(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Proceedings of the San Diego Society of Natural History"

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: I 3 with large root; I, vestigial or missing; cheek teeth 
with strongly inflated principal cusps; P 1 " 3 and P, , with anterior 
and posterior cingular cusps, P 4 also with prominent posterolingual 
cingular cusp; P 4 molariform with large protoconc; M' with large 
conules. lingual cingulum only between principal cusps; M : mark- 
edly smaller than and lying posterolingual to M 1 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 P 1 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 I 3 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 I 2 and I' and right and 
left P : and M 1 , 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 I 3 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 I 3 
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 P 2 . 

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 M 2 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, P 1 4 , 
and M 1 2 . The actual teeth present in the referred cranium are the 
roots of the left I 2 "-' 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. 
I 1 and I 2 are small and have transversely compressed roots. I 1 is 
smaller than I 2 , and both teeth are implanted essentially vertically in 
the palate. The I^'s are much larger, being approximately four times 
the diameter of I 2 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 I 1 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 I 1 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 M 2 is positioned more medially. P 1 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. 

P 2 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. 

P 1 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 P 2 , 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. 

P 4 was a large tooth, nearly as large as M 1 . 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 P 3 and P 4 , the lateral margin of the 
maxilla begins a strong lateral bend, so that P 4 is oriented obliquely 
to the cheek-tooth row. A diastema of approximately 3 mm sepa- 
rates the anterior (paracone) root of P 4 from the posterior root of P\ 

M 1 is a massive tooth. It appears to have been approximately 
30% larger than P 4 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 M 2 alveoli indicate that this tooth was approximately half 
the size of the M 1 . M 2 is positioned lingually opposite the talon of 
M'. Like both P 4 and M 1 , M 2 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 I 3 and C. right P 1 . left P 1 , left P\ right P 4 . 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 P 1 . 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 P 2 and 
severely crushing and comminuting the thin alveolar walls of left 
and right M : and M 1 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 M 1 

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 



Kolponomos 


Kolponomos 


newportensis 


clallamensis 


USNM 


UCMP 


LACM 


2 1 5070 


50056" 


131 148'' 


253.1 





258.4 


107.3 


— 


107.0 


155.2 


— 


153.7 


114.2 


ca. 108 


117.7 


75.2 


— 


(71) 


ca. no 


ca. 98 


119.6 


78.3 


75.2 


74.3 


74.9 


80.9 


(82) 


46.2 


46.8 


45.4 


72.3 


— 


82.3 


179.5 


— 


(178) 


137.0 


— 


141.2 


182.0 


— 


183.7 


116.5 


— 


123.0 


61.5 


— 


74.0 


— 


37.0 


38.5 


— 


36.3 


— 


26.8 


— 


33.3 


18.3 


— 


15.0 


11.1 


11.6 


13.0 


11.3 


20.0 


21.6 



"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. 





K. newportensis 


K. clallamensis 




(USNM 215070) 


(LACM 131148) 


Upper teeth (length x width) 






I' 


12.8 x 10.7 


— 


C 


15.4 x 13.9 


— 


P 1 


7.3 x 11.5 


— 


P : 


16.2x — 
17.2 x 12.5 


15.7 x 10.3 


P' 


ca. 16 x 12 


— 


P 4 


18.5x25.5 


— 


M 1 


22.3 x 28.9 


21.9x31.4 


M : 


16.8 x 16.4 
17.8 x 17.1 


— 


Lower teeth (length x width) 






C 


17.2 x 13.1 
16.5 x 13.5 


— 


P, 


— x ca. 9 

7.6x9.9 


— 


P 2 


13.7 x 10.8 


— 




14.0 x 10.5 




p , 


16.1 x 11.2 

16.2 x ll.l 


— 


P 4 


18.0 x 14.0 
18.7 x 13.6 


— 


M, length 


27.3: 27.5 


— 


M, width trigonid 


24.0; 24.5 


— 


M, width talonid 


— ; 22.7 


— 


M, 


21.5x22.3 


— 




21.3x — 




Mandible 






Length of horizontal ramus. 






condyle to tip of rostrum 


201.5 


— 


Length of base of coronoid process 55.5; 55.7 


— 


Depth of mandible beneath P, 


59.5; 59.8 


— 


Depth of mandible beneath 






anterior part of M, 


43.8; 44.2 


— 


Depth of mandible beneath 






posterior part of M ? 


40.0; 42.2 


— 


Width of mandible atP, 


18.8; 20.5 


— 


Width of mandible at M, 


21.8; 24.2 


— 


Length of symphysis 


53.5 


— 



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 I 3 . 

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 P 2 '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 P 1 , left P 1 , left P\ 
right P 4 , 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 I 3 . I 3 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 P 1 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 P 2 . 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. P 3 is only slightly larger than P 2 and two- 
rooted, the more anterior root passing inside the posterior root of P 2 
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. 

P 4 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 P 4 only an 
encircling band of thin enamel remains. The enamel is broken on 
the posterior side by abrasion between P 4 and the adjacent M 1 . 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 : . P 3 , and M 1 " 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 M 1 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 P 4 . The M : is triangular in form and 
least worn on the left side. It is positioned lingually opposite the M 1 
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 M 1 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 P 4 on the left side. The third and most posterior foramen lies at 
mid-depth of the horizontal ramus beneath the posterior root of P 4 . 
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 I 2 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 P 4 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 P 4 ; 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 M 3 . 

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 



P 4 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 P 4 , and development of a "heel" in M 2 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). 











Taxon" 








Derived state 


AMP 


MUS 


URS 


AMC 


PAC 


ALL 


KOL 


ENA 


1. Basioccipital excavated laterally 


1 







1 


1 


1 


1 


1 


2. Shallow suprameatal fossa 





1 




1 


9 











3. M 3 absent 





1 




1 


1 


1 


1 


1 


4. Basioccipital wide posteriorly 










1 


1 


1 


1 


1 


5. P 4 large protocone 










1 


1 


1 


1 


1 


6. M : with "heel" 










1 


1 


1 


1 


1 


7. M'~ 2 loss parastyle 










1 


1 


1 


1 





8. M'~ : loss paraconule 










1 


1 


7 








9. M 1 lingual metaconule 













1 


7 








10. M, reduced paraconid 










1 


1 


7 





•> 


1 1. M, size metaconid = protoconid 










1 


1 


7 


1 


9 


1 2. Infraorbital canal short 











1 


1 


1 


1 


1 


1 3. Infraorbital foramen large 











1 


1 




1 


1 


14. P 4 short metastyle 











1 


1 




1 


1 


15. Palate posteriorly extended 














1 







1 


16. M, talonid as wide as trigonid 














1 




1 





17. M, metaconid large 














1 




1 





18. Nasolabialis fossa present 



















1 


1 


19. Infraorbital fossa present 



















1 


1 


20. Mastoid process large 



















1 





21. Postorbital process absent 



















1 


1 


22. Alisphenoid "strut" present 




















1 


1 


23. Postglenoid foramen vestigial 














(1 





1 


1 


24. M 2 lingual to M' 




















1 


1 


25. Anterior lacerate foramen and 


















foramen rotundum in common fossa 














(1 





1 


1 


26. M, absent 





1 














1 


1 



"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. M 1, 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 M 1 metaconule. — The ursid metaconule 
is strongly connected to the longitudinally elongated protocone in 
M 1 (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 M 1 metaconule but this cusp is large and well connected to the 
protocone by a crista. Kolponomos also retains a primitive tribos- 
phenic form of M 1 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 ofP 4 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 M 1 . 

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. M 2 lingual to M' . — A peculiar feature of the dentition of 
pinnipedimorphs (and Potamotherium) that retain M 2 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 M 1 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 ). 





Percentage Complete 


Taxon 


Cranial 


Postcranial 


Enaliarctos 


93 


78 


Pteronarctos 


94 





Thalassoleon 


93 


71 


Imagotaria 


91 


60 


Aivukus 


77 


33 


Desmatophoca 


90 





Allodesmus 


86 


93 


Pinnarctidion 


53 





Piscophoca 


77 


58 


At rophoca 


60 


73 


Homiphot a 


90 


45 


Comphotaria 


63 


45 


Alachtherium 


16 





Pontolis 


23 





Pliopedia 





24 


Prorosmarus 


4 





Dusignathus 


26 






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) M 1 : reduced in size relative to premolars. 
(67) M 1 " 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 P 4 and M 1 
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. 







1 


2 


3 


4 


5 


6 


7 


8 


9 


1 


1 1 


1 2 


1 3 


1 4 


1 5 


1 6 


1 7 


1 8 


1 9 


2 


2 1 


2 2 


2 3 


2 4 




















































1 


Amphicyonidae 























9 
































9 

















2 


Mustelidae 






















1 














0/1 

















9 


0/1 














3 


Procyonidae 






















1 














0/1 


0/1 














? 

















4 


Ursidae 









































































5 


Enaliarctos 

























9 




















1 






















6 


Pteronarctos 










0/1 













1 



















1 





















7 


Arctocephalus 










1 







1 


1 


1 






1 









1 


3 





















8 


Callorhinus 










1 







1 


1 


1 






1 









1 


3 





















9 


Otariinae 










1 







1 


1 


1 






1 









1 


3 





















1 


Thalassoleon 










1 







1 


1 


1 






1 









1 


3 





















1 1 


Aivukus 


















1 


9 


1 














9 


1 


2 





















1 2 


Alachitherium 


9 


9 


9 


9 


9 


? 


9 


9 


9 




9 


? 


9 


9 


? 


? 


9 


9 


9 


9 


9 


9 


9 


9 


1 3 


Dusignathus 


9 


? 


9 


9 


9 


? 


9 


9 


? 




9 


9 


9 


9 


? 


9 


9 


? 


9 


9 


9 


9 





9 


1 4 


Gomphotana 


? 


9 





9 


9 


9 


1 




1 






1 







1 


1 








9 














9 


1 5 


Imagotaria 








1 








1 


1 




1 














1 


1 


2 





1 














1 


1 6 


Odobenus 


1 





2 








1 


1 




1 






1 







1 


1 


2 





1 














1 


1 7 


Pliopedia 


9 


9 


9 


9 




9 


? 


9 


9 




9 


9 


9 


9 


9 


9 


? 


9 


9 


9 


9 




? 


9 


1 8 


Pontolis 


9 


9 


9 


9 




? 


? 


9 


? 




9 


? 


? 


9 


? 


9 


? 


9 


9 













1 


1 9 


Allodesmus 


1 


1 










2 


1 




1 






0/1 






1 





1 





1 













2 


2 


Desmatophoca 


1 


1 










2 


1 




1 













1 





1 





1 













2 


2 1 


Pinnarctidion 


9 


9 


9 


9 




2 







? 






? 






? 





1 



















2 


2 2 


Acrophoca 


9 


9 


? 


? 






9 




9 




9 


? 






9 





2 





9 




? 






2 


2 3 


Cystophora 


1 















1 




1 






1 






1 


9 


2 


1 


1 




1 






2 


24 


Erignathus 


1 















1 




1 






1 






1 


1 


2 





1 




1 






2 


2 S 


Homiphoca 


1 





1 









1 




1 






1 






1 


1 


2 





9 











2 


26 


Lobodontini 


1 


1 












1 




1 






1 






1 


1 


2 





1 











2 


27 


Mirounga 


1 


1 












1 




1 






1 






1 


1 


2 





1 











2 


2 8 


"Monachus" 


1 


0/1 












1 




1 






1 






1 





2 





1 











2 


29 


Phocini 


1 


0/ 1 












1 




1 






1 






1 





2 


1 


1 




1 






2 


30 


Piscophoca 


1 


9 












1 




1 






9 






1 





2 





9 




9 






2 







25 


26 


27 


2 8 


2 9 


3 


3 1 


3 2 


3 3 


3 4 


3 5 


3 6 


3 7 


3 8 


3 9 


4 


4 1 


4 2 


4 3 


4 4 


45 


4 6 


4 7 


4 8 




















































1 


Amphicyonidae 


? 


9 


? 


9 








9 


? 














9 




















9 


9 


9 


? 





2 


Mustelidae 





9 


? 








9 
























































3 


Procyonidae 





9 


? 








9 




















0/1 



































4 


Ursidae 






































0/1 


































5 


Enaliarctos 


















9 

















































6 


Pteronarctos 


















9 

















































7 


Arctocephalus 


































































8 


Callorhinus 


































































9 


Otariinae 


































































1 


Thalassoleon 





























































? 


? 


1 1 


Aivukus 




? 


9 










9 


9 













1 













1 












9 


9 


1 2 


Alachitherium 


9 


9 


9 





? 


? 


9 


9 


9 




9 


? 


9 


9 


9 




9 


9 


9 





? 




? 


? 


1 3 


Dusignathus 


9 


1 


9 


9 


9 


? 


9 


9 







9 


? 


9 


9 


9 




9 


9 


9 


9 


9 




9 


9 


1 4 


Gomphotaria 


9 


9 


9 





? 




9 


9 







9 
















? 


9 


9 





9 




? 


9 


1 5 


Imagotaria 


1 


1 


1 










9 


9 













1 

























? 


1 


1 6 


Odobenus 


1 


1 


1 










1 


1 













1 



























1 


1 7 


Pliopedia 


9 


9 


9 


9 


9 


9 


? 


9 


? 




9 


9 


9 


? 


9 




9 


? 


9 


9 


9 




? 


9 


1 8 


Pontolis 


9 


1 


? 









9 


9 













1 





9 



















9 


9 


1 9 


Allodesmus 


1 


1 


1 









9 


9 












1 


9 


2 














1 






9 


20 


Desmatophoca 


1 


2 


1 









9 


? 












1 


9 


1 














1 






9 


2 1 


Pinnarctidion 


? 


2 


9 









9 


9 












1 


? 


1 














1 




? 


9 


2 2 


Acrophoca 


9 


2 


9 


9 




? 


9 


9 


9 






1 


2 


9 


2 


















9 


9 


2 3 


Cystophora 


1 


? 


1 










1 











2 


1 


2 








2 


1 











24 


Engnathus 


1 


2 


1 










1 











2 





2 










1 











25 


Homiphoca 


1 


2 


1 










9 








1 


2 


9 


2 






















26 


Lobodontini 


1 


2 


1 










1 








1 


2 





2 






















27 


Mirounga 


1 


2 


1 










1 











2 





2 






















28 


"Monachus" 


1 


2 


1 




0/1 






1 











2 





2 






















29 


Phocini 


1 


2 


1 






9 




1 











2 


1 


2 


0/1 






2 


1 











30 


Piscophoca 


9 


9 


9 


? 




