was recovered from the surrounding matrix.

Like its counterpart in the maxilla, the P, is anteroposteriorly

compressed to fit between the base of the canine and the overhang-

ing anterior margin of P:. Its crown is heavily worn apically on a

posteriorly slanting surface. There is a posterior cingulum pitted by

breakage. The P, and P, are similar in form; the P, is larger. The

apically worn crowns show a robust principal cusp and anterior and

posterior cingular cusps connected by a lingual cingulum. There is

no labial cingulum. The P4 departs from this form in that a strong

posterolingual cusp is also present; the anterior cingular cusp and

/

-*%a_ F+^r

B

i

i

Figure 11. Cervical vertebrae of holotype of Kolponomos newportensis

n. sp., USNM 215070. A, axis, anterior and left lateral views, lateral view

reversed from right side; B, third cervical, anterior and left lateral views.

posterolabial cingular shelf combine to give this tooth a molariform

appearance.

The M, is offset laterally so that only the anterolingual face of

its paraconid overhangs the posterolabial cingulum of P4; its crown

is only a little longer than wide and worn nearly flat. On the left side

the entoconid region is worn away, but on the right the crown is

complete with its encircling thin enamel. A remnant of the enamel

in the right talonid basin is preserved at this advanced stage of wear.

Filled pulp cavities indicate the full tribosphenic complement of

cusps; the protoconid, metaconid, and hypoconid were subequal in

size, the paraconid and entoconid smaller. There are indentations

internally at the carnassial notch and externally between the talonid

and trigonid. There is no trace of a cingulum.

M, is wider than long, triangular in occlusal outline, and widest

across the trigonid. Wear has removed the anterolingual comer of

the right M, but not the left, which has lost marginal enamel along

the anterior part of the tooth. Filled pulp cavities indicate the

presence of three trigonid cusps and the hypoconid, which is repre-

sented by encircling enamel on the right and has not worn into the

pulp cavity on the left. The pulp cavities seem best interpreted in

comparison with the trigonid of M,: a large protoconid on the

anterolabial corner, a small paraconid directly lingual to it. and a

large metaconid on the median lingual margin. Pulp cavities of the

latter two cusps are connected. There is no M3.

Postcranial .skeleton. — An axis and third cervical vertebra are

available and all structures are preserved on one side or the other of

these bones. The axis (Fig. 1 1 ) has a short centrum and high neural

spine as in pinnipeds and lutrine mustelids. The general form of the

bone most resembles that seen in phoeids or Enhydra, although the

neural spine is larger overall than in the latter. The odontoid process

points anterodorsally as in terrestrial carnivores and in contrast to

most pinnipeds, and there is a shallow groove on its dorsal surface

for the transverse ligament of the atlas, a feature usually missing in

pinnipeds. A broad ridge continues the odontoid process into the

spinal canal. The atlantoaxial articulations are joined beneath the

odontoid process as in ursids, not separated by a notch as in pinni-

peds. The centrum (less the odontoid process) is wider than long. Its

26

R. H. Tedford. L. G. Barnes, and C. E. Ray

It

I?

'

/•'-'

B

r

44

Figure 12. Metapodial and phalanx of holotype of Kolponomos newpor-

tensis n. sp.. USNM 215070. A. metapodial. dorsal and ventral (right)

views; B, phalanx, dorsal and ventral ( right) views.

posterior articulation is dorsoventrally flattened. A marked ventral

keel leads from the joined atlantoaxial articulation to the posterior

articular epiphysis where there is a thickening of the keel into a low

process. Enhydra and phocids show a similar structure. The robust

transverse process sweeps in an arc backward beyond the centrum.

as in pinnipeds. The vertebrarterial canal pierces the base of the

transverse process. It is of large caliber as in pinnipeds and lutrine

mustelids. The neural arch has a narrow base corresponding to the

short centrum. The postzygapophyses project as flanges laterally;

their articulations slant upward posteriorly and laterally. The high

bladelike neural spine hooks posteriorly beyond the neural arch of

the third cervical when in articulation. It is strongly inclined anteri-

orly, ending in a process for the rectus capitis that lies above the

base of the odontoid. The neural spine is thin, wider only at the

rectus capitis origin. Its form most approximates that of phocids

rather than that of otariids or ursids, in which the spine is more

robust and has a marked posterior process. The postzygapophyses

lack dorsal processes for insertion of the axial muscles, but the base

of the arch beneath the overhanging neural spine is excavated for

the more medial elements of this muscle system.

The centrum of the third cervical vertebra is about as long as

wide, and flattened dorsoventrally (Fig. 11) and keeled with a

posterior enlargement, as in pinnipeds. The prominent transverse

processes sweep posterolateral^ and terminate in twin processes.

They seem to lack an anterior spine as in phocids and in contrast to

other carnivores. Large vertebrarterial canals pierce the bases of the

transverse processes. The neural arch is low, and the neural canal is

correspondingly flattened dorsoventrally, as in pinnipeds. Pre- and

postzygapophyses are stout; the latter have low processes for the

axial musculature on their dorsal surfaces. This vertebra lacks a

neural spine as in phocids, Enhydra. and ursids.

Elements of the foot of A", newportensis are represented only by a

much eroded metapodial and a nearly complete median phalanx (Fig.

12). The proportions of the metapodial suggest that it may be a third

metatarsal, but without the proximal end this identification is uncer-

tain. The shaft of this bone is markedly flattened dorsoventrally. as in

pinnipeds. Enough remains of the distal end to indicate that the

articulation was hemispherical and had a well-developed ventral

keel, as in terrestrial carnivores. The median phalanx is also markedly

flattened but not elongated as in pinnipeds or Enhydra. Its distal end

is trochleated. and the proximal articulation indicates that the distal

end of the proximal phalanx was also trochleated. The proximal end

has strong lateral processes for flexor tendons and a dorsomedial

process for extensors, implying powerful movement of the digits.

DISCUSSION

Relationships Between the Species of Kolponomos

Although we have only one nearly complete cranium of each

species, the characters cited in the diagnoses are similar to those

that distinguish other nominal species of arctoid carnivores. More-

over, the two specimens of A", clallamensis are similar in important

particulars that separate them from the specimen here described as

K. newportensis. lessening the possibility that the differences be-

tween the specimens from Washington and Oregon are due to

sexual dimorphism or individual variation. Nor is there evidence

that the individuals differ in ways usually associated with sexual

dimorphism in arctoid carnivores (gross size, size of canines, and

development of muscular processes of the skull).

In some ways the cranium of Kolponomos newportensis seems

the more primitive of the two, having some characters more like

those seen in other arctoids. It has a less highly arched palate, a

shorter paroccipital process, a smaller hyoid fossa, and a smaller

infraorbital foramen. Also, the jugal rims the anteroventral part of

the orbit. Some of its other diagnostic characters, however, such as

the broad snout, lack of an intercondylar notch, and the more

extremely developed mastoid process, are derived relative to K.

clallamensis and other arctoids. K. clallamensis appears to have

had a more specialized feeding mode, a more modified rostrum, and

greater innervation to the fleshy lips.

Relationships of Kolponomos Among the Arctoid Carnivora

In 1960 Stirton compared Kolponomos clallamensis exten-

sively with Allocyon loganensis Merriam. 1930, from the mid-

Arikareean (Oligocene) John Day Formation at Logan Butte, Crook

County, Oregon, and concluded that A. loganensis was "the carni-

vore most closely related to Kolponomos." Among the 29 features

that he delineated, the following similarities seem most informative

cladistically; presence of a nasolabials fossa dorsoanterior to the

orbit, short infraorbital canal and large infraorbital foramen with

infraorbital fossa, and lack of a postorbital process. With the addi-

tional material of Kolponomos now available the following can be

added: the basioccipital is wide posteriorly, the mastoid process is

hypertrophied. and there is a depression anterior to the median

lacerate foramen for the first loop of the internal carotid. The latter

feature is correlated in Allocyon with a deeply excavated lateral

margin of the basioccipital for reception of the carotid artery and

inferior petrosal sinus, typical of ursids and amphicyonids.

Allocyon (Figs. 13, 14) and Kolponomos are similar in the

following dental features that appear to represent synapomorphies:

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematics and Mode of Lite

27

P4 is triangular in outline with a protocone nearly the size of

(Allodesmus) or larger than {Kolponomos) the paracone; M: has a

posteriorly expanded "heel" (the metaconule and posterior cingu-

lum). and the M, talonid is as wide and long as the trigonid.

All of these features seem to support the close phyletic relation-

ship between A llocyon and Kolponomos. In many features Allocyon

is more primitive than Kolponomos, especially those that can be

interpreted as adaptations to molluscivory in the latter. Since the

relationships of Allocyon have not been made explicit heretofore,

we explore the evidence here as a means of placing Allocyon and

thus Kolponomos within the Carnivora.

The synapomorphies listed above uniting Allocyon and

Kolponomos also support their relationship with basal members of

the arctoid clade, especially the amphicyonids and ursids. Particu-

larly important is the presence of the "ursid loop" in which the

internal carotid artery is nested in the inferior petrosal sinus (Hunt

1977). This system has a clear bony signature in the deep marginal

invagination of the basioccipital. Although this feature cannot be

completely seen in Kolponomos. it can in Allocyon, and it is entirely

comparable to the structure in amphicyonids and ursids. A position

closer to ursids is supported dentally by loss of M\ enlargement of

the protocone of P4, and development of a "heel" in M2 by enlarge-

ment of the metaconule and associated posterior cingulum.

Within the Ursidae significant autapomorphies unite the sub-

families Hemicyoninae and Ursinae and exclude Allocyon and

Kolponomos, whose relationships lie near or within the most basal

ursoid group, usually referred to as the Amphicynodontinae

(Simpson 1945). Amphicynodon and Pachycynodon are the most

completely known taxa (Cirot and de Bonis 1992; Cirot 1992)

included in this group. The latter, and larger, form resembles

Allocyon particularly in its posteriorly extended palate (Fig. 14),

but it resembles both Allocyon and Kolponomos in having a short

infraorbital canal, fossa nasolabialis, enlarged infraorbital foramen,

and a similarly reduced postorbital process.

Our conclusions about the relationships of Allocyon and

Kolponomos with the primitive ursoids are similar to those postu-

lated for the most primitive pinnipedimorph, Enaliarctos, by Flynn

et al. (1988). Berta (1991). and Hunt and Barnes (1994, this vol-

ume). In cranial morphology Kolponomos and Allocyon resemble

the pinnipedimorphs in having a fossa for the origin of the

nasolabialis muscle, a short infraorbital canal, a large infraorbital

foramen opening into a fossa, a small optic foramen, and in lacking

a postorbital process. Furthermore, in Kolponomos the lacrimal is

small, fusing early to adjacent bones, and in A", clallamensis the

maxilla forms the anterodorsal orbital rim. In Kolponomos the

foramen rotundum has a common opening with the anterior lacerate

foramen, the postglenoid foramen is vestigial, the posterior carotid

foramen is clearly anterior to the posterior lacerate foramen, and M,

is absent. Some of these features and others noted in the vertebral

column may represent trends parallel to those seen in pinnipeds,

especially with respect to aquatic adaptation (e.g., emphasis on the

internal jugular drainage of the cranium and thus loss of the

postglenoid exit), but the sum total suggests that Allocyon and

Kolponomos represent early offshoots of the same stock that yielded

enaliarctine pinnipedimorphs and that both have their roots in the

earliest differentiation within the Superfamily Ursoidea.

A more explicit hypothesis of the relationships of Kolponomos

to other ursoids and pinnipedimorphs can be made by using the

TABLE 3. Distribution of cranial and dental features discussed in the text (0, primitive state; 1, derived

state).

"AMP, Amphicyonidae; MUS, Mustelida; URS, Ursinae and Hemicyoninae; AMC, Amphicynodon: PAC, Pachycyno-

don; ALL, Allocyon: KOL, Kolponomos: ENA, Enaliarctos.

28

R. H. Tedford. L. G. Barnes, and C. E. Ray

Figure 13. Comparative outline drawings of left side of crania. A. Kolponomos clallamensis Stirton, 1960, referred, LACM 131 148: B, Allocyon

loganensis Merriam. 1930, holotype. UCMP 24106, from Merriam (1930: fig. 1); C, Pachycynodon boriei (Filhol, 1877), holotype, from Filhol

(1877: fig. 59).

sister taxon, the Mustelida (Procyonidae + Mustelidae), and a basal

arctoid group, the Amphicyonidae. as outgroups. For this purpose

we scored 26 of the binary characters discussed above among eight

taxa (Table 3). The taxa are the Amphicyonidae (represented by

Daphoenodon), Mustelida (represented by the archaic forms

Mustelictis, Amphictis, and Plesictis; Schmidt-Kittler 1981).

Ursidae (including the hemicyonine Cephalogale and the ursine

Ursus). Amphicynodon (mostly A. typicus, BM(NH) M749I).

Pachycynodon boriei (Filhol 1877:pl. 58-60, as Cynodictis gryei:

Cirot \992), Allocyon (Merriam 1930, UCMP24106). Kolponomos

(both species), and Enaliarctos (mostly E. meulsi, but also data

from other species described by Berta 1991). The branch-and-

bound algorithm of PAUP(2.4.1 ) found a single most parsimonious

tree (Fig. 15) with a branch length of 36, a consistency index of

0.72, and a retention index of 0.73. Further explanation of the

characters used as synapomorphies are as follows:

1. Basioccipital deeply excavated laterally. — As Hunt (1977)

has shown in living ursids. the large inferior petrosal sinus contain-

ing the intracranial loop of the internal carotid artery lies in a deep

excavation in the lateral margin of the basioccipital that extends to

the posterior lacerate foramen. A morphologically identical struc-

ture occurs in the amphicyonids (including the daphoenines), im-

plying a similar vascular pathway and a synapomorphy for the

Arctoidea. Amphicynodontids and pinnipedimorphs [Enaliarctos;

Hunt and Barnes 1994, this volume) retain this feature. Derived

pinnipeds and members of the Mustelida lack it.

2. Shallow suprameatal fossa present. — All arctoids above the

Amphicyonidae show a suprameatal fossa that may be later trans-

formed into a deep pit in the squamosal dorsal to the external

auditory meatus or may exist as shallow structures floored by the

auditory tube and obliterated in ontogeny by fusion with the meatus

(Schmidt-Kittler 1981). A small shallow fossa excavated dorso-

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematic* and Mode of Life

29

B

Figure 14. Comparative outline drawings of ventral side of crania. A, Kolponomos clallamensis Stirton, 1960. referred. LACM 131148; B.

Allocyon loganensis Merriam. 1930, holotype, UCMP 24106. from Merriam (1930: fig. 3); C, Pachycynodon boriei (Filhol, 1877), holotype. from

Filhol (1877: fig. 60).

posteriorly into the squamosal contribution to the mastoid process

seems to be the most primitive state of this feature. In ursids and

amphicynodontids this structure is shallow primitively and is lost in

derived taxa rather than being obliterated by growth of the tubular

external auditory meatus.

3. M1, absent. — A feature uniting all arctoids above the

Amphicyonidae.

4. Basioccipital wide posteriorly. — The greater width of the

basioccipital across the posterior lacerate foramen versus its width

at the basisphenoid suture is a derived feature of ursids and higher

arctoids. This feature was cited by Wyss (1987) as a pinniped

synapomorphy. later modified by Berta (1991: table 7, no. 44) to

indicate the short and wide basioccipital that describes the condi-

tion of pinnipeds above the Otariidae.

5. Fourth upper premolar with large protocone. — All ursoids

have an upper carnassial with a broad protocone that is usually

shelflike because of incorporation of the lingual cingulum. Further

enlargement of this cusp in Kolponomos relative to the labial cusps

is an autapomorphy that serves to "molarize" the carnassial.

6. Second upper molar with "heel." — All ursoids have an M:

that differs from the tribosphenic form of that of other arctoids by

the posterior shelf or "heel" formed by a well-developed posterior

cingulum, often incorporating the metaconule. This appears to be

the case in Kolponomos and probably Allocyon. Although the M: of

Enaliarctos is very reduced it seems to include a shelflike heel

behind the trigon (as in E. emlongi; Berta 1991 ) and so is coded as

derived in this feature.

7. Reduction and loss of parastyle on M'~:. — Early parastyle

reduction and loss is a feature of ursines and hemicyonines.

Amphicynodontids lost this cusp later in phylogeny as similarly

hypocarnivorous forms (e.g., Pachycynodon) arose.

8. Reduction and loss of paraconule on M'2. — Full loss of the

paraconule took place at different times in the ursine, hemicyonine,

and amphicynodontid lineages, but reduction in the size of this cusp

characterizes the early members of all groups. Curiously,

Kolponomos retains this cusp, inflated like all the molar cusps, to

form the broad grinding surface as in another molluscivore. Enhydra.

9. Lingual position of M1 metaconule. — The ursid metaconule

is strongly connected to the longitudinally elongated protocone in

M1 (and M:). It has lost its primitive connection with the metacone

and occupies a more lingual position on the crown (Cirot and de

Bonis 1992). Amphicynodon retains a primitively labial position of

the M1 metaconule but this cusp is large and well connected to the

protocone by a crista. Kolponomos also retains a primitive tribos-

phenic form of M1 but all cusps are inflated and lack connecting

cristae.

10. Reduction of paraconid of A/,. — Modification of M-, in

ursoids involves enlargement of the talonid relative to the trigonid.

Reduction and loss of the paraconid accompanies attainment of

subequal size of the metaconid and protoconid and their assumption

of a more transverse relationship. Again. Kolponomos appears to

retain a paraconid in its large M, trigonid.

30

R. H. Tedford, L. G. Barnes, and C. E. Ray

P.O. URSIDA

S.F URSOIDEA

PINNIPEDI-

MORPHA

F AMPHICYNODONTIDAE

7 16

22,23,24.25,26

7. 20

8,19,20,21

2? 15, 16,17

12,13.14

4,5,6,7,8,10,11

Figure 15. Phyletic relationships of taxa discussed in the text. For character numbers see text and Table 3. Asterisks indicate reversal to primitive state.

1 1 . Size of metaconid equal to protoconid on A/-,. — The reduc-

tion of the trigonid relative to the talonid in the M, of ursids and

amphicynodontids involves enlargement of the metaconid to the

same size as the protoconid.

12. Infraorbital canal short. — The distance from the anterior

edge of the orbit to the opening of the canal is unusually abbrevi-

ated in the amphicynodontid ursoids and pinnipedimorphs.

13. Infraorbital foramen large. — Amphicynodontid ursoids and

pinnipedimorphs also have a very large anterior opening of the

short infraorbital canal.

14. Metastyle ofP4 short. — In a trend toward hypocamivory the

amphicynodontids shorten the carnassial blade by reduction of the

metastyle.

1 5. Palate posteriorly extended. — The palate is prolonged in the

midline by posterior extension of the palatine bones so that the

internal nares lie at a considerable distance behind the tooth row.

This condition is derived in the Ursoidea but has an independent

distribution that implies some homoplasy (i.e.. it is present in the

Ursinae and some amphicynodontids but not in Kolponomos or

pinnipedimorphs generally!.

1 6. M f talonid as wide as or wider than trigonid. — Correspond-

ing to modification of the M' talon (character 9), amphicynodontids

enlarge the talonid of the lower carnassial to accommodate the

longitudinal protocone and associated metaconule of M1.

17. M j metaconid large. — Modification of the M, trigonid more

for crushing in derived amphicynodontids involved enlargement of

the metaconid relative to surrounding cusps. In volume it comes to

match the paraconid and is nearly as high in the unworn crown as

the protoconid.

18. Nasolabialis fossa present. — A prominent fossa just

dorsoanterior to the orbital rim. presumably for the nasolabialis

muscle, is a derived condition in Allocyon, Kolponomos, and primi-

tive pinnipedimorphs. From the perspective of pinniped evolution

Berta ( 1 99 1 ) coded the loss of this feature in pinnipeds as derived,

i.e., a reversal to the primitive arctoid state.

19. Infraorbital fossa present. — This broad depression lies just

anterior to the opening of the infraorbital canal onto the face. It is

not correlated with the large foramen of amphicynodontids but

characterizes Allocyon, Kolponomos. and pinnipedimorphs.

20. Mastoid process hypertrophied. — This is a synapomorphy

for Allocyon and Kolponomos. although the latter has greatly elon-

gated the process ventrally. The massive backward-pointing

paroccipital process in Allocyon is an autapomorphy for that genus.

2 1 . Postorbital process lacking. — In contrast to other arctoids,

in Allocyon and Kolponomos the postorbital processes of the

frontals are lacking. Low supraorbital ridges at the anterior ends of

the parasagittal crests represent the position of the processes in

these genera and in primitive pinnipedimorphs (Berta 1991 ).

22. Alisphenoid "strut" present. — A reinforced region extends

from the palatine process of the alisphenoid dorsoanteriorly to a

correspondingly reinforced pterygoid process of the palatine. These

elements combine to form a strut bracing the posterior part of the

palate against the braincase. Such a structure is present in

Kolponomos and pinnipedimorphs.

The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Systematica and Mode of Life

31

23. Postglenoid foramen vestigial. — Reduction of this opening

is correlated with greater emphasis on the internal jugular system as

the main venous drainage of the braincase. Although the posterior

lacerate foramen in Kolponomos is not conspicuously enlarged, the

postglenoid foramen is very reduced as in pinnipedimorphs.

24. M2 lingual to M' . — A peculiar feature of the dentition of

pinnipedimorphs (and Potamotherium) that retain M2 is the lingual

position of this tooth adjacent to the talon of M' rather than labial as

in most carnivores (M. Wolsan, pers. comm.). Kolponomos shows

this feature.

25. Foramen rotundum and anterior lacerate foramen lie in a

common fossa. — Berta (1991) discussed the distribution of this

pinnipedimorph synapomorphy. It also clearly occurs in

Kolponomos. but other amphicynodontids have the primitive state

in which these foramina are separated by a bony lamina visible in

lateral view of the skull.

26. My absent. — As in pinnipedimorphs this tooth is absent in

Kolponomos.

Figure 15 summarizes the distribution of these synapomorphies,

indicating the paraphyly of the Amphicynodontidae when the

Pinnipedimorpha (sensu Berta 1991 ) are placed within this group as

the sister taxon of Kolponomos. Some of the characters thought to

be synapomorphies of the Pinnipedimorpha by Berta ( 1991 ) actu-

ally have a wider distribution (e.g.. characters 13. 21. 23, 25. and

26) within the Ursoidea or have ursoid precursor states (character

4). Synapomorphies specifically linking the Pinnipedimorpha with

Allocyon and Kolponomos (characters 18, 19. and 21-26) indicate

that these terrestrial and amphibious arctoids. although dentally

specialized for hypocarnivory. approximate the stem group for the

pinnipedimorphs.

A classification consonant with the phyletic relationship pos-

tulated for the carnivorans discussed above is indicated in Figure

15. The Pinnipedimorpha were not ranked by Berta (1991) and

are so indicated on Figure 15 as an unranked taxon within the

parvorder Ursida of Tedford (1976). Since the traditional

"suborder" Pinnipedia is subsumed in the Pinnipedimorpha it too

must remain unranked in the present attempt to construct a

taxonomy that expresses the phylogenetic conclusions of this

paper.

Speculations About the Mode of Life of Kolponomos

The few postcranial bones presently known indicate that

Kolponomos was not fully aquatic, at least in the sense that the

pinnipeds are. Its foot bones clearly indicate retention of significant

ability for terrestrial locomotion and an amphibious existence. It

was probably littoral in distribution. All known specimens are from

near-shore, shallow-water marine deposits that contain abundant

fossil mollusks, including large mussels and giant pectinids. The

broad, sea-otterlike crushing cheek teeth would have been ideally

suited to a diet of hard-shelled marine invertebrates. The teeth are

well worn, indicating that the diet included very hard-shelled ani-

mals, possibly mussels, limpets, abalone. pectinids. and echinoids.

Coupled with accidentally ingested abrasive sediment, these would

account for the heavily worn condition of the cheek teeth. The

orbits are directed anteriorly rather than laterally as in living bears,

suggesting that Kolponomos probably could view objects directly

in front of its head. This would be of benefit to an animal selectively

eating rock-dwelling benthic or attached (sessile) marine inverte-

brates. The infraorbital foramen is large, quite so in K. clallamensis.

and the mental foramina on the lateral side of the dentary of K.

newportensis are also large. These probably indicate enhanced

tactile sensitivity of the lips and muzzle. Kolponomos might also

have had a large upper lip approaching that in living walruses, and

this would be concomitant with the depth of the premaxilla between

the incisors and the narial opening. Walruses have very sensitive

lips and tactile vibrissas, that apparently aid in distinguishing prey

items when visibility is poor (Fay 1982). Kolponomos might also

have had highly developed tactile vibrissae. The palate is flexed

downward relative to the basicranial plane; the occipital condyles

face ventrally and are positioned posteroventrally relative to the

basicranium, suggesting that the head was carried downward in

relation to the vertebral column. The upper canine and incisor teeth

are large and clustered in thickened bone at the extreme anterior end

of the downturned snout. The nasal opening is retracted posteriorly,

an adaptation that would keep the nostrils away from the substrate.

Large paroccipital and mastoid processes indicate powerful neck

muscles that could have provided strong downward movements of

the skull. All these adaptations suggest that Kolponomos fed on

epifaunal marine invertebrates living on rocky substrates.

Kolponomos probably obtained its food by levering tightly clinging

animals off the substrate and twisting and prying with its head. The

robust median phalanx also suggests that the digits were capable of

powerful movement and these too may have been used to procure

food. This method of feeding is somewhat different from that of

living sea otters. Sea otters actively swim in shallow to moderate

depths and obtain bottom-dwelling animals, largely by pulling and

prying them off rocks with their forelimbs. Sea otters have rela-

tively long digits and strong forelimb musculature. They have large,

flat, crushing cheek teeth, but they do not have enlarged mastoid

and paroccipital processes. This correlates with the fact that they do

not pull their food off rocks by using their heads.

Kolponomos also is unlike the living walrus, which belongs to a

pinniped group that later in the Tertiary became very diverse,

successful, and widespread along the coasts of the northern hemi-

sphere. Modern walruses occupy shorelines, at least part of the

time, when they haul out at specific locations along the shore. When

feeding, however, they are offshore, diving pinnipeds. They typi-

cally dive only to shallow or moderate depths, where they exploit

food resources for the most part different from those of other

pinnipeds, mostly benthic shelled and nonshelled invertebrates.

They do not crush mollusk shells by chewing (Fay 1982) but rather

use the tongue in a pistonlike method to suck the soft parts out of

gaping bivalve shells. They also use the tongue as a piston to direct

a jet of water from the mouth onto the substrate in a method of

hydraulic mining of infaunal prey. Walruses do not chew up shells

of their prey and they do not swallow shells or broken shells, so,

although the general category of food of walruses is the same as that

proposed for Kolponomos. the locating of the food and manner of

gathering and eating it is apparently different.

The specialized dusignathine otariid Gomphotaria pugnax may

be a relatively close functional counterpart to Kolponomos. This

large pinniped is known from upper Miocene rocks of the Califor-

nia coast. Like Kolponomos. Gomphotaria had elongated upper as

well as lower canines, even more fully developed as tusks. Also like

Kolponomos, Gomphotaria had large cheek teeth, which although

not expanded transversely were broken and worn during life from

feeding on resistant prey. Barnes and Raschke ( 1991 ) proposed that

Gomphotaria was a shallow-water or littoral pinniped that pried its

food off rocks, the food presumably being shelled mollusks as we

have postulated for Kolponomos. and that rather than sucking its

food into the mouth it crushed the animals with the cheek teeth.

Kolponomos appears to be an ursid variation on the sea otter

adaptation. On the west coast of North America, middle and upper

Miocene horizons bearing fossil marine vertebrates have been ex-

tensively prospected, much more extensively than have the lower

Miocene deposits. Nothing related to Kolponomos has as yet been

found in these younger deposits. Kolponomos might very well be

the end of its lineage.

32

R. H. Tedford, L. G. Barnes, and C. E. Ray

ACKNOWLEDGMENTS

We thank Albin Zukofsky. II, for his donation of the important

skull of Kolponomos clallamensis from Clallam Bay. We thank

Robert L. (Fritz) Clark for preparing and casting this new skull. The

photographs of the specimen were made by LACM staff photogra-

pher Donald Meyer. We thank Donald E. Savage and J. Howard

Hutchison for making the holotype of Kolponomos clallamensis

available for our work. We thank Jim Goedert for making observa-

tions relating to the collecting sites of the holotype and referred

specimens of Kolponomos clallamensis. for making notes on the

stratigraphy, and for analyzing the associated mollusks to make

inferences about the paleoecology. The holotype of Kolponomos

newportensis was collected by the late Douglas Emlong in two

fragments of one concretion, found on separate occasions more

than six years apart. Emlong recognized that the second, major part

pertained to the first, minor one, even though the latter had no

intelligible bone showing and had not been seen by him for several

years. Assembly of the broken concretion containing the type of K.

newportensis and gross preparation, including separation of the

jammed mandible and skull and loose teeth, was done by Gladwyn

B. (Tut) Sullivan. Final preparation of the type specimens of K.

clallamensis and K. newportensis was skillfully done by Edward

Pedersen of the American Museum of Natural History. Photographs

of these specimens and line drawings were carefully prepared by

Chester Tarka and Lorraine Meeker of the American Museum.

Xiaoming Wang helped with the PAUP analysis of phylogeny.

LITERATURE CITED

Addicott, W.O. 1976a. Neogene molluscan stages of Oregon and Wash-

ington. Pp. 95-115 in A. E. Fritsche, H. Ter Best, Jr.. and W. W.

Womardt (eds.). The Neogene Symposium. Society of Economic

Paleontologists and Mineralogists (Annual Meeting. Pacific Sec-

tion). San Francisco. California.

. 1976b. New molluscan assemblages from the Upper Member

of the Twin River Formation, western Washington: Significance in

Neogene chronostratigraphy. United States Geological Survey

Journal of Research 4:437-447.

-. 1976c. Molluscan paleontology of the lower Miocene Clallam

Formation, northwestern Washington. United States Geological

Survey Professional Paper 976.

Armentrout, J. M., D. A. Hull. J. D. Beaulieu, and W. W. Rau (eds.).

1983. Correlation of Cenozoic stratigraphic units of western Or-

egon and Washington. Oregon Department of Geology and Mineral

Industries Oil and Gas Investigations 7. Portland. Oregon.

Bames, L. G. 1987. Aetiocetus and Chonecetus, primitive Oligocene

toothed mysticetes and the origin of baleen whales. Journal of

Vertebrate Paleontology 7(3) supplement: 10A.

. 1989. Aetiocetus and Chonecetus (Mammalia: Cetacea); primi-

tive Oligocene toothed mysticetes and the origin of baleen whales.

Abstracts of Posters and Papers, Fifth International Theriological

Congress. Rome. Italy, 22-29 August 1989, 1 :479.

, and R. E. Raschke. 1991. Gomphotaria pugnax, a new genus

and species of late Miocene dusignathine otariid pinniped

(Mammalia: Carnivora) from California. Natural History Museum

of Los Angeles County Contributions in Science 426.

, D. P. Domning. and C. E. Ray 1985. Status of studies on fossil

marine mammals. Marine Mammal Science 1:15-53.

Berta, A. 1991. New Enaliarctos* (Pinnipedimorphai from the Oligocene

and Miocene of Oregon and the role of "enaliarctids" in pinniped

phylogeny. Smithsonian Contributions to Paleobiology 69.

Carroll. R. L. 1988. Vertebrate Paleontology and Evolution, W. H.

Freeman, New York, New York.

Cirot, E. 1992. Etude phylogenetique dequelques genres d'Arctoideade

1'Oligocene eurasiatique. Comparaison des donnees morpho-

logiques et moleculaires. Ph.D. Dissertation. Faculte des Sciences

Fondamentales et Appliquees de Poitiers. Poitiers, France.

— , and L. de Bonis. 1992. Revision du genre Amphicynodon,

carnivore de 1'Oligocene. Palaeontographica. Pt. A. 220: 103-130.

Dalrymple, B. 1979. Critical tables for conversion of K-Ar ages from

old to new constants. Geology 7:558-560.

Domning. D. P., C. E. Ray. and M. C. McKenna. 1986. Two new

Oligocene desmostylians and a discussion of tethytherian system-

atics. Smithsonian Contributions to Paleobiology 59.

Fay, F H. 1982. Ecology and behavior of the Pacific walrus. Odobenus

rosmarus divergens Illiger. North American Fauna 74.