9 


9 


9 


9 






1 


2 


9 


2 








1 















38 



A. Berta and A. R. Wvss 



Table 2 (continued). 







49 


50 


5 1 


5 2 


5 3 


54 


55 


5 6 


57 


5 8 


59 


60 


6 1 


6 2 


6 3 


6 4 


6 5 


6 6 


6 7 


6 8 


69 


70 


7 1 


7 2 
















































1 


Amphicyonidae 










































































2 


Mustelidae 


























1 





7 











1 














0/ 1 














3 


Procyonidae 


























7 





7 











1 





























4 


Ursidae 










































































5 


Enaliarctos 


7 


1 











? 

















7 














1 












1 







6 


Pteronarctos 


7 


0/1 











7 

















7 











1 


1 









1 










7 


Arctocephalus 


1 














1 

























3 


2 



















8 


Callorhinus 


1 














1 

























3 


2 









1 









9 


Otaninae 


1 














1 











1 













3 


2 









0/1 









1 


Thalassoleon 


7 














7 


























2 


1 



















1 1 


Aivukus 


? 





7 


7 


7 


7 


1 





7 


1 






1 






3 


7 






1 












1 2 


Alachitherium 


9 


7 


1 





7 


7 


7 


7 


7 


7 


7 




1 




7 


7 


7 


7 


7 


7 












1 3 


Dusignathus 


7 


7 


1 








7 


7 


7 


7 


7 


7 











3 


2 


7 




1 












1 4 


Gomphotaria 


7 





1 








7 


1 





7 


7 


7 




1 




7 


3 


1 


7 


7 


1 





7 


7 


7 


1 5 


Imagotaria 


7 





? 








7 








7 


1 














2 


1 


1 

















1 6 


Odobenus 


1 





1 








1 


1 





7 


7 


7 




1 






3 


2 


1 




1 












1 7 


Pliopedia 


7 


7 


7 




7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


1 8 


Pontolis 


7 


7 


7 




7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


7 


1 9 


Allodesmus 


7 





1 




1 


7 








7 





7 


1 









3 


2 


1 

















20 


Desmatophoca 


7 





1 




1 


7 








7 







7 









2 


1 /2 


1 










1 






2 1 


Pinnarctidion 


7 





7 




7 


7 


7 





7 





7 


7 


7 











1 


7 


7 




7 


7 


7 


7 


2 2 


Acrophoca 


? 





1 




1 


7 


7 


7 


7 


7 


1 


7 










2 


2 


1 


7 














2 3 


Cystophora 


2 





? 







1 


1 





1 


7 














3 


2 


1 
















24 


Erignathus 


2 





7 







1 







1 


7 














3 


2 


1 
















25 


Homiphoca 


? 





1 




7 


7 


1 




1 


7 


7 












2 


2 


1 
















26 


Lobodontini 


2 





1 







1 


1 




1 


1 














3 


2 


1 
















27 


Mirounga 


2 





1 







1 


1 




1 


1 














3 


2 


1 
















28 


"Monachus" 


1 /2 





1 







1 


1 




1 


1 














3 


2 


1 
















29 


Phocini 


2 





7 







1 








1 

















3 


2 


1 
















30 


Piscophoca 


? 





1 




1 


7 


1 


1 


7 


1 














2 


1 


1 












7 


7 







7 3 


7 4 


7 5 


76 


7 7 


7 8 


7 9 


80 


8 1 


8 2 


8 3 


8 4 


8 5 


8 6 


8 7 


8 8 


8 9 


9 


9 1 


| 9 2 


9 3 


9 4 


I 9 5 


9 6 










































1 


Amphicyonidae 











? 


? 



























































2 


Mustelidae 










7 


? 














0/1 





























0/1 











0/1 


3 


Procyonidae 










? 


? 





? 


? 


0/1 





7 


? 


























? 











4 


Ursidae 
























































0/1 

















5 


Enaliarctos 










? 


? 








1 














1 


? 




1 





1 


1 


1 











1 


6 


Pteronarctos 










7 


? 





7 


7 


7 


? 


7 


? 


? 


7 


? 


1 


? 


? 


? 


? 


? 


? 


7 


? 


7 


Arctocephalus 
























1 











1 






1 


1 


1 


1 


1 





1 


1 


1 


8 


Callorhinus 
























1 











1 






1 


1 


1 


1 


1 





1 


1 


1 


9 


Otaninae 












? 











1 











1 






1 


1 


1 


1 


1 





1 


1 


1 


1 


Thalassoleon 









7 


? 


? 


7 


? 


1 











1 






1 


1 


1 


1 


1 





1 


1 


1 


1 1 


Aivukus 









7 


? 


? 


7 


7 


7 


7 


? 


7 


7 


? 




1 


1 


1 


1 


1 


1 


1 


1 


2 


1 2 


Alachitherium 









? 


? 


7 


7 


? 


? 


? 


? 


? 


? 


? 


? 


7 


7 


? 


7 


7 


? 


? 


7 


7 


1 3 


Dusignathus 






1 


? 


? 


7 


7 


? 


? 


7 


7 


? 


? 


? 


? 


7 


? 


? 


? 


? 


7 


? 


7 


? 


1 4 


Gomphotaria 









? 


? 


? 


? 


? 


? 











1 


7 


? 


1 


1 


1 


1 


1 


? 


1 


1 


? 


1 5 


Imagotaria 









7 


? 





7 


7 


7 











1 





1 


1 


1 


1 


1 


1 


1 


1 


1 


2 


1 6 


Odobenus 









1 


1 


1 


1 


1 


0/1 











1 





1 


1 


1 


1 


1 


1 


1 


1 


1 


2 


1 7 


Pliopedia 


7 


7 


7 


? 


? 


? 


7 


7 


7 


7 


? 


? 


? 


7 


1 


1 


1 


1 


1 


1 


1 


1 


1 


2 


1 8 


Pontolis 


7 


7 


7 


7 


? 


? 


? 


? 


7 


? 


7 


? 


7 


7 


7 


? 


? 


7 


? 


? 


? 


? 


? 


7 


1 9 


Allodesmus 





















1 


1 











1 


? 


1 


1 


1 


1 


1 


1 





1 


1 


1 


20 


Desmatophoca 


0/1 






? 


7 


? 


? 


? 


? 


? 


? 


7 


? 


7 


? 


? 


7 


? 


7 


? 


? 


? 


? 


7 


2 1 


Pinnarctidion 


7 


7 





? 


? 


? 


? 


? 


7 


? 


? 


7 


7 


? 


? 


? 


? 


? 


7 


? 


7 


? 


? 


7 


2 2 


Acrophoca 








1 


1 


1 


1 


1 


? 





? 


? 


1 





1 


1 


1 


1 


1 


1 





1 


1 


2 


2 3 


Cystophora 








1 


1 





1 


1 


1 


1 


1 











1 


2 





1 





1 





1 


1 


1 


2 4 


Erignathus 








1 


1 





1 


1 


1 


1 


1 











1 


2 





1 





1 


1 


1 


1 


1 


25 


Homiphoca 






? 


? 


7 


7 


? 


? 


7 


7 


7 


? 


? 


? 


1 


1 


? 


1 





1 


7 


1 


1 


2 


26 


Lobodontini 








1 


1 


1 




1 


1 


1 





2 


1 





1 


1 


1 


1 


1 


1 





1 


1 


1 


27 


Mirounga 








1 


1 


1 




1 


1 








1 


1 





1 


1 


1 


1 


1 


1 





1 


1 


2 


2 8 


"Monachus" 








1 


1 


7 




1 


1 








1 


1 





1 


1 


1 


1 


1 


1 





1 


1 


2 


29 


Phocini 








1 


1 







1 


1 


1 


1 











1 


2 





1 





1 


1 


1 


1 


1 


30 


Piscophoca 








1 


1 


1 




? 


7 


1 


7 


1 


1 





1 


1 


1 


1 


1 


1 





1 


1 


2 



Pinniped Phylogeny 



39 



Tablh 2 (continued). 







97 


9 8 


9 9 


1 00 


1 1 


1 02 


1 03 


1 4Jl 5 


1 6 


1 07 


1 8 


1 09 


1 1 


1 1 1 


1 1 2 


1 1 3 


1 1 4 


1 1 5 


1 1 6 


1 1 7 


1 1 8 


1 1 9 


1 20 


















































1 


Amphicyonidae 


9 











9 


7 


9 


? 





9 


























9 














7 


2 


Mustelidae 

















7 





? 
































9 


9 











? 


3 


Procyonidae 

















7 





? 





























9 








? 


? 


? 





4 


Ursidae 

















? 
























































5 


Enaliarctos 





1 








7 


? 


? 


7 







7 































1 





1 


6 


Pteronarctos 


? 




? 


7 


? 


? 


7 


? 


7 


? 


7 


7 


? 


? 


7 


? 


? 


? 


? 


? 


? 


7 


7 


9 


7 


Arctocephalus 










1 


1 


1 




1 


7 




















0/1 





1 





1 


1 




1 


8 


Callorhinus 










1 


1 


1 




1 






















1 





1 





1 


1 




1 


9 


Otariinae 










1 


1 


1 




1 




























1 





1 


1 




1 


1 


Thalassoleon 










1 


1 


? 


7 


? 


7 




















1 





1 





1 


1 




7 


1 1 


Aivukus 


? 




2 


1 


7 


7 


7 


? 


7 


? 


7 




7 




? 





7 


? 


1 


? 


? 


7 


? 


7 


1 2 


Alachitherium 


? 




? 


7 


7 


7 


7 


? 


7 


7 


7 


7 


? 




? 


? 


9 


? 


7 


? 


7 


9 


? 


7 


1 3 


Dusignalhus 


? 




? 


7 


7 


? 


7 


? 


? 


7 


7 


? 


7 




? 


? 


? 


7 


? 


7 


? 


9 




9 


1 4 


Gomphotana 


? 




? 


1 


7 


7 


7 


7 


? 


? 


? 


1 


9 




7 








7 


1 


7 


1 


1 




1 


1 5 


Imagotaria 


1 




1 


1 


7 


? 


? 


7 


? 


? 


? 


1 


? 




7 





7 


7 


1 





1 


1 







1 6 


Odobenus 







2 


1 


1 


1 




1 







1 


1 












1 


1 


1 




1 


1 




1 


1 7 


Pliopedia 


7 




1 


1 


7 


? 


? 


? 


7 


? 


? 


7 


7 




? 


? 


7 


? 


7 


7 


? 


7 




7 


1 8 


Pontolis 


7 




? 


7 


7 


9 


? 


9 


7 


7 


? 


7 






? 


? 


? 


9 


? 


9 


? 


? 




7 


1 9 


Allodesmus 










1 


1 


1 




7 







? 


1 












1 





1 




1 


1 




1 


20 


Desmatophoca 


7 




? 


7 


7 


7 




? 




? 


7 


? 






-? 


? 


? 


7 


? 


9 


9 


9 




9 


2 1 


Pinnarctidion 


7 




? 


? 


7 


? 




7 




? 


? 


? 






? 


? 


? 


? 


? 


? 


7 


? 




7 


2 2 


Acrophoca 


7 







1 


7 


? 




7 




1 


? 


1 









1 







1 




1 






7 


23 


Cystophora 


1 
























1 















1 







1 




1 









24 


Erignathus 


1 








































1 







1 




1 









2 5 


Homtphoca 


7 




7 


7 


7 


7 


? 


? 




? 


? 


? 









1 


0/1 





1 




1 






7 


2 6 


Lobodontini 










1 


1 


0/1 


1 


1 




1 


1 


1 









1 







1 




1 









2 7 


Mirounga 










1 


1 





1 


1 




1 


1 


1 









1 







1 




1 









28 


"Monachus" 










1 


1 





1 


0/1 




1 


1 


1 









1 


0/1 




1 




1 









29 


Phocini 


1 





































1 


1 







1 




1 









30 


Piscophoca 


7 







1 


? 


? 


? 


7 




? 


? 


1 









1 







7 


? 


? 


? 





7 







1 2 1 


1 22 


1 23 


1 2 4h 2 5 


1 26 


1 27 


1 2 8Jl 29I1 3 


1 3 1 


1 3 2I1 3 3 


1 34 


1 3 5|l 3 6|l 3 7|l 3 8 


1 39 


1 40 


1 4 1 


1 42 


1 43 




































1 


Amphicyonidae 























7 


7 


9 


7 


7 


7 


9 


9 


7 


7 


7 


9 


7 


7 


9 


7 


2 


Mustelidae 



































7 


? 


? 





7 


7 








7 











3 


Procyonidae 


? 

















7 











? 


? 


9 


7 





7 


7 


7 





7 











4 


Ursidae 
































9 


? 





7 





? 





7 





7 











5 


Enaliarctos 








7 


7 








7 


? 


? 


9 


9 


? 





9 


? 


7 


7 


7 


7 


7 


7 


7 


7 


6 


Pteronarctos 


7 


? 


? 


? 


? 


? 


? 


7 


? 


? 


? 


7 


9 


? 


7 


7 


7 


7 


7 


7 


7 


9 


9 


7 


Arctocephalus 





1 


1 



































1 























1 


8 


Callorhinus 








1 



































1 























1 


9 


Otariinae 





1 


1 




















1 














1 























1 


1 


Thalassoleon 





0/1 





7 








? 


? 


? 


? 


? 





? 


7 


7 


7 


7 


7 


7 


7 


7 


7 


9 


1 1 


Aivukus 


? 


? 


? 


7 


7 


? 


? 


? 


? 


? 


7 


? 


? 


? 


7 


7 


7 


? 


7 


7 


7 


7 


9 


1 2 


Alachitherium 


? 


? 


? 


7 


? 


? 


7 


7 


? 


? 


? 


? 


7 


? 


? 


7 


? 


7 


? 


7 


7 


7 


9 


1 3 


Dusignathus 


? 


? 


? 


? 


7 


7 


? 


? 


7 


? 


7 


7 


7 


7 


7 


? 


7 


? 


7 


7 


9 


7 


9 


1 4 


Gomphotaria 


? 





? 


1 


1 


1 


7 


7 


? 


1 


7 


7 


? 


9 


? 


7 


9 


9 


7 


9 


7 


7 


9 


1 5 


Imagotaria 





? 





1 


1 


1 


7 


7 


7 


7 


7 


7 


7 


? 


? 


7 


7 


7 


7 


7 


7 


7 


7 


1 6 


Odobenus 











1 


1 


1 




1 




2 





1 





? 





1 







1 


1 


1 








1 7 


Pliopedia 


? 


? 


? 


? 


? 


7 


? 


? 


? 


? 


? 


9 


? 


? 


7 


7 


? 


7 


7 


9 


7 


7 


7 


1 8 


Pontolis 


7 


? 


7 


? 


? 


? 


? 


? 


? 


? 


7 


7 


? 


? 


7 


? 


7 




? 


7 


7 


7 


7 


1 9 


Allodesmus 











1 





1 




? 


? 


7 


7 


7 


7 


? 


? 


7 


9 




7 


9 


7 


7 


7 


20 


Desmatophoca 


? 


? 


7 


? 


? 


? 


? 


? 




? 


7 


7 


? 


? 


7 


7 


? 




9 


7 


7 


7 


7 


2 1 


Pinnarctidion 


? 


? 


7 


? 


9 


? 


7 


? 




? 


? 


? 


7 


? 


? 


7 


7 




7 


7 


7 


7 


9 


22 


Acrophoca 





1 





1 





2 




? 




7 


7 


? 


7 


? 


7 


7 


? 




7 


7 


7 


7 


9 


2 3 


Cystophora 


1 


1 





1 





2 




1 




1 


1 


1 


1 








1 


1 




1 


1 


1 


1 





24 


Erignathus 


1 


1 





1 





2 




1 




1 


1 


1 











1 







1 


1 


1 


1 





2 S 


Homiphoca 


1 


1 


? 


? 


7 


? 




? 




? 


? 


? 


? 


7 


7 


7 


7 




7 


7 


7 


7 


9 


26 


Lobodontini 


? 


1 





1 





2 




1 




1 


1 


1 


0/1 








1 


1 




1 


1 


1 


1 





27 


Mirounga 





1 





1 





2 




1 


0/1 


2 





1 


1 


1 





1 


1 




1 


1 


1 


1 





28 


"Monachus" 





0/1 





1 





2 




1 




2 





1 


0/1 


1 





1 







1 


1 


1 


0/1 





29 


Phocini 


1 


1 





1 





2 




1 




1 


1 


1 


1 








1 


1 




1 


1 


1 


1 





30 


Piscophoca 


? 


? 


? 


1 





2 




? 




? 


? 


? 


? 


? 


? 


? 


? 




? 


? 


? 


? 


9 



40 



A. Berta and A. R. Wyss 



3 
O 

4-» 

3 

o 



V) 


O 


c 
o 


10 


O 


M 


<D 


3 


Jj 


c 

s 


"3 

V) 

-2 


5 


'B 


<u 


nj 




£ 


£ 


iS 


c3 



■c 





c 



O 

-C 


03 


^5 


Q. 


O 


4^ 


O 


o 


in 


o 


4-1 


-c 


<u 


^~ 


TO 


& 


■a 


nj 


E 


o 


c 


i/i 


k- 




c 


Q 


u 


^ 


£ 


i; 




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) I 1 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) P 4 
protocone shelf absent, (64) P 4 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) I 3 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) P 4 double rooted, fossil lobodontines represent an inter- 
mediate transformation, P 4 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. 

an d 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. 

an d 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. 

an d 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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 P 4 and M 1 . = 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 P 4 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = composed of cancellous bone. 1 = heavily 
pachyostic. A pachyostotic mastoid is unique to phocids ( Burns and 
Fay 1970). 

24. Mastoid process. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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)|. = 
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. = 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. = 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. = 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. = 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. = 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. = 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. = 
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. = 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. = 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. = 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. = 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. = 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. = 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. = 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 I 1 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. = 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. I 1 " 2 , transverse groove. = 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. I 3 . = 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 I 3 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. I 3 , lingual cingulum. = 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. = 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. = 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\ = 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 P 3 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. P 4 , protocone shelf. = 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. P 4 . = 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 P 4 has only a single root (3). 

65. M 1 . = 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 M 1 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 
M 1 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 . = 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 M 1 and reduction or loss of M 2 (see character 68). 

67. M'~ 2 cingulum. = 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. M 2 . = present. I = absent. The occurrence of M 2 varies 
within each of the major pinniped groups. Among walruses, M 2 is 
lacking in Odobenus and Aivukus (Repenning and Tedford 1977). 
In otariids, M 2 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. 

M 2 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. = 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. = 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. = 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. = 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. M 2 . = 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 = 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. = 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. 
= 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 
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. = 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). 
= 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = terrestrial, 1 = aquatic. Odobenus and 
phocids (except Mirounga) copulate in the water, whereas otariids 
and other carnivorans copulate on land. 

130. Pelage. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. = 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. 





Sinus width 


Basioccipital 


Relative sinus 


Taxon and specimen" 


(SW) 


width (BW) 


width'' 


Enaliarctos mealsi 








LACM 4321, holotype 


( lOl- 


37.5 


53.3% 


LACM 5303. referred 


ll J 


36.2 


61.3% 


Enaliarctos mitchelli 








USNM 175637. referred 


9.4 


34.4 


54.6% 


Enaliarctos emlongi 








USNM 250345, holotype 


10.3 


37.8 


54.5% 


Pinnarctidion bishopi 








UCMP 86334. holotype 


(9.8) 


35.1 


55.8% 


LACM 5302, referred 


(8) 


(30) 


53.3% 


Pacificotaria hadromma 








USNM 250320, referred 


(8.4) 


28.6 


58.7% 


Pacificotaria sp. 








USNM 250282, referred 


(13.2) 


46.0 


57.4% 


Desmatophoca oregonensis 








Univ. Oregon F735. holotype 


(13.7) 


57.5 


47.6% 


USNM 250283, referred 


(11.2) 


(46.8) 


47.8% 


Desmatophoca brachycephala 








LACM 120199. holotype 


(13) 


53.5 


48.6% 


undescnhed desmatophocine 








Emlong Coll. 211 


13.7 


47.0 


58.3% 


G. Pierson 84-02-184 


(12.3) 


46.8 


52.6% 


Neotherium minim 








LACM 131950. referred 


13.6 


47.6 


57.19J 


undescribed allodesmine 








LACM 133442 


12.8 


49.2 


52.0% 


undescribed enaliarctine 








LACM 128004 


5.65.9 


24.9 


45-47.4% 


LACM 134394 


10.1 


(39) 


51.8% 


LACM 123817 


(Kill) 


(34.9) 


57.3-63% 



"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. 





Internal carotid 


Basilar 






lensth within the 


length 






subdural venous 


of skull 




Taxon 


sinuses (1CL) 


(BLS) 


ICL/BLS 


Ursus americanus 


7.6 


19.15 


39.7% 


Helarctos malayanus" 


-8.6 


-21 


40.9% 


Melursus ursinus 1 ' 


-7.7 


-19-21 


36.7-40.5% 


Ailuropoda melanoleuca' 


6.8 


-23-24 


28.3-29.6% 


Ailuropoda melanoleuca'' 


8.5 


-24 


35.4% 



"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 ■mo nachines " " monachin es ' Phocina e M onachina e 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. 

Burns. J. J., and F. H. Fay. 1970. Comparative morphology of the skull 
of the ribbon seal, Histriophoca fasciata, with remarks on system- 
atica of Phocidae. Journal of Zoology 161:363-394. 

Fawcett. E. 1918. The primordial cranium of Poecilophoca weddelli 
(Weddell's seal), at the 27-mm. c.r. length. Journal of Anatomy 
52:412-441. 

Fay, F. H. 198 1 . Walrus: Odobenus rosmarus. Pp. 1-23 in S. H. Ridgway 
and R. J. Harrison (eds.). Handbook of Marine Mammals, vol. I. 
Academic Press, New York. New York. 

, V. R. Rausch. and E. T. Felt/.. 1967. Cytogenetic comparison of 

some pinnipeds (Mammalia: Eutheria). Canadian Journal of Zool- 
ogy 45:773-778. 

Fleischer. G. 1 973. Studien am Skelett des Gehororgans der Saugethiere. 
enschliesslich des Menschen. Saugetierkundliche Mitteilungen 
21(2-3):131-239. 

Flower. W. H. 1 88 1 . On the elephant seal, Macrorhinus leoninus (Linn.) 
Proceedings of the Zoological Society of London 10:145-162. 

Flynn, J. J., N. N. Neff, and R. H. Tedford. 1988. Phylogeny of the 
Carnivora. Pp. 73-116 in M. J. Benton (ed.). Phylogeny and Classi- 
fication of the Tetrapods, vol. 2. Systematics Association Special 
Volume 35B. Clarendon Press, Oxford, England. 

Gray, J. E. 1 874. Hand-list of Seals, Morses, Sea-lions, and Sea-bears in 
the British Museum. Dept. of Zoology. British Museum (Natural 
History), London. England. 

Jong. W. W. de 1982. Eye lens proteins and vertebrate phylogeny. Pp. 
75-114 in M. Goodman (ed.). Macromolecular Sequences in Sys- 
tematics and Evolutionary Biology. Plenum. New York. New York. 

Kellogg, R. 1922. Pinnipeds from Miocene and Pleistocene deposits of 
California: A description of a new genus and species of sea lion 
from the Temblor together with seal remains from the Santa 
Margarita and San Pedro Formations as a resume of current theo- 
ries regarding origin of Pinnipedia. University of California Publi- 
cations, Bulletin of the Department of Geology 13(4):23-I32. 

King. J. E. 1964. A note on the increasing specialization of the seal fore 
flipper. Proceedings of the Anatomical Society of Great Britain and 
Ireland 98:476-477. 

. 1965. Giant epiphyses in a Ross seal. Nature 205:515-516. 

. 1966. Relationships of the hooded and elephant seals (genera 

Cystophora and Miroungd). Journal of Zoology 148:385-398. 

. 1972. Observations on phocid skulls. Pp. 81-115 in R J. 

Harrison (ed.). Functional Anatomy of Marine Mammals. Aca- 
demic Press, London, England. 
— . 1983. Seals of the World. 2nd ed. Cornell Univ. Press, Ithaca, 



New York. 
(Cummer, B., and S. Neiss. 1957. Das Cranium eines 103mm langen 

Embryos des sudlichen See-Elephantaen (Miwunga leonina L.). 

Gegenbaurs Morphologishes Jahrbuch 98:288-346. 
Laws. R. M. 1959. Accelerated growth in seals, with special reference to 



the Phocidae. Norsk Hvalfangst-Tidende 9:425^152. 

Leboucq. H. 1904a. Ueber die endlapen der Pinnipedierfinger. 
Verhandlungen der Anatomischen Gesellschaft, pp. 120-124. 
— . 1904b. Organogenie des pmnipedes. I. Lesextremites. Pp. 3-20 
in Resultats du Voyage du S. Y. Belgica en 1897-1898-1899. 
Zoology, vol. 7. J.-E. Buschmann. Antwerp, Belgium. 

Legendre. S., and C. Roth 1988. Correlation of camassial tooth size and 
body weight in Recent carnivores (Mammalia). Historical Biology 
1:85-98. 

Maddison. W. P.. M. J. Donoghue. and D. R. Maddison. 1984. Outgroup 
analysis and parsimony. Systematic Zoology 33:83-103. 

McLaren. I. A. 1960. Are the Pinnipedia biphyletic? Systematic Zool- 
ogy 9:18-28. 

Mitchell. E. 1966. The Miocene pinniped Allodesmus. University of 
California Publications in Geological Sciences 61 : 1—46. 

. 1967. Controversy overdiphyly in pinnipeds. Systematic Zool- 
ogy 16:350-351. 

Muizon, C. de. 1982a. Phocid phylogeny and dispersal. Annals of the 
South African Museum 89(21:175-213. 

— . 1982b. Les relations phylogenetiquesdes Lutrinae(Mustehdae. 
Mammalia). Geobios, Memoire Special 6:259-277. 

Orlov, J. A. 1933. Semantor macrurus (ordo Pinnipedia, fam. 
Semantoridae fam. nova) aus den Neogen-Ablagerungen 
Westsibiriens. Academie des Sciences de l'Union des Republiques 
Socialistes Sovietiques, Travaux de l'lnstitut Palozoologique 
2:1165-1262. 

Queiroz, K. de 1988. Systematics and the Darwinian revolution. Phi- 
losophy of Science 55:238-259. 

Ray, C. E. 1976. Geography of phocid evolution. Systematic Zoology 
25:391-406. 

Repenning. C. A. 1976. Adaptive evolution of sea lions and walruses. 
Systematic Zoology 25:375-390. 

Repenning. C. A. and C. E. Ray. 1977. The origin of the Hawaiian monk 
seal. Proceedings of the Biological Society of Washington 89:667- 
688. 

Repenning, C. A., and R. H. Tedford. 1977. Otarioid seals of the Neo- 
gene. United States Geological Survey Professional Paper 992. 

Ridgway, S. H., and R. J. Harrison (eds.). 1981. Handbook of Marine 
Mammals. Volume 2: Seals. Academic Press, London, England. 

Sarich. V. M. 1969a. Pinniped origins and the rate of evolution of 

carnivore albumins. Systematic Zoology 28:286-295. 
. 1969b. Pinniped phylogeny. Systematic Zoology 28:416—422. 

Schaeffer, B„ M. K. Hecht, and N. Eldredge. 1972. Phylogeny and 
paleontology. Evolutionary Biology 6:31—16. 

Tedford. R. H. 1976. Relationship of pinnipeds to other carnivores 
(Mammalia). Systematic Zoology 25:363-374. 

Turner, W. 1888. Report on the seals collected during the voyage of 
H.M.S. Challenger in the years 1873-76. In Report on the Scien- 
tific Results of the Voyage of H.M.S. Challenger during the Years 
1 873-76. Vol. 26. pt. 2. Neill and Co.. London. England. 

Wiig, 0. 1983. On the relationship of pinnipeds to other carnivores. 
Zoologica Scripta 12:225-227. 

Williston, S. W. 1914. Water Reptiles of the Past and Present. Chicago 
University Press, Chicago. Illinois. 

Wyss, A. R. 1987. The walrus auditory region and the monophyly of 
pinnipeds. American Museum Novitates 2872. 

. 1988. On "retrogression" in the evolution of the Phocinae and 

phylogenetic affinities of the monk seals. American Museum 
Novitates 2924. 