Feldmann. R. M., A. B. Tucker, and R. E. Berglund. 1991. Fossil

crustaceans, paleobathymetry and decapod crustaceans. Washing-

ton. National Geographic Research and Exploration 7:352-363.

Flynn, J. J., N. A. Neff. and R. H. Tedford. 1988. Phylogeny of the

Carnivora. Pp. 73-116 in M. J. Benton (ed.). The Phylogeny and

Classification of the Tetrapods, vol. 2. Systematics Association

Special Volume 35B. Clarendon Press. Oxford, England.

Hunt. R. M.. Jr. 1977. Basicranial anatomy of Cynelos Jourdan

(Mammalia: Carnivora). an Aquitanian amphicyonid from the

Allier Basin, France. Journal of Paleontology 5 1 :826-843.

. and L. G. Barnes. 1994. Basicranial evidence for ursid affinity

of the oldest pinnipeds, in A. Berta and T. A. Demere (eds.) Contri-

butions in marine mammal paleontology honoring Frank C.

Whitmore. Jr. Proceedings of the San Diego Society of Natural

History 29:57-68.

Memam. C. W. 1930. Allocyon, a new canid genus from the John Day

beds of Oregon. University of California Publications. Bulletin of

the Department of Geological Sciences 19:229-244.

Moore. E. J., and W. O. Addicott. 1987. The Miocene Pillarian and

Newportian (Molluscan) stages of Washington and Oregon and

their usefulness in correlations from Alaska to California. United

States Geological Survey Bulletin 1664A.

Olson. S. L. 1980. A new genus of penguin-like pelecaniform bird from

the Oligocene of Washington (Pelecaniformes: Plotopteridae).

Natural History Museum of Los Angeles County Contributions in

Science 382.

Piveteau, J. 1961. Mammiferes: Origine reptilienne, evolution. Traitede

Paleontologie 6:1-1 138.

Rau, W. W. 1964. Foraminifera from the northern Olympic Peninsula,

Washington. United States Geological Survey Professional Paper

374-G.

Ray, C. E., D. P. Domning. and M. C. McKenna. 1994. A new specimen

of Behemotops proteus (order Desmostylia) from the marine Oli-

gocene of Washington, in A. Berta and T. A. Demere (eds.) Contri-

butions in marine mammal paleontology honoring Frank C.

Whitmore. Jr. Proceedings of the San Diego Society of Natural

History 29:207-224.

Romer, A. S. 1966. Vertebrate Paleontology. University of Chicago

Press. Chicago, Illinois.

Schmidt-Kittler. N. 1981. Zur stammesgeschicte der marderverwandten

Raubtiergruppen (Musteloidea. Carnivora). Eclogae Geologicae

Helvetiae 74:753-801.

Simpson, G. G. 1945. The principles of classification and a classifica-

tion of mammals. Bulletin of the American Museum of Natural

History 85: 1-350.

Snavely, P. D.. Jr. 1983. Peripheral rocks: Tertiary geology of the north-

western part of the Olympic Peninsula. Washington. Geological

Association of Canada. Victoria. British Columbia. Field Trip

Guidebook 12:6-31.

— , A. R. Niem. N. S. MacLeod. J. E. Pearl, and W. W. Rau. 1980.

Makah Formation — a deep-marginal-basin sequence of Late Eo-

cene and Oligocene age in the northwestern Olympic Peninsula.

Washington. United States Geological Survey Professional Paper

1162-B.

Stirton. R. A. 1960. A marine carnivore from the Clallam Miocene

Formation, Washington. Its correlation with nonmarine faunas.

University of California Publications in Geological Sciences

36:345-368.

Tabor, R. W., and W. M. Cady. 1978. Geologic map of the Olympic

Peninsula. Washington. United States Geological Survey Miscella-

neous Investigations Series Map 1-994.

Tedford. R. H. 1977. Relationships of pinnipeds to other carnivores

(Mammalia). Systematic Zoology 25 (for 1976):363-374.

Thenius. F. 1969. Stammesgeschichte der Siiugethiere (Einschliesslich

der Hominiden). Handbuch der Zoologie. Band 8. Teil 2. Vol. 1:1-

368.

Wyss. A. R., 1987. The walrus auditory region and the monophyly of

pinnipeds. American Museum Novitates 2871.

Pinniped Phylogeny

Annalisa Berta

Department of Biology, San Diego State University, San Diego. California 92182

Andre R. Wyss

Department of Geological Sciences, University of California. Santa Barbara, California 93106

ABSTRACT. — In our view, presentations inferring pinniped cJiphyly provide inadequate evidence of "otarioid" monophyly and inadequate

evidence that phocids are related to some nonpinniped group. The integrated assessment of higher-level pinniped relationships presented here, based

on cranial, postcranial, and soft-anatomical characters from most living and adequately known fossil pinnipeds, supports pinniped monophyly. We

scored more than 150 character transformations on a generic-level character matrix and used a computer parsimony algorithm (PAUP) to construct

a maximally parsimonious phylogenetic hypothesis for the group. Its major outlines are as follows: (Enaliarctos (Pteronarctos (Otariidae

(Odobenidae (Pinnarctidion, Desmatophoca, Allodesmus (Phocidae))))). Internally, the data are highly consistent. Convergence is much less

pervasive than generally assumed, with reversals being the dominant pattern of homoplasy.

INTRODUCTION

Few mammalian examples more forcefully illustrate the impact

of phylogenetic systematic methods on notions of a particular

group's evolutionary history than does the case of the pinnipeds.

Recent cladistic studies have addressed questions of relationships

within the otariids (fur seals and sea lions) (Berta and Demere

1986) and phocids (true seals) (Muizon 1982a; Wyss 1988b).

odobenid (walrus) affinities (Wyss 1987). and the placement of

certain archaic fossil taxa (Wyss 1987; Berta et al. 1989; Berta

1991). The net result of these efforts is a concept of pinniped

relationships drastically different from what was generally accepted

until relatively recently. We outline here the novel aspects of these

recent proposals and present their methodological basis. In addition

to providing new information on the distribution of morphological

characters, we combine and revise the data sets of these previous

studies in an attempt synthesize what we regard to be the currently

best supported hypothesis of pinniped relationships. We present the

evidence for this hypothesis in the form of a taxon-character matrix

in hopes that it may serve as a starting point for future phylogenetic

analyses of pinnipeds. If this matrix generates debate about charac-

ter coding, or discussion over the in- or exclusion of certain charac-

ters in the analysis, or if it inspires the examination and description

of additional characters that either support or refute the relation-

ships we favor, in short, if it evolves, it will have served its purpose.

HISTORICAL CONSIDERATIONS

All recent workers agree that pinnipeds are members of a

carnivoran subclade. the Arctoidea, that includes among terrestrial

lineages mustelids, ursids. and procyonids (Tedford 1976). Contro-

versy about relationships among the major groups of pinnipeds

centers on the relationship of phocids to the rest of the Arctoidea.

This disagreement reduces to two fundamental questions of mono-

phyly. Before we consider these (to eliminate possible ambiguity)

we must define our usage of "monophyly." We use the term (sensu

Hennig 1966) to denote a group of taxa derived from a common

ancestor and including all of the descendants of that common

ancestor. Evidence for monophyly of a particular group consists of

the shared possession of evolutionary novelties (synapomorphies)

by its members. The two central questions concerning the

phylogenetics of "fin-footed" arctoids are ( 1 ) is the group as a

whole descended from an exclusive common ancestor? (i.e., are

pinnipeds monophyletic?) and (2) do pinnipeds excluding phocids

have a common ancestor not also shared by phocids? (i.e.. are the

Otarioidea, defined as the Otariidae and Odobenidae plus certain

extinct forms, monophyletic?) (Fig. 1 ).

The question of single versus multiple origin! s) dates from

Mivart's ( 1885) suggestion that the group's origin was likely com-

pound, with sea lions and walruses derived from ursids and true

seals derived from mustelids, otters in particular (Fig. IB). Al-

though this view was dismissed by several workers over the next

half century (e.g., Weber 1904; Gregory 1910), it was not dis-

counted altogether by others (e.g., Kellogg 1922; Howell 1929;

Simpson 1945). Thereafter, the notion of multiple pinniped origins

regained wide support in the morphological and paleontological

literature (McLaren 1960; Tedford 1976; Muizon 1982a,b), a shift

influenced particularly by the detailed descriptions of the fossil taxa

Potamotherium (Savage 1957) and Enaliarctos (Mitchell and

Tedford 1973). More recently, one of us argued, on the basis of

anatomical criteria, in favor of a return to the single-origin interpre-

tation (Wyss 1987) (Fig. 1A), a conclusion consistent with but

independent of certain biomolecular and cytologic results (Fay et

al. 1967; Arnason 1986; de Jong 1982, 1986). Several subsequent

studies (Flynnetal. 1988; Berta etal. 1989; Wyss 1988, 1989) have

yielded additional evidence supporting this conclusion, but it con-

tinues to engender debate (Wozencraft 1989; Repenning 1990;

Bonner 1990).

The second question concerns the phylogenetic validity of the

Otarioidea. Since their recognition as distinct groups of mammals,

otariids and odobenids have nearly universally been regarded as

being more closely related to each other than either is to some third

taxon. The observation that the walrus is in many respects more

nearly intermediate between otariids and phocids than had been

previously appreciated (Fay et al. 1967) signaled an important

break from this view. The argument that odobenids are related more

closely related to phocids than to otariids took this suggestion one

logical step further (Wyss 1987). This proposed phocid-odobenid

linkage opened the broader question of where this pair should be

placed relative to the other arctoids. Either a special link between

mustelids and phocids (plus now odobenids) could continue to be

recognized (rendering the Otarioidea polyphyletic), or the associa-

tion of phocids and odobenids could be recognized within the

context of a monophyletic Pinnipedia (rendering the Otarioidea

paraphyletic). Thus the questions of otarioid and pinniped mono-

phyly are to some degree interwoven, yet if care is taken both may

be evaluated with considerable independence.

To two phylogenetic questions, are pinnipeds monophyletic and

are otarioids monophyletic, there are four alternative pairs of re-

sponses. All except one of these have recent historical precedent:

affirmative, affirmative (no recent proponents); affirmative, nega-

tive (Fig. 1A) (Wyss 1987; Berta et al. 1989; Flynn et al. 1988;

Wyss and Flynn 1993); negative, negative (Mitchell and Tedford

In A. Berta and T. A. Demere (eds. i

Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore. Jr.

Proc. San Diego Soc. Nat. Hist 29:33-56, 1994

34

A. BertaandA. R. Wyss

Terrestrial

Arctoids

Enaliarctos Otariidae

Desmatophoca

Odobenidae Allodesmus Phocidae

. Otarioidea

Mustelidae

Potamotherium

Figure 1. Two competing hypotheses regarding phylogenetic relation-

ships among pinnipeds. A, Current yiew of pinniped monophyly proposes

common ancestry for all pinnipeds from a terrestrial arctoid and supports a

sister-group relationship between walruses, seals, and their extinct relatives

(Wyss 1987; Berta et al. 1989; Flynn et al. 1988). B. Alternative view of

otarioid monophyly proposes independent origins of otarioid and phocid

lineages from different terrestrial artcoid groups and supports a sister-group

relationship between walruses, sea lions, and their supposed fossil relatives

(Barnes 1989. and others).

1973: 205, 278); negative, affirmative (Fig. IB) (Barnes 1989;

Wozencraft 1989; Repenning et al. 1979). We begin with the

otarioid question because in some senses it is less complex and

easier to address in isolation. We emphasize again that to test the

hypothesis of otarioid monophyly apart from the question of pin-

niped monophyly we must restrict our attention to those derived

features characteristic of otarioids not also occuring in phocids.

OTARIOID MONOPHYLY

Were pinnipeds demonstrably polyphyletic both the otarioids

and phocids could be diagnosed with characters each shares but

acquired independently. But because we view otarioid monophyly

as a hypothesis in legitimate need of testing we regard claimed

otarioid synapomorphies occurring also in phocids as questionable,

at least for the initial part of the analysis.

Barnes (1989: fig. 9) presented the most recent, most detailed,

and, so far as we are aware, only nominally cladistic diagnosis of

the Otarioidea (= his Otariidae). He listed on a branching diagram

20 characters diagnostic of the Otarioidea.

Following the procedure of Wible ( 1 99 1 ), we evaluated these 20

features, grouping them as follows: ( 1 ) characters for which the

reported derived state occurs in relevant outgroups (i.e.. nonpin-

niped Arctoidea) and are therefore primitive and phylogenetically

uninformative, (2) characters for which the derived state occurs

also in phocids and thus are of uncertain value. (3) characters

dubiously described, and (4) characters for which descriptions and

distribution are correct but for which we offer comment. Characters

are labeled with letters corresponding to the order in which they

were listed by Barnes ( 1989: fig. 9) at node 1, the basal node of his

branching diagram.

Barnes' otarioid "synapomorphies" occurring also in relevant

outgroups. — The derived state of the following characters occurs

elsewhere among the Arctoidea and thus represents a level of

generality broader than supposed by Barnes ( 1989).

(a) Neck lengthened. Wyss (1987: 11) discussed previously

problems associated with this character. Even if it did characterize

otarioids primitively (among odobenids, it doesn"t characterize at

least Odobenus, the only odobenid for which this character can

currently be scored), it is not a derived feature among the Carnivora

(Bisaillonet al. 1976).

(e) Foramen ovale and posterior opening of alisphenoid canal

joined in an elongated recess. The arrangement in ursids is identical

(Davis 1964) and almost certainly represents the ancestral condi-

tion of the Arctoidea.

(h) Embayment formed in lateral edge of basioccipital for loop

of median branch of internal carotid artery. This character is well

known in ursids (Hunt 1974) and amphicyonids (Hunt 1977) and is

unquestionably derived at a level broader than the Otarioidea.

Although this embayment is absent in all living pinnipeds, in cer-

tain phocids (e.g., Monachus) sharp crests on the dorsal surface of

the basioccipital may represent an osseous vestige of it.

(j) Basal whorl of cochlea directed posteriorly. As discussed by

Wyss (1987), this condition characterizes all therian mammals

except phocids, which are uniquely specialized in having a more

transversely oriented basal whorl.

(1) Auditory ossicles not enlarged. This feature is obviously

primitive; the derived condition (ossicles enlarged) originated three

times among otarioids according to Barnes' scheme and again

(presumably independently) in phocids (see below under Assessing

the Pattern of Homoplasy).

(m) Entotympanic restricted to medial part of bulla around

carotid canal. This condition corresponds to the type "A" bulla of

Hunt ( 1974), which characterizes ursids, some amphicyonids, some

mustelids, and perhaps arctoids ancestrally. It is not uniquely diag-

nostic of otarioids and is therefore not relevant to the question of

otarioid monophyly.

(n) Internal acoustic meatus round. A round internal auditory

meatus is widespread among terrestrial carnivorans and is unques-

tionably primitive at the level suggested here. The partially or

completely divided condition seen in certain "otarioids" and

phocids is derived (see discussion of character 24 in analysis be-

low). The meatus is not round in odobenids, Pinnarctidion,

Desmatophoca. or Allodesmus.

(p) Bony tentorium in braincase closely appressed to dorsal

surface of eminence containing semicircular canals and floccular

fossa. As discussed by Wyss (1987: 24), this is the typical

carnivoran condition and is undoubtedly primitive. Because the

bony tentorium varies widely among the Carnivora (Nojima 1990:

table 2) and is difficult to identify, we excluded it from our analysis.

Barnes' otarioid "synapomorphies" that occur also in

phocids. — The derived state of the following characters occurs also

in phocids, a group not included in Barnes" ( 1989) analysis.

(b) Proximal limb elements shortened. Phocids have previously

been recognized as possessing short proximal limb bones (Weber

1904; Howell 1929), and this character has been identified at a

Pinniped Phylogeny

35

more general level, the Pinnipedimorpha, comprising all pinnipeds

plus Enaliarctos (Berta et al. 1989).

(c) Maxilla forms part of wall of orbit. Wyss ( 19S7) reported

that the derived state in whieh the maxilla makes a significant

contribution to the medial orbital wall and forms the anterior orbital

rim also occurs in phocids.

(d) Foramen rotundum and anterior opening of alisphenoid

canal combined into one large orbital fissure. Barnes' diagram fails

to indicate that Pinnarctidion and Desmatophoca are exceptions.

Phocids also share this derived condition (see discussion of charac-

ter 19. Appendix 1 ).

(f) Sphenopalatine foramen enlarged. This derived state also

occurs in phocids (see character 12, Appendix 1).

(g) Petrosal isolated from surrounding cranial bones. Repenning

( 1972) discussed this feature as occurring in phocids also. We have

not analyzed it because of the difficulty of quantifying it. We

observe only subtle differences in this feature among pinnipeds and

terrestrial carnivorans.

(o) Posterior lacerate foramen enlarged, not expanded trans-

versely. The posterior lacerate foramen is enlarged in all phocids as

well. In some, however, it is also expanded transversely, but this is

apparently a secondary transformation. The condition likely primi-

tive for phocids (e.g.. that seen in Monachus) is indistinguishable

from the supposed "otarioid" condition.

(q) Postglenoid foramen reduced. Phocids are also character-

ized by having reduced or lost the postglenoid foramen (see charac-

ter 40, Appendix 1 ).

(r| Entepicondylar foramen lost from humerus. An entepi-

condylar foramen is variably present among phocids. From a previ-

ous phylogenetic study of the group (Wyss 19881 and our present

analysis, presence of this foramen in phocines and most early fossil

"monachines" is probably secondary for the group.

(s) Olecranon process of ulna enlarged. As illustrated by Howell

( 1929: fig. 10). phocids possess a condition of the olecranon pro-

cess similar to that seen in otariids and odobenids.

Barnes' description dubious. — (k) Head lost from incus. The

loss of a head on the incus presupposes that a head was once

present, which to us seems highly unlikely. By comparison with the

outgroups identified here, the head on the incus is a phocid

autapomorphy. absent in all other carnivorans.

Barnes' (1989) descriptions require modification. — (i) Mastoid

process large and cubic. The size and shape of the mastoid process

in nonphocid pinnipeds is not significantly changed over the condi-

tion in ursids. Wyss (1987) critiqued the use of this feature at

greater length.

(t) Aquatic propulsion by fore- and hindlimbs, principally the

forelimbs. Living pinnipeds swim in two different ways. Otariids

generate propulsion principally by use of the forelimbs, whereas

phocids and odobenids use principally the hindlimbs (English 1976;

Gordon 1981, 1983). It has been argued that the ancestor of pinni-

peds (or even the ancestor of "otarioids," if this group should prove

monophyletic) likely generated propulsion by using all four limbs,

as Enaliarctos probably did (Berta et al. 1989). This argument

applies equally to the ancestor of phocids even if they are not

related to other pinnipeds. That some distant ancestor of phocids

was a four-limb swimmer is indicated by the phocids' forelimbs'

being highly modified (used in steering) despite their propelling

themselves by the hindlimbs. If phocids had evolved hindlimb

swimming directly from a terrestrial ancestor, the forelimbs should

not be as highly transformed as they are.

In summary, Barnes' analysis does little to bolster the case for

otarioid monophyly: indeed, it fails to reveal a single persuasive

synapomorphy for the group. We recognize that a proposed otarioid

synapomorphy is not automatically invalidated by its appearance in

phocids. Plausibly, the Phocidae and "Otarioidea" could be diag-

nosed with some of the same (convergently acquired) characters,

provided that additional characters demonstrated a phylogenetic

separation between the two groups.

Historically, the assumed linkage between phocids and

mustelids provided this separation, but, as discussed below, recent

reviews of characters previously cited in support of this pairing call

it into question. The weakness of the evidence supporting the

relationship of one pinniped subgroup (phocids) to a terrestrial

arctoid lineage (mustelids) to the exclusion of other pinnipeds (a

requisite for the acceptance of convergence between otarioids and

phocids) leads us to dismiss at least initially apomorphies occurring

in both "otarioids" and phocids as necessarily indicating "otarioid"

monophyly. Certain "otarioid" synapomorphies might represent

convergences: however, we fail to see the logic of accepting this

claim in the absence of phylogenetic evidence substantiating a

linkage of phocids to some terrestrial lineage of arctoids.

Citing Repenning and Tedford ( 1977), Wozencraft (1989: 516)

argued that there are "many" synapomorphies supporting a walrus-

otariid clade. yet he did not list a single shared derived feature in

support of this contention. Of the 1 1 features listed as diagnostic of

otarioids in the earlier study, all are primitive or of otherwise

unclear phylogenetic significance (Wyss 1987). Thus neither

Barnes' nor Wozencraft's analyses identify synapomorphies cor-

roborating otarioid monophyly.

PINNIPED MONOPHYLY

Before addressing the question of pinniped monophyly. we first

examine the recent arguments in favor of pinniped diphyly. Accep-

tance of pinniped diphyly requires that two criteria be satisfied:

evidence of otarioid monophyly and evidence that phocids are

related to some nonpinniped terrestrial group. We concluded in the

previous section that otarioid monophyly was not well founded.

With respect to the second question. Wyss (1987) reviewed the

characters used by Tedford (1976) and Muizon (1982a) to unite

phocids and mustelids, concluding that no strong case could be

made for a mustelid-phocid pairing. Wozencraft ( 1989) argued in

favor of a mustelid-phocid link but did not discuss the synapo-

morphies supporting nodes on his maximally parsimonious trees.

To consider all possible pinniped-terrestrial arctoid pairings we

include as outgroups the Ursidae, Mustehdae, Procyonidae, and

extinct Amphicyonidae. The monophyly of these groups is gener-

ally accepted (Flynn et al. 1988). Principal references for these taxa

are as follows: Amphicyonidae, Hunt (1974), Hough (1948);

Ursidae, Davis (1964), Beaumont (1965); Mustelidae. Savage

(1957), Schmidt-Kittler (1981): Procyonidae, Baskin (1982).

Wozencraft and Decker ( 1 99 1 ).

METHODS AND MATERIALS

Our assessment of relationships among pinnipeds relies upon

outgroup comparison. Flynn et al. ( 1988) reviewed the relationship

of pinnipeds to other arctoids. proposing two principal hypotheses:

pinnipeds as the sister group of ursids and pinnipeds as part of a

polytomy with other arctoid families. Berta ( 1991 ) used the Ursidae

and the Amphicyonidae as the first and second outgroups to

pinnipedimorphs on the basis of their retaining the excavated

bassioccipital and presumed loop of the internal carotid artery, a

synapomorphy (see Hunt and Barnes 1994. this volume). It is worth

mentioning that no extant pinnipeds have the internal carotid loop,

and the excavated basioccipital. most extreme in Enaliarctos. is

presumably lost. Fortunately, strong postcranial evidence that

Enaliarctos is related to pinnipeds supports the presumed loss of

this feature at the level of the Pinnipedia. Proponents of both di- and

36

A. Berta and A. R \V\ss

monophyly must accept this loss. Thus this feature can in no way be

judged to favor a monophyletic Otarioidea. Four synapomorphies

link ursids and pinnipedimorphs: (1) shelflike anteromedially

placed I*4 protocone. (2) narrow M1 with longitudinally elongated

protocone (Flvnn et al. 1988). (3) knoblike acromion process of

scapula, and (4) robust olecranon process on ulna (Berta 19911.

Wyss and Flynn (1993) used similar evidence to support a sister-

group relationship between the Ursoidea (defined as the common

ancestor of ursids and amphicyonids plus all of its descendants ) and

the Pinnipedia.

In addition to living representatives of the three pinniped fami-

lies, we include, as terminal taxa, their extinct relatives and indicate

their degree of completeness (Table 1). With two exceptions the

monophyly of these taxa is generally accepted. On the basis of

comparative anatomical evidence Wyss (1988) questioned the

monophyly of "Monachus" (indicated by quotes). Berta (1991)

recognized Enaliarctos as a metataxon [term formulated by

Gauthier (1986); see also Gauthier et al. (1988) and Donoghue

( 1985)] since there is no unambiguous evidence supporting either

its monophyly or paraphyly. Initially we included all fossil taxa and

in later runs of the data selectively removed them to determine their

effect on the tree.

PAUP ANALYSIS

We scored 143 skeletal character transformations on a taxon- ■

character matrix (Table 2). Of these characters, 73 were

craniodental (64 binary and 11 multistate), 52 were postcranial (48

binary and 4 multistate), and 15 were soft anatomical. Some 160

character transformations were possible.

We subjected the data to Swofford's ( 1991 ) computer algorithm

PAUP, version 3.0s, using the heuristic search option. In all runs

multistate characters were entered as unordered. In the initial PAUP

run eight characters, 8, 13, 37, 47, 63, 74, 82, and 138. were

excluded since they could not be unambiguously polarized. Our

initial PAUP analysis considering all fossil taxa resulted in over 100

most parsimonious trees. Principal differences among the 100 trees

were in the position of poorly known odobenids including

Alachtherium, Dusignathus, Pliopedia, and Pontolis. Later analy-

ses excluded these taxa, including only those at least 53% complete.

Table I. Completeness of fossil taxa

studied as a percentage of the number

of characters scored (Appendix 1 ).

We obtained the same result. 100+ most parsimonious trees. Each

cladogram had a branch length (BL) of 239, a consistency index

(CI) of 0.640, and a rescaled consistency (RC) index of 0.554. The

RC excludes autapomorphies from the analysis as well as totally

homoplastic characters (see Wiley, et al. 1991 ). A strict-consensus

tree is presented in Fig. 2.

DISCUSSION

Pinniped monophyly is supported by diverse anatomical data.

We discuss below the major groupings shown in Fig. 2. The various

characters are numbered as in Appendix I . Diagnostic characters

for the nodes and terminal taxa in Fig. 2 are listed in Appendix 2.

Pinnnipedimorpha

Berta et al. (1989) proposed the name Pinnipedimorpha as a

term for the monophyletic group including Enaliarctos and the

Pinnipediformes. Postcranial (Berta and Ray 1990) and cranial

features (Berta 1991 ) have been used to diagnose the group (Table

2, Figs. 3-5 ). We recognize 1 8 unequivocal characters and 6 equivo-

cal characters diagnosing the Pinnipedimorpha. Among unequivo-

cal synapomorphies are 10 craniodental features: (11) infraorbital

foramen large (Fig. 3), (15) anterior palatine foramina anterior of

maxillary-palatine suture, (25) round window large with round

window fossula developed, (27) basal whorl of scala tympani en-

larged, (40) postglenoid foramen vestigial or absent, (43) jugular

foramen enlarged, (48) processus gracilis and anterior lamina of

malleus reduced. (66) M1 : reduced in size relative to premolars.

(67) M1"2 cingulum reduced or absent, and (72) M, metaconid

reduced or absent.

Unequivocal postcranial synapomorphies of pinnipedimorphs

include structural details of the flippers such as (87) greater and

lesser humeral tuberosities enlarged (Fig. 4), (88) deltopectoral crest

strongly developed (Fig. 4). (90) humerus short and robust. (92)

olecranon fossa shallow. (98) digit I on the manus emphasized (Fig.

4). ( 105) digits I and V on the pes emphasized (Fig. 5). ( 1 10) ilium

short, and (118) and femoral condyles strongly inclined medially.

An additional 16 equivocal synapomorphies might be diagnos-

tic of this clade but are subject to equally parsimonious alternative

distributions. Seven of these characters, not preserved in

Enaliarctos, we assigned to the less inclusive level of the

Pinnipedia: (31) cochlear aqueduct large, (49) middle ear cavity

and external auditory meatus with distensible cavernous tissue, (54)

deciduous teeth fewer. (60) number of lower incisors reduced.

(101) metacarpal I longer than metacarpal II, (103) foreflipper

claws short, and (104) manus, digit V, intermediate phalanx strongly

reduced.

The oldest known pinnipedimorph, Enaliarctos, described on

the basis of crania and isolated teeth (Mitchell and Tedford 1977;

Barnes 1979), is now known from a nearly complete skeleton

collected from the late Oligocene or early Miocene Pyramid Hill

Sandstone Member of the Jewett Sand in central California (Berta

et al. 1989; Berta and Ray 1990). Other crania and associated lower

jaws and postcranial elements referred to this taxon are described

from deposits of similar age in coastal Oregon (Berta 1991 ).

Pinnipediformes

The name Pinnipediformes encompasses the ancestor of

Pteronarctos and all its other descendants, the Pinnipedia. This

group can be diagnosed on the basis of 14 characters, two of which

are unequivocal: (14) embrasure pit on palate between P4 and M1

shallow to absent and (24) mastoid process close to paroccipital

process, the two connected by a high continuous ridge (state I of

multistate character).

Pinniped Phytogeny

37

TABLE 2. Distribution among pinniped taxa and relevant outgroups of characters scored. Numbers in heading correspond to numbers of

characters in Appendix 1 . 0. Ancestral state; 1 , 2. and 3, derived states; '.'. character missing or not preserved.

38

A. Berta and A. R. Wvss

Table 2 (continued).

Pinniped Phylogeny

39

Tablh 2 (continued).

40

A. Berta and A. R. Wyss

3

O

4-»

3

o

■c

0D0BENIDAE

PH0C0M0RPHA

PINNIPEDIA

PINNIPEDIFORMES

PINNIPEDIMORPHA

Figure 2. Stnct-consensus cladogram of 100 equally parsimonious trees identified by PAUP analysis. Character distributions are listed in Table 2. Short

bars, convergences; long bars, reversals.

An additional 12 characters are identified as equivocal synapo-

morphies at this node. Because 6 of these are unknown in Ptero-

narclos we considered them diagnostic of the Pinnipedia. the least

inclusive level at which their distribution can be confirmed (see

below h

Pteronarclos, the oldest known member of the Pinnipediformes,

has been described on the basis of crania and lower jaws (Barnes

1989: Berta. in press) from the Miocene Astoria Formation of

coastal Oregon. This taxon provides the first definitive evidence of

development of the pinnipeds' unique orbital wall, to which the

maxilla contributes significantly (9). and a lacrimal that fuses early

in ontogeny and does not contact the jugal (10). Both of these

features may be present in Enaliarctos, but available material is not

sufficiently well preserved to determine this.

Pinnipedia

Illiger (1811) proposed the name Pinnipedia to unite the otariids.

odobenids, and phocids. Of the nine characters diagnosing this

group only three craniodental ones are unequivocal synapo-

morphies: (30) pit for tensor tympani absent, (59) I1 lingual ungu-

ium absent, and (71) M, , trigonid suppressed. Potential synapo-

morphies with other equally parsimonious explanations are

(7) fossa nasolabialis absent. (S) fossa muscularis absent.

( 16) antorbital processes large and well developed. (63) P4

protocone shelf absent, (64) P4 double or single rooted, and (73) M,

absent.

These characters' being relatively few should not be interpreted

as weakness of the case for pinniped monophyly. If only living

forms were considered all pinnipedimorph and pinnipediformes

synapomorphies described above would represent pinniped syna-

pomorphies. For example, the following synapomorphies, equivo-

cal at the level of the Pinnipedimorpha or Pinnipediformes. are

unambiguous at the level of the Pinnipedia: (81) lumbar vetehrae

five, (94) olecranon process laterally flattened with expanded distal

half, (95) radius with marked anteroposterior flattening and ex-

panded distal half. ( 109) pubic symphysis unfused, (115) fovea for

teres femoris ligament strongly reduced or absent, and (117) greater

femoral trochanter large and flattened. These are in addition to the

two confirmed pinnipediform synapomorphies and the 18

pinnipedimorph synapomorphies listed in Appendix 1.

Otariidae

Relationships among the fur seals and sea lions based on cladis-

tic analysis (Berta and Demere 1986) are being reanalyzed by Berta

using two different, more appropriate outgroups. Pteronarclos and

the Phocomorpha. Our analysis here supports the latter study in

Pinniped Phylogeny

41

Figure 3. Lateral views of the skulls of representative pinnipeds and a generalized terrestrial arctoid. Extent of maxilla indicated by stippling. En.

Enaliarctos emlongi; OD. Odobenidae (Odobenus wsmarus). OT, Otariidae (Zalophus califomianus); PH, Phocidae {"Moncwhus" schauinslandi"); Pt,

Pteronarctos goedertae; UR. Ursidae (Ursus americanus). Pinnipedimorph synapomorphies: 10. lacrimal greatly reduced; 1 1. infraorbital foramen large.

progress in recognizing a sister-group relationship between the

southern fur seals, Arctocephalus and the sea lions, the Otariinae.

The northern fur seal. Callorhinus, and the extinct taxon

Thalassoleon are positioned as sequential sister taxa to this clade.