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

Thomas A. Demere 

Department of Paleontology, San Diego Natural History Museum. P. O. Box 1390, San Diego. California 92112. and 
Department of Biology, University of California, Los Angeles, California 90024 

ABSTRACT. — Two new species of fossil walruses (Family Odobenidae) from the San Diego Formation (upper Pliocene; Blancan correlative) 
of San Diego County, California, are referred to the extant Odobeninae and the extinct Dusignathinae. The humerus of the new odobenine taxon 
shares several apomorphies with the type humerus of Valenictus from the late Miocene of southeastern California and is assigned to this formerly 
problematic genus. Valenictus chulavistensis, n. sp., is a tusked walrus closely related to modern Odobenus but more derived in possessing an 
entirely edentulous mandible and lacking all postcanine maxillary teeth. The toothlessness of V. chulavistensis is unique among known pinnipeds but 
parallels the condition seen in modern suction-feeding beaked whales (family Ziphiidae) and the narwhal (Monodon). The new dusignathine is 
assigned to the genus Dusignathus primarily because of synapomorphies of the lower jaw. Dusignathus seftoni, n. sp., possesses enlarged upper and 
lower canines and a shortened rostrum. The co-occurrence of these taxa in the San Diego Formation indicates that odobenid diversity in the eastern 
North Pacific continued to be greater than at present at least into late Pliocene time. 



INTRODUCTION 

The discovery and description of new species of fossil and 
living organisms is always an illuminating event, as it supplies 
new data points in the "history of life." Such discoveries are 
especially important to researchers attempting to reconstruct the 
phylogeny of groups as divergent as walruses, whose lack of 
modern diversity contrasts with their greater fossil diversity 
(Repenning and Tedford 1977). Not only do these discoveries fill 
out the taxonomic membership of known branches, they may 
supply the first evidence of previously unknown but related 
groups. Moreover, they provide insights into the morphological 
diversity within a clade and help define the taxonomic distribu- 
tion of specific character states. 

This report describes two new species of fossil walruses (family 
Odobenidae. sensu Repenning and Tedford 1977) from the marine 
upper Pliocene San Diego Formation of San Diego County, Califor- 
nia. The new taxa are assignable to two monophyletic (sensu 
Hennig 1966) lineages of odobenids, one to the extinct 
Dusignathinae (sensu Barnes and Raschke 1990). the other to the 
Odobeninae (sensu Repenning and Tedford 1977), the clade that 
includes the living arctic walrus, Odobenus rosmarus. 

This report is part of a more general study of the higher system- 
atic relationships of odobenids (Demere 1994, this volume) and 
builds upon the earlier work of Repenning and Tedford ( 1977). A 
rapidly improving fossil record for odobenids has contributed much 
to this study. 

GEOLOGY 

The majority of the new fossil material reported here was col- 
lected from marine sandstones of the San Diego Formation (Demere 
1983; Domning and Demere 1984) as exposed at various localities 
in the eastern portion of the city of Chula Vista, southwestern San 
Diego County, California. The San Diego Formation in this area 
consists of approximately 50 m of interbedded pebble conglomer- 
ates, fine-grained massive sandstones, tine-grained laminated sand- 
stones, and shelly sandstones. This sequence of sedimentary rocks 
was deposited in shoreface to middle-shelf environments (Demere 
1983). 

The San Diego Formation has produced abundant and well- 
preserved remains of marine invertebrates and vertebrates. The 
marine invertebrate assemblage includes foraminifers. brachiopods, 
molluscs, crustaceans, and echinoderms (Hertlein and Grant 1960, 
1972). The marine vertebrate assemblage includes sharks and rays, 
bony fishes, sea birds (Howard 1949, 1958: Chandler 1990), ceta- 
ceans (Barnes 1973; Demere 1986), pinnipeds (Berta and Demere 



1986), and sirenians (Domning and Demere 1984). Remains of 
terrestrial mammals have also been collected from this rock unit 
(Table 1 ). 

The co-occurrence of Stegomastodon sp., Titanotylopus sp., 
Equus sp., Platygonus sp., and Megalonyx sp. in the San Diego 
Formation indicates correlation with the Blancan North American 
Land Mammal Age (NALMA), late Pliocene. In addition, the asso- 
ciated marine invertebrate assemblage indicates correlation with 
the "San Joaquin" molluscan stage of Addicott ( 1972), provincial 
late Pliocene, estimated to be 2-3 million years old ( Demere 1 983 ). 

METHODS AND MATERIALS 

The fossil material described in this report is housed at the San 
Diego Natural History Museum, San Diego, California (SDSNH). 
Comparisons were made with specimens at other institutions in- 
cluding the Natural History Museum of Los Angeles County, Los 
Angeles, California (LACM) and the National Museum of Natural 
History. Smithsonian Institution, Washington, D.C. (USNM). Of 
special note is new undescribed material of Neotherium mirum 
examined at the LACM and undescribed material of Pontolis 
magnus examined at the USNM. Additional specimens cited in this 
report are housed at the Museum of Paleontology. University of 
California, Berkeley, California (UCMP); Department of Geologi- 
cal Sciences, University of California, Riverside, California (UCR); 
Museum of Comparative Zoology, Harvard University. Cambridge, 
Massachusetts (MCZ); and Institut Royal des Sciences Naturelles 
de Belgique, Brussels, Belgium (IRSNB). 

Cranial measurements follow Siversten (1954) and Barnes 
(1979), mandibular measurements follow Repenning and Tedford 
(1977), and postcranial measurements follow Kellogg (1931 ). 

SYSTEMATICS 

Class Mammalia Linnaeus, 1758 

Order Carnivora Bowdich. 1821 

Family Odobenidae Allen, 1880 

Subfamily Odobeninae Mitchell, 1968 

Valenictus Mitchell, 1961 

Type species. — Valenictus imperialensis Mitchell. 1961. 
Emended diagnosis. — An odobenine walrus distinguished from 
other taxa by the following apomorphies of the humerus: greatly 



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

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

Proc. San Diego Soc. Nat. Hist. 29:77-98. 1994 



78 



Thomas A. Demere 



enlarged entepicondyle, short and robust shaft, large lesser tuberos- 
ity, and narrow bicipital groove. 

Distribution. — Late Miocene to late Pliocene of southern and 
central California. 

Included species. — V. imperialensis Mitchell, 1961; V! chula- 
vistensis. n. sp. 

Valenictus chuluvistensis n. sp. 

Figures 1-7 

Diagnosis. — A species of Valenictus distinguished from V. 
imperialensis by the following features of the humerus: larger over- 
all size, more sigmoidal posterior profile, sharply keeled supinator 
ridge, more robust and rectangular entepicondyle. and more obtuse 
angle between the shaft and the axis of the distal trochlea. Also 
diagnosed by the following autapomorphies: edentulous dentary. 
edentulous premaxilla and postcanine maxilla, osteosclerotic long 
bones, astragalus with broad sulcus calcanei. very reduced collum 
tali, and coalesced navicular and sustentacular facet. Shares the 
following apomorphies with tusked odobenines: enlarged, ever- 
growing upper canines with three layers (globular orthodentine, 
orthodentine. and cementum), palate narrow and arched longitudi- 
nally as well as transversely, enlarged mastoid processes as widest 
part of skull, shortened temporal fossa with blunt zygomatic arches, 
and dorsoventrally expanded postorbital process of jugal. 

Type material. — Holotype: SDSNH 36786, a partial skeleton 
preserving the left side of the skull (maxilla, jugal, squamosal, 
mastoid, occipital condyle), nearly complete mandible, partial ver- 
tebral column (5 cervical, 12 thoracic, 2 lumbar, 4 sacral, and 5 
caudal vertebrae), partial right and left scapulae, left humerus, 
radius, ulna, and manus, partial right manus. left femur, right pes. 
and partial left pes. Collected by Richard A. Cerutti. 

Paratype: SDSNH 38227, a nearly complete skull with both 
canines but lacking the nasals, premaxillae, and middle ear ossicles. 
Collected by Richard A. Cerutti. 

Etymology. — The specific name is for the city of Chula Vista, 
San Diego County. California, where the remains of this, and many 
other Pliocene marine mammals, have been found. 

Holotype and paratype locality. — SDSNH locality 3551, 
Rancho Del Rey, city of Chula Vista, San Diego County, California. 

Horizon and age. — San Diego Formation, "lower member" 
(Demere 1983); late Pliocene (Blancan NALMA correlative). 

Referred material. — The following SDSNH specimens were all 
collected from the San Diego Formation (complete locality infor- 
mation is available to interested researchers upon request): 38228, 
partial rostrum preserving the left maxilla and premaxilla; 15162. 
partial left C; 38225, right C; 38226, partial right C; 35284, 
partial left C; 25180, partial right C; 30796, fragmentary right C; 
35276. partial atlas vertebra; 36796, fused sacral vertebrae (3); 
38676, fused sacral vertebrae (3); 38394, left scapula; 38308, par- 
tial left humerus; 38307, right humerus, distal end; 38286, left 
humerus, distal end; 35263. partial left humerus; 38315, left 
humerus; 35275. right humerus; 38312, left humerus; 38300, par- 
tial left humerus; 38230, right ulna; 32790, right radius; 38288, left 
radius; 38324, right radius; 38650, right radius; 36799, left 
scapholunar; 36800, left unciform; 38201, left magnum; 38208. left 
metacarpal I; 42694, left metacarpal 1; 38206, left metacarpal IV; 
38339, associated right and left innominates and partial baculum; 
38310, left innominate; 25145, partial left innominate; 38291, par- 
tial right innominate; 38325, partial left innominate; 32770, frag- 
mentary right innominate; 36798, right innominate; 42751, partial 
left hindlimb with femur, tibia, fibula, calcaneum. navicular, 
mesocuneiform, and metatarsals I, II, and III; 25394. left femur; 
25074. left femur; 25076, left femur, distal end; 42690, right femur; 
38245, right femur, proximal end; 32767. left femur, distal end; 



32777, left fibula, distal end; 42654, right fibula; 42655, right 
fibula; 22290, right tibia; 25087, left tibia, distal end: 35296. right 
tibia, distal end; 33935. left tibia, distal end; 38633. partial left tibia; 
35273. left astragalus; 21 130. right calcaneum; 22412, right calca- 
neum; 32765, left calcaneum; 25179. associated ribs and hindlimb 
bones with right patella and navicular, and left entocuneiform and 
metatarsal II; 38209. left metatarsal I; 38261, left metatarsal I. 

Cranium. — The holotype partial skull (SDSNH 36786) is from 
a mature adult male, as indicated by closure of all preserved sutures 
and the narrowing of the proximal end of the upper canine (Rutten 
1907). 

In contrast, the paratype skull (SDSNH 38227, Figs. 1A, B; 2A, 
B) is from an immature male (F. H. Fay, pers. comm.). The tusks in 
this skull taper continuously from the root to the distal end. and 
many sutures are distinct. This skull was shortened anteroposteri- 
orly by lithostatic load, with the palate displaced against the audi- 
tory bullae and beneath the basioccipital. Portions of the zygomatic 
arches and braincase were partially etched by root action and soil 
acidity. The premaxillae were largely destroyed by a bulldozer at 
the time of their discovery. 

A referred left rostral fragment (SDSNH 38228. Figs. 2C, D) is 
from an immature female (F H. Fay. pers. comm.) and of the three 
cranial specimens preserves the smallest canine alveolus (Table 2). 
The maxillary portion of SDSNH 38228 is nearly complete (sur- 
faces are preserved for the maxilla/jugal. maxilla/frontal, maxilla/ 
palatine, and maxilla/nasal sutures) and occurred with the only 
premaxilla known for this taxon. 

The cranium of V. chulavistensis preserves many features char- 
acteristic of the tusked odobenines, including a narrow longitudi- 
nally and transversely arched palate (as in Odobenus rosmarus and 
Alachtherium cretsii), mastoid processes as the widest part of the 
skull, diagonally oriented orbitosphenoid with small, funnel-shaped 
optic foramen, lack of a sagittal crest, telescoping of palate beneath 
anterior portion of basicranium. and posterior elongation of the 
hard palate to the level of the glenoid fossae. In addition, the skull 
of this new species also preserves more generalized odobenid fea- 
tures including lack of supraorbital processes and large antorbital 
processes constructed from both maxilla and frontal. 

In lateral aspect the premaxilla of SDSNH 38228 is shaped 
roughly like an acute right triangle, the hypotenuse being the 
external narial border. Anteriorly the premaxillae terminate in a 
conspicuous nasal process, as in Zalophus californianus (see 
Howell 1929) and most fossil odobenids (e.g., Neotherium mirum 
and Imagotaria downsi). The nasal process is elevated above the 
canine alveolus as in Odobenus and Alachtherium; however, the 
vertical dimension between the nasal process and the incisive 
border of the premaxilla is short. In Odobenus the nasal process is 
very reduced (in adults) and the narial opening is elevated well 
above the incisive margin. The incisive region of the premaxilla 
in the new taxon is edentulous, lacking all traces of alveoli (Figs. 
IB, 2C). The incisive foramina are distinct and oriented nearly 
horizontally as they extend posterodorsally into the narial open- 
ing. The external narial opening in SDSNH 38228 would have 
been a transversely compressed oval, measuring approximately 
38 mm high by 29 mm wide. In lateral aspect the narial opening 
makes an angle of approximately 40° with the horizontal axis of 
the skull. In Odobenus this angle is approximately 70° and in 
Alachtherium approximately 55°. 

Unlike the derived condition in Odobenus, the ascending pro- 
cesses of the premaxillae of Valenictus chulavistensis overlapped 
the nasal bones externally for approximately one-half the length of 
the nasals. This assessment is based on the configuration of the 
sutures, as the nasals themselves are not preserved. 

The maxillae of all three rostral specimens are conspicuously 
swollen in the region of the canine root (Fig. 3). However, this 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



79 



Table 1. Composite faunal list of mammals from the San Diego 
Formation. 

Rodentia 

Heteromyidae 

Heteromyidae sp. 
Cricetidae 
Neotoma sp. 
Lagomorpha 
Leporidae 
Leporidae sp. 
Artiodactyla 

Tayassuidae 

Platygonus sp. 
Camelidae 

Titanotylopus sp. 
cf. Hemiauchenia sp. 
Cervidae 
Cervidae sp. 
Perissodactyla 
Equidae 

Equus sp. 
Tapiridae 
Tapirus sp. 
Carnivora 
Felidae 

Felis sp. cf. F. rexroadensis 
Mustelidae 

Spilogale sp. 
Canidae 

Caninae sp. 
Otariidae 

Callorhinus gilmorei Berta and Demere 
Otariidae sp. 
Odobenidae 

Valenictus chulavistensis n. sp. 
Dusignathus sefioni n. sp. 
Cetacea 
Mysticeti 

"Cetotheriidae" 
Herpetocetus sp. 1 
Herpetocetus sp. 2 
Balaenopteridae 

Balaenopiera davidsonii Cope 
Balaenopteridae sp. 1 
Balaenopteridae sp. 2 
Balaenopteridae sp. 3 
Balaenopteridae sp. 4 
Balaenidae 

Balaenidae sp. 1 
Balaenidae sp. 2 
Odontoceti 
Pontoporiidae 

Parapontoporia Sternberg) Gregory and Berry 
Phocoenidae 

Phocoenidae sp. 1 
Phocoenidae sp. 2 
Monodontidae 

Delphinapterinae sp. 
Delphinidae 

Delphinidae sp. 1 
Delphinidae sp. 2 
Sirenia 

Dugongidae 

Hydrodamalis cuestae Domning 
Proboscidea 

Gomphotheriidae 
Stegomastodon sp. 
Edentata 

Megalonychidae 
Megakmyx sp. 



swelling is not as great as in Odobenus and is expressed more 
anteroposteriorly than transversely. In Odobenus, the maxillae are 
so swollen that the infraorbital foramina are almost completely 
hidden when the skull is viewed in anterior aspect. The inclination 
of the canine root in relation to a vertical transverse plane (Fay 
1982:111) is more procumbent (Fig. 2B) than that of Odobenus 
(36°-56° in V. chulavistensis, compared with 9°-20° in Odobenus, 
and 25°-48° in Alachtherium). As in all pinnipeds, the maxillae 
form the anterior walls of the orbits (Wyss 1987). The infraorbital 
foramen is large ( 19 mm wide by 33 mm high in SDSNH 38227: 12 
by 28 mm in SDSNH 36786: 19 by 20 mm in SDSNH 38228), with 
a delicate dorsal strut and more robust ventral strut. The ventral 
surface of the latter is marked by a conspicuous fossa, which opens 
postero ventral ly to accommodate the sharply keeled dorsal margin 
of the dentary. The ventral margin of the maxilla, posterior to the 
large canine alveolus, is keeled between the lateral and palatal 
surfaces, and is continuous with the lateral keeled margin of the 
ventral strut of the infraorbital foramen (Fig. 3). In lateral aspect, 
the ventral margin of the maxilla in Odobenus is continuous 
lingually with an alveolar shelf, and is continuous labially with the 
ventral strut of the infraorbital foramen as in V. chulavistensis. This 
is unlike the condition in Neotherium, Imagotaria, Pontolis, 
Dusignathus, Aivukus. and Gomphotaria, in which the ventral strut 
of the infraorbital foramen is conspicuously elevated above the 
ventral (alveolar) margin of the maxilla. 

The jugal is relatively longer and more transversely compressed 
in V. chulavistensis than in Odobenus. However, as in Odobenus, 
the jugal in the new species contacts the maxilla in a transversely 
compressed peg-and-socket joint. The postorbital process of the 
jugal of V. chulavistensis is large and dorsoventrally expanded as in 
Odobenus. The orbit (i.e., diameter between the maxillary border of 
the orbit and the postorbital process of the jugal) is small as in 
Odobenus, relatively smaller than in Aivukus, Pontolis, Imagotaria, 
and Neotherium. 

The zygomatic portion of the squamosal is robust and shortened 
as in Odobenus and Alachtherium, not slender and elongated as in 
all other fossil odobenids (including Aivukus). The squamosal fossa 
at the root of the zygoma is short and narrow and continuous 
posteriorly with a narrow shelf above the external auditory meatus. 
The external auditory meatus is open broadly externally and not 
restricted by the closeness of the mastoid and postglenoid pro- 
cesses, as it is in adult crania of Odobenus. 

In dorsal aspect, the temporal fossae are oval openings antero- 
posteriorly shortened relative to those of Neotherium, Imagotaria, 
Gomphotaria, and Aivukus. They are not as shortened, however, as 
those of Odobenus. 

The squamosal/parietal suture is horizontal and positioned near 
the base of the broadly convex braincase. The cranial vertex is 
broadly rounded transversely (as in Odobenus) and marked by a 
weakly raised interparietal suture. There is no indication of the 
sagittal sulcus described by Rutten (1907) for the holotype cranial 
fragment of Alachtherium antverpiensis (= A. cretsii), nor of the 
parasagittal cristae seen on adult crania of Odobenus. 

The parietal/frontal suture is partially preserved on the right 
side of SDSNH 38227 at the level of the intertemporal constriction. 
The suture indicates that the frontals extended posteriorly between 
the parietals at the midline, as in Odobenus. Anterior to the 
intertemporal constriction the frontals widen dramatically, termi- 
nating anterolaterally in large antorbital processes. The frontal/ 
maxilla suture is obscure but seems to have been transversely 
oriented. It is clear, however, that the suture bifurcates the antorbital 
processes, so both maxilla and frontal form the processes. 

Near the midline, the lambdoidal crest is developed medially as 
a distinctive, anterodorsally inclined, transverse crescentic shelf 
(convex border anteriorly placed) that joins with its more lateral 



80 



Thomas A. Demere 




Figure 1. Valenictus chulavistensis, new species. A, B. SDSNH 38227. paratype skull. A, dorsal view (stereophotographs); B, ventral view 
(stereophotographs). C. SDSNH 38225, referred right C, medial view. Scale bar, 5 cm. 



portions as they descend toward the mastoid processes. This inclined 
shelflike crest (the site of insertion of neck extensor muscles) is also 
seen in Odobenus, the holotype of Alachtherium antverpiensis 
(Rutten. 1907), the holotype of Alachtherium antwerpiensis Hasse, 
1910. and the referred skull of Alachtherium antverpiensis (Rutten 
1907) described by Erdbrink and van Bree (1990). The lambdoidal 
crest does not project posterodorsally to overhang the occipital 
shield as in Neotherium, Imagotaria, Pontolis, and Gomphotaria. 
The occipital shield is vertically oriented with a distinct sagittal 
crista as in Odobenus and Alachtherium. In posterior aspect (Fig. 
2A) the occiput is hemispherical as in Odobenus and differs from the 



rectangular shield characteristic of Alachtherium (see Hasse 1910). 
The occipital condyles are widely separated dorsally, do not reach 
the roof of the foramen magnum, and are not exceptionally large. 

The basioccipital is broad and roughly pentagonal (Figs. IB: 3). 
The posterior portion of the basioccipital bears a strong sagittal 
ridge, bounded anterolaterally on either side by rugose circular 
areas for insertion of the rectus capitis ventralis muscles. A portion 
of the basioccipital/basisphenoid suture is preserved at the antero- 
lateral corners of the basioccipital. 

The auditory bulla completely fills the portion of the basi- 
cranium between the basioccipital. mastoid, and postglenoid pro- 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



81 




' 






\ 



D 



% 



9 



Figure 2. Valenictus chulavistensis, new species. A, B. SDSNH 38227. paratype skull. A, posterior view; B, lateral view. C. D. SDSNH 38228, referred 
left maxilla and premaxilla. C. ventral view (stereophotographs); D. lateral view. Scale bar, 5 cm. 



cess. The bulla is not inflated except near the posterior opening of 
the carotid canal. This opening is rather steeply inclined 
anterodorsally as in adult crania of Odobenus and is closely ap- 
pressed to a small posterior lacerate foramen. There is no clear 
distinction between the entotympanic and ectotympanic. Anteri- 
orly, the bulla is closely appressed to, and overrides, the posterior 
portion of the postglenoid process. Several distinct and irregular 
bullar processes lie anterior to the hyoid fossa and stylomastoid 
foramen. The latter foramen is a large cylindrical opening well 
separated from the large, slitlike hyoid fossa by a smooth continu- 
ous surface extending from the bulla laterally out onto the ventro- 
medial surface of the mastoid process. This condition is like that in 
Odobenus and Alachtherium and unlike that in Imagotaria and 
Pontolis, in which the two openings are connected by a continuous 



groove, or Neotherium, in which the two openings lie very close to 
one another. 

The foramen ovale and alisphenoid foramen are closely ap- 
pressed and recessed in a common fossa, which is directed ventrally 
as in Odobenus. The anterior opening of the carotid canal and bony 
eustachian tube are obscured by the diagenetically telescoped 
basicranium. 

The palate is narrow and arched both longitudinally and trans- 
versely (Figs. IB, 2C) as in Odobenus. There is no alveolar shelf 
along the lingual border of the upper canines, as the entire 
postcanine portion of the maxilla is edentulous, without any trace of 
alveoli. The maxilla/palatine suture on the palate meets the margin 
of the temporal fossa at the apex of a small but distinct pterygoid 
process. Together the horizontal laminae of the palatine bones form 



82 



Thomas A. Demere 



Table 2. Skull measurements of Valenictus chulavistensis and Dusignathus seftoni 

(in mm). 



Total length (condylobasal length) (0)'" 

Rostral tip to middle of occipital crest 

Length of tooth row, P'-M - 

Width of rostrum across canines (12) 

Width of rostrum across base of I- 

Width of palate at P 4 

Depth of greatest palatal arch 

Width across antorbital processes (5) 

Width between infraorbital foramina 

Width across intertemporal constriction 

Width of braincase (8) 

Zygomatic width (17) 

Auditory width ( 19) 

Mastoid width (20) 

Paroccipital width 

Greatest width across occipital condyles 

Greatest width of anterior nares (3) 

Greatest height of anterior nares 

Greatest width of nasals 

Greatest length of nasals (4) 

Width of zygomatic root of maxilla (14) 

Greatest width of foramen magnum 

Transverse diameter of infraorbital 

foramen 









Dusignathus 


Valenictus chulavistensis 


seftoni 


SDSNH 


SDSNH 


SDSNH 


SDSNH 


36786" 


38227* 


38228 


38342" 


410'' 


222'' 




320 
83 


152'' 


146 


112'' 


140'' 
56 
94'' 




56 


45 






147 J 


116'' 


108'' 




127 


96'' 


118'' 




82 




51 
106 


232 c 


206 

171 




222'' 


306'' 


247 
154 






110' 


95 








46'' 


30'' 


50 






36'' 


52 






26'' 


40 

73 


38 


29 


21 


21 


42'' 


48 







26 



29 



23 



"Holotype. 

''Paratype. 

'Numbers in parentheses refer to measurements in Siversten ( 1954). 

''Estimate based on bilaterally symmetrical feature. 

''Estimate on broken features. 



a broad trapezoid, the longest side represented by the internal narial 
opening and the palatine/pterygoid sutures. The hamular processes 
of the pterygoids are constructed as in Odobenus (i.e., dorsoven- 
trally compressed and transversely expanded flanges that hook 
laterally at their flattened distal extremities) and are unlike the 
delicate transversely compressed processes of Imagotaria. The me- 
dial wall of the pterygoid is preserved within the narial passage, 
with a clearly defined pterygoid/basisphenoid suture. The hard 
palate is elongated, extending to the postglenoid fossae as in 
Odobenus and Alachtherium, and lacks any hint of the horizontal 
pterygoid strut (Barnes 1989) that characterizes the lateral borders 
of the internal narial opening in almost all other pinnipeds. In this 
configuration the site for origination of the internal pterygoid 
muscle is moved to the temporal fossa. 

The mastoid processes are greatly enlarged, constructed inter- 
nally from cancellous bone, and extend ventrally to a level below 
the hamular processes and auditory bullae, as in Alachtherium and 
Odobenus. In lateral aspect, the mastoid process presents a broad 
"teardrop" form that is more convex posterolaterally than in 
Odobenus. The ventral portion of the mastoid is slender and antero- 
posteriorly compressed to produce a transversely elongate process. 
In Odobenus, the ventral portion of the mastoid is more of a swollen 
knob, with a conspicuously roughened area for origination of the 
digastricus muscle. In V. chulavistensis the mastoid and paramastoid 
processes are closely appressed, with the latter forming the thin, 
delicate process characteristic of odobenids. As in Odobenus, the 
mastoid/paramastoid suture is not fused. 



The right orbital wall of SDSNH 38227, although damaged, 
provides details on the structure of this region. The optic foramen is 
funnel-shaped and positioned dorsally within a diagonally oriented 
orbitosphenoid that lacks a conspicuous horizontal plate of bone 
anterior to the foramen. This unusual configuration is also seen in 
Odobenus. In other pinnipeds, the optic foramen is typically a 
vertical slot positioned ventrally within a horizontally oriented 
orbitosphenoid that has a relatively long plate of bone anterior to 
the foramen. As in Odobenus. the orbitosphenoid in V' chula- 
vistensis appears to be bounded anteriorly by a relatively large and 
posteriorly placed orbital vacuity unlike the more anteriorly placed 
vacuities seen in extant otariids and at least one fossil odobenid 
(Imagotaria sp.. USNM 335599). In Odobenus the palatine bone 
forms the entire ventral border of the orbital vacuity, including both 
the anteroventral and posteroventral portions (i.e.. the maxilla/ 
palatine suture meets the maxilla/frontal suture anterior to the vacu- 
ity). In otariids (e.g., Zalophus. Eumetopias, and Otaria) and at 
least one fossil odobenid (Imagotaria sp., USNM 335599) the 
anteroventral border of the vacuity is formed from the maxilla (i.e., 
the maxilla/palatine suture does not reach to the maxilla/frontal 
suture but instead contacts the vacuity directly). 

A portion of the left petrosal was recovered with SDSNH 38227 
and is odobenidlike in its relatively large size, enlarged apex ante- 
rior to the promontorium, and broad internal auditory meatus. This 
meatus has passages for the facial and vestibulocochlear nerves 
separated by a low transverse crest. The roof of the meatus is not 
preserved, and much of the promontorium and all of the cochlea 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



83 




Figure 3. Reconstruction of Valenictus chulavistensis, new species, 
ventral aspect. Based largely on SDSNH 38227. ac, alisphenoid canal; Bo. 
basioccipital; Bs. basisphenoid; cc, carotid canal; fh, hypoglossal foramen; 
fi, incisive foramen; fio, infraorbital foramen; tip. posterior lacerate fora- 
men; fsm, stylomastoid foramen; hf, hyoid fossa; J, jugal; m, mastoid; Mx, 
maxilla; Pa. palatine; Pth. hamular process of pterygoid; Pm. premaxilla; tb. 
tympanic bulla. Scale bar, 5 cm. 



and floccular fossa are missing. Thus the fenestra ovale can not be 
measured. The petrosal apex is anteriorly elongated as in Aivukus 
and not shortened as in Odobemts. 