Two unequivocal osteological characters diagnose the otariid clade:

(17) supraorbital processes large and shelflike, particularly among

adult males (state 3 of multistate character), and (86) secondary

spine on scapula present. Two additional unequivocal synapo-

morphies based on soft-anatomical characters diagnose the extant

Otariidae: (135) pelage units uniformly spaced and (143) trachea

with bifurcation of bronchi posterior. An additional character possi-

bly diagnostic of this group is (4) frontals extending anteriorly

between the nasals, but this feature is also incipiently developed in

some species of Pteronarctos (Barnes 1989; Berta, in press).

Basal members of this clade (e.g.. Pithanotaria and

Thalassoleon) are known from the late Miocene in California and

Baja California (Repenning and Tedford 1977).

Phocomorpha

We propose the name Phocomorpha for the monophyletic group

that includes the most recent common ancestor of the odobenids

and phocoids plus all its descendants (see Fig. 2). A sister-group

alliance between the odobenids and phocids was originally pro-

posed by Wyss (1987) and endorsed by Flynn et al. (1988) and

Berta et al. (1989). Our analysis provides additional characters

supporting this hypothesis. We identified nine unequivocal synapo-

morphies: (26) canals for facial and vestibulocochlear nerves in-

cipiently or completely separated (state 1 of multistate character),

(32) canal for cochlear aqueduct merged or nearly merged with

round window, (34) petrosal visible in posterior lacerate foramen,

(42) basioccipital short, broad, and widened posteriorly. (46) audi-

tory ossicles enlarged. (51) angular process on mandible reduced

and elevated above the base of ascending ramus, ( 124) calcaneal

tuber short. ( 126) caudally directed process of astragalus at least

incipiently present (state 1 of multistate character), and (127)

baculum enlarged. An additional seven soft-anatomical and behav-

ioral synapomorphies diagnose extant members of this clade: ( 128)

testes abdominal, (129) copulation aquatic. (132) primary hair

nonmedullated, ( 136) subcutaneous fat thick, ( 139) external pinnae

absent, (140) opening of sweat duct proximal, and (141) venous

system with inflated hepatic sinus, well-developed caval sphincter.

large intervertebral sphincter, duplicate vena cava, and gluteal route

for hindlimbs.

In addition, nine equivocal synapomorphies were identified at

this level. Six of these potential osteological synapomorphies re-

quire reversals within some phocoids or independent evolution in

odobenids and phocids: (76) cervical vertebrae with small trans-

verse processes and neural spines, (77) cervical vertebrae smaller

than thoracic or lumbar vertebrae with spinal canal nearly as large

as centrum, (79) neural spines on thoracic vertebrae low, (107)

hindflipper claws reduced. (114) ischial spine large, and (116)

lesser femoral trochanter reduced or absent.

Odobenidae

Relationships among walruses are the subject of current study

(see Demere 1994, this volume). Although our data do not resolve

relationships at the generic level, we identified six characters as

supporting monophyly of the group. Two of these are unequivocal

synapomorphies of the postcranial skeleton: (99) pit or rugosity on

metacarpal I and (125) medial process on calcaneal tuber. An

additional three equivocal synapomorphies may diagnose this clade

but are subject to other equally parsimonious interpretations. These

are (17) supraorbital processes completely absent (state 2 of

multistate character), (58) I3 terete and caniniform, and (93) diam-

eter of humeral trochlea larger than that of distal capitulum.

Odobenids are first recognized from the middle Miocene of

California. Not included in our study and undoubtedly representing

a significant position in odobenid evolution is Neotherium minim,

new material of which is being studied by L. G. Barnes.

Phocoidea

As defined by Wyss and Flynn ( 1 993 ), the Phocoidea are a clade

including the common ancestor of phocids and desmatophocids

plus all of its descendants. Seven unequivocal synapomorphies

diagnose this clade: (5) posterior termination of nasals posterior to

frontal-maxillary contact, (22) pterygoid process with fiat, concave

lateral margin, (24) mastoid process distant from paroccipital pro-

cess (state 2 of multistate character), (26) internal auditory meatus

absent and canals for vestibulocochlear nerve completely separated

(state 2 of multistate character), (35) auditory bulla underlaps basi-

occipital, (52) bony flange below ascending ramus, and ( 142) peri-

42

A. Bena and A. R. Wyss

Figure 4. Left forelimb of representative pinnipeds and a generalized

terrestrial arctoid in dorsal view. En, Enaliarctos emlongi; OD. Odobenidae

(Odobenus ros marus); OT, Otariidae (Zalophus californianus); PH, Phocidae

("Monachus" Khauinslandi); UR. Ursidae {Ursus americanus). Pinmpedi-

morph synapomorphies: 87. greater and lesser humeral tuberosities enlarged;

88. deltopectoral crest strongly developed; 90, humerus short and robust; 94,

olecranon process laterally flattened and posteriorly expanded; 95, distal halt

of radius expanded; 1(11. digit I of manus enlarged (see Appendix 1 ).

cardial plexus well developed. Two of 13 equivocal synapomor-

phies most likely represent reversals near the base of the clade: (3)

nasal processes prominent and (16) antorbital process of frontal

large and well developed.

Phocidae

Relationships among extant phocids have recently been ana-

lyzed cladistically (Muizon 1982a; Wyss 1988b). We identify as a

monophylctic clade the archaic seals Piscophoca, Homiphoca, and

Acrophoca and hypothesize a sister-group relationship between

ot yy od £&

Figure 5. Left hindlimb of representative pinnipeds and generalized

terrestrial arctoids in dorsal view. En, Enaliarctos emlongi; OD, Odobenidae

(Odobenus rosmarus); OT, Otariidae (Zalophus californianus); PH,

Phocidae ("Monachus" schauinslandf); UR. Ursidae (Ursus americanus).

Pinnipedimorph synapomorphy: 105, digits I and V of pes elongated (see

Appendix 1 1.

Piscophoca and Homiphoca. Elephant and monk seals (Mirounga

and "Monachus") and extant lobodontines are more closely related

to the archaic seal clade than they are to phocine seals. The Phocinae

(consisting of Erignathus, Cystophora, and the Phocini) are recog-

nized as a monophylctic group in agreement with Burns and Fay

(1970), Muizon (1982a), and Wyss (1988b). The Phocidae are

diagnosed by 26 derived characters, eight of which are unequivocal

synapomorphies for the group: (20) alisphenoid canal absent. (23)

mastoid heavily pachyostotic, (28) basilar cochlear whorl directed

transversely. (29) dorsal region of petrosal expanded, (33) opening

of cochlear fenestra outside tympanic cavity, forming a cochlear

foramen, (41) pit for tympanohyal anterior to stylomastoid fora-

men, (112) insertion for ilial psoas muscle on ilium, and (126)

Pinniped Phylogeny

43

process of astragalus directed caudally (slate 2 ofmultistate charac-

ter). Eleven other characters are potential synapomorphies.

This phylogeny implies many character reversals at the base ol

the phocine seal clade, a pattern discussed further below.

Our apparent endorsement of "monachine" monophyly is based

on our not treating "Monachus" schauinslandi as a separate taxon.

Outside the subfamily Phocinae. we don't attribute much signifi-

cance to the intraphocid relationships depicted in Figure 2. For

example, our results reveal a puzzling arrangement in which re-

ported fossil lobodontines (Homiphoca, Acrophoca, and Pisco-

phoca) are not nested among extant lobodontines. Of the three

synapomorphies linking fossil lobodontines, none is unequivocal.

One character. (53) mandibular condyle elevated above tooth row.

requires a reversal among modern lobodontines. For another char-

acter. (64) P4 double rooted, fossil lobodontines represent an inter-

mediate transformation, P4 becoming single rooted among modern

lobodontines and other phocids. A third character, (36) mastoid lip

covering or partially covering external cochlear foramen, either

reverses in "Monachus" and Mirounga or originated independently

among fossil and modern lobodontines. Thus, this arrangement

separating fossil from extant lobodontines is poorly supported.

Although phocids have a temporal range extending back into

the middle Miocene, much of the material is fragmentary. Later

archaic phocids are represented by abundant well-preserved mate-

rial of Acrophoca and Piscophoca from the early Pliocene of Peru

(Muizon 1981) and of Homiphoca from late Miocene or early

Pliocene of South Africa (Hendey and Repenning 1972; Muizon

and Hendey 1980).

EXPERIMENTAL MANIPULATIONS OF DATA

We performed several experimental manipulations of the data

set. In one run we forced otarioid monophyly. The strict-consensus

tree that resulted from 100 trees was 34 steps longer than our

preferred tree. In another run to address the question of diphyly we

forced the monophyly of musteloids and phocids. The strict-con-

sensus tree that resulted from 100 trees was 77 steps longer than our

preferred tree. Finally, in an attempt to determine the role of fossils

in pinniped phylogeny, we excluded all fossil taxa. The resulting

tree showed no major change in topology.

ASSESSING THE PATTERN OF HOMOPLASY

There has been a widespread recent tendency among carnivoran

systematists to assume pervasive convergence of pinnipeds (Wyss

1989). sometimes when such assumptions are unnecessary.

Wozencraft (1989:504) saw the controversy of pinniped mono-

phyly vs. diphyly as centering "on the treatment of parallel and

convergent characters." suggesting that monophyly is favored only

if aquatic adaptations are not excluded. In our view this line of

reasoning is flawed in two respects: ( 1 ) it assumes convergence at

the outset, something for which one needs a phylogeny to uncover,

and (2) it assumes that because a particular structure has some

"adaptational" or functional significance it probably originated in-

dependently and is therefore irrelevant phylogenetically. The im-

plausibility of the latter view is patent: taken to its logical extreme

one would have difficulty in defending the monophyly of even

noncontroversial groups of pinnipeds. The posterior process on the

phocid calcaneum. for example, has important functional implica-

tions in keeping the hindlimb posteriorly extended, yet it has never

been rejected as supporting a common origin for the group. We

regard the distinctively reduced fifth intermediate phalanx on the

pinniped manus (among numerous other features) as equally de-

serving of serious phylogenetic consideration.

Barnes ( 1989: fig. 9) assumed convergence even when such an

assumptions was unnecessary. For example, he viewed enlarged ear

ossicles as originating independently in the Desmatophocinae, wal-

ruses (though he showed these taxa as sister groups), and in a clade

including the Allodesminae and Pinnarctidion, though no charac-

ters bar a linkage between this clade and his desmatophocine-

w alius clade. Thus three originations of this character are proposed

where one would have sufficed. Additionally, the assumption of

convergence implies that the ossicles enlarged independently in the

phocids, too. Thus assuming convergence may violate parsimony.

There is nothing to prevent one from suggesting that any character

has originated independently in every terminal taxon.

We mapped patterns of homoplasy on our strict-consensus tree

(Fig. 2). Reversals exceed convergences 48 to 41. Our analysis

establishes that the majority of reversals, excluding those confined

to terminal taxa. occurred among the phocine seals, a pattern previ-

ously noted by Wyss (1988b) and referred to by Howell ( 1929) as

"retrogressive" evolution. Reversals are more than twice as com-

mon here as at any other place on the tree. Nearly all of these

reversals occurred at the base of the phocine clade rather than

among terminal taxa. These reversals are confined largely to details

of flipper structure (see Wyss 1994, this volume, for further discus-

sion) and include (85) supraspinous fossa slightly larger than

infraspinous fossa. (89) supinator ridge of humerus well developed,

(91) entepicondylar foramen present. ( 100) metacarpal heads keeled

with trochleated phalangeal articulations, ( 101 ) metacarpal I equal

in length to others, ( 103) foreflipper claws long, ( 104) intermediate

phalanx of digit V unreduced, ( 107) hindflipper claws unreduced,

and (108) pes with short, rounded metatarsal shafts with rounded

heads, associated with trochleated phalangeal articulations.

MOLECULAR DATA

Studies of DNA hybridization, amino acid sequences, and chro-

mosomes support pinniped monophyly [see Wyss ( 1987) for a more

detailed review], although there is disagreement as to which group

of terrestrial arctoids pinnipeds are most closely related, or. in the

case of chromosomal work, to which pinnipeds the walrus is most

closely related. Arnason and Widegren (1986) demonstrated that

pinnipeds share four highly repetitive DNA components unique to

pinnipeds or shared with mustelids (with the exception of Mephi-

tis). De Jong (1982) found that in the eye-lens protein alpha lens

crystallin there are two amino acid replacements uniquely shared

by phocids and otariids (the walrus was not sampled). In addition

these workers discovered a similarity between the mustelid Mustela

and the procyonid Bassariscus in two amino acid replacements that

differ from replacements seen in the other carnivores sampled (de

Jong 1986). Significantly, phocids do not share these similarities,

thus failing to support a phocid-mustelid alliance. More recently,

sequencing by Keith et al. (1991) of the milk protein beta

lactoglobin supports a close relationship between phocids and

otariids. Their results indicate that these groups are closer to canids

than to ursids. Neither the walrus nor mustelids have yet been

sampled (Keith, pers. comm.l.

The karyological similarity among pinnipeds supports pinniped

monophyly. Fay e! al. ( 1967) supported Odobenus as a karyological

intermediate between phocids and otariids. Arnason (1974). how-

ever, disputed this conclusion, arguing for a stronger similarity

between otariids and phocids.

We hope future molecular and karyological data will be ana-

lyzed cladistically.

CONCLUSIONS

In summary, we believe that the cases for otarioid monophyly

and pinniped diphyly have not been established. We urge that those

44

A. Berta and A. R. Wyss

continuing to defend these hypotheses analyze the data explicitly

and include character distributions among appropriate outgroups

and all members or potential members of the ingroups. Pinniped

monophvlv is supported overwhelmingly by diverse anatomical

data and is strongly suggested by biochemical data as well.

The historical expectation of convergence among pinnipeds has

clouded the interpretation of their relationships. In the context of a

well-corroborated phylogenetic hypothesis it seems that the pattern

of homoplasy argues for character reversals occurring as commonly

as convergences.

ACKNOWLEDGMENTS

We thank C. A. Repenning and Richard Tedford for critical

readings of the manuscript. Line drawings were skillfully prepared

by Conine Petti (Fig. 3) and Mary Parrish (Figs. 4 and 5). Berta

gratefully acknowledges support by the National Science Founda-

tion (BSR 9006535).

LITERATURE CITED

Arnason. U., and B. Widegren. 1986. Pinniped phylogeny enlightened

by molecular hybridizations using highly repetitive DNA. Molecu-

lar Biology and Evolution 3:356-365.

Barnes L. G. 1972. Miocene Desmatophocidae (Mammalia: Carmvora)

from California. University of California Publications in Geologi-

cal Sciences 89:1-76.

1979. Fossil enaliarctine pinnipeds (Mammalia: Otamdae)

from Pyramid Hill. Kern County, California. Natural History Mu-

seum of Los Angeles County Contributions in Science 318.

1987. An early Miocene pinniped of the genus Desmatophoca

(Mammalia: Otariidae) from Washington. Natural History Mu-

seum of Los Angeles County Contributions in Science 382.

-. 1989. A new enaliarctine pinniped from the Astoria Formation.

Oregon, and a classification of the Otariidae (Mammalia: Car-

nivora). Natural History Museum of Los Angeles County Contri-

butions in Science 403.

Baskin. J. A. 1982. Tertiary Procyoninae (Mammalia: Carmvora) ot

North America. Journal of Vertebrate Paleontology 2:71-93.

Beaumont, G. de. 1965. Contribution a I'etude du genre Cephalogale

Jourdan (Camivora). Schweizerische Palaontologische Abhand-

lungen 82:1-34.

Berta, A. 1991 . New Enaliarctos* (Pinnipedimorpha) from the Ohgocene

and Miocene of Oregon and the role of "enaliarctids" in pinniped

phylogeny. Smithsonian Contributions to Paleobiology 69.

In press. New specimens of the pinniped Pteronarctos from the

Miocene of Oregon. Smithsonian Contributions to Paleobiology.

and t. A. Demere. 1986. Callorhinus gilmorei n. sp., (Car-

nivora: Otariidae) from the San Diego Formation (Blancan) and its

implications for otariid phylogeny. Transactions of the San Diego

Society of Natural History 21:111-126.

— , and C. E. Ray. 1990. Skeletal morphology and locomotor

capabilities of the archaic pinniped Enaliarctos mealsi. Journal ot

Vertebrate Paleontology 10:141-157.

and a. R. Wyss. 1990. Reply to oldest pinniped. Science

248:499-500.

-, C. E. Ray, and A. R. Wyss. 1989. Skeleton of the oldest known

pinniped, Enaliarctos mealsi. Science 244:60-62

Bisaillon. A., and J. Pierard. 1981. Osteologie du morse de l'Atlantique

{Odobenus rosmarus L.. 1758). Ceintures et membres. Zentralblatt

fur Veterinarmedi/in. Reihe C, Anatomia, Histologia, Embryologia

10:310-327.

_ and N. Lariviere. 1976. Le segment cervical des carnivores

(Mammalia: Carnivora) adaptes a la vie aquatique. Canadian Jour-

nal of Zoology 54:43 1 -436.

Bonner, W. N. 1990. The Natural History of Seals. Facts on File, New

York. New York.

Brown. J. C, and D. W. Yalden. 1973. The description ot mammals— 2.

Limbs and locomotion of terrestrial mammals. Mammal Review

3:107-134.

Burns. J. J., and F. H. Fay. 1970. Comparative morphology of the skull

of the nbbon seal. Histricophocafasciata, with remarks on system-

atica of Phocidae. Journal of Zoology 161:363-394.

Davis. D. D. 1964. The giant panda: A morphological study of evolu-

tionary mechanisms. Fieldiana. Zoology 31:285-305.

Demere, T. A. 1994. The family Odobenidae: A phylogenetic analysis of

fossil and living taxa. In A. Berta and T. A. Demere (eds.). Contri-

butions in marine mammal paleontology honoring Frank C.

Whitmore. Jr. Proceedings of the San Diego Society of Natural

History 29:99-124.

Donoghue. M. J. 1985. A critique of the biological species concept and

recommendation for a phylogenetic alternative. Bryologist 88:172-

181.

Doran, A. H.G. 1878. The mammalian ossiculaauditus. Transactions ot

the Linnean Society, 2nd ser„ Zoology (1879), 1:371-497.

English, A. 1975. Functional anatomy of the forelimb in pinnipeds.

PIi'd. dissertation. University of Illinois at the Medical Center,

Chicago. Illinois.

1976. Limb movements and locomotor function in the Califor-

nia sea lion {Zalophus californianus). Journal of Zoology, London

178:341-364.

Fay, F. H. 1981 . Walrus: Odobenus rosmarus. Pp. 1-23 in S. H. Ridgway

and R. J. Harrison (eds.). Handbook of Marine Mammals. Vol. 1.

Academic Press. New York, New York.

1982. Ecology and biology of the Pacific Walrus, Odobenus

rosmarus divergent Illiger. North American Fauna 74.

- V R Rausch, and E. T. Felitz. 1967. Cytogenetic comparison of

some pinnipeds (Mammalia: Eutheria). Canadian Journal of Zool

ogy 45:773-778.

Fleischer. G. 1973. Studien am Skelett des Gehororgans der Saugetiere,

einschhesslich des Menschen. Saugetierkundliche Mitteilungen

21(2-31:131-239.

Flynn. J. J.. N. A. Neff, and R. H. Tedford. 1988. Phylogeny ot the

Carnivora. Pp. 73-116 in M. J. Benton (ed.). The Phylogeny and

Classification of the Tetrapods, vol. 2. Clarendon Press, Oxford,

England.

Gauthier, J.A. 1986. Saurischian monophyly and the origin ot birds. In

K. Padian (ed.). The origin of birds and the evolution of flight.

Memoirs of the California Academy of Sciences 8: 1-55.

A. G. Kluge. and T. Rowe. 1988. The early evolution of the

Amniota. Pp. 103-155 in M. J. Benton (ed.). The Phylogeny and

Classification of the Tetrapods, vol. 1. Clarendon Press, Oxford,

England.

Gordon, K. 1981. Locomotor behavior of the walrus {Odobenus). Jour-

nal of the Zoological Society of London 195:349-367.

. 1983. Mechanics of the limbs of the walrus {Odobenus

rosmarus) and the California sea lion {Zalophus californianus).

Journal of Morphology 175:73-90.

Gregory. W. K. 1910. The orders of mammals. Bulletin of the American

Museum of Natural History 27: 1-524.

Harrison, R. J., L. Harrison Matthews, and J. M. Roberts. 1952. Repro-

duction in some Pinnipedia. Transactions of the Zoological Society

of London 27:437-540.

and J. D. W. Tomlinson. 1956. Observations on the venous

system in certain Pinnipedia and Cetacea. Proceedings of the Zoo-

logical Society of London 126:205-233.

Hendey. Q. B.. and C. A. Repenning. 1972. A Pliocene phocid from

South Africa. Annals of the South African Museum 59:7 1-98.

Hennig, W. 1956. Phylogenetic Systematics. University of Illinois Press.

Urbana, Illinois.

Hough. J. R. 1948. The auditory region in some members ot the

Procyonidae, Canidae and Ursidae. Its significance in the phylog-

eny of the Carnivora. Bulletin of the American Museum of Natural

History 92:67-1 18.

Howell, A. B. 1929. Contribution to the comparative anatomy ot the

eared and earless seals (genera Zalophus and Phoca). Proceedings

of the United States National Museum 73 (15): 1-142.

1930. Aquatic mammals: Their adaptations to life in the water.

Charles C. Thomas. Sprinfield. Illinois.

Hunt. R. M. 1974. The auditory bulla in Camivora: An anatomical basis

for reappraisal of carnivore evolution. Journal of Morphology

143:21-76.

Pinniped Phylogeny

45

— . 1977. Basicranial anatomy of Cynelos Jourdun (Mammalia:

Carnivora), an Aquitanian amphicyonid from the Allier Basin.

France. Journal of Paleontology 5 1 :826-843.

— , and L. G. Barnes. 1994. Basicranial evidence for ursid affinity

of the oldest pinnipeds, In A. Berta and T. A. Demere (eds.).

Contributions in marine mammal paleontology honoring Frank C.

Whitmore, Jr. Proceedings of the San Diego Society of Natural

History 29:57-68.

Illiger, J. C. W. 1811. Prodomus Systematis Mammalium et Avium. C.

Salfeld. Berlin. Germany.

Jong. W. W. de. 1982. Eye lens proteins and vertebrate phylogeny. Pp.

75-1 14 in M. Goodman ted.). Macromolecular Sequences in Sys-

tematics and Evolutionary Biology. Plenum. New York. New York.

. 1986. Protein sequence evidence for the monophyly of the

carnivore families Procyonidae and Mustelidae. Molecular Biol-

ogy and Evolution 3:276-281.

Keith. E. O., J. Grobler, S. Pervais, and K. Brew. 1991 . P. 28 in Pinniped

phylogenies deduced from protein sequence data. Abstracts of the

9th Biennial Conference on the Biology of Marine Mammals (Chi-

cago, Illinois). Society for Marine Mammalogy.

Kellogg. R. 1922. Pinnipeds from Miocene and Pleistocene deposits of

California. University of California Publications in Geology 1 3:23-

132.

King. J. E. 1966. Relationships of the hooded and elephant seals (genera

Cystophora and Mirounga). Journal of Zoology. London 148:385-

398.

. 1969. Some aspects of the anatomy of the Ross seal.

Ommatophoca rossi (Pinmpedia: Phocidae). British Antarctic Sur-

vey Science Papers 63:1-54.

. 1971. The lacrimal bone in the Otariidae. Mammalia 35:465-

470.

. 1977. Comparative anatomy of the blood vessels of the sea

lions Neophoca and Phocarcios; with comments on the differences

between the otariid and phocid vascular systems. Journal of the

Zoological Society of London 181:69-94.

. 1983. Seals of the World. 2nd ed. Cornell University Press,

Ithaca. New York.

, and R. J. Harrison. 1961. Some notes on the Hawaiian monk

seal. Pacific Science 15:282-293.

Ling. J. K. 1965. Functional significance of sweat glands and sebaceous

glands in seals. Nature 208:560-562.

Lyon. G. M. 1937. Pinnipeds and a sea otter from the Point Mugu shell

mound of California. University of California Los Angeles Publi-

cations in Biological Sciences 1:133-168.

McLaren, I. A. 1960. Are the Pinnipedia biphyletic? Systematic Zool-

ogy 9:18-28.

Meijere, J. C. H. de 1894. fiber die Haare der Saugethiere. besonders

iiber ihre Anordnung. Morphologisches Jahrbuch 21:312—124.

Mitchell, E. D. 1966. The Miocene pinniped Allodesmus. University of

California Publications in the Geological Sciences 61 : 1 — 46.

, and R. H. Tedford. 1973. The Enaliarctinae: A new group of

extinct aquatic Camivora and a consideration of the origin of the

Otariidae. Bulletin of the American Museum of Natural History

151:201-284.

Mivart, G. 1885. Notes on the Pinnipedia. Proceedings of the Zoological

Society of London, pp. 484-500.

Muizon.C. de 1981. Les vertebresfossilesde la formation Pisco (Perou).

Recherche sur les grandes civilisations. Instituts Francais d'Etudes

Andines Memoire 6.

. 1982a. Phocid phylogeny and dispersal. Annals of the South

African Museum 89:175-213.

. 1982b. Les relations phylogenetiquesdes Lutrinae (Mustelidae.

Mammalia). Geobios. Memoire Speciale 6:259-277.

, and Q. B. Hendey. 1980. Late Tertiary seals of the South

Atlantic Ocean. Annals of the South African Museum 82:91-128.

Murie, J. 1871. Researches upon the anatomy of the Pinnipedia. Part I:

On the walrus ( Trichechus rosmarus). Transactions of the Zoologi-

cal Society of London 7:41 1^464.

Noback, C. R. 1951 . Morphology and phylogeny of hair. Annals of the

New York Academy of Sciences 53:476-491 .

Nojima, T. 1990. A morphological consideration of the relationships of

pinnipeds to other carnivorans based on the bony tentorium and

bony falx. Marine Mammal Science 6:54-74.

Novacek. M. J. 1986. The skull of lepticlid insectivorans and the higher-

level classification of eutherian mammals. Bulletin of the Ameri-

can Museum of Natural History 183: l-l 12.

Repenning, C. A. 1972. Underwater hearing in seals: Functional mor-

phology. Pp. 307-331 in R. J. Harrison (ed.). Functional Anatomy

of Marine Mammals, vol. I . Academic Press. London. England.

— . 1990. Oldest pinniped. Science 24X4W

— , R. S. Peterson, and C. Huhbs. 1971. Contributions to the

systematica of the southern fur seals, with particular reference to

the San Fernandez and Guadalupe species. Pp. 1-34 in W. H. Burt

(ed.). Antarctic Pinnipedia. Antarctic Research Series 18.

— , and C. E. Ray. 1977. The origin of the Hawaiian monk seal.

Proceedings of the Biological Society of Washington 89:667-688.

— , and R. H. Tedford. 1977. Otarioid seals of the Neogene. United

States Geological Survey Professional Paper 992.

— . C. E. Ray, and D. Grigorescu. 1979. Pinniped biogeography.

Pp. 357-369 in J. Gray and A. J. Boucot (eds.). Historical Biogeog-

raphy, Plate Tectonics, and Changing Environment. Oregon State

University Press, Corvallis, Oregon.

Robinette, H. R.. and H. J. Stains. 1970. Comparative study of the

calcanea of the Pinnipedia. Journal of Mammalogy 51 :527-541.

Savage. R. J. G. 1957. The anatomy of Potamotherium, an Oligocene

lutnne. Proceedings of the Zoological Society of London 129:151-

244.

Scheffer. V. B. 1964. Hair patterns in seals (Pinnipedia). Journal of

Morphology 115:291-304.

, and K. W. Kenyon. 1963. Baculum size in pinnipeds. Zeitschnft

fur Saugetierkunde 28:39— 11.

Schmidt-Kittler, N. 1981. Zur stammegeschichte der marderverr-

wandten Raubtiergruppen (Musteloidea. Camivora). Ecologae

Geologicae Helvetiae 74:753-801.

Segall, W. 1943. The auditory region of the arctoid carnivores. Zoology

Series, Field Museum of Natural History 29:33-59.

Simpson, G. G. 1945. The principles and a classification of mammals.

Bulletin of the American Museum of Natural History 85:1-350.

Swofford. D. L. 1991. PAUP: Phylogenetic analysis using parsimony.

Version 3.0s. Computer program distributed by the Illinois Natural

History Survey. Champaign. Illinois.

Tarasoff, F. J. 1972. Comparative aspects of the hind limbs of the river

otter, sea otter, and seals. Pp. 333-359 in R. J. Harrison (ed.).

Functional Anatomy of Marine Mammals, vol. 1 . Academic Press.

New York, New York.

Tedford, R. H. 1976. Relationship of pinnipeds to other carnivores

(Mammalia). Systematic Zoology 25:363-374.

, L. G. Barnes, and C. E. Ray. 1994. The early Miocene littoral

ursoid camivoran Kolponomos: Systematica and mode of life. In A.

Berta and T. A. Demere (eds.). Contributions in marine mammal

paleontology honoring Frank C. Whitmore. Jr. Proceedings of the

San Diego Society of Natural History 29: 11-32.

Weber. M. 1904. Die Saugetiere. Einfurung in die Anatomie und

Systematik der Recenten und Fossilien Mammalia. Jena, Germany.

Wible, J. 1991. Origin of Mammalia: The craniodental evidence reex-

amined. Journal of Vertebrate Paleontology 11:1-28.

Wiley, E. O., D. Siegel-Causey, D. R. Brooks, and V A. Funk. 199 1 . The

compleat cladist. A primer of phylogenetic procedures. University

of Kansas Museum of Natural History Special Publication 19.

Wozencraft, C. 1989. The phylogeny of the Recent Carnivora. Pp. 495-

535 in J. L. Gittleman (ed). Carnivore Behavior, Ecology, and

Evolution. Cornell University Press, Ithaca, New York.

, and D. M. Decker. 1991. Phylogenetic analysis of recent

procyonid genera. Journal of Mammalogy 42:55.

Wyss. A. R. 1987. The walrus auditory region and monophyly of pinni-

peds. American Museum Novitates 2871.

— . 1988a. Evidence from Hipper structure for a single origin of

pinnipeds. Nature 334:427^128.

. 1988b. On "retrogression" in the evolution of the Phocinae and

phylogenetic affinities of the monk seals. American Museum

Novitates 2924.

. 1989. Flippers and pinniped phylogeny: Has the problem of

convergence been overrated? Marine Mammal Science 5:343-375.

. 1 994. The evloution of body size in phocids: Some ontogenetic

46

A. Berta and A. R. Wyss

and phylogenetic observations. In A. Berta and T. A. Demerefeds.).

Contributions in marine mammal paleontology honoring Frank C.

VVhitmore, Jr. Proceedings of the San Diego Society of Natural

History 29:69-76.

-. and J. J. Flynn. 1993. A phylogenetic analysis and definition of

the Carnivora. Pp. 32-52 in F. S. Szalay, M. J. Novacek. and M. C.

McKenna (eds.). Mammal Phytogeny. Springer- Verlag. Berlin.

Germany.

Yalden, D. W. 1970. The functional morphology of the carpal bones in

carnivores. Acta Anatomica 77:481-501.

APPENDIX 1

Craniodental. postcranial. and soft-anatomical characters exam-

ined among recent and fossil pinnipeds. The discussion of a

character's hypothesized sequence of transformation is an a poste-

riori assessment based on the distribution of that feature on our

strict-consensus tree.

Skull

1. Premaxilla-nasal contact. 0 = extensive, 1 = reduced. In

Odobenus, Allodesmus, and the Phocidae (Wyss 1987:7. 15. fig. 5)

the contact between the premaxilla and nasal is short and narrow.

Wozencraft (1989) incorrectly identified phocids and lutrines as

sharing a reduced premaxilla-nasal contact. Among lutrines, the

premaxilla-nasal contact is reduced only in the sea otter, Enhydra.

Wyss noted that the premaxillae of Odobenus differ from those of

phocids in being broadly sutured with the nasals inside the nasal

cavity; in phocids no such internal contact occurs. An undescribed

fossil odobenid of the genus Imagotaria (LACM 118675) shows

the primitive condition of a broad contact between the premaxilla

and nasals. This derived feature we judge diagnostic of Allodesmus

+ Desmatophoca + phocids (Phocoidea) and as an autapomorphy of

Odobenus or a reversal in Imagotaria.

2. Premaxilla. 0 = ascending process visible laterally along entire

length, 1 = ascending process dips into nasal aperture. According to

Muizon ( 1982a: 186, 187, fig. 4), in "monachines" the premaxilla-

maxilla suture is, in its medial part, located inside the nasal aper-

ture." This condition applies strictly to neither "M." monachus nor

Homiphoca and according to our most parsimonious tree is likely

primitive for phocids. Although most phocines show the primitive

condition of this character, variation exists with Histricophoca and

Pagophilus posessing the derived "monachine" condition (Muizon

1982a). Allodesmus and Desmatophoca show a similar derived

condition (see Barnes 1972, fig. 4; 1987, fig. 1 ). The derived state is

a synapomorphy for phocoids with several reversals.

3. Nasal processes of premaxilla. 0 = not prominent, 1 = promi-

nent, protrude dorsal and anterior to alveolar margin, 2 = well

elevated anterior and dorsal to alveolar margin. As Howell ( 1929)

first noted, there is a well-defined process formed by the premaxil-

lary tips in Zalophus that is absent in Phoca. In Odobenus, the nasal

processes are elevated well above the alveolar incisor margin,

owing to the great modifications of the snout. As noted by

Repenning and Tedford (1977:18), this condition distinguishes

Odobenus from other odobenids.

Prominent nasal processes do not occur in ursids, Enaliarctos,

Allodesmus, Desmatophoca, or phocids. An intermediate condition

in which the nasal process are prominent and protrude (but are not

elevated) dorsal and anterior to the incisor alveolar margin distin-

guishes Pteronarctos. otariids. odobenids, and phocids primitively

(i.e., Imagotaria, Aivukus, and Homiphoca). Hence the presence of

prominent nasal processes is most parsimoniously interpreted as

having originated at the level of Pinnipediformes and having been

lost among phocoids. Its presence in Homiphoca is regarded as an

independent derivation.

4. Frontals. 0 = do not extend anteriorly between nasals, 1 =

extend anteriorly between nasals. Otariids display a characteristic

W-shaped nasofrontal contact, in which the frontals extend anteri-

orly between the nasals (King 1983:151, fig. 6.4). In other pinni-

peds and most terrestrial carnivorans the frontals and nasals do not

show this relationship. Wozencraft (1989:521) incorrectly main-

tained that odobenids and otariids share the derived condition, a W

or divergent shape. Both juveniles and adults of Odobenus, as well

as Imagotaria. maintain a horizontal line of contact between the

nasals and the frontals.

The derived condition is an autapomorphy for all taxa more

closely related to living otariids than to other pinnipeds. However, it

should be noted that in at least one nonotariid, Pteronarctos

goedertae (see Barnes 1989: figs. 1,2). the frontals extend slightly

between the nasals, which might be interpreted as incipient devel-

opment of the derived condition; accordingly, we scored the condi-

tion scored in this taxon as variable.

5. Posterior termination of nasals. 0 = at or near frontal-maxil-

lary contact, 1 = posterior to frontal-maxillary contact. The nasals'

narrowing greatly posteriorly and terminating far posterior of the

frontal-maxillary contact is a synapomorphy uniting Desmato-

phoca. Allodesmus, and phocids (Berta 1991). In terrestrial carni-

vorans, Enaliarctos. Pteronarctos. otariids, and odobenids the

nasals terminate at or near the broad frontal-maxillary contact.

6. Palatine process of maxilla. 0 = terminates at last molar. I =

extends behind last molar, 2 = developed as a shelf (pterygoid

process of maxilla. Barnes 1987). Barnes (1979:23) noted that in

Pinnarctidion bishopi a "wide, thin, squared posterolateral^ pro-

jecting shelf of the palate is beneath each orbit." Barnes (1987)

described this structure, better developed in Desmatophoca

brachycephala. as an expansive pterygoid process of the maxilla

that forms a thin infraorbital shelf with a prominent posterolateral

corner. He observed that this structure is more prominent in D.

brachycephala than in Allodesmus packardi and D. oregonensis.

Pinnipedimorphs are distinguished ancestrally from terrestrial

carnivorans by having an intermediate condition ( 1 ) in which the

palatine process of the maxilla extends posterior to the last molar.

Berta (1991) recognized the presence of a palatine shelf in

Pinnarctidion, Desmatophoca, Allodesmus as a second derived con-

dition (2).

7. Nasolabialis fossa. 0 = present, 1 = absent. The nasolabialis

fossa, described in Enaliarctos by Mitchell and Tedford ( 1973:220,

234) as a "rather deep fossa for the quadratus labii superioris

muscle," is "located on the rostrum, just anterior to the antorbital

rim." Among terrestrial carnivorans the nasolabialis fossa is present

in the archaic ursids Allocyon and Cephalogale. It is present in

Enaliarctos. Pteronarctos. and Pinnarctidion and absent in all other

pinnipedimorphs (Berta 1991). a distribution suggesting that ab-

sence of the nasolabialis fossa is a pinniped synapomorphy. We

consider its presence in Pinnarctidion a reversal to the primitive

condition.

8. Fossa muscularis. 0 = present. 1 = absent. In ursids, "immedi-

ately behind the lacrimal fossa is a shallow pit. the fossa muscularis,

in which the inferior oblique muscle of the eye arises; the thin dry

floor of this pit is usually broken through on dry skulls, and then

resembles a foramen. ... In Ursus it is relatively enormous, as large

as the lacrimal fossa" (Davis 1964:49). The fossil ursid

Cephalogale has behind its lacrimal fossa a slight depression de-

limited by a ventrally floored ridge, possibly the precursor of the

deep, posteriorly positioned fossa muscularis seen in Enaliarctos

and Pteronarctos.

Because this character could not be unambiguously polarized

from our outgroups we excluded it from the initial run of characters.

Pinniped Phylogeny

47

9. Maxilla. 0 = does not contribute significantly to medial

orbital wall. I = contributes significantly to orbital wall and forms

anterior orbital rim. In terrestrial carnivorans the maxilla is usually

limited in its posterior extent by contact of thejugal or palatine with

the lacrimal (Wyss 1987). Sutures in the orbital region of available

specimens of Enaliarctos are fused, hence the arrangement of bones

in this region cannot be determined. An undescribed species of

Pteronarctos (USNM 335432) shows sutures in this region; al-

though a lacrimal is clearly present it contacts neither the palatine

nor thejugal.

Therefore, we identify the derived condition as a synapomorphy

of Pteronarctos plus the pinnipeds (= Pinnipediformes). Additional

specimens of Enaliarctos may demonstrate this to be a

pinnipedimorph synapomorphy. Barnes ( 1989) used this feature as

an "otarioid" synapomorphy.

10. Lacrimal. 0 = distinct, contacts jugal. 1 = fuses early in

ontogeny to maxilla and frontal, greatly reduced or absent: does not

contact jugal. Associated with the pinniped configuration of the

maxilla is the great reduction or absence of the lacrimal. King

( 1971 (demonstrated the presence of a lacrimal in all extant otariids.

showing that in them, unlike terrestrial carnivorans, the lacrimal

tends to fuse relatively early in ontogeny to the maxilla and frontal,

obscuring it. In no otariid. however, does it contact the jugal or

palatine. As observed by Wyss (1987). a lacrimal is difficult to

identify in phocids and odobenids. Wozencraft ( 1989:522) argued

that the lacrimal, including in otariids and odobenids an orbital

flange, is present in these groups. As noted above, however, this

condition is fundamentally different from that in terrestrial

carnivorans. In his discussion of a related character, Wozencraft

incorrectly argued that lack of contact between thejugal and lacri-

mal also characterizes ursids and mustelids. On the contrary, terres-

trial carnivorans can be distinguished from pinnipeds by their

lacrimal's contacting thejugal or being separated from it by at most

a thin sliver of the maxilla. The distinctiveness of the orbital mosaic

in "otarioids" was highlighted even by a proponent of otarioid

monophyly (Barnes 1989); it occurs, however, in phocids also.

Presence of a lacrimal in Enaliarctos cannot be determined. In

Pteronarctos repenningi (USNM 335432) the lacrimal is distinct

but fails to contact the palatine or thejugal. The derived condition is

a synapomorphy linking Pteronarctos and pinnipeds (Berta 1991 ).

1 1 . Infraorbital foramen. 0 = small, 1 = large. A large infraorbital

foramen is a pinnipedimorph synapomorphy (Berta 1991).

Infraorbital foramina are small in most terrestrial carnivorans ex-

cept amphicynodont ursids (seeTedford et al., 1994, this volume).

12. Orbital vacuities. 0 = absent, 1 = present. Wyss (1987:16,

fig. 5) noted in pinnipeds an unossified space (orbital vacuity) in the

ventral orbital wall near the juncture of the frontal, maxilla, and

palatine. Such orbital vacuities characterize pinnipedimorphs ex-

clusive of Enaliarctos, Pteronarctos, Imagotaria, Aivukus, and

Desmatophoca (Berta 1991).

Wozencraft ( 1989:522) distinguished differences among pinni-

peds in the formation of orbital vacuities. According to him, an

enlarged sphenopalatine foramen eclipses the orbitosphenoid (cre-

ating an orbital vacuity) in otariids and odobenids but not phocids.

Phocids do, however, possess an orbital vacuity that variably in-

cludes the sphenopalatine foramen.

The distribution of this character suggests that orbital vacuities

evolved independently in the three major pinniped groups, among

otariids, in Odobenus, and among some phocoids {Allodesmus +

phocids).

13. Palate. 0 = parallel-sided, I = posteriorly widening. In

phocids. Allodesmus. and to a lesser degree Pinnarctidion, unlike

otariids and odobenids, the palate widens posteriorly, a derived

condition (Wyss 1987). In contrast. Wozencraft (1989:521) identi-

fied a posteriorly broad palate as the primitive condition among

carnivorans. While we recognize that the palate diverges widely in

most terrestrial carnivorans, it does not in ursids or amphicyonids.

Enaliarctos and otariids retain what we interpret to be the ancestral

condition for the Pinnipedimorpha. This condition is a synapo-

morphy uniting phocoids. Because of the variability among out-

groups we polarized this character on a second run of characters.

14. Embrasure pit between P4 and M1. 0 = deep, I = shallow or

absent. Enaliarctos is distinguished from Pinnarctidion, Ptero-

narctos (Barnes 1979, 1989) and all other pinnipedimorphs by its

deep embrasure pit for the crown of M, between P4 and M'. Barnes

further noted that reduction of this pit indicates a corresponding

reduction in the size of the lower carnassial. Terrestrial carnivorans

typically possess a deep embrasure pit on the palate. We here regard

this character as a Pteronarctos + pinniped synapomorphy.

15. Anterior palatine foramina. 0 = on or slightly posterior to

maxillary-palatine suture. 1 = anterior of maxillary-palatine su-

ture. The major palatine foramen (= anterior palatine foramen or

anterior opening of the palatine canal; see Novacek 1986) is situ-

ated anterior of the maxillary-palatine suture in otariids, phocids,

and odobenids but lies on the suture in other arctoids (Davis 1964;

Burns and Fay 1970). Wozencraft ( 1989) incorrectly argued that the

primitive condition is characteristic of all three extant pinniped

families.

Although Barnes (1979) distinguished Enaliarctos mitchelli

from E. inealsi by its paired posterior palatine foramina ( = anterior

palatine foramina), additional material of the latter shows the de-

rived condition to characterize all pinnipedimorphs (Berta 1991 ).

16. Antorbital process of the frontal. 0 = absent or small, 1 =

large and well developed. Barnes (1979) noted that the antorbital

process (often referred to as the lacrimal process) of Pinnarctidion

is not as broadly based as in Enaliarctos and that it protrudes farther

from the side of the skull. We maintain that well-developed

antorbital processes do not occur in Enaliarctos or Pinnarctidion.

As coded here the derived condition occurs in otariids and

odobenids. We interpret it as a convergence of those two families or

as having been present in the Pinnipedia primitively then reversed

in the Phocoidea. Although the antorbital process is lacking in

"Monachus" it occurs among some phocids {Erignathus, Mirounga,

Lobodontini), we infer as a secondary derivation.

17. Supraorbital process. 0 = distinct and blunt, 1 = reduced to a

supraorbital ridge. 2 = completely absent, 3 = large and shelflike.

The primitive condition seen in terrestrial carnivorans is a frontal

with a small, rounded supraorbital process. State 1 in which the

supraorbital process is reduced is seen in Enaliarctos, Pteronarctos,

Allodesmus, Desmatophoca, and Pinnarctidion. Phocids and

odobenids except Gomphotaria have lost the process completely

(2). The large shelflike supraorbital processes (3) of otariids we

consider an autapomorphy of the group.

Barnes ( 1989: 18) argued that absence of supraorbital processes

is primitive for otarioids. He noted that Pteronarctos can be distin-

guished from Enaliarctos on the basis of its larger supraorbital

processes. Additional specimens of these taxa (Emlong collection),

however, show that this distinction does not hold. Enaliarctos and

Pteronarctos both possess small tuberosities or ridges in this re-

gion. Therefore we interpret the reduction of supraorbital ridges in

Enaliarctos, Pteronarctos, and archaic phocoids as independently

derived. Alternatively, the condition in Enaliarctos and Ptero-

narctos may be primitive for pinnipedimorphs.

18. Least interorbital width. 0 = occurs in posteriormost portion

of interorbital septum, 1 = occurs in the anterior half of the

48

A. Berta and A. R Wyss

interorbital seplum. Burns and Fay ( 1970) noted that in Cystophora

and the Phoeini. the interorbital distance is least in the anterior half

of the interorbital septum. In other pinnipedimorphs and in other

carnivorans the interorbital region is narrowest in the posteriormost

section.

19. Foramen rotundum. 0 = separate from anterior lacerate

foramen. I = merged with anterior lacerate foramen [see also

discussion of Barnes" character (d) under Otarioid Monophyly].

Pinnarctidion can be distinguished from Enaliarctos and other

pinnipeds by its having the foramen rotundum separate from the

anterior lacerate foramen. Canids and ursids also share the condi-

tion of a separate foramen rotundum (Barnes 1979, 1987).

Our phylogeny implies that Pinnarctidion is an exception

among pinnipedimorphs in displaying the primitive condition. We

suggest that the derived condition is a pinnipedimorph

synapomorphy. reversed in Pinnarctidion.

20. Alisphenoid canal. 0 = present, 1 = absent. Since the

alisphenoid canal is widespread among terrestrial carnivorans.

including all ursoids, its presence in pinnipeds is undeniably primi-

tive. Thus presence of an alisphenoid canal in the "Otarioidea" does

not support the unity of this group, as argued by Barnes (1989).

Absence of the alisphenoid canal among phocids has long been

regarded as a synapomorphy of that group or as a synapomorphy of

mustelids + phocids.

21. Mastoid visible in dorsal view of skull. 0 = no, 1 = yes. A

lateral swelling of the mastoid is visible in a dorsal view of the skull

in phocine but not "monachine-" phocids (King 1966; Burns and

Fay 1970). This is regarded as the derived condition because it does

not occur in terrestrial carnivorans or other pinnipedimorphs.

22. Pterygoid process. 0 = rounded with convex lateral margin,

1 = flat with concave lateral margin. According to Barnes ( 1989)

Pinnarctidion. Allodesmus, and Desmatophoca are distinguished

from other "otarioid" pinnipeds by their flat pterygoid strut with a

concave lateral margin. We found that phocids also possess the

derived condition.

23. Mastoid. 0 = composed of cancellous bone. 1 = heavily

pachyostic. A pachyostotic mastoid is unique to phocids ( Burns and

Fay 1970).

24. Mastoid process. 0 = close to paroccipital process, the two

connected by a low discontinuous ridge. 1 = close to paroccipital

process, the two connected by a high continuous ridge, 2 = distant

from paroccipital process. In ursids and other arctoids the mastoid

process fails to form a complete ventral ridge that extends back to

the paroccipital process as it does it otariids and odobenids. A high,

continuous ridge joins these processes in Ptewnarctos, otariids, and

odobenids and is thus likely primitive for pinnipedimorphs. In

Pinnarctidion the two processes are separated and not broadly

continuous, although they are connected by a crest (Barnes 1979).

The plesiomorphic condition occurs in terrestrial carnivorans.

including ursids (Mitchell and Tedford 1973:248) and Enaliarctos.

State 1 occurs in Ptewnarctos, otariids. and odobenids. State 2

occurs in phocids, Pinnarctidion. Allodesmus. and Desmatophoca.

25. Round window. 0 = unenlarged, 1 = large, with round

window fossula. In pinnipeds the round window is large and the

fossula apparently serves to shield the secondary tympanic mem-

brane from the distensible cavernous tissue of the middle ear

(Repenning 1972). This fossula is absent in other carnivorans ex-

cept perhaps Potamotherium and the lutrines, in which a very

shallow fossula may incipiently (and variably) be present (Tedford

1976; pets, obs.). In these latter forms the round window is not

greatly enlarged. Among pinnipeds the round window is most

expanded in phocids hut it is also very large in odobenids. The

derived condition is a pinnipedimorph synapomorphy.

26. Internal auditory meatus. 0 = present and canals for

vestibulocochlear and facial nerves closely adjacent. 1 = present

and canals for vestibulocochlear and facial nerv es incipiently sepa-

rated. 2 = absent and canals for vestibulocochlear and facial nerves

completely separated. Derived state 1 occurs in Odobemts and

Imugotaria (Repenning and Tedford 1977); state 2. in phocids.

Allodesmus. Desmatophoca. and Pinnarctidion.

27. Basal whorl of scala tympani. 0 = unenlarged, 1 = very

enlarged. The basal whorl of the cochlea is greatly enlarged in

width and diameter in all pinnipeds (Repenning 1972). This expan-

sion appears to be most marked in odobenids and phocids. Pending

comparative measurements, it may actually prove to be a phocoid +

odobenid synapomorphy.

28. Basal cochlear whorl. 0 = posterolateral to long axis of skull,

1 = transversely directed. In phocids, the basal whorl of the cochlea

runs transverse to the long axis of the skull, rather than posterolater-

ally as in other carnivores including otariids and odobenids

(Repenning 1972). The derived condition is thus a phocid

synapomorphy.

29. Dorsal region of petrosal. 0 = unexpanded. 1 = expanded.

Repenning and Ray (1977) observed that "Monachus" schauin-

slandi can be distinguished from all other phocids by its having a

relatively unexpanded dorsal petrosal region (see also discussion in

Wyss 1988: fig. 2).

30. Pit for tensor tympani. 0 = present. 1 = absent. In terrestrial

carnivorans the tensor tympani originates from a small pit in the

petrosal anterior to the oval window. In pinnipeds this pit is lost and

the muscle originates with the bony eustachian tube (Repenning

1972). Among pinnipedimorphs. this pit is present in Enaliarctos

(Mitchell and Tedford 1973) and Ptewnarctos (Berta 1991 ). Hence,

the derived condition is a pinniped synapomorphy.

3 1 . Cochlear aqueduct. 0 = small, 1 = large. As noted by

Fleischer (1973) the pinnipeds' cochlear aqueduct is greatly en-

larged. Pending a quantified survey among the carnivores of co-

chlear aqueduct dimensions, we tentatively regard this character as

a pinniped synapomorphy.

32. Canal for cochlear aqueduct. 0 = separate from round win-

dow, 1 = merged or nearly merged with round window. In pinnipeds

the cochlear aqueduct is only narrowly separated from the round

window, and in phocids at least the canal for the aqueduct strictly

speaking does not exist (Fleischer 1973). In terrestrial carnivorans

the cochlear aqueduct is a very narrow canal that passes about half

the width of the promontorium through the petrosal itself. In

phocids and Odobemts the connection of the cochlear aqueduct to

the cochlea is via the round window, although Odohenus may still

have a narrow, bony separation between the round window and the

cochlear aqueduct. Otariids retain a condition more primitive than

in other pinnipeds in that their cochlear aqueduct still pierces the

petrosal. Accordingly, the derived condition is an odobenid +

phocid synapomorphy.

33. External cochlear foramen. 0 = absent, 1 = present. Phocids

display a unique external cochlear foramen (Burns and Fay 1970).

an opening at the bulla-mastoid junction just posterior to the au-

ricular foramen and stylomastoid foramen.

34. Petrosal. 0 = not visible in posterior lacerate foramen, 1 =

visible in posterior lacerate foramen. Burns and Fay (1970) and

King ( 1966) have discussed the visibility in phocids of the petrosal

in ventral view through the posterior lacerate foramen. The petrosal

is also visible in Odobemts and its fossil allies (Repenning and

Tedford 1977). Pinnarctidion. Desmatophoca. and Allodesmus

(Wyss 1988b; Berta 1991 ). In the primitive condition seen in terres-

trial carnivorans. Enaliarctos, Ptewnarctos, and otariids the petro-

sal is not visible in ventral view from the posterior lacerate foramen.

Pinniped Phylogeny

49

35. Auditory bulla. 0 = abuts basioccipital, I = underlaps basi-

occipital. In ventral view, the bulla abuts the basioccipital in ursids,

Enaliarctos, Pteronarctos, otariids, and odobenids. In the derived

condition seen in Pinnarctidion, Desmatophoca, Allodesmus, and

phocids, the bulla underlaps the basioccipital (Berta 1991 ).

36. Mastoid lip. 0 = not extensive. 1 = covers or partially covers

external cochlear foramen. As noted by numerous workers

(Repenning and Ray 1977; Muizon and Hendey 1980; Muizon

1982a) extant lobodontine phocids uniquely show a mastoid lip

overlapping the posterior bullar wall and covering the external

cochlear foramen. The derived condition is also seen in fossil

lobodontines, Homiphoca (Muizon and Hendey 1980), Acrophoca,

and Piscophoca (Muizon 1981). The primitive condition in which

the mastoid lip does not cover the external cochlear foramen is seen

in Mirounga, "Monachus" and phocines (Wyss 1988).

37. Caudal entotympanic. 0 = uninflated, 1 = inflated. 2 =

greatly inflated. Odobenus, Allodesmus, and Pinnarctidion show an

intermediate condition, a bulla slightly inflated, whereas in phocids

the bulla is greatly inflated (Wyss 1987:24). Wozencraft (1989:523)

identified inflation of the caudal entotympanic as a feature shared

by canids. procyonids, some mustelids, and phocids. We agree that

some mustelids and procyonids possess an inflated entotympanic

(although not necessarily primitively), but in ursids the caudal

entotympanic is not inflated. From this distribution we interpret the

inflation of the entotympanic in some mustelids and procyonids as

an independent acquisition.

We excluded this character from the initial run of characters

because of the variability among mustelid and procyonid outgroups.

38. Posterior opening of carotid canal. 0 = visible in ventral

view, posteromedial process present, 1 = not visible in ventral view,

posteromedial process absent. In the phocines excluding Erignathus

the posterior opening of the internal carotid canal is not visible in

ventral view owing to prominent bullar inflation (Burns and Fay

1970). In other pinnipeds a bony shelf projects from the dorsal and

or medial margin of the aperture toward the posterior lacerate

foramen. Hence we recognize the derived condition as a synapo-

morphy of the Phocini plus Cystophora.

39. Squamosal-jugal articulation. 0 = splintlike, 1 = mortised,

2 = exaggeratedly mortised. Barnes (1979:23) described in Enali-

arctos and otariids a splintlike arrangement of squamosal and jugal

in which the jugal tapers to a sharp point and the squamosal does

not touch the postorbital process of the jugal. In Pinnarctidion

bishopi the squamosal does not taper but ends in a blunt, vertically

expanded tip. It not only touches the postorbital process of the jugal

but fits into a shallow notch on its posterior side. Barnes further

observed that the mortised articulation in which both the postorbital

process of the jugal and the zygomatic process of the squamosal are

expanded dorsoventrally is more greatly developed in Allodesmus

than in phocids.

Condition 1 unites Pinnarctidion and Desmatophoca; condition

2 unites Allodesmus and the phocids (Berta 1991 ). Barnes ( 1989: 18)

argued that a mortised squamosal-jugal articulation occurs in the

Odobenidae also, an observation with which we disagree [e.g..

Imagotaria (Repenning and Tedford 1977: fig. 4)].

40. Postglenoid foramen. 0 = large, 1 = vestigial or absent. The

primitive condition occurs in Cephalogale, Allocyon, and amphi-

cyonids. In Enaliarctos the postglenoid foramen is small (Mitchell

and Tedford 1973:249). It is absent in "Monachus" but relatively

large in Phoca, suggesting it may be secondarily derived among

some phocids. We have coded the Phocini as polymorphic for this

character since this foramen was present in most but not all speci-

mens of Histricophoca examined by Burns and Fay ( 1970). Berta

(1991) identified the derived condition as a pinnipedimorph

synapomorphy.

41. Pit for tympanohyal [= vagina processus styloidei of

Mitchell and Tedford (1973:227, fig. 9) and Mitchell (1968)|. 0 =

closely associated with stylomastoid foramen. 1 = anterior to

stylomastoid foramen. In ursids (including Cephalogale) the pit for

the tympanohyal lies with the stylomastoid foramen in a common

fossa (Mitchell and Tedford 1973:246), contradicting Wozencraft's

( 1989) statement that ursids are characterized by the derived condi-

tion. In all pinnipedimorphs except phocids the tympanohyal pit

lies very close and posteromedial to the stylomastoid foramen. By

contrast, in phocids the tympanohyal lies ventral and anterior to the

stylomastoid foramen.

42. Basioccipital. 0 = long and narrow, 1 = short, broad, and

widened posteriorly. The derived condition unites odobenids.

phocids, Allodesmus, and Desmatophoca (Wyss 1987; Berta 1991 ).

43. Jugular (= posterior lacerate) foramen. 0 = unenlarged. 1 =

enlarged, 2 = further enlarged medial to basioccipital. Enlargement

of the jugular foramen is a pinnipedimorph synapomorphy (Wyss

1 987; Berta 1 99 1 ). We disagree with Wozencraffs ( 1 989:523 ) claim

that a large posterior lacerate foramen also characterizes canids and

ursids.

A secondary modification of this feature in which the posterior

lacerate foramen extends medial to the tympanic bulla unites the

phocines (see Wyss 1988b: 16).

Barnes' ( 1989) use of an expanded posterior lacerate foramen

as an "otarioid" synapomorphy substantiates our recognition of the

derived condition as distinct from that seen in terrestrial carnivorans

and validates its use in phylogenetic analysis. If this feature is

reliable enough to diagnose "otarioids," it is equally valid in diag-

nosing pinnipedimorphs.

44. Basioccipital-basisphenoid region. 0 = strongly concave, 1

= flat to convex. Burns and Fay ( 1970) noted that in all phocines the

basioccipital-basisphenoid region is flat to convex. In ursids.

Enaliarctos. "monachines," otariids. odobenids, Allodesmus. and

Desmatophoca this region is strongly concave (Davis 1964; Barnes

1972: Repenning and Tedford 1977; Barnes 1987; Wyss 1988b).

Hence, the derived condition is a phocine synapomorphy.

45. Paroccipital process. 0 = small. 1 = enlarged posterolater-

al^. Related to conformation of the mastoid process (character 24)

is the morphology of the paroccipital process. In ursids. Enaliarctos.

Pteronarctos. otariids, odobenids, and phocids the paroccipital pro-

cesses are small. In Desmatophoca. Allodesmus. and Pinnarctidion

(Berta 1991 ) these processes are enlarged posterolaterally.

46. Auditory ossicles. 0 = unenlarged, 1 = enlarged. Enlarged

ossicles unite odobenids. Allodesmus. phocids, Desmatophoca, and

Pinnarctidion. Related to this character is the size of the

epitympanic recess containing the ossicles. Many mustelids, how-

ever, have large epitympanic recesses (sinuses) without enlarged

ossicles.

47. Muscular process of malleus. 0 = present, I = very reduced or

absent. Among terrestrial carnivorans only ursids have lost the

muscular process (site for insertion of tensor tympani)(Doran 1878).

This process is absent in all pinnipeds also. Wozencraft (1989)

incorrectly argued that absence of the muscular process is primitive.

Flynn et al. (1989:94) followed Segall (1943). who reported that

ursids possess at most a very reduced muscular process. Because

Segall united ursids and procyonids on the basis of the reduced

muscular process Flynn et al. did not treat this feature as an ursid-

pinnipedimorph synapomorphy. Wozencraft (1989:524) distin-

guished ursids, melines, mephitines, and lutrines from canids.

procyonids, and mustelines by their small rather than large muscu-

lar processes. Wyss has rechecked Segall's carnivoran ossicle col-

lection at the Field Museum of Natural History (Chicago) and

reaffirmed that the muscular process is indeed invariably absent in

bears and present in procyonids.

50

A. Berta and A. R. Wyss

The derived condition, extreme reduction or loss of the muscu-

lar process on the malleus, we recognize as an ursid-pinnipedi-

morph synapomorphy. Because of the variation in the outgroups,

this character was polarized on subsequent runs of the data.

48. Processus gracilis and anterior lamina of malleus. 0 =

unreduced. 1 = reduced. As observed by Doran (1878). in terrestrial

carnivorans as in most mammals there is a slender process and a

broad lamina extending between the head region and the manubrial

base. In phocids. otariids and Odobenus the processus gracilis and

associated lamina are greatly reduced or absent. Wozencraft (1989)

reversed the polarity of this character.

Berta (1991) used the derived condition to unite all pinnipeds

excluding Enaliarctos. Without further quantification the condition

in Enaliarctos cannot be judged significantly different from that of

other pinnipedimorphs.

49. Middle ear cavity and external auditory meatus. 0 = cavern-

ous tissue absent, 1 = cavernous tissue developed. 2 = unique

pattern of tissue development. The middle ear cavity of pinnipeds is

filled by a distensible tissue thought to inflate with blood in re-

sponse to increasing external pressure during diving (Repenning

1972). Phocids (exclusive of at least "Monachus" schauinslandi

show a unique (at least among pinnipeds) pattern of distribution of

the cavernous tissue, thickest near the floor and roof of the middle

ear cavity, thinning near the eustachian tube, across the tympanic

membrane, and in the epitympanic recess (Wyss 1988b).

50. Pseudosylvian sulcus. 0 = weakly present or absent. 1 =

strongly developed. In Enaliarctos mealsi the "sylvian fossa (or

more correctly pseudosylvian sulcus) is enlarged to a broad and

deep crease down the side of the brain, effectively separating the

cerebrum into front and back halves. Sunken within the fossa is the

gyrus arcuatus primus" (Mitchell and Tedford 1973:237). Accord-

ing to Barnes ( 1979) the "Enaliarctinae" can be distinguished from

other pinnipeds by their prominent pseudosylvian sulcus. He distin-

guished Pteronarctos from Enaliarctos by its shallower pseudo-

sylvian sulcus (Barnes 1989). Our comparisons with additional

specimens of Pteronarctos show that P. goedertae (USNM 335432 )

has strongly developed pseudosylvian sulci.

The pseudosylvian sulcus does not appear in amphicyonids or.

from the skull and endocranial cast, strongly in Cephalogale. The

derived condition occurs in Enaliarctos and variably in Ptero-

narctos (Berta 1991).

Mandible

51. Angular (= pterygoid) process. 0 = unreduced and located

near base of ascending ramus, 1 = reduced and elevated above base

of ascending ramus. A well-developed angular process positioned

near the base of the ascending ramus characterizes terrestrial

carnivorans, Enaliarctos, and otariids. The derived condition

occurs in "monachine" phocids, odobenids. Allodesmus, and

Desmatophoca (Emlong specimens).

52. Flange below ascending ramus. 0 = absent, 1 = present. A

thinning and ventral extension of the posterior end of the mandibu-

lar ramus to form a bony flange below the angular process unites

Allodesmus, Desmatophoca. and phocids (Berta 1991 ). Terrestrial

carnivorans and other pinnipedimorphs do not develop this flange.

53. Mandibular condyle. 0 = at or slightly above level of tooth

row. 1 = well elevated above tooth row. The mandibular condyle in

Allodesmus, Desmatophoca. Piscophoca, and Acrophoca is el-

evated above the tooth row. In most terrestrial carnivorans and all

other pinnipedimorphs the condyle is low.

Dentition

54. Deciduous dentition. 0 = unreduced, 1 = reduced. Numerous

authors (e.g.. King 1983) have noted that in pinnipeds the size of the

deciduous teeth is reduced.

55. Upper incisors. 0 = six. 1 = four. Living and fossil

monachines have reduced the upper incisors to four from the typical

pinniped and terrrestrial carnivoran number of six (King 1966;

Muizon 1982). The apparently reduced I1 in Allodesmus may indi-

cate a trend toward incisor reduction early in phocoid evolution

(Wyss 1988b). This tooth is reduced or lost in the odobenines

(Barnes 1989).