Dentition. — The dentition of Valenictus chulavistensis consists 
solely of the upper canines, which are elongated, curved, ever- 
growing tusks constructed as in Odobemts rosmarus and O. huxleyi 
(see Ray 1960). Internally, the tusks of Odobemts and V. 
chulavistensis consist of three layers, a central column of globular 
orthodentine, a surrounding ring of dense compact orthodentine, 
and a thin outer layer of cementum (Ray 1960). The thicknesses of 
these layers as measured on a referred partial tusk (SDSNH 38226) 
are 7.6, 8.9, and 1 .5 mm, respectively, for the globular orthodentine 
(radius), compact orthodentine, and cementum (see Table 3 for 
additional measurements). This specimen has a bluntly rounded 
crown and well-worn anterior border. The cementum layer is well 
preserved proximally and thins toward the anterodistal edge, prob- 
ably because of wear. The medial surface has two broad longitudi- 
nal grooves, the more anterior being more prominent and extending 
nearly to the distal tip of the tooth. The lateral surface is faintly 
fluted, with one particularly strong longitudinal groove near the 
posterior margin. The cementum layer is completely worn away on 
the tip of the crown and in a longitudinal band running proximally 
along the anterior margin from the tip back toward the proximal 
end, where the wear band rolls medially. In a few places where the 
cementum layer has been broken away, a pattern of very fine 



transverse lines (growth lines) is preserved in the compact 
orthodentine layer, as noted by Ray ( 1960) for O. huxleyi. 

SDSNH 38225 is a beautifully preserved complete right canine 
(Fig. 1C). In contrast to the broadly rounded anterior margin of 
SDSNH 38226. this specimen has a sharply beveled margin (aver- 
aging about 10 mm wide) extending from the distal end 235 mm up 
the anterior circumference of the tooth. Another beveled surface 
occurs along the distolateral surface from the tip proximally for a 
distance of 162 mm. Preserved along the medial border of the 
anterior half of the tooth, a conspicuous wear surface is bounded by 
cementum on its margins but lacks cementum itself. This large wear 
surface is irregularly concave proximally and distally. An intact 
cementum layer is preserved along the entire posterior margin of 
the tusk. The intralveolar portion of the tusk is characterized by 
numerous closely spaced fine longitudinal grooves that terminate 
abruptly distal to where they encounter the extralveolar surface. 
This specimen does not preserve any of the broad longitudinal 
grooves on its medial surface as seen in SDSNH 38226. A faint 
groove, however, is preserved on the lateral surface within 20 mm 
of the anterior margin. A second, more conspicuous longitudinal 
groove occurs just lateral to the posterior margin but does not 
continue onto the extralveolar surface. On this tusk, cementum is 
preserved only on the proximolateral surface, the posterior margin, 
and the distomedial surface. These wear patterns (i.e., abraded and 
worn anterior surfaces at the distal ends of the crowns) are similar 
to those reported by Fay (1982) for tusks of O. rosmarus. The wear 
on the distal end of SDSNH 38225 is so extensive that the dense 
orthodentine layer has been abraded away on the anterior surface to 
reveal the central globular orthodentine core (a condition also seen 
in tusks of O. rosmarus). The medial surfaces of the tusks of V. 
chulavistensis have broad longitudinal grooves as in Odobemts and 
in contrast to the more numerous and distinct longitudinal grooves 
preserved on the fluted tusks of Gomphotaria pugnax (see Barnes 
andRaschke 1991). 

The tusks in the paratype skull (Fig. 2B) as well as in two nearly 
complete referred tusks (SDSNH 38225, 38284) are arched in the 
parasagittal plane (Table 3), as in Odobemts. Radii of this arc as 
measured along the posterior surface of the two referred tusks are 
403 and 398 mm, respectively; in a fossil tusk of O. rosmarus, 270 
mm (Rutten 1907). Fay ( 1982:1 1 1 ) noted that in living walruses the 
radius of the longitudinal arc is variable, with ranges of 456 to 
>5000 mm for males and 226 to 1425 mm for females. The fossil 
tusks from Chula Vista fall within these ranges and point out the 
taxonomic weakness of this feature as discussed by Erdbrink and 
van Bree(1990). 

Mandible. — A nearly complete mandible (Fig. 4) was collected 
with the holotype skeleton (SDSNH 36786) and consists of a left 
mandibular ramus (lacking only the coronoid process) strongly 
fused at the symphysis to a partial right horizontal ramus (Table 4). 
In lateral aspect, the mandible presents a slender profile unlike that 
of any known pinniped. A slender and strongly upturned symphy- 
seal region forms an angle between the anterior margin and the 
ventral margin of the horizontal ramus of about 125°. The posterior 
margin of the upturned symphysis forms an angle of about 130° 
with the ventral margin. In Alachlherium (IRSNB M. 1 70) the sym- 
physeal portion is also upturned but more massive, with the anterior 
margin of the symphysis forming an angle of only about 1 12° with 
the ventral margin and about 143° with the posterior (alveolar) 
margin. 

The ascending, symphyseal portion of the mandible (right and 
left rami) of V. chulavistensis is slender and triangular in cross 
section (not swollen and massive as in Odobemts). with the apex of 
the triangle corresponding to the anteroventral margin of the sym- 
physis. In Alachtherium the ascending symphyseal portion of the 
mandible is also somewhat triangular. The "incisive" border of the 



S.4 



Thomas A. Demere 










' 







^ 



y 




Figure 4. Valenictus chulavistensis, new species, SDSNH 36786, holotype mandible. A, dorsal view (stereophotographs); B, lateral view. Scale bar. 



5 cm. 



mandible is characterized by a highly vascularized, laterally ex- 
panded bony pad (as in Prorosmarus and Odobenus). This bony pad 
is continuous posteriorly with a broad trough that runs postero- 
ventrally to the point of divergence between the right and left rami. 
The ridges that form the dorsolateral borders of this symphyseal 
trough run down and out onto the horizontal rami to become the 
sharply keeled dorsal margins of the rami. Both rami are entirely 
edentulous, with no trace of alveoli or alveolar shelves. The horizon- 
tal rami are transversely compressed and dorsoventrally shallow 
(Table 4), in contrast to the deep rami of Alachtherium. 

On the lateral surface of the horizontal ramus, at the point of 
divergence of the rami, are a pair of opposing mental foramina, one 
opening posteriorly, the other anteriorly. Both are set in a deeply 
excavated and elongated oval fossa. Alachtherium possesses a simi- 
lar pair of mental foramina within an open oval fossa. In 
Prorosmarus and Odobenus a single nearly circular mental foramen 
penetrates deeply into the ramus. At the back of this foramen are a 
pair of opposing smaller foramina. A pair (right and left) of large 
longitudinal nutrient foramina lie at the extreme anterior tip of the 
mandible, just below the bony pad. Similar foramina are seen on the 
mandibles of Prorosmarus, Alachtherium, and Odobenus. 



In lateral aspect, the ventral margin of the ramus between the 
rugose inferior genial tuberosity and the marginal process is concave. 
The marginal process (Davis 1964) is well developed and divisible 
into dorsal and ventral components. The ventral portion of the process 
is a keeled ridge, set off from the ventral margin of the ramus as a 
posteriorly directed pointed process. In ventral aspect, the axis of this 
process diverges medially from that of the ramus. The dorsal portion 
of the marginal process lies immediately above the posterior end of the 
ventral portion and is a conspicuous, anteroposteriorly elongated, 
knoblike eminence. The two portions of the marginal process are 
separated by an anteroposteriorly oriented sulcus. Dcntarics of mature 
individuals of Otaria hyronia display a similar divided marginal 
process (personal observation), while dentaries of immature individu- 
als of Otaria display intermediate conditions, from a single conspicu- 
ous flangelike marginal process to one that shows incipient division. 
Prorosmarus, Alachtherium, and Odobenus also possess well-devel- 
oped marginal processes, although without any obvious division into 
dorsal and ventral components. The structure of the marginal process 
in V. chulavistensis suggests a large digastricus muscle with an anteri- 
orly placed insertion on the ramus. The horizontal ramus is widest at 
the level of the marginal process. 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



85 



Table 3. Measurements of upper canines (tusks) of Valenictus chulavistensis (in mm). 



SDSNH 

36786" 



SDSNH 

38225 



SDSNH 
38226 



SDSNH 
38227' 



Radius of curvature, anterior surface 
Radius of curvature, posterior surface 
Total length, arc of anterior surface 
Total length arc of posterior surface 
Length, tangent of arc of posterior surface 
Length, base of root to intra-alveolar margin 
Anteroposterior diameter, base of root 
Transverse diameter, base of root 
Anteroposterior diameter, intra-alveolar margin 
Transverse diameter, intra-alveolar margin 
Anteroposterior diameter, mid-crown 
Transverse diameter, mid-crown 



265 


355 


351 




312 


403 


520 






485 




340' 




419 




320 c 




405 




311' 


136 


145 




125'' 


74 


71 




64 


36 


44 




34 


81 


69 




53-54 


49 


47 




36-39 


68 


64 




49-49 


45 


42 




31-33 



"Holotype. 

''Paratype (two tusks). 
'Bstimate on broken feature. 



The pterygoid process is large and robust, ventromedially di- 
rected, hooklike, and well separated from the marginal process. In 
Alachtherium and Odobenus the two processes are positioned close 
together, with the pterygoid process as a low, anteroposteriorly 
elongated knoblike projection closely appressed to the ramus and 
not medially extended. 

The mandibular condyle is a robust and transversely elongated 
cylinder, similar in size and thickness to that of an Odobenus of 
comparable size. The coronoid process is not preserved, but its 
broken base indicates it was slender and anteroposteriorly elon- 
gated. In this respect the coronoid process was probably similar in 
form to that of Alachtherium cretsii and quite different from the 
short, stout, and broad-based coronoid process of Odobenus. In 
dorsal (occlusal) aspect the mandible has a "'wish-bone" or furcula 
shape, with the left mandibular ramus preserving a distinctive 
sigmoidal outline, laterally concave between the tip of the jaw and 
the posterior border of the symphysis and laterally convex from 
there to the posterior border of the condyle. This sigmoidal outline 
is also characteristic of mandibles of Alachtherium, Prorosmarus, 
and Odobenus (see Berry and Gregory 1906) and serves to accom- 
modate the greatly enlarged upper canines (tusks). 

Postcrania. — The holotype includes all major portions of the 



postcranial skeleton except the tibia and innominate. Fortunately, 
these elements are represented in additional, referred material. It is 
beyond the scope of this report to describe each of these skeletal 
elements. The unique morphology of the humerus, calcaneum, and 
astragalus, however, calls for discussion of these elements. 

Humerus. — The current sample includes five complete and five 
partial humeri (Table 5). The following description focuses prima- 
rily on the holotype (Fig. 5). 

The humerus of Valenictus is striking in its overall stockiness 
relative to the more slender and elongated humeri of Odobenus and 
Alachtherium. Stockiness, expressed as the ratio of proximal width 
(measured at the widest part of the lesser tuberosity) to total hu- 
meral length, is significantly greater (p > 0.05) in Valenictus than in 
other odobenids. 

In V chulavistensis the humerus is constructed from very dense 
osteosclerotic bone, as in sirenians. In other marine mammals, it 
consists of spongy, cancellous bone. This greater bone density also 
characterizes all other limb bones of the holotype, including car- 
pals, tarsals, and metapodials. Interestingly, in contrast to sirenians. 
osteosclerotic bone does not occur in the axial skeleton (i.e., verte- 
brae and ribs) of V. chulavistensis. The nature of the internal struc- 
ture of the holotype humerus of V. imperialensis is unknown. 



Table 4. Mandibular measurements of fossil odobenids (in mm) 



Greatest length 

Length of tooth row, P,_, 

Depth of horizontal ramus at P, 

Width of horizontal ramus at P, 

Depth of horizontal ramus at P 4 

Width of horizontal ramus at P, 

Width of horizontal ramus at shallowest point 

along ramus 
Minimum depth of horizontal ramus 
Height, pterygoid process to coronoid process 
Length of symphysis 
Minimum width of symphysis 
Greatest width of condyle 



Valenictus chulavistensis 


Alachtherium cretsii 


Dusignalhus seftoni 


Dusignathus santacruzensis 


SDSNH 


IRSNB 


SDSNH 


UCMP 


36786 


M.168 


20801 


27131 


310 


364 


344 


215" 




98 


107 


72 




117 


76 


64 




30 


39 


17 




93 


86 


55" 




35 


39 


17 


20 


31 


35 


17 


50 


76 


87 


53 




157 


145 


83 


135 


184 


135 


70 


45 


58 


61 


29 


58 


75 


75 





"Estimate on broken feature. 



86 



Thomas A. Demere 







m 




r 



B ' 




Figure 5. Valenictus chulavistensis, new species, SDSNH 36786. holotype left humerus 
(stereophotographs); C, posterior view. Scale bar, 5 cm. 



A, anterior view (stereophotographs); B, lateral view 



The proximal end of the humerus of V. chulavistensis is charac- 
terized by a relatively large and well-rounded capitulum (head) 
positioned only slightly below a thickened greater tuberosity. In 
V. imperialensis, O. rosmarus, and a humerus (USNM 187328) 
referred by Repenning and Tedford (1977: pi. 17) to the problem- 
atic odobenid Pliopedia pacifica, the greater tuberosity is also low 
relative to the head. In V. imperialensis the head is relatively larger 
than in the new species. The lesser tuberosity of both taxa is 
distinctly thickened and, with the greater tuberosity, encloses a 
narrow and proximally inset bicipital groove. The lesser tuberosity 
is positioned only slightly below the proximal capitulum. In 
Odobenus the lesser tuberosity is relatively smaller and placed 
more distally and the bicipital groove is broader and less inset. In 
the transverse plane, the greater tuberosity of both species of 
Valenictus is distinctly elongated, deflected medially, and of nearly 



constant width to its anterior extremity. This medial deflection of 
the lesser tuberosity provides the proximal end of the humerus with 
a very broad profile in anterior aspect. This feature also character- 
izes the humeri of V. imperialensis and Pliopedia pacifica (USNM 
187328). 