56. Upper incisor roots. 0 = transversely compressed, 1 = round.

As noted by King ( 1966), the roots of the upper incisors, particu-

larly the first two, generally are extremely compressed transversely

among carnivorans (including otariids. Enaliarctos. Pteronarctos,

early odobenids, Desmatophoca, Allodesmus, Cystophora, and the

Phocini). "Monachines" and Erignathus are characterized by roots

rounder in cross-section. We recognize the derived condition as a

phocid synapomorphy with a reversal in phocines.

57. I1"2, transverse groove. 0 = present, 1 = absent. In otariids

the first two upper incisors have a deep transverse groove (King

1983:165). This "double cusping" is also present in ursids, canids,

amphicyonids. Enaliarctos, Pteronarctos, and early odobenids. The

derived condition unites phocids.

58. I3. 0 = incisiform with oval cross-section, 1 = eaniniform

with circular cross-section. In fur seals the lateral incisor is

incisiform with an oval cross-section, whereas in sea lions it is

eaniniform with a circular cross-section (Repenning et al. 1971).

Berta and Demere (1986) identified Enaliarctos and the fossil

otariids Thalassoleon and Pithanotaria as sharing the primitive

condition. The Otariinae {Zalophus, Otaria. Eumelopias.

Neophoca, and Phocarctos) and Odobenidae share the derived

condition.

In Desmatophoca I3 is procumbent and oval in cross-section

(Barnes 1987). This tooth is absent from reported specimens of

Allodesmus, although Barnes ( 1972:14) mentioned its procumbent

roots.

59. I3, lingual cingulum. 0 = present, 1 = absent. A simple lateral

incisor lacking a lingual cingulum characterizes most fossil and

modern otariids (Berta and Demere 1986). Pithanotaria starri

shows the primitive ursid condition, in which the crown of I

broadens posteriorly near the base and has a distinct posteromedial

lingual cingulum (Repenning and Tedford 1977). The derived con-

dition also occurs in phocids, odobenids. and Desmatophoca (Berta

1991). This character is usually but not always associated with

"double cusping" of I ' ": and is a further extension of it on the lateral

incisor.

60. Number of lower incisors. 0 = three, 1 = two or none.

Pinnipeds have two lower incisors (King 1983); ursids, amphicyo-

nids, and canids have three. Because the number of lower incisors is

unknown for Enaliarctos or Pteronarctos we tentatively regard this

character as a pinniped synapomorphy, recognizing that it might be

as general as the Pinnipedimorpha.

6 1 . Upper canines. 0 = same size as lower, 1 = larger than lower.

Dusignathus and Imagotaria can be distinguished from other

odobenids by their upper and lower canines of similar sizes

(Repenning and Tedford 1977). In contrast, odobenines (Aivukus,

Alachtherium. Gomphotaria, Odobenus) have elongated upper ca-

nines, as a derived condition. Enaliarctos, Pteronarctos, otariids,

and phocoids share the primitive condition.

62. P\ 0 = double rooted. I = single rooted. The third premolar

of terrestrial carnivorans and Enaliarctos bears two separate roots.

Barnes (1989) distinguished Pteronarctos from Enaliarctos by the

former's bilobed posterior root. Primitively in otariids, as judged

from Pithanotaria and Thalassoleon (Repenning and Tedford

1977). P' is double rooted. In odobenids. Allodesmus, and

Pinniped Phylogeny

51

Desmatophoca P3 has a single root with two or three lobes. The

double-rooted condition of this tooth among phocids represents an

apparent reversal to the primitive condition (Berta 1991), or. if

odobenids are monophyletic, it could be a convergence in

Desmatophoca. Allodesmus, and odobenids.

63. P4, protocone shelf. 0 = present 1 = absent. The presence of

a protocone shelf on the upper carnassial has been used to distin-

guish Enaliarctos and Ptemnarctos from all other pinnipedimorphs

(Barnes 1979, 1989). The shelflike protocone is an ursid-

pinnipedimorph synapomorphy (Flynn et al. 1989; Berta et al.

1989). The occurence of a protocone shelf in Pinnarctidion we

regard as a reversal. Because this character could not be unambigu-

ously polarized it was excluded from the initial run of characters.

64. P4. 0 = three-rooted, 1 = three-rooted with posterior root

bilobed, 2 = double rooted, 3 = single rooted. Enaliarctos and

apparently Pinnarctidion (Barnes 1979:24) possess three separate

roots on the upper carnassial. the primitive condition seen in terres-

trial carnivorans. Three derived states may be recognized. In

Ptemnarctos the posterior root is bilobed ( 1 ). Otariids (e.g.,

Thalassoleon, Pithanotaria), odobenids (e.g., Iinagolaria), and

Desmatophoca primitively possess the double-rooted condition of

this tooth (2). In other otariids, other odobenids. most phocids, and

Allodesmus P4 has only a single root (3).

65. M1. 0 = three-rooted, 1 = double-rooted, 2 = single-rooted.

Although M' of Enaliarctos mealsi was originally described as

having three roots, Barnes (1979) determined, in part from addi-

tional material, that it had only two roots. In Cephalogale, Allocyon,

and amphicyonids M1 is three-rooted. Primitively in otariids this

tooth is double-rooted, as in the fossil otariids Pithanotaria and

Thalassoleon (Repenning and Tedford 1977).

The double-rooted (including bilobed and trilobed) condition of

M1 occurs in Ptemnarctos, Desmatophoca oregonensis (Barnes

1989), Pinnarctidion, Enaliarctos (Barnes 1979). the archaic

odobenid Imagotaria, and phocids. In Desmatophoca brachy-

cephala and Allodesmus this tooth is single-rooted (Barnes 1989),

as it is in most phocids and some odobenids.

66. M'~2. 0 = unreduced in size relative to premolars, 1 =

reduced relative to premolars. According to Mitchell and Tedford

( 1973) the degree of reduction of the upper molars in Enaliarctos is

greater than that of any known early arctoid; Berta ( 1991 ) identified

it as a derived condition. Later pinnipedimorphs also show a re-

duced M1 and reduction or loss of M2 (see character 68).

67. M'~2 cingulum. 0 = unreduced. I = reduced or absent.

Archaic "musteloids," Cephalogale, and amphicyonids show the

primitive condition, large external cingulae on the upper molars;

Enaliarctos and other pinnipedimorphs display the derived state in

which the external cingulum is reduced or absent (Mitchell and

Tedford 1973).

68. M2. 0 = present. I = absent. The occurrence of M2 varies

within each of the major pinniped groups. Among walruses, M2 is

lacking in Odobenus and Aivukus (Repenning and Tedford 1977).

In otariids, M2 is lacking in Pithanotaria (although as noted by

Repenning and Tedford this may be an artifact of preservation).

Eumetopias, Neophoca, and variably in Zalophus (King 1983).

Among phocoids Desmatophoca and Allodesmus possess this tooth

but phocids do not.

M2 is present in all ursids and amphicyonids and variably

present among archaic "musteloids" such as Mustelictis, Amphictis,

Amphicticeps, and Plesictis robustus but not P. genettoides; see

Hough (1948).

69. Lower cheek-tooth row. 0 = long, I = short. A short row,

defined relative to the distance from P, to the ascending ramus,

occurs in Callorhinus, Otaria (Berta and Demere 1986), and Ptem-

narctos (Berta, in press). The distribution of this feature suggests it

originated separately in each taxon.

70. Lower premolars, large anterior cusp. 0 = absent. 1 =

present. A large anterior cusp on the lower premolars occurs in

Enaliarctos and Desmatophoca (Berta 1991 ). The primitive condi-

tion, lack of this cusp, characterizes Cephalogale, amphicyonids,

and "musteloids" (Beaumont 1964; Baskin 1982). We suggest inde-

pendent acquisition of this feature in Enaliarctos and

Desmatophoca.

71. M,_,, trigonid and talonid. 0 = present, 1 = suppressed. The

lower molars of amphicyonids. Cephalogale. Enaliarctos, and

Ptemnarctos possess a trigonid. Among all pinnipedimorphs except

Enaliarctos and Ptemnarctos the trigonid has been suppressed. In

Cephalogale. a crestlike entoconid and distinct hypoconid occur,

whereas in Enaliarctos and Pternarctos only the hypoconid is present,

an intermediate condition (Mitchell and Tedford 1973; Berta, in

press). In all other pinnipedimorphs the talonid has been suppressed.

72. M|, metaconid. 0 = present, 1 = reduced or absent. In

amphicyonids the metaconid is variable, being large (Daphoenus,

Daphoenocyon) or reduced (Daphoenictis) (Hunt 1974). In

Cephalogale the metaconid is subequal to the paraconid but is

progressively reduced through the lineage (Beaumont 1965:6. 33).

Enaliarctos is characterized by a greatly reduced metaconid

(Mitchell and Tedford 1973). and this cusp is suppressed in all

pinnipeds except Enaliarctos and Ptemnarctos (Berta. in press).

73. M2. 0 = present. 1 = absent. All pinnipedimorphs except

Enaliarctos and Ptemnarctos lack M, (Berta 1991). This tooth is

consistently present among terrestrial carnivorans, in one species of

Desmatophoca, D. oregonensis (Berta, in press), and perhaps in

Pithanotaria (see Repenning and Tedford 1977:58).

74. M_v 0 = present, 1 = absent. The third lower molar is absent

in all pinnipedimorphs but present in Amphicynodon, Pachy-

cynodon. Allocyon, Cephalogale. and amphicyonids. Tedford

( 1976) united mustelids, procyonids, and phocids (Mustelida) partly

on the basis of the loss of M,. He interpreted the loss of M, in

"otarioids" as independent. We interpret this tooth to have been lost

independently among "musteloids" and pinnipedimorphs.

75. Cheek tooth crowns. 0 = compressed. 1 = bulbous. Bulbous

cheek-tooth crowns characterize Allodesmus, Desmatophoca, and

Dusignathus (Barnes 1989) as well as phocids.

Axial Skeleton

76. Cervical vertebrae, transverse processes and neural spines. 0

= large. 1 = small. Howell (1929:20) noted that well-developed

transverse processes and neural spines on the cervical vertebrae

characterize otariids, whereas the cervical vertebrae are smaller and

the transverse processes less stout in phocids. The cervical verte-

brae of Odobenus more nearly resemble those of phocids in their

small size (see comments below). The primitive condition charac-

terizes ursids (Davis 1964:78). The condition in Enaliarctos is

unknown. The derived condition is thus either a synapomorphy

uniting odobenids and phocoids. with Allodesmus representing a

reversal, or originated independently in odobenids and phocids.

77. Cervical vertebrae. 0 = larger than thoracic and lumbar, with

spinal canal less than one-half the diameter of the centrum. 1 =

smaller than thoracic and lumbar, with spinal canal nearly as large

as centrum. Odobenus and phocids share cervical vertebrae that are

smaller than the thoracics and lumbars (Fay 1981:10). In otariids

the cervical vertebrae are larger than the thoracics, a condition we

regard as primitive on the basis of outgroup comparison. The

condition in Enaliarctos is unknown (Berta and Ray 1990). In

Allodesmus the cervical vertebrae appear larger than the thoracics

(Mitchell 1966:8, pi. 7).

Like the previous character, this one is an odobenid + phocoid

52

A. Berta and A. R. Wyss

synapomorphy. with Allodesmus representing a reversal, or a

feature independently derived in odobenids and phocids.

78. Atlas, vertebrarterial (= transverse) foramen. 0 = visible in

posterior view. 1 = visible in dorsal view. Among phocids two

conditions occur (King 1966). In "Monachus" monachus the fora-

men is visible only in posterior view, as in most terrestrial

carnivorans (except canids), otariids. the fossil odobenid Imago-

taria, Allodesmus. and phocines. In Mirounga and lobodontine

phocids, the transverse foramen is visible dorsally. In "Monachus"

tropicalis and Odobenus the foramen is partially visible in dorsal

view (Wyss 1988).

79. Thoracic vertebrae, neural spines. 0 = high, 1 = low. In

contrast to phocids and Odobenus, which have low neural spines,

otariids show high neural spines on the thoracic vertebrae (King

1983: 156). Ursids, Enaliarctos, and Allodesmus possess high neural

spines (Davis 1964; Berta and Ray 1990; Mitchell 1966: 10. pi. 10).

80. Lumbar vertebrae, transverse processes. 0 = short, 1 = long.

Otariid lumbar vertebrae show small transverse processes and closely

set zygopophyses, while those of phocids show larger transverse

processes and more loosely fitting zygopophyses (King 1983:156).

In Odobenus. as in the Phocidae, the transverse processes are two or

three times as long as wide (Fay 1981 : 10). whereas in otanids these

processes are about as long as wide. In ursids the transverse processes

are relatively short (Davis 1964). The transverse processes of the

lumbar vertebrae of Allodesmus (Mitchell 1968: pi. 11) and

Enaliarctos (Berta and Ray 1990) are longer than wide. We interpret

the derived condition to have arisen independently in Enaliarctos and

in a group including odobenids, Allodesmus. and phocids or as

primitive for pinnipedimorphs with a reversal in otariids.

81. Lumbar vertebrae. 0 = six, 1 = five. Five lumbar vertebrae

are present in most pinnipeds, although six are more usual in

walruses (Fay 1981; King 1983:154). In Ursus the number of

lumbars is six in 79% of specimens and five in the remaining 21%

(Davis 1964:74, table 9). Enaliarctos had six lumbar vertebrae

(Berta and Ray 1990), whereas Allodesmus had five (Mitchell

1966).

As we have coded this character, it diagnoses the pinnipeds with

a reversal in walruses.

Pectoral Girdle and Forelimb

82. Scapula, hooklike process of teres major. 0 = absent, 1 =

present. This process is common to all phocids except Mirounga

and "Monachus." The shape of the caudal angle in Mirounga and

"Monachus" more nearly resembles that in odobenids. otariids, and

Allodesmus. Hence we regard the hooklike process as an

apomorphy of phocines and lobodontines (Wyss 1988b).

83. Scapula, acromion process. 0 = knoblike, 1 = reduced. The

acromion process is reduced in the phocines. A knoblike acromion

occurs in ursids, Allodesmus. Odobenus, otariids, and Enaliarctos

and is therefore likely primitive for pinnipeds (Wyss 1988b).

84. Scapula, scapular spine. 0 = unreduced, 1 = slightly reduced,

2 = very reduced. Phocids exemplify three distinctive patterns of

scapular spine development (Wyss 1988b: 17): a strongly developed

spine that may extend to the vertebral scapular border, as in

phocines. an intermediate condition in which the spine reaches or

nearly reaches the scapular margin but is less prominent than in

phocines. a condition seen in "Monachus" and Mirounga, and spine

extremely reduced, serving only as a support of the acromion

process, as in lobodontines. The scapular spine is large and well

developed in ursids (Davis 1964) and amphicyonids, e.g.,

Daphoenocyon (Hough 1948).

85. Supraspinous fossa. 0 = slightly larger than infraspinous

fossa, 1 = considerably larger than infraspinous fossa. A large

supraspinous fossa is a constant feature in otariids. Odobenus.

Allodesmus. and Enaliarctos (Mitchell 1966; Bisaillon and Pierard

1981; Berta and Ray 1990: fig. 3). In relation to the infraspinous

fossa, the supraspinous fossa tends to become substantially re-

duced, particularly among the phocines. As a result the scapula of

these taxa could be interpreted as more closely resembling that

typical of terrestrial carnivorans than that of any other pinniped

(Wyss 1988b). Berta and Ray ( 1990) identified the derived condi-

tion as a pinnipedimorph synapomorphy. as it is considered here. It

is one of a very few possible otarioid synapomorphies (accepting a

monophyletic Monachinae and convergence between that group

and "otarioids"). but is contradicted by overwhelming evidence of

pinniped monophyly.

86. Secondary spine of scapula. 0 = absent, 1 = present. A ridge

subdividing the supraspinous fossa is present in otariids (King

1983) but not in walruses or phocids (English 1975). The condition

of the spine in Enaliarctos. Pteronarctos, and Allodesmus is un-

known. Accordingly, we regard the secondary scapular spine as an

otariid synapomorphy.

87. Greater and lesser tuberosities of humerus. 0 = unenlarged, 1

= enlarged. In pinnipeds, the greater and lesser tuberosities are very

prominent relative to the primitive camivoran condition, although

the greater is considerably more enlarged in otariids and the lesser

more enlarged in phocids (Howell 1929). Enaliarctos has enlarged

humeral tuberosities (Berta and Ray 1990), thus the derived condi-

tion is a pinnipedimorph synapomorphy.

88. Deltopectoral crest of humerus. 0 = not strongly developed,

1 = elongated and strongly developed. 2 = short and strongly

developed. Pinnipedimorphs are distinguished from terrestrial

carnivorans by having strongly developed deltopectoral crests. In

"monachine" phocids, otariids, odobenids, and Allodesmus the del-

toid crest is elongated, extending two-thirds to three-quarters the

length of the shaft at which point the crest and shaft merge

smoothly. In phocines the deltoid crest extends less than one half

the length of the shaft and ends abruptly, in lateral view nearly

overhanging the shaft. The insertion of the pectoralis is then more

proximally restricted. The shorter, more abruptly ending crest in

phocines does not represent the generalized phocid condition but is

more likely a secondary derivation (Wyss 1988b). a conclusion

supported by our analysis.

89. Supinator ridge of humerus. 0 = well developed, 1 = absent

or poorly developed. The supinator ridge, absent in otariids,

odobenids, and A I lode smus(Repenrimg andTedford 1977; Mitchell

1968) is well-developed in terrestrial carnivorans, including ursids,

procyonids, some mustelids (Davis 1964), and Enaliarctos (Berta

and Ray 1990). As noted by King (1966), this ridge is strongly

developed in phocines and absent in "monachines."

90. Humerus. 0 = long and slender. 1 = short and robust.

Following Wyss (1989). Berta and Ray (1990) identified a short,

robust humerus as a pinnipedimorph synapomorphy. In terrestrial

carnivorans, the humerus is longer and more slender than that in

pinnipeds (English 1975:90).

91. Entepicondylar foramen. 0 = present, 1 = absent. An

entepicondylar (= supracondylar) foramen is usually found in

phocines but not in "monachines" (some exceptions have been

reported among fossil "monachines") or other pinnipeds (King

1983: 157). An entepicondylar foramen is absent in Enaliarctos. It is

present in Ailuropoda and Tremarctos but otherwise absent in the

Ursidae. It is large in Polamotherium (Savage 1957) and

amphicyonids. Absence of an entepicondylar foramen is the ances-

tral pinnipedimorph condition (Berta and Ray 1990). Uncertainties

Pinniped Phylogeny

53

in polarity notwithstanding, absence of this foramen cannot effec-

tively be used to diagnose "otarioids" (Barnes 1989) because this

condition also likely pertains ancestrally to phocids.

92. Olecranon fossa. 0 = deep. 1 = shallow. The humerus of all

pinnipedimorphs including Enaliarctos is characterized by a shal-

low olecranon fossa. The olecranon fossa of terrestrial carnivorans

is deep (Berta and Ray 1990). Hence we regard this feature as a

pinnipedimorph synapomorphy.

93. Diameter of humeral trochlea. 0 = same as diameter of distal

capitulum. 1 = considerably larger than diameter of distal capitu-

lum. Repenning and Tedford ( 1977) used this feature to distinguish

odobenids from otariids. In odobenids the anteroposterior diameter

of the trochlea is considerably larger than that of the distal capitu-

lum. In Allodesmus the trochlea is approximately the same diameter

as the distal capitulum. In Erignathus and the Phocini the trochlea is

larger than the distal capitulum. From this distribution, we interpret

this character as having arisen independently in the Odobenidae

and Phocinae, then having been lost in Cystophora.

94. Olecranon process. 0 = knoblike and unexpanded, I = later-

ally flattened and posteriorly expanded. The pinniped condition, in

which the olecranon process is laterally flattened and posteriorly

expanded, is not seen elsewhere in the Camivora or in other aquatic

mammals (Wyss 1989). As identified by Berta and Ray (1990). the

derived condition unites pinnipeds; it does not occur in Enaliarctos.

95. Radius. 0 = convexly arched and unexpanded. 1 = markedly

flattened anteroposterior^, with expanded distal half. The derived

condition characterizes a group at least as inclusive as the

Pinnipedia (Howell 1929; King 1983; Wyss 1988a) and it may be

found to characterize the Pinnipediformes once a Pteronarctos

radius becomes known. It is approached slightly in Potamotherium

(Savage 1957). In terrestrial carnivorans. the radius is convex and

bent in a sigmoid curve in the lateral plane.

96. Pronator teres process. 0 = absent, 1 = present, proximal. 2 =

present, distal. Howell (1929) described a well-defined "pronator

teres process" on the shaft of the medial side of the radius in

pinnipeds. This feature is not strongly marked among terrestrial

carnivorans except Potamotherium (Savage 1957, fig. 24).

Repenning and Tedford (1977) used the position of the pronator

teres process to distinguish otariids. in which the process is more

proximal, from odobenids, in which it is more distal. A more distal

pronator teres process also characterizes "Monachus" Mirounga,

and the fossil lobodontines Acmphoca, Homophoca, and Pisco-

phoca: in Allodesmus, phocines. and extant lobodontines the prona-

tor teres process is positioned proximally.

We consider the pronator teres process a pinnipedimorph

synapomorphy. State 1, a proximally positioned process, is com-

mon to Enaliarctos and otariids. A more distal process, state 2.

unites odobenids and phocids primitively, with the condition in

Allodesmus, lobodontines. and phocines representing reversals.

97. Distally projecting ledge on cuneiform. 0 = absent, 1 =

present. King (1966) considered the distally projecting process

(palmar process) that arcs over the palmar surface of the fifth

metacarpal head as distinctively phocine. This process is absent in

otariids, odobenids (except Imagotaria), Allodesmus. "mona-

chines." and other phocids. Terrestrial carnivorans lack a palmar

process (Yalden 1970).

98. Manus. 0 = central digits (II-IV) more strongly developed. 1

= digit I emphasized, digits II-V progressively smaller. In the hand

of pinnipedimorphs digit I (metacarpal I and proximal phalanx) is

elongated, whereas in other carnivorans the central digits are the

most strongly developed (Wyss 1987:18, fig. 6; Wyss 1989). The

manus of pinnipedimorphs is ectaxonic (Brown and Yalden 1973),

the digits of the pollical side being the longest and those of the ulnar

side being smallest (English 1975:110). Terrestrial carnivorans

show a more symmetrically arranged mesaxonic manus with digit

III the longest, the second and fifth the next longest, and the pollex

the shortest (English 1976:3. table I ).

Berta and Ray (1990) considered digit length individually and

collectively (i.e., progressive decrease in size of digits I-V) as

separate characters.

The derived condition occurs in Enaliarctos (Berta and Ray

1990), so we interpret it as a pinnipedimorph synapomorphy.

99. Metacarpal I, pit or rugosity. 0 = absent, 1 = pit present, 2 =

rugosity present. According to Barnes ( 1989) the pit or rugosity on

the proximal dorsal surface of metacarpal I for attachment of the

pollicle extensor muscle distinguishes odobenids from other

"otarioid" pinnipeds. He identified the imagotariines Imagotaria

and Pliopedia as bearing a pit, the odobenines Aivukus and

Odobenus as bearing a rugosity. Repenning and Tedford (1977)

found the condition in Dusignathus similar to that in Imagotaria.

There is no pit or rugosity in Allodesmus, otariids (Mitchell 1968),

or phocids (Murie 1 87 1 ). Therefore we interpret the pit or rugosity

on metacarpal I as an odobenid synapomorphy.

100. Metacarpal heads. 0 = keeled with trochleated phalangeal

articulations, 1 = smooth, with phalanges fiat, articulations

hingelike. King ( 1966) noted that in phocines (as in most terrestrial

mammals) a longitudinal ridge divides the distal and palmar sur-

faces of the metacarpal head. Coinciding with this arrangement, the

proximal articulation surfaces of the proximal phalanges are marked

by a deep notch on their palmar margins accomodating these

metapodial ridges. By contrast, in other phocids the metacarpal

heads are smooth and the metacarpophalangeal and interphalangeal

articulations are flatter, broader, and hingelike. The "monachine"

configuration closely resembles that seen in otariids, odobenids.

and Allodesmus (Wyss 1988b). As judged from Enaliarctos (Berta

and Ray 1990), the ancestral pinnipedimorph condition is one in

which the metacarpal heads are keeled and the phalangeal articula-

tions are trochleated. The phocine condition thus represents a rever-

sal to the primitive condition.

101. Metacarpal I and II. 0 = approximately equal in size, 1 =

metacarpal I longer. Pinnipeds except phocines are characterized by

having the first metacarpal greatly elongated and thicker in com-

parison to metacarpal II (King 1966; Wyss 1988b: fig. 5). Among

terrestrial carnivorans these elements are approximately equal in

size. Therefore we regard the phocine condition as a reversal to the

primitive condition.

102. Digits, cartilaginous extensions. 0 = absent, 1 = present.

Cartilaginous rods distal to each digit serve to support an extension

of the flipper border: they occur and are long on both the fore- and

hindflippers of otariids. Short cartilaginous extensions are present

in walruses (Fay 1981) and Allodesmus (Mitchell 1966:15). King

( 1969) reported dimunitive cartilaginous extensions in the phocid

Ommatophoca and suggested they probably exist in Hydruga as

well.

As Wyss ( 1987:23) wrote, "it seems conceivable that the primi-

tive pinniped flipper was approximated by that of the walrus (short

cartilaginous extensions present), that in otariids with their empha-

sis on forelimb propulsion these extensions have become greatly

elongate, and that in phocids with their emphasis on hindlimb

propulsion the extensions have become secondarily lost. "

The probable development of cartilaginous extensions in

Enaliarctos (as judged from the flat distal articular surface of the

terminal phalanges on both hands and feet) implies they are primi-

tive for pinnipedimorphs.

103. Foreflipper claws. 0 = long, 1 = short. As noted by King

54

A Berta and A. R. Wyss

( 1966) the fore- and hindflippers of phocines are characterized by

well-developed claws; in other phocids the claws tend to be poorly

produced. In otariids and Odobenus the claws of the manus are

reduced, as was probably the case in Allodesmus. As long claws on

the manus are found among terrestrial carnivorans, we interpret

them as primitive.

104. Manus, digit V, intermediate phalanx. 0 = unreduced, 1 =

strongly reduced. King (1966) distinguished "monachines" from

phocines by the strong reduced fifth intermediate phalanx of their

manus. Yet this condition occurs in all other pinnipeds for which the

region is known. Wyss (1988b. 1989) and Berta and Ray (1990)

listed the derived condition as a synapomorphy of pinnipeds with a

reversal in phocines.

105. Pes. 0 = central digits elongated, 1 = digits I and V

emphasized. Pinnipedimorphs including Enaliarctos have elon-

gated digits I and V (metatarsal I and proximal phalanx) in the pes

whereas in other carnivorans the central digits are the most strongly

developed in the pes (Wyss 1987:18, fig. 6; Wyss 1988a; Berta and

Ray 1990).

106. Metatarsal III. 0 = approximately equal to the others; 1 =

much shorter. Among "monachines" and Cystophora the third meta-

tarsal is considerably shorter than the others (Wyss 1988b, fig. 7).

In other pinnipeds and terrestrial carnivorans the metatarsals are

approximately equal. Thus the "monachine" condition is derived,

with the lengthening of this element among phocines (exclusive of

Cystophora) a reversal to the primitive condition, or a convergence

in "monachines" and Cystophora.

107. Hindflipper claws. 0 = unreduced, 1 = reduced. 2 = mark-

edly reduced. As noted by King (1966) reduced claws on the

hindflipper are common among "monachines." Because the

hindlimb claws (at least on the central three digits) of other pinni-

peds tend to be strongly developed. Wyss ( 1988b) interpreted this

condition as a potential "monachine" synapomorphy. Terrestrial

carnivorans show the primitive condition, well-developed claws on

both the manus and pes.

108. Pes. 0 = short, rounded metatarsal shafts with rounded

heads, associated with trochleated phalangeal articulations, 1 =

long, flattened metatarsal shafts with flattened heads, associated

with nontrochleated, hingelike phalangeal articulations. Correlated

with the morphology of the hand is that of the foot. Pinnipeds

(except phocines) are characterized by relatively long, flattened

metatarsal shafts with flattened heads associated with smooth,

hingelike phalangeal articulations (Wyss 1988, 1989). The ances-

tral pinnipedimorph condition, seen in Enaliarctos, resembles that

of terrestrial carnivorans, in which the metatarsal heads are keeled

and associated with trochleated phalangeal articulations (Berta and

Ray 1990). Therefore we regard phocines as having reverted to the

primitive condition.

109. Pubic symphysis. 0 = fused, 1 = unfused. In terrestrial

carnivorans the pubic symphysis forms a fully ossified union,

whereas in pinnipedimorphs only a ligament binds adjoining bones

(Savage 1957). Berta and Ray (1990) identified the derived condi-

tion as occurring in all pinnipeds except Enaliarctos.

1 10. Ilium. 0 = relatively long, 1 = short. Compared with that of

terrestrial mammals, the pinnipeds' pelvis has a shortened ilium

and an elongated ischium and pubis (King 1983; see Tarasoff

1972:340, table 4 for comparisons among Cards, Littra. Pagophilus

and Zalophus). Berta and Ray ( 1990) identified the derived condi-

tion as a synapomorphy uniting pinnipedimorphs.

111. Ilium. 0 = anterior termination simple, 1 = strongly everted,

laterally excavated anteriorly. Living phocines except Erignathus

are characterized by a lateral eversion of the ilium accompanied by

a deep lateral excavation (King 1966). Terrestrial carnivorans and

other pinnipedimorphs possess the primitive condition in which the

ilium is not strongly excavated laterally.

1 1 2. Insertion for ilial psoas muscle. 0 = on femur. 1 = on ilium.

In all phocids the psoas major muscle inserts on the ventral edge of

the ilium. In all other pinnipeds and terrestrial carnivorans this

muscle inserts on the lesser trochanter of the femur (Muizon 1982:

fig. 183). The derived condition is a phocid synapomorphy.

1 13. Separate foramen in innominate for obturator nerve. 0 =

absent. I = present. A separate foramen for passage of the obturator

nerve, the obturator foramen, occurs in Thalassoleon mexicanus

and Allodesmus, variably in the arctocephaline otariids and

Offo/wu/.vl Repenning and Tedford 1977: Mitchell 1966: Fay 1982).

Phocids (exclusive of "Monacluts" schauinslandi, Piscophoca, and

Acrophoca) lack this foramen (Repenning and Ray 1977; Muizon

1981). The absence of the foramen in terrestrial carnivorans and

Enaliarctos suggests that its absence in phocids (except "XT',

schauinslandi. Piscophoca. and Acrophoca) represents a reversal; a

high degree of variability, however, makes this difficult to judge.

1 14. Ischial spine. 0 = unenlarged. 1 = large. A large dorsally

directed ischiatic spine is present in phocids and odobenids. Ac-

cording to King ( 1983:160. fig. 6.24). muscles attached to this spine

help elevate the hindflippers and produce the phocids' characteris-

tic posture.

Ursids have a small ischial spine, the primitive condition (Davis

1964). The ischial spine is small in Allodesmus also (Mitchell 1966:

pi. 20). Accordingly, this feature is most parsimoniously interpreted

either as a synapomorphy uniting odobenids and phocoids [except

Homiphoca (Muizon and Hendey 1980: fig. 12) and Allodesmus) or

as independently derived in odobenids and phocids.

1 15. Fovea for teres femoris ligament (= lig. capitus femoralis).

0 = present and well developed, 1 = strongly reduced or absent.

Pinnipeds share the derived condition of the position of the fovea on

the head being barely visible and the ligament lacking (King

1983:161). Enaliarctos retains the primitive condition, a well-de-

fined pit on the head for the teres femoris ligament (Berta and Ray

1990).

1 16. Lesser femoral trochanter. 0 = present, 1 = very reduced or

absent. According to King (1983:161) "the lesser trochanter is

present only as a small knob distal to the head in otariids and is

absent in phocids." Allodesmus has a rugose raised area represent-

ing the lesser trochanter (Mitchell 1966). and Odobenus has a

similar scar. The lesser trochanter is extremely well developed in

the fossil walrus Imagotaria, more so than in living otariids. and

contrasting even more strongly with the living walrus (Repenning

andTedford 1977:38). Since walruses primitively possess a distinct

lesser trochanter, apparently its reduction or loss occurred indepen-

dently in later walruses and the phocoid clade [Allodesmus +

phocids), unless it reappeared as an autapomorphy in Imagotaria.

117. Greater femoral trochanter. 0 = small and rounded, 1 =

large and flattened. The derived condition is a pinniped synapo-

morphy, as Berta and Ray ( 1990) identified it in all pinnipedimorphs

except Enaliarctos. In terrestrial carnivorans and Enaliarctos the

greater trochanter is separate from the lateral femoral border rather

than being broadly continuous with it as in pinnipeds.

1 18. Medial inclination of condyles. 0 = slight. 1 = strong. The

angle of inclination of the femoral condyles is the angle formed

across the condyles to a line perpendicular to the shaft (see Tarasoff

1972: table IV for comparisons). A small angle of inclination

(approximately 10°) was noted for Potamolherium (Savage 1957)

and is the common condition for terrestrial carnivorans. With this

femoral specialization is associated an increased angle of slope on

Pinniped Phylogeny

55

the condyles of the tibia. Berta and Ray ( 1990) used the derived

condition to diagnose pinnipedimorphs including Enaliarctos.

1 1 9. Trochanteric fossa of femur. 0 = unreduced. 1 = reduced or

absent. According to King (19X3), the trochanteric fossa is small

but present in phocines and otariids but absent in "monachines."

But. as Muizon (1981) has pointed out, some "monachines" (i.e.,

Homiphoca and Piscophoca) have a trochanteric fossa. The primi-

tive condition, a deep trochanteric fossa, is present among ursids

(Davis 1964). Potamotherium (Savage 1957), and Enaliarctos

(Berta and Ray 1990). The derived condition unites otariids,

odobenids, Allodesmus, and phocids (i.e., the Pinnipedia, with re-

versals characterizing the taxa noted above).

1 20. Patella. 0 = flat, 1 = conical. According to King ( 1 983: 1 6 1 ).

the patella of phocids is flatter, that of otariids and walruses, more

conical. The flat patella of the fossil walrus Imagotaria indicates

that the flattened condition may be primitive for walruses.

Allodesmus possesses a conical patella (Mitchell 1966: pi. 20).

Ursids are characterized by a relatively flat patella (Davis 1964).

Since the patella of Enaliarctos is conical, this condition may be

primitive for pinnipeds (Berta and Ray 1990), with the flattened

condition representing a reversal, occurring once among early wal-

ruses and once among phocids.

121. Post-tibial fossa. 0 = weak, 1 = strong. The post-tibial (=

intercondyloid) fossa is more strongly developed in phocines than

in "monachines" (King 1966). This fossa is shallow in otariids.

Odobenus, Allodesmus, Enaliarctos, and most terrestrial carni-

vorans. Hence, the derived condition is a phocine synapomorphy.

122. Tibia and fibula. 0 = unfused. 1 = fused proximally. The

tibia and fibula are fused at their proximal ends in otariids (except

Callorhinus and the fossil Thalassoleon mexicanus) and phocids

(except "Monachus" schauinslandi). In walruses these elements are

separate, even in old animals (King 1983:161). In Callorhinus the

tibia and fibula are unfused (Lyon 1937). Thalassoleon macnallyae

(in contrast to T. mexicanus) has a proximally fused tibia and fibula

(Repenning and Tedford 1977). These elements are unfused in

Allodesmus (Mitchell 1966), Enaliarctos (Berta and Demere 1986),

and terrestrial carnivorans. This distribution suggests that the an-

cestral pinnipedimorph condition is unfused (Berta and Ray 1990).

123. Calcaneal secondary shelf. 0 = absent, 1 = present. All

living otariids possess a well-developed secondary shelf of the

sustentaculum (Robinette and Stains 1970). According to

Repenning and Tedford (1977:39), this shelf is not seen in

Imagotaria. Nor have we seen it in Odobenus. It is "essentially

lacking" in the fossil otariid Thalassoleon mexicanus and only

slightly developed in Hydrarctos lomasiensis (Muizon 1978). Thus

the derived condition is an autapomorphy of otariids above the level

of Thalassoleon.

124. Calcaneal tuber. 0 = long. 1 = short. In terrestrial

carnivorans, when the calcaneum is in articulation with the astraga-

lus the calcaneal tuber extends far proximal of the astraglar head.

This also tends to be the case in otariids, but in phocids the calca-

neal tuber is shortened and projects posteriorly only as far as the

process of the astragalus. Similarly, in odobenids and Allodesmus

the calcaneal tuber is short and extends only slightly beyond the

head (from Mitchell 1966: pis. 21, 22). In agreement with Wyss

(1987), Berta and Ray (1990) identified the derived condition as a

synapomorphy uniting odobenids. Allodesmus, and phocids.

125. Medial process on calcaneal tuber. 0 = absent, 1 = present.

Repenning and Tedford ( 1977) noted that walruses are character-

ized by a prominent tuberosity on the medial side of the calcaneal

tuber. This process is absent in other pinnipedimorphs and terres-

trial carnivorans. Hence we interpret the derived condition as an

odobenid synapomorphy.

126. Caudally directed process (calcaneal process) of astraga-

lus. 0 = absent. I = present. 2 = well developed. The phocid

astragalus is distinguished by a strong caudally directed process

over which the tendon of the flexor hallucis longus passes. This

arrangement prevents anterior flexion of the foot, resulting in seals'

inability to bring their hindlimbs forward during locomotion on

land. In the living walrus there is at least a tendency toward devel-

opment of a calcaneal process (better developed in Imagotaria;

Repenning and Tedford 1977), and Allodesmus appears to be simi-

lar (see Mitchell 1966: pis. 21. 22). A calcaneal process is absent in

terrestrial carnivorans, Enaliarctos. and otariids (Howell 1930)

We interpret presence of a calcaneal process on the astragalus as

a multistate character. An intermediate condition ( I ) occurs in

walruses and Allodesmus; the second condition (2) is unique to

phocids.

127. Baculum. 0 = unenlarged. 1 = enlarged. Scheffer and

Kenyon (1963), Wyss (1987), and Berta and Ray (1990) showed

odobenids, phocids, and Allodesmus to share the derived condition

of large bacula; otariids retain the primitive unenlarged condition.

Soft-Anatomical and Behavioral

128. Testes. 0 = scrotal. 1 = abdominal (i.e., inguinal). The

testes of otariids and terrestrial carnivorans lie outside the inguinal

ring. In contrast, in phocids and Odobenus the testes are inguinal

(Harrison et al. 1952; Fay 1981, 1982; Davis 1964).

129. Copulation. 0 = terrestrial, 1 = aquatic. Odobenus and

phocids (except Mirounga) copulate in the water, whereas otariids

and other carnivorans copulate on land.

130. Pelage. 0 = abundant, 1 = sparse, 2 = secondary hairs

absent. Berta and Demere ( 1 986 ) used lack of underfur as a deri ved

condition to diagnose sea lions. Sparse underfur is also diagnostic

of Odobenus and phocids (Scheffer 1958), in which it evolved

independently from sea lions. Secondary hairs occur in otariids and

the majority of phocids but are virtually absent in "Monachus,"

Mirounga, and Odobenus (Scheffer 1964; Fay 1982).

131. Natal coat. 0 = black, I = gray or white. Phocids exclusive

of "Monachus" and Mirounga have a first pelage paler than that of

otariids, odobenids. and most terrestrial carnivorans, a condition

that Wyss (1988b) interpreted as a potential synapomorphy uniting

phocines and lobodontines.

132. Primary hair. 0 = medullated, 1 = nonmedullated. Scheffer

( 1964) observed that the primary hairs of otariids have a medulla

but those of phocids and Odobenus do not. Since medullated hair

has been documented for Canis and Mustela (Noback 1951). the

derived condition has been interpreted as a synapomorphy uniting

phocids and Odobenus (Wyss 1987).

133. Mystacial whiskers. 0 = smooth, 1 = beaded. Beaded

mystacial whiskers diagnose all phocids except "Monachus,"

Erignathus (Wyss 1988b). Ommatophoca, and Hydrurga (Ling,

pers. comm.), which have retained or reverted to the primitive

smooth condition.

1 34. Molt. 0 = cornified tissue and hair do not form sheets. 1 =

cornified tissue and hair form sheets during molt. As noted by Wyss

( 1988b). an unusual pattern of molting characterizes Mirounga and

"Monachus" schauinslandi (only species of that genus whose molt

has yet been examined). In these seals, the primary hairs become

fused to the stratum corneum so that when the pelage is shed it

forms large continuous patches held together by this thin layer of

cornified epidermal tissue. Wyss interpreted this feature as an

apomorphy of these two species.

56

A. Berta and A. R. Wyss

135. Pelage units. 0 = arranged alternately, 1 = spaced uni-

formly. Scheffer ( 1964) pointed out that in Odobenus and phocids

the pelage units are arranged in groups of two to four or in rows. In

otariids the pelage units are uniformly spaced. Because the pelage

units of Ursus and Cams are arranged alternately (Meijere 1884).

Wyss (1987:10) considered their uniform arrangement in otariids a

synapomorphy of that family.

1 36. Subcutaneous fat. 0 = thin, 1 = thick. Tarasoff (1972) noted

that walruses and phocids are characterized by thick layers of

subcutaneous fat. These layers are less well developed in otariids

and lacking in other terrestrial carnivorans. including lutrines.

137. Mammary teats. 0 = four, 1 = two. Ursids (Davis 1964).

otariids. and Odobenus have two pairs of nipples, whereas phocids

except "Monachus" and Erignathus have only one pair, thought to

correspond to the posterior pair of other pinnipeds (King 1983).

138. Grooming. 0 = extensive, 1 = lacking. Associated with

sparse pelage is the lack of grooming observed in walruses and

phocids (Tarasoff 1972). Since grooming is recorded for lutrines we

tentatively interpret lack of grooming as the derived condition.

139. External pinnae. 0 = present, 1 = absent. Odobenus and

phocids lack external ear pinnae, the presence of which character-

izes otariids and other terrestrial carnivorans.

140. Sweat-duct orifice position. 0 = distal, I = proximal. In the

adult walrus and phocids sweat ducts open proximal to the opening

of the sebaceous glands. By contrast, in otariids the sweat duct is

more distal (Ling 1965. Fay 1982).

141. Venous system. 0 = hepatic sinus uninflated, caval sphinc-

ter absent, intervertebral sinus small, posterior vena cava single, 1 =

hepatic sinus inflated, caval sphincter well developed, interverte-

bral sphincter large, posterior vena cava duplicate, route for

hindlimbs gluteal. Walruses and phocids share the specialized ve-

nous system outlined above (Fay 1981 ). In contrast, otariids have a

less specialized venous system that more closely approximates the

typical mammalian pattern. Wyss( 1987) used the derived condition

to unite Odobenus and the phocids.

142. Pericardial plexus. 0 = poorly developed. I = well devel-

oped. Another structure of the venous system, a well-developed

pericardial plexus, distinguishes phocids exclusive of "Monachus"

schauinslandi from otariids and Odobenus (Harrison and

Tomlinson 1956; King and Harrison 1961; King 1977; Fay 1981).

143. Trachea. 0 = bifurcation of bronchi anterior, 1 = bifurcation

of bronchi posterior. Fay ( 1981 ) and King (1983: fig. 9.2) observed

that in the walrus and phocids the trachea divides into the two

primary bronchi immediately outside the lung. A similar condition

occurs in ursids and canids. By contrast, in otariids the bifurcation

is more anterior, at the level of the first rib. and the two elongated

bronchi run parallel until they diverge to enter the lungs dorsal to

the heart. Hence the derived condition represents one of the very

few synapomorphies of the Otariidae.

APPENDIX 2

Diagnostic characters for the nodes and terminal taxa in Figure

2 are summarized below according to conventions used by Gauthier

et al. (1988). These diagnoses were obtained from the consensus

topology by means of the "describe-tree" option in PAUP version

3.0s (Swofford 1991 ). Synapomorphies are placed at the level! s) of

generality at which they are observed. Some characters may be of a

more general distribution; these are placed in brackets. Reversals

are designated by a minus sign preceding the character number.

Ambiguous character assignments (including convergences) are

designated by an asterisk following the character number. Only

terminal taxa that could be autapomorphously characterized are

listed.

Pinnipedimorpha: [9], [10], 11, 15, 17*. 19*. 25. 27. [31], 40. 43.

47*. 48. [49], [54], [60], 65*. 66, 72. 80*, 85*, 87, 88. 90, 91*, 92,

96*. 98, 101*. 103*. 104*. 105. 110. 118, 120*

Enaliarctos: 50, 70*

Pinnipediformes: 3*. 9. 10. 14, 24. 64*. [81], 89*, [94], [95], 100*.

108*. [109]. 113*. [115], [117], 119*

Pteronarctos: 69*

Pinnipedia: 7*, 8*. 16*, 30, 59, 63*. 64* (1 to 3), 71. 73*. 81, 94,

95, 115, 117, 119

Otariidae: 4, 12*. 17* ( 1 to 3). -80. 86, 135, 143

Thalassoleon: 64

Unnamed node (Arctocephalus + Callorhinus + Otariinae): 62*,

65* (1 to 2), 123

Unnamed node (Arctocephalus + Otariinae): -113, 122*

Callorhinus: 69*

Otariinae: 58*, 130*

Phocomorpha: 26*. 32. 34. 37*. 42. 46. 51, 57*. 76*. 77*, 79*. 96

(I to 2). 107*. 114*. 116*. 124, 126, 127. 128. 129, 130* (0 to 2),

132. 136. 138. 139, 140, 141

Phocoidea: 1*. 2*. -3, 5. 6* ( 1 to 2). 13.-16,22,24(1 to 2), 26(1

to 2). 35. 39*. 45*. 52. 53*. 65* ( 1 to 2). 75. 133*. 137*. 142

Allodesmus: 26 (2 to 1 ). 39* ( 1 to 2), 62*. -73, -76, -77, -79, -1 14

Desmatophoca: 62*. 64 (3 to 2), 70*

Pinnarctidion: -7,-19, -63, 64* (3 to 0). 65* (2 to 1 ), 68*. -75

Phocidae: 6 (2 to 1 ), 12*, 17* ( 1 to 2), 20, 23, 28, 29. 33, 37* ( 1 to

2). 39* (1 to 2). 41, -45. -53. 56*. 68*. -102. 112, -113. -120.

122*. 126(1 to2)

Unnamed node {Acrophoca + Homiphoca + Piscophoca +

"Monachus," Mirounga + Lobodontini): 55*, 58*, 78*. 84, 96* (1

to 2), 106*. 134*

Unnamed node (Acrophoca + Homiphoca + Piscophoca): -2, 36*,

53*, 64* (3 to 2)

Unnamed node (Homiphoca + Piscophoca): 82, -1 19, 121*

Piscophoca: 65* (2 to 1)

Homiphoca: 3. 16. -91, -1 14

Mirounga: 16

"Monachus": -137

Lobodontini: 16*. 36*. 82*. 84 ( 1 to 2), 96* (2 to I ). 130* (2 to 1 ).

131*. -134

Phocinae: (Erignathus + Cystophora + Phocini): -2, 21, 44, 82*,

83, -85, 88 ( 1 to 2). -89, -91, *93, *97, -100, -101, -103. -104, -

105. -107. -108, 121. 130* (2 to 1). 131*

Unnamed node (Cystophora + Phocini): 18, 38, 43 ( 1 to 2), -56

Cystophora: 55*, -93, 106*

Erignathus: 1 6. - 1 33, - 1 37

Phocini: 1 1 1

Odobenidae: 17* ( I to 2), 58*. 93*. 96* ( 1 lo 2), 99. 125

Imagotaria: 64* (3 to 2). 97*. -116. -120

Unnamed node (Aivukus + Gomphotaria + Odobenus): 55*. 61.

62*. 68*. 78*. 99 ( 1 to 2)

Unnamed node (Gomphotaria + Odobenus): 1*. -3, 12*

Odobenus: 3* (0 to 2). *65 ( 1 lo 2)

Gomphotaria: 1 7* ( 2 to 0). -37. - 1 1 3

Basicranial Evidence for Ursid Affinity of the Oldest Pinnipeds

Robert M. Hunt, Jr.

Division of Vertebrate Paleontology, University of Nebraska, Lincoln, Nebraska 68588-0514

Lawrence G. Barnes

Section of Vertebrate Paleontology, Natural History Museum of Los Angeles County, 900 Exposition Blvd.. Los Angeles,

California 90007

ABSTRACT. — Marine camivorans of the genera Pirmarctidion and Enaliarctos (late Oligocene and early Miocene), acknowledged to be

among the geologically oldest pinnipeds in the fossil record, are now known from crania that supply detailed information on basicranial structure.

These fossils reveal that the basioccipital bone in these genera is deeply excavated on its lateral margins by large embayments that occupy 53.3 to

61.3% of the basioccipital width. Such embayments have not been reported in living pinnipeds but have been identified in ursid and amphicyonid

camivorans. The soft tissues that occupied these sinuses in fossil amphicyonids and ursids remain conjectural, but dissection of the embayed

basioccipital in living ursids demonstrates that this pocket contains a loop of the internal carotid artery nested within a large venous (inferior

petrosal) sinus. Nesting of the artery within the sinus may be a countercurrent heat-exchange mechanism to cool arterial blood flowing to the brain.

Presence of these basioccipital sinuses in Enaliarctos, Pirmarctidion, and other early pinnipeds and their apparent absence in living pinnipeds

suggest they have been lost or modified during evolution in Neogene marine environments. We speculate that during prolonged exercise there may

be less need to cool the brain's blood supply in aquatic environments than in terrestrial habitats. The existence of the basioccipital embayment in the

geologically oldest pinnipeds, coupled with the ursid morphology of their upper carnassial. supplements other evidence indicating that pinnipeds are

derived from an ursid ancestor and does not support the view that pinnipeds are most closely related to mustelids.

INTRODUCTION

In their initial report on the pinniped Enaliarctos, Mitchell and

Tedford ( 1973) described cranial material of three individuals re-

ferred to Enaliarctos mealsi: ( 1 ) the nearly complete holotype skull

(LACM 4321 ). (2) a cranial endocast, and (3) a more poorly pre-

served skull comprising rostral and caudal parts. They also assigned

three upper and three lower isolated cheek teeth to the genus. By the

time Barnes (1979) reviewed the group, all Enaliarctos crania

reported had been discovered on the southern and western slopes of

Pyramid Hill. Kern County, south-central California, in marine

rocks of late Oligocene to early Miocene age. The holotype skull

was found in 1961 by Harold Meals, who, in company with Richard

Bishop, also found the important isolated cheek teeth. The second

(bipartite) skull and the endocranial cast were discovered a decade

earlier in 1950 by paleontologist Chester Stock. Mitchell and

Tedford (1973) created the pinniped subfamily Enaliarctinae for

Enaliarctos, recognizing its significance as an important morpho-

logical link between aquatic pinnipeds and terrestrial arctoid Car-

nivore.

Since these early discoveries, new enaliarctine fossils have been

found at a number of localities along the Pacific coast of the United

States in Washington, Oregon, and California. The late Douglas

Emlong collected enaliarctines now conserved in the National Mu-

seum of Natural History (Smithsonian Institution), Washington.

D.C. (USNM); James and Gail Goedert and Guy Pierson have

contributed new enaliarctines to the Natural History Museum of

Los Angeles County (LACM). These fossils are primarily of early

Miocene to early middle Miocene age ( 16.3 to 23.3 Ma, Harland et

al. 1990), but some probably come from rocks as old as latest

Oligocene. Recent publications on these new enaliarctines have

described important fossils from the Pyramid Hill localities in

California (Barnes 1979; Berta and Ray 1990) and new crania from

the Oregon coast (Barnes 1989, 1990. 1992; Berta 1991). These

discoveries demonstrate not only the diversity of latest Oligocene

to early Miocene pinnipeds along the Pacific coast but also show

that the postcranial skeleton of Enaliarctos mealsi had already

evolved many aquatic specializations characterizing the skeletons

of living otariids.

Although Mitchell and Tedford (1973) erected the subfamily

Enaliarctinae for Enaliarctos mealsi only, Barnes (1979. 1989,

1992) placed additional genera (Pirmarctidion, Pteronarctos,

Pacificotaria) in this subfamily, which he considered a basal otariid

stock from which arose a number of otariid lineages. Recently, most

authors have come to regard the subfamily Enaliarctinae (or its

familial equivalent. Enaliarctidae) as a paraphyletic taxon (Wyss

1 987; Barnes 1 989; Berta 1 99 1 ), and Berta ( 1 99 1 ) has attempted to

derive a cladistic scheme of relationships for these "enaliarctine"

fossils. Despite the recent discovery and description of numerous

early pinnipeds, disagreement as to whether a paraphyletic

Enaliarctinae is an acceptable systematic category persists.

In this study we do not attempt to resolve the complex question

of the phylogenetic relations of the "enaliarctine" pinnipeds. Our

intent is to demonstrate the broad distribution of the embayed

basioccipital in the oldest known pinnipeds, to present measure-

ments of various taxa quantifying its size, to suggest the presence of

a carotid loop within the embayment. and to argue that, when both

dental and basicranial evidence is considered, a sister-group rela-

tionship of pinnipeds and ursids is highly probable. The taxonotnic

generality of the embayed basioccipital argues for its presence in

the common ancestor of "enaliarctine" pinnipeds. Thus we employ

the paraphyletic term "enaliarctine" in the sense of Barnes (1992)

for the genera Enaliarctos, Pirmarctidion. Pteronarctos, and

Pacificotaria without prejudging their phyletic relationships, to be

determined by future cladistic studies.

Recent debate focusing on the derivation of pinnipeds from

terrestrial camivorans has generally agreed that pinnipeds must

have evolved from an arctoid. There is less unanimity in the selec-

tion of the arctoid branch from which pinnipeds might have sprung.

Mitchell and Tedford (1973) emphasized the multiple anatomical

features of Enaliarctos that they interpreted as transitional between

those of terrestrial arctoids and those of marine pinnipeds: the

fissiped heterodont cheek teeth (particularly the upper and lower

carnassials and upper molars), the pattern of convolutions on the

anterior surface of the brain (based on cranial endocasts), and the

structure of the basicranium (chiefly the auditory region). The form

of the upper carnassial. the neuroanatomy of the endocasts. and the

structure of the auditory bulla are in agreement, suggesting deriva-

tion of Enaliarctos from hemicyonine ursids within or near the

genus Cephalogale (Mitchell and Tedford 1973). Although the

placement of Cephalogale in the Hemicyoninae might be debated,

most paleontologists familiar with basicranial anatomy and denti-

tions of arctoid camivorans concur with this assessment of ursid

In A. Berta and T. A. Demere (eds.)

Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore. Jr.

Proc. San Diego Soc. Nat. Hist. 29:57-67, 1994

58

R M. Hunt, Jr. and L. G. Barnes

affinities. Arnason and Widegren ( 1986). however, on the basis of

molecular hybridization, claimed a mustelid ancestry for all living

pinnipeds. Consequently, we wish to document a feature of the

basicranial anatomy among the enaliarcline pinnipeds that bears

importantly on this question and has largely escaped attention in

previous analyses of these animals. This feature, termed the

embayed basioccipital, was earlier noted by Barnes in Enaliarctos

mealsi (Barnes 1979: 9). Pinnarctidion bishopi (Barnes 1979: 26).

Pteronarctos goedertae (Barnes 1989: 13), and Allodesmus

packardi (Barnes 1979: 9). He suggested on this basis a relationship

between enaliarctines and ursids or amphicyonids. Hunt and Barnes

(1991) surveyed and measured the basioccipital embayments in

enaliarctines and pointed out that these structures are found in all

enaliarctines known to date.

MATERIAL AND METHODS

We examined the original enaliarctine material described by

Mitchell and Tedford (1973) as well as additional fossil crania

discovered subsequently: (1) Enaliarctos mealsi. LACM 4321,

genoholotype, nearly complete skull, Pyramid Hill Member of

Jewett Sand, probably from the lower nodule-bearing "grit zone,"

LACM locality 1627. Kern Co., California; (2) E. mealsi, LACM

5303. bipartite skull (rostral portion and associated endocast). Pyra-

mid Hill Member of Jewett Sand, exact stratigraphic level uncer-

tain. LACM (C1T) locality 481, Kern Co., California; (3)

Pinnarctidion bishopi. University of California, Berkeley. Museum

of Paleontology (UCMP) 86334. genoholotype, nearly complete

skull. Pyramid Hill Member of Jewett Sand, in place in the upper

fossiliferous concretion-bearing bed on south face of Pyramid Hill,

UCMP locality V6916 (= LACM locality 1628), Kern Co., Califor-

nia; (4) P. bishopi, LACM 5302, cranial endocast. Pyramid Hill

Member of Jewett Sand, exact stratigraphic level uncertain, LACM

(CIT) locality 481, Kern Co.. California; (5) undescribed pinniped.

LACM 128004, nearly complete skull, Pysht Formation, LACM

locality 5561, Merrick's Bay, Clallam Co., Washington; (6)

undescribed pinniped, LACM 134394 (J. L. Goedert 258), cranial

endocast. Pysht Formation. LACM locality 5561, Merrick's Bay,

Clallam Co.. Washington; (7) cf. Pteronarctos piersoni, LACM

123817. posterior cranium. Astoria Formation, LACM locality

4851, Moloch Beach. Lincoln Co., Oregon; (8) Pteronarctos

goedertae, LACM 123883, genoholotype, complete skull, basal

part of Astoria Formation. LACM locality 5058. north end of Nye

Beach. Newport. Lincoln Co., Oregon; (9) Pteronarctos piersoni,

LACM 127972, holotype, complete skull. "Iron Mountain bed."

Astoria Formation, LACM locality 4851, Moloch Beach, Lincoln

Co.. Oregon; (10) Pteronarctos piersoni, LACM 128002. paratype.

complete skull, "Iron Mountain bed," Astoria Formation, LACM

locality 4851, Moloch Beach, Lincoln Co.. Oregon.

These fossils have been previously discussed and illustrated by

Mitchell and Tedford ( 1973) and Barnes (1979, 1989, 1990), with

the exception of numbers 5, 6, and 7, which have not yet been

described in the paleontological literature. Number 5 is the subject

of a manuscript in preparation by Barnes and Hunt; number 7 is

illustrated for the first time in this report.

Preservation of some of these pinniped crania in a variety of

indurated sedimentary matrices has conserved the three-dimen-

sional geometry of the skulls, particularly the complex structure of

the basicranium. The hardness of the rock, however, has also neces-

sitated tedious, painstaking preparation to reveal details of the

basicranial anatomy. A number of these skulls and endocasts expe-

rienced weathering as concretions in outcrops and/or on beaches

prior to their collection. In some cases weathering has fortuitously

removed basicranial bone in such a way as to reveal the extent of

the basioccipital embayments better than normally could be seen in

an undamaged skull.

Usually the basioccipital embayments have filled with sediment

so that the volume and dimensions of the sinuses in life are pre-

served and can be measured. In some skulls, however, after the

sinus filled with sediment, compression of the skull by the weight

of overlying rock has compacted the basioccipital bone adjacent to

the sinus to a greater degree than the bone enclosing the sediment-

supported sinus itself. Hence the sinus protrudes from the skull, its

surface expression slightly exaggerated by differential compaction.

This should be kept in mind during viewing of the stereophoto-

graphs of pinniped basicrania. Dimensions of the sinuses were

measured in millimeters with dial calipers.

BASIOCCIPITAL SINUSES IN EARLY MIOCENE PINNIPEDS

A deeply excavated lateral margin of the basioccipital bone is a

characteristic feature of the basicranium of enaliarctines (sensu

Barnes 1992) and primitive members of the subfamilies Desmato-

phocinae, Imagotariinae. and Allodesminae (all sensu Barnes 1 989).

Development of the sinuses is bilateral, one occurring on each edge

of the bone. Each sinus is situated directly medial to the

entotympanie ossification of the auditory bulla that transmits the

internal carotid artery and anteromedial to the posterior lacerate

foramen; they penetrate deeply into the basioccipital but do not

reach the midline of the bone, hence they do not communicate.

Table 1 indicates the relative depth of penetration of the sinus

into the basioccipital in various species. Depth of penetration is

measured as the greatest transverse width of the sinus, at about the

midpoint along the length of the entotympanie tube housing the

internal carotid artery. In previously described enaliarctines the

greatest transverse width approximates 1 cm in most animals.

To express depth of penetration of the sinuses relative to basioc-

cipital dimensions, we combined the transverse width of both si-

nuses and expressed this value as a percentage of the total basioc-

cipital width (Table 1 ). These values range from a low of about 45%

in a small undescribed enaliarctine (LACM 128004) to a high of

6 1 .3% in a referred specimen of Enaliarctos mealsi ( LACM 5303 ).

Included in this sample are not only enaliarctines but also early

desmatophocines (Desmatophoca), allodesmines, and imago-

tariines (Neotheriitm). These values are similar to those measured

for the same dimensions in living ursids (e.g., Ursus arctos, Univ.

Nebr. State Mus. ZM-191, sinus width, 12.9 mm; basioccipital

width, 50.4 mm; sinus width as percentage of basioccipital width,

5 1 .2%. The same set of measurements in a young Ursus americanus

is 6 mm, 26.3 mm, and 45.6%). Both pinnipeds and living ursids

possess a wide basioccipital bone.

Figure 1 illustrates the surficial expression of the basioccipital

sinuses in the basicranium of the bipartite skull referred to

Enaliarctos mealsi by Mitchell and Tedford (1973: 229). The pair

of sinuses occupies 61.3% of the basioccipital width. Fine-grained

dark gray quartz sand fills the sinuses and the posterior lacerate

foramina. Thin (basioccipital) bone covers the ventral surface of the

sinuses; the bone has been broken away from the posterior floor of

the left sinus and the medial and posterior part of the right sinus,

revealing the sediment plug within. However, the prominent ventral

protrusion of the sinuses below the basicranium in LACM 5303 is

the result of differential crushing of the basioccipital. The basicra-

nium of the holotype of Enaliarctos mealsi, which is not crushed,

fails to show this exaggerated bulging of the sinuses (Mitchell and

Tedford 1979: 221, fig. 5). Nevertheless, the position and shape of

the sinus in the holotype and LACM 5303 clearly correspond, as

both display a swollen posterior terminus and a marked medial

extension. The sinuses arc shaped very like those in the basioccipi-

tal of the Aquitanian terrestrial ursid Cephalogale gracile (early

Miocene. Allier Basin. France).

Basicranial Evidence lor Ursid Affinity of the Oldest Pinnipeds 59

Table 1. Measurements (in mm) of maximum transverse width of the

basioccipital sinus in enaliarctine and other archaic pinnipeds, and a

comparison of relative development of the sinuses in various species.

"LACM. Natural History Museum of Los Angeles County. Los Angeles; UCMP,

University of California Museum of Paleontology, Berkeley; USNM, National

Museum of Natural History, Smithsonian Institution, Washington.

Calculated as the combined width of both right and left sinuses expressed as a

percentage of total basioccipital width (2(SW]/BW).

'Parentheses denote estimated measurements.

To illustrate the internal structure of these basioccipital sinuses, the basioccipital and petrosal. The lumen of the inferior petrosal

we prepared carefully the cranial endocast (LACM 5302) referred venous sinus nested within this dense connective tissue tube is very

by Barnes (1979: 17) to Pinnarctidion bishopi and previously as- difficult to dissect; a high-speed drill is required to cut away the

signed by Mitchell and Tedford 1973) to Enaliarctos mealsi basioccipital ventral to the sinus, exposing the floor of strong

(Fig. 2). Fine-grained sediment filling the sinus was removed, and fibrous connective tissue. A scalpel can then be used to cut through

the surrounding bones of the left auditory region were cleaned. the connective tissue to enter and open the sinus.