The pectoral crest of the humerus of V. chulavistensis is elon- 
gate, like that of Odobenus, and extends as a broad ridge distally 
almost to the trochlea. By contrast, in Imagotaria, Gomphotaria, 
and Aivukus, the pectoral crest is strongly developed as a keeled 
ridge. In Valenictus, Odobenus, Alaclitherittm, and Pliopedia the 
pectoral crest gradually joins with the anterodistal surface of the 
humerus. In Aivukus and Gomphotaria, the pectoral crest displays 
an abrupt distal deflection and descends sharply to the anterodistal 
surface of the humerus. The deltoid tuberosity in V. chulavistensis is 
separate from the pectoral crest and positioned posterolateral to the 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



87 



Table 5. Measurements of humeri of fossil odobenids (in mm). 



Valenictus chulavistensis 



Valenictus Dusignathus 
imperialensis seftoni 













LACM 




SDSNH 


SDSNH 


SDSNH 


SDSNH 


SDSNH 


(C1T) 


SDSNH 


36786" 


38312 


38315 


35275 


38300 


3926" 


43873 


326 


315 


306 


263 


300 


253 


346 


325 


306 


310 


270 


296 


252 


321 


313 


300 


396 


254 


288 


245 


295 


120 


120 


119 


109 


112 


102 


96 


91 


88 


85 


77 


79 


69 


86 


56 


55 


56 


61 


63 


51 


62 


75 


94 


86 


76* 


74* 


88 


97 


162 


159 


168 


137 


153 


125 




58* 


68 


61 


60 


59 


53 


65 


50* 


52 


42 


52 


46 


39 




84 


74 


74 


68 


85 


67 


75* 


53 


50 


63 


38 


59 


47 


34 



Greatest length, greater tuberosity to radial capitulum 
Length, proximal capitulum to radial capitulum 
Length, lesser tuberosity to radial capitulum 
Transverse width across tuberosities 
Greatest transverse width of proximal capitulum 
Transverse width at narrowest part of shaft 
Anteroposterior width at midshaft 
Greatest width across epicondyles 
Greatest anteroposterior diameter of medial 

edge of trochlea 
Greatest anteroposterior diameter of radial capitulum 
Greatest width of distal articulation 
Transverse width of entepicondyle 



"Holotype. 
Estimate on broken feature. 



crest on the lateral surface of a convex shaft. This configuration is 
like that of Odobenus. V. imperialensis, and Pliopedia (USNM 
187328) and unlike that of Imagotaria. Gomphotaria, Pontolis, and 
otariids, in which the insertion for the deltoid muscle appears as a 
ridge confluent with the pectoral crest. An intermediate condition is 
seen on humeri of Alachtherium creisii (van Beneden 1877: pi. .3, 
fig. 1) and Prorosmarus alleni (MCZ 7713 in Repenning and 
Tedford 1977). in which, the deltoid insertion, although still on the 
pectoral crest, is more posterolaterally placed (i.e.. the crest is 
transversely broadened). 

The distal end of the humerus of both species of Valenictus is 
very broad, primarily because of the greatly enlarged entepicondyle. 
In the holotype of V. chulavistensis the width of the entepicondyle is 
16% of the total length of the humerus. This measure varies from 14 
to 20% (N = 4) in the referred humeri. A least-squares regression 
analysis revealed no significant correlation between enlargement of 
the entepicondyle and body size (R 2 = 0.31, p > 0.05). In 
V. imperialensis, the entepicondyle falls within the range of 
V. chulavistensis at 1 9% of the total humeral length. In Odobenus 
the measure is only 8%, in Alachtherium about 8%, and in 
Imagotaria approximately 10%. In V. chulavistensis, the 
entepicondyle is extremely large and robust (Fig. 5C) and antero- 
posteriorly compressed with an outline roughly rectangular in both 
medial and anterior aspects. The proximodistal axis of this rectan- 
gular process is rotated posteriorly at its distal end. The 
entepicondyle of V! imperialensis is also enlarged but more distally 
placed, rounded, and knoblike, rather than rectangular and rotated 
posteriorly. A partial humerus (USNM 13643) collected from the 
lower Pliocene San Joaquin Formation. Kettleman Hills, Califor- 
nia, shares many features with humeri of V. chulavistensis, includ- 
ing the large and robust rectangular entepicondyle rotated posteri- 
orly and the osteosclerotic internal bone structure. Repenning and 
Tedford ( 1977) illustrated this specimen (pi. 16. fig. 7) and referred 
it to V. imperialensis. From the features discussed above, however. I 
tentatively refer USNM 1 3643 to the new species from Chula Vista. 

The ectepicondyle of V. imperialensis is conspicuously reduced 
relative to the more enlarged condition in V. chulavistensis, 
Odobenus, and Alachtherium. The humerus of Valenictus 
chulavistensis also differs from that of V. imperialensis in possess- 
ing a distinctly embayed olecranon fossa set medial to a distinctly 



keeled supinator ridge. In V. imperialensis, the olecranon fossa has 
a more convex surface adjacent to a broadly rounded supinator 
ridge. In fact, the entire posterior profile of the shaft of 
V. imperialensis is planar, that of V. chulavistensis, sigmoidal. 

As in all odobenids, the greatest anteroposterior diameter of the 
medial lip of the trochlea of V chulavistensis is greater than that of 
the distal radial capitulum. However, the distal trochlear axis forms 
an angle of 90° (N = 5) with the humeral shaft's axis. In 
V. imperialensis (N = 1 ) and Alachtherium (N = 1 ) this angle is 83°, 
while in Odobenus (N= 2) the angle is even more acute at 77°. This 
suggests that the antebrachium of the new species was not as 
medially directed as in Odobenus. 

The important differences that distinguish the humerus of 
V. chulavistensis from that of V. imperialensis include larger size, 
sigmoidal posterior profile, sharply keeled supinator ridge, robust 
and rectangular entepicondyle, more prominent ectepicondyle, and 
more obtuse angle between the shaft axis and distal trochlear axis. 

Calcaneum. — The calcaneum of V. chulavistensis is unique. It 
is much broader distally than proximally. In Odobenus and 
Imagotaria (USNM 23862) these two dimensions are nearly equal. 
In dorsal or astragalar aspect (Fig. 6A). the sulcus calcanei between 
the sustentacular and ectal facets is broad and unlike the narrow 
sulcus of Imagotaria and Odobenus. Correlated with this broaden- 
ing is a sustentacular facet that is positioned well distad, almost 
parallel with the distal cuboid facet. This distal placement of the 
sustentacular facet coupled with a well-developed lateral trochlear 
process (peroneal tubercle of Kellogg 1931 ) gives the calcaneum of 
V. chulavistensis its extremely broad distal end. The cuboid facet is 
an elongate rectangle, in contrast to the quadrate cuboid facet of 
Odobenus and the short rectangular cuboid facet of Imagotaria 
downsi (Repenning and Tedford 1977, USNM 23862). As in 
Imagotaria (USNM 23862), and in contrast to Odobenus and 
Prorosmarus (USNM 215236), the sustentaculum lacks a second- 
ary shelf (Robinette and Stains 1970). The ectal facet is nearly 
planar, not convex as in Odobenus, Imagotaria. and otariids. The 
calcaneal tuber is long, with a prominent medial tuberosity 
(Fig. 6B) similar in size and form to that of Odobenus but less 
medially elongated than that of Imagotaria (USNM 23862). The 
cuboid facet forms an angle of between 14° and 21° with the 
longitudinal calcaneal axis. In Imagotaria this measure is between 



88 



Thomas A Demere 



A 







>.?-■ 



fev 










■' 



B 





D 



Figure 6. Valenictus chulavistensis, new species. A, B, SDSNH 36786. holotype right calcaneum. A, astragalar view (stereophotographs); B, palmar 
view. C. SDSNH 36786, holotype left astragalus, calcanear view (stereophotographs). D, SDSNH 35273. referred left astragalus, proximal view. Scale bar, 
5 cm. 



10° and 15°, while in Odobenus it is between 30° and 35°. Like 
other skeletal elements, the calcaneum is constructed of osteo- 
sclerotic bone. 

Astragalus. — This element also has an extremely unusual mor- 
phology (Figs. 6C. D). The capitulum is not set off from the stocky 
trochlear portion of the astragalus by a distinct neck as is in all other 
pinnipeds. The medial trochlear ridge (maleolar tibial facet) is 
distinctly longer than the lateral trochlear ridge (trochlear tibial 
facet), not shorter or of equal length as in Odobenus and Imagotaria 
(USNM 23867). In fact, the medial trochlear ridge extends so far 
proximally that it meets the medial plantar tuberosity. In Odobenus 
and Imagotaria both tibial articular trochlea are well separated 
from the medial plantar tuberosity by a distinct sulcus for the flexor 
hallucis longus tendon. In V. chulavistensis, the medial side of the 
medial trochlear ridge has a well-developed sulcus and there is no 
prominent lateral process (collum tali), only a flexure in the lateral 
outline of the astragalus (Fig. 6C). In Odobenus and otariids the 
lateral process is prominent and well separated from the capitulum. 
The capitulum of V. chulavistensis is directed medially at an angle 
of approximately 40' to the long axis of the astragalus. In plantar 
aspect (Fig. 6C), the medial sustentacular facet is confluent with the 



navicular facet, in sharp contrast to the distinct and well-separated 
navicular and sustentacular facets of other pinnipeds. In 
V. chulavistensis, the region between the sustentacular and eetal 
facets is a very broad sulcus calcanei. correlated with the corre- 
sponding broad sulcus of the calcaneum. The ectal facet is broadly 
J-shaped and extends laterally to meet the plantar border of the 
vertical fibular facet (i.e., there is no proximolateral shelf between 
the ectal facet and the fibular facet as is seen in other pinnipeds). As 
in all odobenids, the astragalus of V. chulavistensis has a postero- 
medial calcaneal process (medial plantar tuberosity); however, the 
process in this taxon is a broadly rounded structure, less distinct 
than the prominent process of Imagotaria, as discussed by 
Repenning and Tedford ( 1977). 

Phxlogenetic relationships. — Valenictus chulavistensis is an 
odobenine walrus closely related to modern Odobenus rosmarus 
and the fossil walruses Alachtherium cretsii, Prorosmants alleni, 
and Pliopedia pacifica (Fig. 7). Odobenine synapomorphies (num- 
bers refer to characters as discussed by Demere 1994, this volume) 
supporting this relationship include ( I ) external narial opening 
elevated above incisive margin, (9) palate narrow and arched trans- 
versely and longitudinally, ( 10) hard palate elongated, (II) palatine 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



89 






J? J* 



jr / f f J 



^ 



iT J? 



f S J? J* * 



53> 



^ .&' 



Odobenlnae 




Figure 7. Phylogenetic relationships of dusignathine and odobenine walruses. 



telescoped beneath alisphenoid, (12) hamular processes broad. (13) 
pterygoid strut lost, (17) mastoid processes as widest part of cra- 
nium, (19) cranial vertex with distinct flattened traction surface, 
(20) sagittal crest lost (also seen in some phocids). (21) zygomatic 
arches blunt and robust. (23) temporal fossae shortened, (24) optic 
foramen funnel-shaped. (25) orbital vacuity posteriorly placed. (29) 
upper canine with well-developed globular orthodentine column. 
(43) vesicular mandibular terminus, and (47) deltoid tubercle of 
humerus posterior to pectoral crest. 

The humeri of Alachtherium and Prorosmarus possess features 
more plesiomorphic than those of Odobenus and Valenictus. Sev- 
eral autapomorphies of the new fossil species (e.g., edentulous 
lower jaw and nearly edentulous upper jaw, osteosclerotic long 
bones, and numerous features of the humerus, astragalus, and calca- 
neuml suggest that V. chulavistensis diverged from its common 
ancestor with Odobenus prior to pursuing its own unique evolution- 
ary path toward its derived edentulous condition. 

Recognition of Valenictus chulavistensis as a tusked odobenine 
walrus settles a long-standing question about the relationships of 
Valenictus imperialensis. When Mitchell ( 1961 ) first described this 
species he considered it to be a specialized odobenid. Later, he 
(Mitchell 1968) implied that V. imperialensis was distantly related 
to the Odobeninae. Repenning and Tedford (1977) and Barnes 
(1989) assigned this species to the Dusignathinae. with reserva- 
tions. 

Functional morphology. — A complete discussion of the func- 
tional aspects of the skeleton of Valenictus chulavistensis is beyond 
the scope of this report. Three aspects, however, are discussed here: 
development and function of elongated ever-growing canines 
(tusks), feeding behavior as it relates to tooth loss, and locomotor 
implications of the humerus. 

Tusks and behavior. — Possession of homologous enlarged up- 
per canines in V. chulavistensis, Alachtherium, Odobenus, and prob- 



ably also Prorosmarus suggests that the common ancestor of these 
odobenine taxa had tusks and that modern Odobenus inherited 
them. In this light, adaptational scenarios explaining tusk evolution 
in Odobenus must also explain the development of tusks in all fossil 
walruses of temperate latitudes. The tusks of Odobenus must be 
considered not solely as adaptations for an arctic existence but as 
structures with a history. Fay (1982) showed that walruses do not 
use their tusks directly in benthic feeding, as erroneously suggested 
by other workers. The wear patterns noted by Fay ( 1982) on tusks 
of Odobenus, also preserved on tusks of V chulavistensis, are the 
product of incidental abrasion during benthic feeding. As a walrus 
forages with its muzzle against the substrate, it drags its tusks 
passively through the bottom sediments, wearing their anterior 
margins. The anatomical and behavioral data of Fay (1982:137), 
when combined with the phylogenetic data presented here, suggest 
that tusks are not the product of viability selection but rather 
evolved for social display, most probably under the pressures of 
sexual selection. Walrus tusks, like cervid antlers, are structural 
adaptations for social interactions (e.g., intraspecific dominance) 
rather than as sea-floor "plowshares" or arctic "ice tongs." Presum- 
ably, Valenictus chulavistensis used its tusks for social display as 
does the living Odobenus rosmarus. 