Figure 2A provides a full view of the entire posterior cranium. In extant ursids the interior of the sinus is lined by venous

showing that only a small amount of basicranial bone still adheres epithelium that is smooth and reflective, almost glassy in appear-

to the cranial endocast. Figure 2B, a closer view of the left auditory ance, and contains a conspicuous elongated loop of the internal

region, reveals the anatomical detail of the sinus and surrounding carotid artery. Upon exiting the auditory bulla at the anterior carotid

structures. foramen, the internal carotid makes a sharp, nearly 180° turn in

Although the petrosal promontorium has been broken, and much order to enter the venous sinus. In ursid and amphicyonid carni-

of it lost, the internal margin of the petrosal remains intact, display- vores the turning point is registered as a conspicuous depression in

ing a shallow anteroposterior groove defining the lateral margin of the basisphenoid bone immediately posterior to the foramen ovale;

the sinus. The medial limit of the sinus is defined by the basioccipi- this depression is also found at the same location in the basisphenoid

tal embayment. The roof of the sinus is without a bony covering of the enaliarctine Pinnarctidion (LACM 5302, Fig. 2B). Although

because the petrosal margin is separated from the basioccipital by a we cannot be certain whether the enlarged venous sinus of early

wide gap filled with fine-grained sediment. In living ursids. this gap pinnipeds contained an internal carotid loop, the depression in the

is likewise not covered by bone but is spanned by a thin but basisphenoid of LACM 5302 suggests that it did because the de-

extremely strong sheet of connective tissue made up of the fused pression demonstrates the 180° reversal in the direction of the

dura mater and endocranium (which separates the cranial cavity artery necessary to create such a loop.

from the lumen of the inferior petrosal sinus). The lateral and In June 1985. a small enaliarctine skull (basilar length approxi-

medial walls and floor of the sinus are reinforced by periosteum of mately 10.2 cm), believed to be of latest Oligocene age, wasdiscov-

60

R M. Hunt. Jr. and L G Barnes

Figure 1. Ventral view of basicranium of Enaliarctos mealsi. LACM 5303, Pyramid Hill. Kern County, California. Crushing has accentuated the

surface expression of sediment plugs filling the inferior petrosal sinus (ips). a subdural venous channel running between the basioccipital (BO) and

petrosal bones that continues forward to become the cavernous sinus of the basisphenoid (BS). Black arrows mark the medial edges of these sinuses,

demonstrating their maximum penetration into the basioccipital. The posterior entrance of the internal carotid artery (ica) into the auditory bulla

identifies the tubular ossified entotympanic element in this early pinniped, fused to the medial edge of ectotympanic (T). Note the enlarged posterior

lacerate foramen (plf). Scale bar = 1 cm.

ered by J. L. Goedert in the Pysht Formation. Clallam County.

Washington (Fig. 3). This remarkable skull (LACM 128004) re-

tains a well-preserved basicranium with an intact auditory region.

Careful preparation of the basicranium revealed important details

of the bulla and basioccipital sinuses. The bulla is made up of two

unfused elements: the ectotympanic. covering the middle ear, and

the entotympanic. which forms a tube surrounding the internal

carotid artery. The nearly complete ectotympanic is preserved on

the right side; a complete entotympanic tube is preserved on the

left. Medial to the entotympanic ossification, the basioccipital is

broken fortuitously to reveal a conspicuous sinus in the lateral part

of the basioccipital. Figure 3A shows the bilateral development of

the sinus within the basioccipital; both sinuses together occupy

about 45—47% of the basioccipital width (Table I). Figure 3B

demonstrates that the sinus is large, two and a half times as wide as

the arterial tube within the entotympanic. There appears to be a

prominent depression situated in the anterointemal corner of the

auditory region in the basisphenoid where the internal carotid ar-

tery, after exiting the anterior carotid foramen in the entotympanic

tube, turns sharply backward to enter the basioccipital sinus; this is

comparable in form and position to the same depression described

above in Pinnarctidion bishopi (LACM 5302, see Fig. 2B).

Despite its small size, this skull belonged to an adult (based

upon a suture age of at least 28 according to the method of Sivertsen

(1954)). Its size and distinctive morphology indicate that it repre-

sents a new taxon with auditory region and basioccipital sinuses

configured in the same basic manner as the described Pyramid Hill

enaliarctines. Thus, the shared presence of these basioccipital si-

nuses in several genera of early pinnipeds suggests they were

present primitively within the Enaliarctinae and can be inferred to

have existed in their common ancestor.

This hypothesis receives additional support from the discovery

of a posterior cranium (Fig. 4) of an early middle Miocene pinniped

(LACM 12.3817) found in 1983 by Guy Pierson in the Astoria

Formation, Moloch Beach, Lincoln Co., Oregon. The fossil comes

from LACM locality 485 1 , the same site that produced the holotype

skull of Pteronarctos piersoni (Barnes 1990: 3) and the dome-

headed chalicothere Tylocephalonyx (Munthe and Coombs 1979:

78-79). Barnes ( 1990) estimated the age of the concretion-bearing

horizon (the "Iron Mountain bed") within the Astoria Formation

that produced these fossils to be about 16 Ma.

It is difficult to assign this fossil to a taxon confidently because

it is abraded and lacks the rostral part of the skull; however, it is

similar in size and general structure to Pteronarctos piersoni and

we tentatively identify it as cf. Pteronarctos piersoni.

Both auditory bullae have been lost from this basicranium,

revealing the robust petrosal bones. Directly medial to these

petrosals are large sediment-filled basioccipital sinuses that occupy

57.3 to 63% of basioccipital width. The sinuses appear to penetrate

the basioccipital most deeply at the midpoint along the medial edge

of the petrosal. The configuration of the sinus is particularly well

displayed on the left side.

Barnes (1989) described the skull of Pteronarctos goedertae

from near the base of the Astoria Formation (LACM locality 5058),

Nye Beach. Lincoln Co., Oregon, estimating an early Miocene age

of about 19 Ma for these sediments. The basioccipital of the holo-

type cranium (LACM 123883) has been cut open by Barnes, who

reported (1989: 13) a small embayment for the sinus. Thus, the

genus Pteronarctos provides evidence for basioccipital embay-

ments over the approximately 3-million-year history of the lineage

in Oregon ( 16 to 19 Ma).

Primitive members of other pinniped subfamilies also have an

embayed basioccipital. The most primitive Imagotariinae, the earli-

est ancestors of dusignathines and odobenine walruses (Repenning

and Tedford 1977), are the middle Miocene Japanese species of the

genus Prototaria (see Barnes 1989; Kohno et al. 1992). The holo-

type cranium of Prototaria primigena has a large, convex protuber-

ance on the surface of the basioccipital on each side medial to the

auditory bulla. This specimen has not been dissected, so the pres-

ence of a vacuity filled with sediment cannot be demonstrated

unequivocally. Another species in the same genus, however,

Prototaria Kohno n. sp. (in press), also has similar protuberances in

Basicranial Evidence for Ursid Affinily of the Oldest Pinnipeds

61

Figure 2. Crania] endocast and partial basicranium referred to Pinnarctidion bishopi, LACM 5302, Pyramid Hill, Kern County, California: A,

ventrolateral view of complete endocast; B, detailed view of bone remaining in left auditory region. A deep embay ment of the basioccipital (BO) houses an

enlarged inferior petrosal venous sinus (ips) medial to petrosal (P). The venous sinus probably contained an elongated loop of the internal carotid artery, as

implied by the form of a small depression (d) in the basisphenoid (BS), posterior to the foramen ovale (to). In living ursids with the internal carotid loop, a

similar depression is the location of a sharp 1 80° bend in the artery where it changes course to run posteriorly in the sinus, forming the carotid loop. Scale

bar = 1 cm.

the same location. This cranium was sagittally sectioned by erosion

prior to its discovery. The braincase has been cleaned of sediment,

revealing a prominent deep embayment in the basioccipital medial

to the petrosal. Thus the most primitive Imagotariinae retained the

venous sinus within the basioccipital.

An embayed basioccipital is also present in a slightly more

evolved imagotariine, Neotkerium mirum, from the middle Mio-

cene Sharktooth Hill Bonebed in California and long known only

by postcranial bones (Kellogg 1931; Repenning andTedford 1977).

A virtually complete skull (LACM 1 3 1 950) that undoubtedly repre-

sents this species, recently excavated from the Sharktooth Hill

Bonebed, was fortuitously broken through the braincase. A deeply

embayed basioccipital lies medial to the petrosal in this specimen

(Fig. 5).

All later imagotariines for which we have data on the floor of

the cranium have no embayment in the basioccipital. The floor of

the braincase is smooth, and there is only a slight depression where

the embayment exists in more primitive species. This is the case in

the holotype of Pontolis magnus (see Repenning and Tedford 1 977:

pi. 10, fig. 2) and in a referred skull of Imagotaria downsi (see

Repenning and Tedford 1977: pi. 10, fig. 1).

Data for any dusignathine or any fossil odobenine walrus do not

exist. The floor of the basioccipital in the Recent Odobenus

rosmarus lacks a typically developed basioccipital embayment but

shows evidence of two small pockets that may represent vestiges of

the embayed condition.

The braincase of the holotype of the early middle Miocene

Desmatophoca oregonensis from the Astoria Formation in Oregon

has been prepared, demonstrating a deeply embayed basioccipital

very much like that of Neotherium mirum. D. oregonensis is the

type species of Desmatophoca Condon, 1906. The only other spe-

cies presently known in this subfamily is Desmatophoca

62

R. M. Hunt, Jr and L. G Barnes

Figure 3. Nearly complete skull of an undescribed otanid pinniped, LACM 128004, Pysht Formation, Merrick"s Bay, Clallam County, Washington: A,

ventral view; B, detailed view of left auditory region showing the enlarged inferior petrosal sinus (ips) within the margin of basioccipital (BO). The sinus is

enclosed by basioccipital (BO), entotympanic (E), and petrosal (P). Diameter of the sinus is more than twice that of the bony tube for the internal carotid

artery (ica). Entotympanic (E) and ectotympanic (T) elements of the auditory bulla are both fully ossified but remain unfused where they are in contact

(asterisk). BS, basisphenoid; to, foramen ovale; gf, glenoid fossa; SQ. squamosal. Scale bar = I cm.

Basicranial Evidence for Ursid Affinity of the Oldest Pinnipeds

63

Figure 4. Ventral view of posterior cranium of cf. Pteronarctos piersoni, LACM 1 238 1 7, Astoria Formation, Moloch Beach, Lincoln County, Oregon,

ventral view. Weathering and erosion of the cranium prior to its collection abraded the basicranium. exposing the enlarged inferior petrosal sinuses (ips)

situated between the petrosal (P) and embayed basioccipital (BO) bones. Scale bar = 1 cm.

brachycephala Barnes, 1987, from the late early Miocene Astoria

Formation of Washington. In many features, this species is more

primitive basicranially than D. oregonensis, and preparation of the

holotype revealed a well-developed basioccipital embayment.

An embayed basioccipital is also present in the earliest known

member of the subfamily Allodesminae. The evidence for this is an

undescribed braincase (LACM 1 33442) from the early middle Mio-

cene part of the Astoria Formation in Lincoln County, Oregon, the

same horizon that produced Pteronarctos piersoni and Desmato-

phoca oregonensis (see Barnes 1987, 1989, 1990). This cranium

has the following allodesmine features: cuboid mastoid process,

large and posteriorly projecting paroccipital process, large

lambdoidal and nuchal crests, well-developed sagittal crest, flat

tympanic bulla, and facet for the tympanohyal in the tympanohyal

pit. This specimen also has a deeply embayed basioccipital, very

much as in the holotype of Desmatophoca oregonensis and the

referred skull of Neotherium mirum.

The embayment appears to have been lost in later, more highly

evolved Allodesminae. Evidence for this can be found in the middle

Miocene Allodesmus kernensis, the type species of the genus

Allodesmus. A cranium of a young adult male referred to this

species (LACM 21097) has been found in the Sharktooth Hill

Bonebed in California, the same horizon that produced the type

material of this species and of Neotherium mirum, the primitive

imagotariine. The braincase is open dorsally and reveals the de-

tailed internal structure of the endocranium. Where the other pinni-

peds discussed above have a deeply embayed basioccipital, this

specimen has only a broad, flat, and slightly concave sulcus, sug-

gesting that the basioccipital embayment was lost in the Allodesmus

lineage by middle Miocene time.

BASIOCCIPITAL SINUSES IN LIVING URSIDS

From analogy with living ursids, the basioccipital embayments

of extinct ursid and amphicyonid Carnivora are believed to have

contained in life an artery nested within a subdural venous sinus that

functioned as a countercurrent heat-exchange device to cool arterial

blood flowing to the brain (Hunt 1974, 1977). This interpretation is

based upon dissections of the basioccipital embayments of extant

ursids (Hunt and Joeckel 1989: Hunt 1990). in which the internal

carotid artery becomes greatly lengthened by doubling back on itself

within the subdural inferior petrosal venous sinus situated in the

lateral margin of the basioccipital bone (Fig. 6). Histological study

of the internal carotid artery both within and outside the sinus has

supported the heat-exchange hypothesis (Hunt 1990).

The carotid loop of ursids was originally identified by Tandler

(1899) in his classic investigation of mammalian cranial arteries.

Many years later, Davis ( 1964) drew attention to Tandler's discov-

ery in his description of a similar convoluted internal carotid in the

cavernous sinus of the giant panda, Ailuropoda melanoleuca. Davis

also found a carotid loop in the American black bear, Ursus ameri-

canus. Subsequently, injection of radio-opaque material into the

cranial arteries of 107 species of mammals representing 49 families

allowed Boulay and Verity (1973) to produce a series of radio-

graphs of representative species of the major carnivoran families,

permitting a preliminary survey of the morphology of the internal

carotid artery in arctoid, aeluroid, and cynoid Carnivora. Although

they made no mention of the elongated and looped carotid of ursids.

their radiographs plainly show an enormous internal carotid loop in

the sloth bear. Melursus ur sinus, a probable loop in the Asiatic-

black bear, Selenarctos thibetanus, and a well-defined loop in the

giant panda, confirming the earlier work of Davis (1964). These

initial descriptions of the looped carotid gave no attention to its

probable function.

Tandler's (1899: 721-722) initial description of the loop formed

by the internal carotid artery in the polar bear, Thalarctos maritimus.

corresponds to our current observations in other living ursids (Hunt

1990). He wrote, "After the [internal] carotid has perforated the

bony basicranium, it lies subdurally in the wide, caudally extended

cavernous sinus. Here the artery takes the form of a double loop,

whose individual legs appear to be twisted about their long axis. By

64

R. M. Hunt, Jr. and L. G. Barnes

Figure 5. Internal view of the posterior cranium (LACM 13 1950) of the early imagotariine Neotherium mirum, Sharktooth Hill Bonebed, Kern County,

California. Note the large embayed basioccipital bone housing the inferior petrosal venous sinus (ips).

means of this characteristic pattern, the subdural portion of the

carotid attains considerable length; in the case that I investigated,

the length of the vessel from its entrance through the bony basicra-

nium to its exit through the dura at the sella turcica amounted to

about 16 cm."

From Tandler's description, it is more likely that in the polar

bear the carotid loop in fact lies primarily within the inferior petro-

sal venous sinus (between the basioccipital and petrosal), extending

anteriorly from there into the cavernous sinus on the dorsal surface

of the basisphenoid. The inferior petrosal venous sinus and cavern-

ous sinus together form a single linear channel running from the

posterior lacerate foramen forward to the sella turcica of the

basisphenoid. Such an enormous loop of necessity requires access

to the full length of this subdural sinus. Tandler reported that the

internal carotid loop had been illustrated prior to 1899, but implied

that it had not been discussed: "Barkow no doubt perceived this

somewhat complicated relationship [of the arterial loop], based

upon an obvious figure (Plate 4) of this feature given in Part 4 of his

Comparative Morphology."

Davis (1964: 252) described and illustrated clearly a similar

convoluted vessel in his detailed anatomical study of the giant

panda: "Emerging from the carotid canal, the (internal carotid]

artery enters the cavernous sinus. Immediately after entering the

sinus it forms a tight knot by arching first posteriorly, then anteri-

orly upon itself. This is followed in the vicinity of the sella turcica

by a tight S-loop, all of which greatly increases the length of the

vessel; while the distance traversed within the sinus (from the

carotid foramen to the anterior border of the sella) is only 22 mm.,

the length of the vessel is 68 mm." Davis (1964: 277) also identified

an internal carotid loop in the American black bear, mentioning

Tandler's earlier description of the artery in the polar bear: "A

striking example of the close agreement between Ailuropoda and

the Ursidae is the elongation and looped arrangement of the

subdural part of the internal carotid. In all other carnivores the

carotid passes straight through the sinus cavernosus. but in a speci-

men of Thalarctos described by Tandler the vessel immediately

arched caudad in the sinus, forming a long U-shaped loop twisted

around its own long axis, along the medial border of the petrosal. I

found an identical situation in a specimen of Ursus americanus, in

which the subdural part of the carotid measured 60 mm while the

linear distance traversed by this part of the vessel was only 12 mm.

a ratio of 1:5."

A radiograph of the basicranium of the sloth bear in Boulay and

Verity's (1973: 162) catalogue shows an artery particularly elon-

gated in the auditory region. Using their scaling for this radiograph,

we estimated the length of the carotid loop to be 7.7 cm, measured

from the entrance of the artery into the inferior petrosal sinus to its

exit from the cavernous sinus. The same measurement is difficult to

determine for their radiograph of the Asiatic black bear because the

arterial path is obscured, although looping of the vessel is evident.

Radiographs of the giant panda clearly demonstrate a convoluted

arterial path just as Davis ( 1964) described: the internal carotid loop

lies primarily within the cavernous sinus; its subdural length is

about 8.5 cm (Boulay and Verity 1973: 168, 171 ).

Hunt (1990) dissected the basioccipital sinus in the American

black bear and the sun bear. Helarctos malayanus, and discovered a

similar anatomical arrangement in which an elongated loop of the

internal carotid, twisted upon itself, is nested within the large

saclike inferior petrosal venous sinus. The subdural length of the

carotid loop in the former species measured 7.6 cm. in the latter

about 8.6 cm. These measurements of the subdural length of the

internal carotid are compared in Table 2. where they are also

presented as a percentage of the basilar length of the skull in various

living ursids.

Thus, an internal carotid loop nested within a venous subdural

sinus is now known in five of the seven species of living bears and

in the giant panda. Because the basioccipital embayment has been

identified in dried skulls of all of the remaining living ursids. the

Basicranial Evidence for Ursid Affinity of the Oldest Pinnipeds

65

Figure 6. Left auditory region of a living ursid, showing the looped internal carotid artery nested within the inferior petrosal venous sinus. The sinus is

emplaced in the lateral margin of the basioccipital bone. In the illustration the side of the basioccipital has been removed to show the artery-vein complex,

and the venous sinus has been opened to show the arterial loop within. Small circles around the artery show the location of the anterior and posterior carotid

foramina: the segment of the carotid between the circles lies within the medial wall of the auditory bulla (from Hunt 1977). BO. basioccipital; BS,

basisphenoid; AL, alisphenoid; PE, petrosal.

presence of the carotid loop in all living members of the family

Ursidae is probable. All living ursids in which the internal carotid

arterial loop has been dissected or identified in a radiograph show

the loop resting within the inferior petrosal venous sinus; in fact, the

most posterior extent of the loop reaches to the posterior termina-

tion of the sinus. In Ailuropoda melanoleuca, however, the internal

carotid loop is contained primarily in the cavernous sinus, hence

anterior to its location in other living ursids.

A SUBDURAL INTERNAL CAROTID LOOP

IN OTHER CARNIVORA

The radiographs published by Boulay and Verity (1973) make

possible a survey of the subdural internal carotid in 31 camivoran

species. Not all radiographs clearly portray the artery during its

entire course en route to the Circle of Willis, owing to failure of the

injected medium to penetrate the artery or to an unfavorable orien-

tation of the head during radiography. The aeluroid Carnivora,

however, lack a looped carotid, many having an artery reduced and

nearly nonfunctional. Aeluroids are well known for their tendency

to bypass the internal carotid and to rely upon an external carotid

blood supply, in which an orbital rete is interposed between the

orbit and the brain (Davis and Story 1943; Hunt 1974).

Table 3 indicates the state of the subdural internal carotid in the

cynoid and arctoid carnivorans injected by Boulay and Verity.

There is no evidence of a carotid loop in canids. Neither do we find

a carotid loop in Procyon lotor. Story (1951) found no internal

carotid loop in any of the living procyonids.

Boulay and Verity were able to inject a large number of

. mustelids, including species of Mustela, Manes, Meles, Lutra, and

Gulo. In Mustela and Martes. there is no evidence of a looped

carotid artery. In both Meles meles and Lutra lutra. however, al-

though no loop occurs in the inferior petrosal sinus, the artery

displays a sinuous bend or small loop within the cavernous sinus

just before reaching the Circle of Willis. These loops appear to be

analogous to the loop of the internal carotid developed in the same

position in Ailuropoda. but are not as developed.

Most interesting of all these cranial radiographs of mustelids,

however, is that of the wolverine, Gulo gulo. which has a remark-

ably well-developed loop of the internal carotid. Careful compari-

son of landmarks on the radiograph with dissected wolverine skulls

demonstrates that the carotid loop is within the cavernous sinus

(Fig. 7), developed in the same location and having the same

configuration as in the giant panda. There is no carotid loop in the

66

R. M. Hunt, Jr. and L. G. Barnes

Table 2. Lengths (in cm) of and ratios between the subdural

internal carotid artery and basilar length of the skull in living ursids.

"Measured during dissection by R. M. Hunt.

''Measured from radiograph (Boulay and Verity 1973:162, 168, 171).

'Data from Davis ( 1964:252).

basioccipital's inferior petrosal sinus nor is there any deep embay-

ment of that bone of the ursid type. The condition in Meles and

Lutra is possibly an initial stage in the development of a subdural

internal carotid loop within the cavernous sinus like that of the

wolverine. We regard the hypothesis that an internal carotid loop

was present primitively in the Mustelidae and was subsequently

lost in many living mustelid lineages as improbable and without

basis; furthermore, a carotid loop in the cavernous sinus does not

appear to register in bone and hence cannot be detected in fossils.

These observations have important implications: (1) The ca-

rotid loops of mustelids and ursids must have been independently

derived — they occur in different subdural locations in the basicra-

nium and so cannot be derived from a common ancestral condition.

Many living mustelids entirely lack such a loop and, we presume,

never possessed one. (2) The similar carotid loops of the giant

panda (Ursidae) and wolverine (Mustelidae) are surely parallelisms

and indicate that arctoid carnivorans may independently develop

such loops in the cavernous sinus of the skull. (3) Only ursid and

amphicyonid Carnivora and archaic pinnipeds share the same type

of deep basioccipital embayments in their skulls and therefore are

presumed to possess subdural carotid loops within the inferior

petrosal sinus. Despite Tandler's description, we doubt that the

polar bear has a significant carotid loop in its cavernous sinus — its

loop is probably always within the basioccipital, as it is in all other

ursine bears. (4) Arctoid Carnivora appear to have the potential to

develop convoluted internal carotid arteries within the subdural

sinuses that lie along the sides of the basicranial axis of Huxley.

Such convoluted arteries may have evolved in various lineages in

parallel, hence the similarity of the basioccipital in ursids and

amphicyonids may exemplify convergence, or it may indicate rela-

tionship between the two families (a determination will require

TABLE 3. Status of the internal carotid artery within the inferior

petrosal and cavernous sinuses of arctoid and cynoid Carnivora,

from radiographs published by Boulay and Verity (1973).

additional evidence). The similar basioccipital embayments and

their contained arteries found in the living ursine bears are probably

all derived from a common ancestral taxon, probably the Miocene

European Ursavus. and do not represent parallel evolution within

the modern Ursidae, as implied by the high degree of similarity

among the species so far studied.

It is probably no accident that these looped carotids occur in the

two families of large-bodied arctoid Carnivora, the Ursidae and

Amphieyonidae. Dissipating heat is more difficult for larger mam-

mals (Taylor 1980), and these large terrestrial carnivorans could

well have benefited from a device to cool the blood flowing to the

brain. It is interesting that among the mustelids a looped carotid

artery, although of a different nature, occurs in the largest terrestrial

representative of the family living today, the wolverine. We do not

expect to find carotid loops in large aquatic mustelids such as

Pteronura and Enhydra, because of the amount of time these ani-

mals spend in the water, but we anticipate that such a loop may have

been present in the large extinct terrestrial mustelid Megalictis, an

early Miocene carnivore that attained the size of a small bear.

SUMMARY

The presence of embayed basioccipital sinuses (associated with

a common basicranial anatomical pattern) in the late Oligocene to

early middle Miocene enaliarctine pinnipeds of California. Oregon,

and Washington indicates the taxonomic generality and wide geo-

graphic distribution of this anatomical trait among the earliest

known pinnipeds of the eastern North Pacific margin. All

enaliarctines of late Oligocene to early middle Miocene age for

which basicrania are known have the basioccipital embayment. All

primitive members of subfamilies (Imagotariinae, Desmato-

phocinae. Allodesminae) believed to have evolved from enaliarc-

tines also have an embayed basioccipital. These include the three

most primitive middle Miocene imagotariines (Prototaria primi-

genia, P. n. sp., and Neotherium mirum), the earliest known

allodesmine (a cranium from the early middle Miocene Astoria

Formation), and the holotypes of the early middle Miocene

desmatophocines Desmatophoca oregonensis and D. brachy-

cephala. None of the later, more derived otariid pinnipeds has an

embayed basioccipital like those of archaic fossil pinnipeds and

living ursids. The structure is absent in the modern fur seals and sea

lions (Olariinae) and modern walrus (Odobeninae). in which we

Gulo gulo (wolverine)

Group A: Carotid follows a straight course within subdural venous

sinuses: Cuius lalrans, Canis familiaris, Nyclereules procyonoides.

Procyon lotor, Mustela erminea, Mustela putorius, Maries flavigula.

Group B: Carotid follows a straight course with slight sinuous bend in

cavernous sinus: Meles meles. Ultra Intra, Zalophus califomianus,

Pusa sibirit a

Group C: Carotid does not follow a straight course: elongate loop

developed either in inferior petrosal (Melursus ursinus, Selenarctos

thibetanus) or cavernous sinus (Ailuropoda melanoleuca, Gulo gulo).

internal

carotid artery

Figure 7. Radiograph of the subdural loop of the internal carotid

artery within the cavernous sinus of the wolverine, Gulo gulo (from

Boulay and Verity 1973).

Basicranial Evidence for Ursid Affinity of the Oldest Pinnipeds

67

noted that unusual unidentified depressions on the basioccipita]

margin may be vestiges of the embayment.

Among living pinnipeds the embayed basioccipita] must have

been lost or so altered that it is not easily recognized. Presumably

the venous sinus and its carotid arterial loop have been modified

through time. The usefulness of a cranial mechanism to cool arterial

blood flowing to the brain seems limited in aquatic mammals. We

think it significant in this regard that living pinnipeds subjected to

high temperatures quickly become uncomfortable and return to the

water (King 1983:146-149).

The only other Carnivora in which this basioccipital sinus is

conspicuous are the terrestrial ursids and extinct amphicyonids. The

ursid (not amphicyonid) upper carnassial teeth (Mitchell and

Tedford 1973). type A (plesiomorphic arctoid) auditory bullae

(Hunt 1974, 1977), and embayed basioccipital sinuses of

enaliarctines strongly suggest ursid ancestry. Among known terres-

trial fossil carnivorans, the most probably ancestral taxa are small

species of the Eurasian ursid Cephalogale and Amphicynodon.

Both hemicyonine and ursine ursids, as well as pinnipeds, appear to

have originated within the amphicynodontine radiation. The

basicranial evidence makes a sister group between pinnipeds and

mustelids implausible, if enaliarctines are the ancestors of otariids

or particularly if they are considered broadly representative of the

basal pinniped stock from which both otariids and phocids were

derived.

An internal carotid loop, within either the inferior petrosal or

cavernous subdural venous sinuses of the basicranium. has evolved

in parallel in several arctoid lineages. Extant ursids possess a

subdural carotid loop nested in the inferior petrosal sinus. Extinct

ursids and amphicyonids are believed to have had similar loops

because they possess deep basioccipital embayments like those

found in living bears. The giant panda and wolverine have indepen-

dently evolved a subdural carotid loop within the cavernous sinus;

these loops' being located differently from those of ursids and

amphicyonids indicates that they must be parallel developments.

The function of these carotid loops nested within a subdural venous

sinus seems best explained as a device to cool warm arterial blood

flowing to the brain in large exercising mammals.

No living pinniped insofar as we can determine possesses an

embayed basioccipital or a carotid loop within the inferior petrosal

venous sinus. Some extant pinnipeds (Zalophus californianus, Pusa

sibirica, Phoca vitulina; Tandler 1899; Boulay and Verity 1973) do

show a slightly sinuous bend of the internal carotid artery within the

cavernous sinus; however, this sinuosity is not as pronounced as in

the mustelids Meles and Lutra. Yet the basioccipital embayment of

enaliarctines is pronounced.

We conclude from this character's taxonomic distribution that

pinnipeds lost these brain-cooling devices early in their history, for

most lineages in middle Miocene time, and that their subdural

carotids and sinuses returned, by evolutionary reversal, to a more

normal configuration.

LITERATURE CITED

Arnason, U., and B. Widegren. 1986. Pinniped phylogeny enlightened

by molecular hybridizations using highly repetitive DNA. Molecu-

lar Biology and Evolution 3:356-365.

Barnes, L. G. 1979. Fossil enaliarctine pinnipeds (Mammalia: Otariidae)

from Pyramid Hill, Kern County. California. Natural History Mu-

seum of Los Angeles County Contributions in Science 318.

. 1987. An Early Miocene pinniped of the genus Desmatophoca

(Mammalia, Otanidae) from Washington. Natural History Mu-

seum of Los Angeles County Contributions in Science 382.

1989. A new enaliarctine pinniped from the Astoria Formation,

— . 1990. A new Miocene enaliarctine pinniped of the genus

Pteronarctos (Mammalia: Otariidae) from the Astoria Formation.

Oregon. Natural History Museum of Los Angeles County Contri-

butions in Science 422.

1992. A new genus and species of Middle Miocene enaliarctine

pinniped (Mammalia, Camivora. Otariidae) from the Astoria For-

mation in coastal Oregon. Natural History Museum of Los Angeles

County Contributions in Science 431.

Berta, A. 1991. New Enaliarctos* (Pinnipedimorpha) from the Oligo-

cene and Miocene of Oregon and the role of "enaliarctids" in

pinniped phylogeny. Smithsonian Contributions to Paleobiology

69:1-33,

— , and C. E. Ray. 1990. Skeletal morphology and locomotor

capabilities of the archaic pinniped Enaliarctos mealsi. Journal of

Vertebrate Paleontology 10:141-157.

Boulay. G. du. and P. Verity. 1973. The Cranial Arteries of Mammals.

Whitefriars Press. London, England.

Davis. D. D. 1964. The Giant Panda: A morphological study of evolu-

tionary mechanisms. Fieldiana: Zoology Memoir 3:1-339.

, and E. Story. 1943. The carotid circulation in the domestic cat.

Zoological Series, Field Museum of Natural History 28: 1—17.

Harland, W. B., R. L. Armstrong, A. V. Cox. L. E. Craig, A. G. Smith,

and D. G. Smith. 1990. A Geologic Time Scale 1989. Cambridge

University Press, New York, New York.

Hunt. R. M.. Jr. 1974. The auditory bulla in Camivora: An anatomical

basis for reappraisal of carnivore evolution. Journal of Morphology

143:21-76.

. 1977. Basicranial anatomy of Cynelos Jourdan (Mammalia:

Camivora), an Aquitanian amphicyonid from the Allier Basin,

France. Journal of Paleontology 5 1 :826— 843.

. 1990. Vascular countercurrent cooling mechanisms in Car-

nivora. Journal of Vertebrate Paleontology 10 (3) supplement: 28 A

(abstract).

, Jr., and L.G. Barnes. 1991. Basicranial evidence for ursid

affinity of the oldest pinnipeds (Mammalia. Camivora). Journal of

Vertebrate Paleontology 11 (3) supplement: 37A (abstract).

, and R. M. Joeckel. 1989. Anatomical evidence for basicranial

Oregon, and a classification of the Otariidae (Mammalia: Car-

nivora). Natural History Museum of Los Angeles County Contri-

butions in Science 403.

vascular heat exchange cooling blood flowing to the brain of

arctoid Carnivora (Mammalia, Ursidae). Annalen van de

Koninklijke Belgische Vereniging voor Dierkunde 119(1) supple-

ment: 90.

Kellogg. R. 1931. Pelagic mammals from the Temblor Formation of the

Kem River region, California. Proceedings of the California Acad-

emy of Sciences 19:217-397.

King, J. E. 1983. Seals of the World. Cornell University Press, Ithaca,

New York.

Kohno, N..L. G. Barnes, and K. Hirota. 1992. Miocene pinnipeds of the

genera Prototaria and Neotherium in the North Pacific Ocean;

relationships and distribution. Abstracts, 29th International Geo-

logical Congress, Kyoto, Japan, August, 1992, Vol. 2. p. 349.

Mitchell, E.. and R. H. Tedford. 1973. The Enaliarctinae: A new group

of extinct aquatic Camivora and a consideration of the origin of the

Otanidae. Bulletin of the American Museum of Natural History

151(31:201-284.

Munthe, J., and M. C. Coombs. 1979. Miocene dome-skulled

chalicotheres (Mammalia. Perissodactyla) from the western United

States: A preliminary discussion of a bizarre structure. Journal of

Paleontology 53:77-91.

Repenning. C. A., and R. H. Tedford. 1977. Otarioid seals of the Neo-

gene. U.S. Geological Survey Professional Paper 992.

Sivertsen, E. 1954. A survey of the eared seals (family Otariidae) with

remarks on the Antarctic seals collected by M/K "Norvegia" in

1928-1929. Del Norske Videnskaps-Akademii Oslo 36:1-76.

Story. E. 1951. The carotid arteries in the Procyomdae. Fieldiana: Zool-

ogy 32(8):477-557.

Tandler, J. 1 899. Zur vergleichenden Anatomie der Kopfarterien bei den

Mammalia. Denkschriften der kais. Akademie der Wissenschaften,

Mathematisch-Naturwissenschaftliche Klasse (Wien) 67:677-784.

Taylor. C. R. 1980. Responses of large animals to heat and exercise.

Pp. 79-89 in Horvath and Yousef (eds.). Environmental Physiol-

ogy: Aging, Heat and Altitude. Elsevier North Holland.

Amsterdam. Netherlands.

Wyss, A. R. 1987. The walrus auditory region and monophyly of pinni-

peds. American Museum Novitates 2871.

The Evolution of Body Size in Phocids: Some Ontogenetic

and Phylogenetic Observatons

Andre R. Wyss

Department of Geological Sciences, University of California, Santa Barbara, California 93106

ABSTRACT. — Large body size is generally regarded as having arisen relatively late in phocid history. Evaluation of the question of the size of

the ancestral phocid in a phylogenetic context reveals, however, that large size is most likely the ancestral condition, and small size among some

members of the subfamily Phocinae is best regarded as secondary. A pattern of widespread character reversal in phocid evolution coincides roughly

with this inferred decrease in size. Several features diagnostic of the family Phocidae and subfamily Phocinae appear at least partly attributable to

ontogenetic juvenilization.

INTRODUCTION

Extant phocids vary widely in size, from Pitsa sibirica, with a

nose-to-tail length of 1 .3 m, to the two species of Mirounga, whose

adult males measure 4-5 m (King 1983). This degree of diversity

raises the question of the size of the ancestral phocid. a topic rarely

addressed rigorously. Large body size among marine mammals is

generally considered an adaptive response to the physiological

demands imposed by a heat-dissipating aquatic environment

("large" applies to species whose adult females are 2.3 m long).

The widely held notion that large size among pelagic taxa repre-

sents an evolutionary advancement follows directly from this view

and finds strong support in the fossil record of certain marine

Carnivora. The trend among otariids and odobenids, for example,

seems to be toward increasing body size (Repenning 1976). Thus a

similar shift in size during the evolution of phocids also seemed

likely. Alternatively, were phocids primitively large, some mem-

bers of the group only secondarily attaining more diminutive pro-

portions? The comments presented below have a dual aim; first, to

determine which of these two alternatives is supported by phylo-

genetic evidence, and second, to evaluate the conclusion in relation

to known patterns of character evolution within the group, taking

into account possible ontogenetic modifications.

The currently most widely held view of phocid systematics

recognizes two subfamilies: the Phocinae, including the tribe

Phocini, together with Erignathus and Cystophora, and the

"Monachinae," including the monk, elephant, and Antarctic seals

(Fig. 1). As a result of their presently nearly disjunct geographic

distributions, phocines and "monachines"' have been informally

dubbed "northern" and "southern" phocids, respectively. Because

the monophyly of the "monachine" assemblage is questionable

(Wyss 1988). I place its name in quotation marks, as above.

The notion that phocids were small primitively has considerable

historical precedent. Most arguments favoring this view and having

more than a simply intuitive basis do not. however, fare well under

scrutiny.

Cystophonnae Monachinae Phocinae "Monachinae" Phocinae

/

Lobodontinae

/

Monachinae

Phocinae

Figure 1. History of thought on phocid interrelationships. A. Laws'

(1959) depiction of phocines as representing an early stage in phocid

evolution; B, bipartite division of phocids recognized by most workers since

King (1966); C, scheme proposed by Wyss (1988) in which one of these

divisions, the "Monachinae," is considered paraphyletic. Triangles denote

monophyletic groups with unspecified internal branching arrangements.

Comments by Kellogg (1922: 98) both reflect the consensus

view and give an impression of its imprecise foundation: "It has

been stated by Williston (1914) that 'it seems to be a law of

evolution that no large creatures can give rise to races of small

creatures,' and that 'the largest sea animals have been the final

evolution of their respective races.' As the history of the animals in

the past appeared to confirm this, it was assumed by some that the

sea lions, walruses, and elephant seals therefore represent a higher

degree of specialization than do smaller seals and that the latter

approximate more nearly in size the ancestral group." Although

Kellogg did not go on to state whether he agreed with this assump-

tion, other workers have been more forthright in expressing their

opinion on the question of primitive phocid size.

Flower (1881: 156) considered Mirounga to combine "in itself

in the fullest degree all the characters by which the Seals are

distinguished from the terrestrial Carnivora." Barrett-Hamilton

(1902) echoed this view, considering the Phocinae to represent the

least, the Cystophorinae (a formerly recognized association of

Mirounga and Cystophora) the most "specialized" phocid subfami-

lies.

This interpretation was carried over into the more recent litera-

ture by Laws ( 1959: 430-431 ), who argued — partially on the basis

of prior acceptance of progressive increase in body size — that "the

series Phocinae-Monachinae-Lobodontinae-Cystophorinae shows

increasing specialization to an aquatic life, with Phoca the most

primitive and Mirounga the most specialized genera" (Fig. 1A).

Laws did, however, provide additional anatomical information in

support of his phylogenetic series; that evidence will be considered

below.

In his influential consideration of pinniped biogeography,

McLaren ( 1960: 20) likewise argued that the smaller species of the

subfamily Phocinae are "anatomically the most primitive and least

aquatically adapted," (emphasis in original) but offerred no mor-

phological evidence in support of this contention.

King (1965) considered Pagophilus and Cystophora (the only

phocines she treated) to represent "less adapted phocids" and later

reaffirmed this view, considering the smallest phocids to be gener-

ally "less advanced" and "presumably closer to the ancestral phocid"

(King 1972: 1 1 1 ). Finally. Mitchell ( 1966: I (judged Mirounga "the

most specialized and advanced phocid," subsequently advancing the

view (Mitchell 1967) that Pusa (the smallest phocid) represents the

"most logical" phocid structural ancestor, because of its highly

generalized morphology. He did not indicate, however, whether size

was part of this qualification. Other workers have regarded Pusa

(and the other constituents of the tribe Phocini) as a rather late-

diverging phocid clade but have not considered the implications of

this as they relate to size change (Ray 1976; Muizon 1982b).

Much of the impetus for construing phocines to be "basal"

phocids derives from the earlier incorrect paleontological practice

of allocating fragmentary fossil phocids of uncertain affinities to

the genus Phoca. This procedure may have been the outgrowth of

a nomenclaturally bound preconception that Phoca somehow

In A. Berta and T. A. Demere (eds.)

Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore. Jr.

Proc. San Diego Soc. Nat. Hist. 29:69-75, 1994

70

Andre R Wyss

represents an "archetypal" or "average" phocid (an assessment

inconsistent with recent studies), or it may have been a holdover

from the Linnaean practice of lumping all pinnipeds under this

name. Whatever its motivation, this practice was commonly em-

ployed at the end of the last century, when much fossil phocid

material was first being described. The systematics and nomencla-

ture of the living taxa have since been variously updated, but

information concerning fossils failed to keep pace. The result has

been, as discussed by Ray (1976), a paleontological literature now

badly out of date. Thus it is not true (though it is often stated) that

material properly referable to the genus Phoca itself is known from

the middle Miocene. Nevertheless, the impression that an extant

genus (whose living members are small) has ancient representa-

tives, combined with the view (widespread until recent years) that

phylogenetic relationship is virtually dictated by evidence from the

fossil record, strongly shaped notions of phocid evolution. Paleon-

tological hegemony in systematics has ended, and in the case of

phocids comparative studies indicate that some living forms (spe-

cies of the genus Monachus) actually represent the most persis-

tently conservative and earliest diverging members of the family

(Ray 1976; Repenning and Ray 1977). Indeed, the earlier view that

Phoca "typified" the family significantly hindered attempts to elu-

cidate relationships among phocids and the relationship of phocids

to other carnivorans.

One additional factor contributing to acceptance of the "small

equals primitive" concept of phocid evolution merits attention.

During the past three decades it has been widely supposed in

anatomically based studies that the Pinnipedia represent an unnatu-

ral phylogenetic assemblage, comprising on one hand otariids and

odobenids, thought to be related to ursids, and, on the other,

phocids, thought to be related to mustelids (Potamotherium, an

enigmatic late Oligocene through Miocene otterlike taxon, and

Semantor, a closely related Pliocene form, in particular) (Orlov

1933; McLaren 1960;Tedford 1976;Muizon 1982a,b).

PHYLOGENETIC CONSIDERATIONS

If the pinnipeds had multiple origins, the presumption that

phocids were small ancestrally seems credible given the size of

their perceived close relatives. The body weight of Potamotherium

(based on regression of dental dimensions in living carnivorans)

has been estimated as 7.3 kg (Legendre and Roth 1988); typical

mustelids are several times less massive than even the smallest

phocids. It is generally acknowledged, however, that support for the

diphyletic ancestry of pinnipeds has eroded significantly during the

past several years. Consensus is growing that the pinnipeds share an

exclusive common ancestor, on the basis of morphology (Flynn et

al. 1988). karyology. (Fay et al. 1967), immunology (Sarich 1969,

a,b), and biochemistry (Jong 1982; Arnason and Widegren 1986).

The weight of present evidence argues against the venerable notion

of a phocid-mustelid alliance (Wiig 1983; Wyss 1987), with broad

implications for theories of relationship within and between the

major lineages of pinnipeds. At the intrafamilial level, the accep-

tance of pinniped monophyly is perhaps most disruptive of conven-

tionally accepted views of relationships among phocids (Wyss

1988).

I have assumed the monophyly of pinnipeds and a close rela-

tionship among odobenids, two Miocene lineages that include

Desmatophoca and Allodesmus, and the phocids. Odobenids.

allodesmids. and desmatophocids have traditionally been consid-

ered closely allied to the otariids sensu stricto (e.g., Repenning and

Tedford 1977). The conclusions of this study, however, are not

strictly dependent on acceptance of either these assumptions, sup-

ported in detail by Wyss (1987, 1988).

Because their precise usage is critical to much of the discussion

that follows, some of the descriptors used in the preceding text, in

particular such terms as "primitive," "derived." and "advanced,"

bear comment. As highlighted by attempts such as Laws' ( 1959) to

place taxa within linear arrays, elements of such outdated notions as

the scala naturae or great chain of being still influence current

evolutionary thought (Queiroz 1988). In the contemporary litera-

ture one still frequently sees reference to taxa as "advanced" or

"highly evolved." the implication being that these are "direct line"

descendants of groups occupying a "lower" evolutionary rung.

Because evolution is not a process of sequential progression or

linear advancement, and because an organism may be progressive

in some respects and conservative in others, attributes of organisms

rather than the organisms themselves are properly regarded as

primitive or advanced. Thus the question is not whether "primitive"

phocids were small, but whether being small was primitive for

phocids.

Three lines of evidence have been used previously to assess

ancestral phocids' body size. First, for reasons of presumed physi-

ologic advantage, large body size is assumed to represent a derived

condition. Despite its intuitive appeal, this proposal remains diffi-

cult to evaluate critically in the absence of corroboration from

independent lines of evidence.

Second, evidence from the fossil record has been considered to

support the notion of small size in the ancestral phocid. Beyond the

inadvisability of interpreting the oldest known member of a group

as necessarily representing the ancestor of that group or its earliest

offshoot (Schaeffer et al. 1972), the fossil record implies the nearly

simultaneous appearance of disparate lineages of phocids (Ray

1976). Thus the stratigraphic record documenting the early history

of true seals is either highly incomplete, or the appearance of the

group was marked by a rather rapid pulse of morphologic change.

Beyond indicating that the earliest known phocids are similar in

size to such large modern forms as the monk seals (see Ray 1976),

fossils shed little light on the question of ancestral size.

Third is independent anatomical information, particularly of the

appendicular skeleton. Laws ( 1959) cited the apparent coincidence

of the assumed increase of size with changes in flipper morphology.

He noted that the phocine hindflipper bears large nails, that of other

phocids, vestigial or no nails. The digits of the phocine foreflipper

are nearly equal in length; in other phocids the first is markedly

longer and the succeeding four are progressively reduced. Using a

similar anatomical basis but arguing by analogy. King ( 1964) hy-

pothesized that the "trend in the Phocidae is towards the develop-

ment of a 'flipper,' like that of otariids[s] and cetaceans[s], from a

'paw.'" She noted the elongated first digit, reduced fifth digit, and

reduced claws as contributing to a "specialized" flipper shape as in

Ommatophoca (a large Antarctic form).

The pedal features cited by Laws and King as occurring in the

generally larger nonphocine phocids are indeed structurally ad-

vanced relative to the conditions seen in terrestrial carnivorans, and

those characterizing phocines (the smaller phocids) are by all ap-

pearances more conservative. Critical to determining the possible

phylogenetic implications of these features, however, is an evalua-

tion of their generality of distribution. The seemingly "specialized"

architecture of the foreflipper of Ommatophoca may be primitive at

a more general level. Indeed, enlargement of the first digit, reduc-

tion of the fifth, and reduction of claws lose their systematic impor-

tance among the phocids when it is recalled that very similar

conditions prevail in otariids, odobenids, and in all fossil pinniped

lineages for which adequate material is known. These attributes are

diagnostic of a more inclusive group, the Pinnipedia. and therefore

do not represent advanced features of the phocids. Phocines are

progressive among pinnipeds in secondarily reacquiring such oth-

erwise terrestrial carnivoran attributes as an entepicondylar fota-

men, a distinct supinator crest on the humerus, strong claws, and

The Evolution of Body Size in Phocids: Some Ontogenetic and Phylogenetic Observatons

71

more typically developed first and fifth digits of the manus —

including a strongly produced intermediate phalanx on the fifth

(Wyss 1988).

The only remaining seemingly advanced pedal character as-

cribed by Laws and King to the larger phocids is reduction of the

third digit of the hindflipper. This feature characterizes "mona-

chine" phocids (the largest members of the Phocidae) and in itself

appears to support monophyly of this assemblage. This feature also

occurs in Cystophora, however, a taxon generally regarded as a

phocine (Wyss 1988). Nonetheless, the shortened third digit repre-

sents a potential synapomorphy of the "Monachinae," monophyly

of which in turn implies that large size likely appeared at some time

subsequent to the origin of phocids. Broader comparisons do not

substantiate this view, however. Rather, a comparative review of 39

osteological and soft anatomical features (Wyss 1988) revealed that

shortening of the third digit is more reasonably interpreted as

primitive for phocids. with a reversal to a more typically carnivoran

form occurring among phocines. The earlier determination that

reduction of claws and enlargement of the first and reduction of

fifth digits on the hand represent generalized phocid conditions is

also supported independent of the question of a single versus mul-

tiple pinniped origin(s). In summary, apart from an insubstantial

functional rationalization, we are left without published evidence to

suggest that phocids are secondarily large.

Notions of character evolution are predicated on notions of

phylogeny. theories of transformation for any given feature (size,

for instance) being determined by superimposing its distribution on

a branching diagram derived from unrelated comparative informa-

tion. A key element of formulating such an independent phylogeny

is the assessment of character polarity, a procedure involving the

census of features in taxa outside the immediate group of concern

(Maddison et al. 1984). However, this process, outgroup compari-

son, may become problematic for taxa such as phocids, where there

is disagreement about higher-level affinities. Since opinion on the

question of the phocids' closest allies differs as widely as mustelids

and odobenids, I shall initially sidestep the controversy and attempt

to demonstrate that ancestral phocid size may be securely inferred

without recourse to comparison with any particular outgroup. In

addition, the question of phocid size change may be resolved unam-

biguously irrespective of which of the currently debated internal

arrangements of the group is adopted.

Acceptance of the two commonly recognized subfamilies of

phocids as monophyletic groups does not permit unequivocal as-

sessment of primitive phocid size. As depicted in Figure 2a, the

phocines' size varies, but "monachines" are uniformly large. It may-

be argued equally parsimoniously that ( 1 ) phocids were small

ancestrally and large size has originated among "monachines" and

some phocines independently, or (2) large size is primitive for

phocids and small size represents a secondary innovation among

some phocines.

As alluded to earlier, however, a dichotomy between the

Phocinae and "Monachinae" may be unjustified phylogenetically,

as it appears that the "Monachinae" are paraphyletic. Note that if

"monachines" are taken to be even minimally paraphyletic (i.e.,

divisible into two monophyletic subgroups as in Fig. 2b), the primi-

tive condition for phocid size is most economically interpreted as

large. The logic of outgroup analysis (see Maddison et al. 1984)

implies that this decision is not sensitive to the pattern of relation-

ship accepted for phocines. A "Monachinae" more highly

paraphyletic than the one depicted in Figure 2b would argue even

more persuasively for this interpretation.

Acceptance of even a limited degree of phylogenetic resolution

among phocines dispenses with the need for "monachine"

paraphyly as the basis for inferring large size as the ancestral phocid

condition. It is generally agreed that Cystophora and Erignathus are

successively more distant from the tribe Phocini (i.e., Phoca, Pusa,

Halichoerus, Histriophoca, and Pagophilus). Anatomical and cyto-

logical evidence supporting this branching pattern, from King

(1966), Fay et al. (1967), Burns and Fay (1970), and Muizon

( 1982a), were summarized by Wyss ( 1988). That Erignathus and

Cystophora fall within the size range of "monachines" (although

Cystophora less centrally so) establishes with reasonable assurance

that phocines are characterizable as primitively large. Thus, irre-

spective of whether "monachines" are paraphyletic, if Erignathus is

acknowledged as the sister taxon of other phocines (a concept

having strong anatomical and karyological support) the ancestral

phocid must have been large and the smallness of some phocines

must be secondary.

Acceptance of either "monachine" paraphyly or the placement

of Erignathus as the sister taxon of other phocines (two premises

supportable even in the context of pinniped diphyly) conflicts with

the judgment that phocids were ancestrally small. Acceptance of

both premises makes the case for size decrease even more secure.

Similarly, acceptance of pinniped monophyly increases confidence

in this conclusion. Pinnipeds exclusive of phocids are, in general,

rather large, particularly those here regarded as closely associated

with phocids: odobenids, desmatophocids, and allodesmids. Fig-

ure 3 presents observed ranges of standard lengths of adult females

and neonates for most living species of phocids and Odobenus. As

for extinct lineages, the standard length of a single male specimen

of Allodesmus kentensis has been estimated as 260 cm (Mitchell

1966), clearly placing this species within the cluster of large-bodied

pinnipeds (Fig. 3). Standard lengths for other closely related but

less well known species probably differed only slightly from this

figure. Condylobasal length of Desmatophoca oregonensis is re-

ported as 32.5 cm; that of D. brachycephala, 28.3 cm (Barnes

1987). Correspondingly, Desmatophoca was presumably shorter

than A. kernensis; nevertheless Desmatophoca represents a large

pinniped, clearly excluded from the cluster of small phocines

formed in the lower left half of Figure 2.

Phocinae

some some .. .. r

Monachinae Phocinae ■monachines" "monachines' Phocinae Monachinae Eng Cysto Phpc-ini

S.L?

Figure 2. Alternative interpretations of ancestral phocid body size. Recognition of two monophyletic subfamilies (A) precludes unambiguous

assessment of primitive phocid condition. Admittance of either "monachine" paraphyly ( B) or recognition of some resolution of phocine interrelationships

(C) results in acceptance of primitive condition as large. L, large; I, intermediate; S, small; Erig, Erignathus; Cysto, Cystophora.

72

Andre R. Wyss

1.6 1

* 1-4

1.2

(0

2

£

CD

To

c

o

c

CD

co

c

CO

1.0

2 0.8

CO

0.6

4-0,

1.0

-1 — r-

1.5

■ i

2.0

■* — r-

2.5

"■ — r-

3.0

_,— i —

3.5

4.0

Standard Length Adult Female (meters)

Figure 3. Ranges of adult female standard length versus neonatal standard length for most Recent phocid species and Odobenus. I, Hydrurga leptonyx,

2. Mirounga (range encompasses both species); 3, Odobenus rosmants: 4. Leptonychotes weddelli: 5. Monachus (range encompasses both surviving

species); 6. Erignathus barbatus; 7, Lobodon carcinophagus; 8, Ommatophoca rossi; 9, Cystophora crista; 10, Halichoerus grypus; 11, Pagophilus

groenlandica; 12, Histriophoca fasciata; 13, Phoca vitulina: 14, Pusa hispida; 15, P«.sa sibirica. Polygon separates members of the Phocini. Arrows

indicate species of which only maximum adult female size has been reported. Data from Ridgway and Harrison ( 1981 ). Fay (1981 ). and King ( 1983).

DISCUSSION

Several phylogenetic grounds, therefore, favor the hypothesis

of phocid size decrease. Given this, we may consider whether the

decrease is correlated with any other known patterns of character

evolution. In the Phocidae there is evidence of massive character

reversion (Wyss 1988). Examples of this phenomenon in phocines,

in addition to lengthening of the third digit of the pes, a relatively

unabbreviated fifth digit on the manus. de-emphasis of the first

digit of the manus, and development of strong claws, include

trochleated interphalangeai articulations, development of strong

keels on metapodial heads, presence of an entepicondylar foramen

and salient supinator crest on the humerus, strong scapular spine

with emphasis of the infraspinous fossa, and large hook-shaped

insertion of the teres major, and perhaps the resumed development

of a third upper incisor and lateral compression of the upper incisors

(Wyss 1988). None of these features characterized phocids

ancestrally and none is found elsewhere among pinnipeds, yet all

except the last are widely distributed among terrestrial arctoid

carnivorans.

Could these character reversals be related to changes in size,

which in turn might be related to general shifts in timing during

ontogeny? Modes of such shifts have been the subject of much

commentary (e.g., Albrecht et al. 1979). If patterns of character

change among phocids are indeed related to ontogenetic perturba-

tions, they are not related in any straightforward manner. The

distribution of observed character change in phocid phylogeny does

not seem easily accommodated by any single heterochronic trajec-

tory discussed by Albrecht et al. and other authors. Several ana-

tomical features illustrate these points.

One well-known uniquely derived phocid attribute is the lack of

fusion (even in maturity) between the paroccipital (jugular) process

of the exoccipital and the mastoid region of the petrosal (Burns and

Fay 1970), a fusion common to adult terrestrial carnivorans and

otariids (Fig. 4). Other phocid cranial sutures also appear to be late

in closing or never close tightly, for example, the one between the

basioccipital and the medial margin of the auditory region — a pat-

tern carried to extreme among phocines (see below). As in most

pinnipeds, but in contrast to terrestrial carnivorans, the antero-

ventral part of the orbital wall in phocids fails to ossify, resulting in

the persistence into adulthood of large vacuities. Similar morphol-

ogy may be seen in the fetal stages of some terrestrial carnivorans.

As discussed by Burns and Fay (1970). the basicranial region of

phocines is distinctive for the variability of numerous perforations,

including one or two pairs in the presphenoid, a large vacuity just

posterior of center in the basioccipital, and one or more on each

exoccipital between the condyle and paroccipital process. These

vacuities are most persistently developed among the Phocini but are

not uncommon among juvenile "monachines." Likewise, pterygoid

canals are often very large, even in fully mature individuals, par-

ticularly in Monachus, Leptonychotes, Lobodon. and Halichoerus

(Fig. 4), whereas in other pinnipeds and terrestrial carnivorans

these are by adulthood reduced to minute openings.

Two additional conditions of the phocid auditory region seem to

indicate developmental juvenilization. Embryologically in mam-

mals, the round window (fenestra cochlea) and the canal housing

the cochlear aqueduct (cochlear canaliculus) arise from a common

aperture on the posterior surface of the pars cochlearis of the

petrosal. During development, growth of a cartilaginous eminence

on the posteromedial rim of this aperture, the processus recessus,

delimits separate openings into the scala tympani for both of these

structures. In phocids, however, the growth of this structure is

suppressed, and an osseous division between the entrance of the

perilymphatic duct and the round window does not develop

(Kummer and Neiss 1957; Fleischer 1973). Similarly, embryonic

phocids fail to develop a prefacial commissure (= suprafacial com-

The Evolution of Body Size in Phocids: Some Ontogenetic and Phylogenetic Observatons

73

Figure 4. Ventral view of skull of "Monachus" albiventer displaying

several anatomical features discussed in text. Note lack of fusion between

the paroccipital process and mastoid region of the petrosal, and perforations

in the presphenoid. pterygoid, and basioccipital. From Gray (1874), re-

versed left to right from original.

missure), a cartilaginous rod typical of mammals (including terres-

trial carnivorans) that bridges the facial nerve on the dorsal surface

of the petrosal and contributes to the formation of the interna!

auditory meatus.

The apparent loss of cartilaginous extensions of the digits in

most adult phocids also seems to be readily accounted for by some

relatively "simple" ontogenetic truncation. Cartilaginous exten-

sions are otherwise present in all pinnipeds, including Enaliarctos,

the sister taxon of the remaining pinnipeds (Wyss 1987; Berta et al.

1989). Confidence in this assessment would be enhanced by de-

tailed ontogenetic and histologic investigation of the ends of the

digits, particularly among phocines. To date, such studies have been

carried out only on two lobodontines. Lobodon and Leptonychotes

(Leboucq 1904a,b).

Together, these features suggest that the origin of phocids may

have involved neotenic retention of embryonic traits in a group

stemming from a more generalized pinniped ancestry. Enthusiasm

for such an all-encompassing notion of developmental transforma-

tion is tempered, however, by the realization that other aspects of

phocid morphology represent products of a uniquely accelerated

ontogeny. One outstanding example of this concerns the develop-

ment of the auditory region. In otariids, at birth, the elements

constituting the auditory bulla remain unexpanded and unfused (as

is typical of eutherian mammals possessing an ossified auditory

bulla). At this ontogenetic stage the ectotympanic maintains its

primitive crescentic form, and the entotympanics remain in initial

stages of ossification. In phocids, in contrast, the auditory region is

essentially fully formed at birth, the bulla being completely fused.

Even at this early stage the deposition of thick layers of pachyostotic

bone in the temporal region — a diagnostic phocid attribute — is

highly advanced. Also, the massive ear ossicles of phocids appear

extremely early in ontogeny, having been noted, for example, in an

embryo of Leptonychotes 27 mm long (Fawcett 1918). Thus, if the

origin of phocids did involve neoteny, the effects of such a shift

clearly failed to extend to several components of their morphology,

most notably details of the auditory region. Other indications of a

developmental acceleration in phocids include the suppression or

early replacement of the deciduous dentition and an extremely short

period of lactation (King 1983).

In the subfamily Phocinae. neoteny of certain features is carried

even further than in the family Phocidae as a whole. Most obvious,

of course, is the dramatically smaller size of members of this

subfamily, previously discussed. Also, as noted above, the lack of

closure during ontogeny of several vacuities of the basi- and

exoccipital bones is most marked among the Phocinae, as is the

degree of separation between the auditory complex and the basioc-

cipital. Typically (and primitively in mammals, including pinni-

peds) the posterior lacerate foramen becomes defined during ontog-

eny as a roughly circular aperture between the posteromedial bor-

der of the auditory region and the exoccipital. Earlier in ontogeny

the presumptive "foramen" is confluent anteriorly with the

petrobasilar fissure, effecting a broad unossified region between the

auditory complex and its medially bordering bones; subsequent

obliteration of the petrobasilar fissure results in the typical adult

configuration of the foramen. As first noted by King (1966) and

confirmed by Burns and Fay (1970). the petrobasilar fissure rarely

closes among the Phocini. The latter authors found that the fissure

closed in less than 25% of their sample of any species of the tribe

Phocini, in 50% of their sample of Cystophora. in 2% of their

sample of Erignathus, and in 0% of their sample of other phocids.

King ( 1972) presented morphometric evidence interpretable as

indicating the juvenilized form of the phocine cranium. Comparing

the changes in size of the cranium, snout, and orbits of younger

(smaller) and older (larger) skulls of a single species (Mirounga

leonina), King noted the proportionally larger (longer, wider, and

higher) crania, shorter snouts, and larger orbits of the smaller skulls.

Proceeding to the comparison of skull shape among adults of differ-

ent species. King (1972: 96) found that "changes in proportions of

cranium, snout and orbit between the smaller and larger skulls are

just those that would be evident if young and adult skulls of the

same species were being compared. Thus the skulls of the smaller

seals of the Family, although adult, present a more juvenile appear-

ance than do skulls of larger animals." Recent proposals of phocid

interrelationships and the probable patterns of size change imply

that the juvenile appearance of the adult skulls of smaller species

(Fig. 5) more likely represents a secondary derivation, as King (on

the basis of incorrect paleontological evidence) had gone on to

suggest.

These data might suggest that the smallest phocids (Pnsa,

Phoca, Histriophoca, and Pagophilus) are neotenic derivatives of a

group (other phocids) that in some respects is itself already neo-

tenic. It is difficult to reconcile, however, such a simple develop-

mental scenario for the origin of small phocids with the known

distribution of other characters within the group. Why should the

origin of phocines appear to coincide with an episode of widespread

character reversal, and why are these reversals maintained in taxa

(small phocines) whose ontogeny is apparently truncated?

At present it seems unrealistic to attempt to explain the origin

and early diversification of phocids in terms of any absolute, cohe-

sive, or unidirectional model of developmental transmutation.

Rather it appears that phocid morphology is best viewed as the

product of a complex interplay between multiple, seemingly incon-

gruent patterns of developmental modification, including both ac-

celeration and retardation.

74

Andre R. Wyss

Figure 5. Comparative dorsal views of phoeid skulls. A, Mirounga leonina, juvenile. From Gray ( 1 874), reversed left to right. B, Mirounga leonina,

adult. From Turner ( 1 888), reversed top to bottom. C, Phoca vitulina, adult. From Blainville ( 1 839-64), reversed top to bottom. Changes made on negative

and print of C to make lighting appear to be from upper left. Note close resemblance in overall cranial proportions, particularly with respect to size of orbits

and "swollenness" of cranial vault.

ACKNOWLEDGMENTS

I thank Annalisa Berta and Thomas Demere, organizers of this

symposium, for the generous offer to participate. To them and to an

anonymous reviewer I owe numerous improvements to the final

manuscript. Francis H. Fay provided, as is his custom, an extremely

thorough and helpful commentary on the manuscript, for which I

am most grateful. For their superb efforts with Figures 4 and 5, I

thank Lorraine Meeker and Chester Tarka.

LITERATURE CITED

Albrecht. P.. S. J. Gould. G. F. Oster, and D. B. Wake. 1979. Size and

shape in ontogeny and phylogeny. Paleobiology 5:296-317.

Arnason, U.. and B. Widegren. 1986. Pinniped phylogeny enlightened

by molecular hybridizations using highly repetitive DNA. Molecu-

lar Biology and Evolution 3:356-365.

Barnes, L. G. 1 987. An early Miocene pinniped of the genus Desmatophoca

(Mammalia: Otariidae) from Washington. Natural History Museum

of Los Angeles County Contributions in Science 382.

The Evolution of Body Size in Phocids: Some Ontogenetic and Phylogenetic Observatons

75

Barrett-Hamilton. G. E. H. 1902. I. Mammalia. Pp. 1-66. in Report on

the Collections of Natural History Made in the Antarctic Regions

during the Voyage of the "Southern Cross." William Clowes &

Sons, London, England

Berta, A.. C. E. Ray, and A. R. Wyss. 1989. Skeleton of the oldest known

pinniped, Enaliarctos mealsi. Science 244:60-62.

Blainville, H. M. D. de. 1839-64. Osteographie. ou. Description

Iconographique Comparee du Squelette et du Systeme Dentaire des

Mammiferes Recents et Fossiles pour Servir de Base a la Zoologie

et a la Geologic Atlas; vol. 2. J. B. Bailliere et Fils (etc.], Paris.

France.