Jaws and feeding. — Possession of an edentulous lower jaw and 
nearly edentulous upper jaw begs the question, "how did Valenictus 
chulavistensis feed?" Fay ( 1982) has shown that modern walruses 
are suction-feeders, specializing on soft-bodied and thin-shelled 
benthic invertebrates (e.g., polychaetes, tunicates, and molluscs). 
According to Fay (1982) Odobenus does not use its peglike cheek 
teeth to crush prey but rather relies on a strong oral suction to ingest 
prey whole. He suggested that any function the cheek teeth retain is 
related to aquatic communication, supported by the observation 
that submerged walruses produce a loud clacking sound by percus- 
sive tooth occlusion. When feeding on thin-shelled pelecypods 



90 



Thomas A. Demere 




Figure 8. Dusignalhus seftoni, new species. SDSNH 38342, hololype 
skull, computer tomography scan three-dimensional images. A. oblique 
lateral view; B. oblique anterior view. Scale bar, 3 cm. 



Odobenus rosmarus sucks the mollusks from their shells before 
ingesting them (i.e., shells are not crushed by the teeth) (Fay 1982). 
With a strategy of feeding by oral suction, teeth are vestigial, 
implying that their loss in V. chulavistensis is not a "preadaptation" 
for starvation but a derived condition related to a unique feeding 
strategy. A test of this hypothesis is provided by the bearded seal, 
Erignathus barbatus, which as a strong suction-feeder and part- 
time henthic browser, frequently loses its teeth in old age (F. H. Fay, 
pers. comm.). In addition, the living monodontid, Monodon mono- 
ceros, and ziphiid odontocetes (e.g., Mesoplodon spp.) have lost all 
postcanine teeth and are oral-suction feeders, specializing on squid 
and small schooling fishes. Thus tooth loss following adoption of 
suction feeding is a derived condition at which several different 
groups of suction-feeding marine mammals have arrived inde- 
pendently. Since the common ancestor of V. chulavistensis and 
Odobenus rosmarus obviously had postcanine teeth, the edentulous 



condition of V. chulavistensis is more derived than the retention of 
teeth by Odobenus and represents the first case among pinnipeds of 
loss of all teeth but tusks. 

Humerus and locomotion. — The humerus of V. chulavistensis 
preserves an interesting mosaic of characters. Most conspicuous is 
the overall stockiness of the humerus and the greatly enlarged 
entepicondyle. Enlargement of the entepicondyle is correlated with 
an increase in the mass of the forelimb's flexor and pronator muscu- 
lature (English 1980) and suggests that V. chulavistensis relied 
more on forelimb flexion and pronation during swimming than 
does Odobenus. As discussed by English (1980), however, the 
strong muscles suggested by the large entepicondyle might have 
served to oppose supination and passive forelimb extension rather 
than to impose pronation and flexion actively. This opposition to 
supination and extension are important actions in maintaining a 
rigid pectoral "rudder." 

Gordon (1981) divided extant pinnipeds into three general 
groups by mode of aquatic locomotion: forelimb swimmers (i.e., 
otariids), hindlimb swimmers (i.e., phocids). and forelimb/hindlimb 
swimmers (i.e.. odobenids). Forelimb-swimming otariids rely pri- 
marily on adduction and abduction of the forelimb rather than on 
flexion and extension (Howell 1929; English 1980; Gordon 1981), 
suggesting that pronation and supination are important muscle ac- 
tions. Hindlimb-swimming phocids rely primarily on abduction and 
adduction of the hindlimb (Howell 1929). The forelimb/hindlimb 
swimming mode proposed for odobenids may be misleading, as 
walruses' primary source of aquatic propulsion is supplied by the 
hindlimbs; they use the forelimbs only for steering and stabilization 
(F. H. Fay, pers. comm.). 

Berta and Ray ( 1990) suggested that forelimb/hindlimb aquatic 
locomotion is the primitive condition for pinnipeds, as presumed in 
Enaliarctos. Whether or not the condition in Odobenus is homolo- 
gous with that in Enaliarctos requires further analysis. It does seem, 
however, that V. chulavistensis adopted a locomotor strategy in- 
volving more forelimb pronation, or suppression of pronation, than 
that of Odobenus. This implies a greater degree of forelimb in- 
volvement in aquatic locomotion in Valenictus. 

Giffin (1992) independently assessed the swimming behavior 
of Valenictus chulavistensis, including 10 vertebrae from the holo- 
type, in her analysis of pinniped locomotion. Assuming a correla- 
tion between neural canal anatomy and locomotor ability, she con- 
cluded that V chulavistensis was a forelimb/hindlimb (hindlimb- 
dominated) swimmer like modern Odobenus and not primarily a 
forelimb swimmer like modern otariids. Importantly, Giffin found a 
close similarity between Odobenus and phocids in terms of axial 
innervation and correlated muscle activity. 

The humerus of Alachtherium, like that of Odobenus, does not 
have an enlarged entepicondyle (van Beneden 1877: pi. 3, fig. 1), 
suggesting that the condition in Valenictus is uniquely derived, 
while that of Odobenus is a shared primitive feature retained from 
the common ancestor of all tusked odobenines. 

The osteosclerotic nature of the limb bones of V. chulavistensis 
is unique among known fossil and living pinnipeds and is conver- 
gent with the condition in sirenians. Functionally, this may have 
reduced buoyancy for the species' presumed benthic feeding in 
temperate latitudes. The fact that Odobenus rosmarus lacks 
osteosclerotic bone is a puzzle but may be related to its arctic 
habitat of cold, dense bottom waters. 

Discussion. — Valenictus chulavistensis is possibly the most 
completely known fossil odobenine. represented by essentially ev- 
ery major skeletal element. This species was relatively large, simi- 
lar in overall size to modern Odobenus rosmarus, but smaller than 
the great fossil walrus Alachtherium cretsii from the early Pliocene 
of the eastern North Atlantic. 

The genus Valenictus has long been considered a problematic 
taxon. in large part because the type species is based on a single 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



91 



postcranial element not readily comparable with other more com- 
pletely known taxa. Referral of the new San Diego Formation 
species to this genus offers a solution to this taxonomic problem by 
supplying important new information that confirms the odobenine 
relationships of Valenictus. 

Valenictus imperialensis is also unusual because it occurs in the 
Imperial Formation of the Colorado Desert. Imperial County, Cali- 
fornia. The Imperial Formation was deposited during the late Mio- 
cene and early Pliocene in the proto-Gulf of California. As now, the 
Gulf had no direct connection with the temperate eastern North 
Pacific but instead extended south into tropical latitudes along a 
tectonic lineament characterized by crustal thinning and extension 
(Mammerickx and Klitgord 1982). The occurrence of tropical and 
subtropical molluscan taxa in the Imperial Formation, some with 
Caribbean affinities (Kew 1914; Vaughan 1917; Hanna 1926; 
Schremp 1981; Kidwell 1988). and their total absence in the well- 
studied marine Neogene deposits of coastal southern California, 
supports a strictly tropical connection and also implies an equato- 
rial connection between the Caribbean and eastern tropical Pacific 
before the raising of the Isthmus of Panama. The invertebrate and 
vertebrate faunas of the proto-Gulf and temperate eastern Pacific 
were rather isolated from each other, implying a certain degree of 
endemism for the Imperial Formation faunas. Thus V. imperialensis, 
possibly confined to the subtropical proto-Gulf of California, may 
have been the result of late Miocene allopatric speciation. Further- 
more, V. chulavistensis may represent a secondary late Pliocene 
dispersal of this clade into the temperate eastern North Pacific 
following emergence of the Isthmus of Panama. 

The holotype humerus and only known specimen of Valenictus 
imperialensis shares several apomorphies with the humerus of 
V. chulavistensis. Although the possibility that the two species are 
conspecific cannot be ruled out altogether, the morphological differ- 
ences presented above coupled with the late Miocene age of V 
imperialensis and its apparent restriction to the proto-Gulf of Cali- 
fornia suggest that synonymy is unlikely. If all of the features shared 
by the two taxa represent synapomorphies inherited from a common 
ancestor and the additional apomorphies of V. chulavistensis repre- 
sent uniquely derived features, V. imperialensis may not be diagnos- 
able at the species level (i.e.. it may represent a nomen dubium). This 
is a problem inherent in the questionable practice of describing fossil 
taxa from isolated skeletal elements of dubious diagnostic value. 
The discovery of additional material of Valenictus imperialensis 
and/or new material of other related odobenine species with the same 
synapomorphic features will help to resolve these questions. 

Subfamily Dusignathinae Mitchell, 1968 

Dusignathus Kellogg. 1927 

Type species. — Dusignathus santacruzensis Kellogg, 1927. 

Distribution. — Late Miocene and late Pliocene of California 
and Baja California. 

Included species. — D. santacruzensis Kellogg, 1927, and 
D. seftoni, n. sp. 

Emended diagnosis. — Dusignathine walruses distinguished 
from other taxa by the following apomorphies: mandibular sym- 
physis narrowly V-shaped in occusal aspect, lower canines closely 
appressed to each other, left and right dentaries forming acute angle 
of 60° at symphysis, rostrum shortened, and mandibular rami deep 
(relative to Comphotaria). 

Dusignathus seftoni n. sp. 

Figures 8-1 1 

Diagnosis. — A species of Dusignathus distinguished from D. 
santacruzensis by the following autapomorphies: upper and lower 



cheek teeth forming a laterally convex arch in occlusal aspect, 
postcanine teeth in upper and lower jaws with medially rotated 
anteroposterior axes of roots, roots of all cheek teeth closely ap- 
pressed, and dentary with deeply excavated masseteric fossa. Shares 
the following apomorphies with other dusignathines: nasal/frontal 
suture posteriorly directed and V-shaped; upper and lower canines 
enlarged as tusks. 

Type material.— SDSNH 38342, a skull lacking the basi- 
cranium. Collected by Richard A. Cerutti and Matthew W. Colbert, 
12 May 1989. 

Etymology. — The species is named in honor of Thomas W. 
Sefton. who has generously supported the collection and study of 
fossil marine mammals from San Diego County. 

Type locality.— SDSNH locality 3468, city of Chula Vista, San 
Diego County, California. 

Horizon and age. — San Diego Formation, "lower member" of 
Demere ( 1983), late Pliocene (Blancan NALMA correlative). 

Referred specimens. — SDSNH 20801. right dentary preserving 
part of the symphyseal region of the left dentary; SDSNH 38256, 
damaged left humerus; SDSNH 43873, left humerus (all collected 
from the San Diego Formation). Complete locality information is 
available to interested reserachers upon request. 

Cranium. — The holotype cranium was damaged by earth-mov- 
ing equipment. The left side of the braincase is missing, as is the left 
zygomatic arch. Also missing is the entire basicranium, including 
both auditory regions, mastoids, and postglenoid fossae. The major- 
ity of the occipital shield, including the paramastoid processes, is 
also missing. Anteriorly, the left tooth row is obliterated, and with it 
the posterior border of the palate and internal narial opening. The 
right I\ C, and P 1 " 2 are sheared off just distad of the alveoli. The 
pattern of suture closure (see Sivertson 1954) indicates a subadult 
individual. 

The cranium (Figs. 8, 9, 10A, B) preserves many general fea- 
tures characteristic of odobenids. including a low sagittal crest (as 
in Neotherium and Imagotaria), lack of supraorbital processes of 
frontals (as in Neotherium. Imagotaria. Comphotaria, cf Pontolis. 
Aivukus, Alachtherium, Valenictus, and Odobenus), prominent 
antorbital processes (as in Imagotaria, Comphotaria, Pontolis, 
Aivukus, Alachtherium, Valenictus, and Odobenus), and enlarged 
infraorbital foramen (as in Imagotaria, Pontolis. Gomphotaria, 
Aivukus, Alachtherium, Valenictus, and Odobenus). 

The rostrum is short and broad relative to that of Gomphotaria 
(Table 2) and houses a pair of enlarged canines (tusks). The inclina- 
tion of the canine root in relation to a vertical transverse plane is 
33°, which is more vertically inclined (Figs. 8A, B) than the canines 
of Gomphotaria pugnax and approaches the condition in tusked 
odobenines. The frontal/maxilla suture forms an acute angle (ap- 
proximately 60°) with the sagittal plane of the skull and is 
continuous with the nasal/frontal suture (Fig. 10B). The antorbital 
processes are split by the frontal/maxilla suture and are thus con- 
structed from both frontal and maxilla. The nasals project posteri- 
orly to form a wedge between the frontals (as in Gomphotaria and 
Pontolis, USNM 314300) and, anteriorly, are roughly rectangular 
(Figs. 9A. 10B). Thin ascending processes of the premaxillae over- 
lap the nasals along 68% of their lateral margins. The anterior narial 
opening is more vertically oriented than that of Comphotaria and 
ends in a prominent nasal process of the premaxillae. The floor of 
the narial opening is only slightly elevated (25 mm) above the level 
of the incisive margin. In contrast to D. santacruzensis, the maxillae 
are swollen to accommodate the roots of the enlarged canines, but 
not to the extent that they obscure the infraorbital foramina when 
the skull is viewed in anterior aspect (Fig. 8B). 

The maxillary root of the zygomatic arch is delicately con- 
structed, with a very slender dorsal strut. The ventral strut is di- 
rected dorsolaterally. in contrast to the more horizontally directed 
struts seen in Imagotaria and Gomphotaria. The effect of this is to 



92 



Thomas A. Deniere 





Figure 9. Reconstructions of Dusignalhus seftoni. new species. A, dorsal view; B, lateral view. Al, alisphenoid; ap. antorbital process; fio, infraorbital 
foramen; fsp. sphenopalatine foramen; Fr, frontal; J. jugal; Mx, maxilla; Na, nasal; ov, orbital vacuity; Pa, palatine; Pr, parietal; Pm. premaxilla; Sq, 
squamosal. Scale bar. 5 cm. 



give the infraorbital foramen a rounded triangular shape and a long 
axis inclined dorsolaterally, in contrast to Imagotaria and 
Gomphotaria, in which the long axis is directed horizontally. The 
ventral surface of the ventral strut is marked by a distinctive fossa 
(origin of the maxillo-naso-labialis muscle; Howell 1929) that is 
more prominent than that in Aivukus. The dorsal and ventral struts 
are positioned one above the other, in contrast to the condition seen 



in Gomphotaria. whose the dorsal strut lies anterior to the ventral. 
The jugal is also delicately constructed, and has a small triangu- 
lar postorbital process (Fig. 8A). The orbit is large relative to that of 
Gomphotaria. The squamosal fossa forms a shelf over the missing 
external auditory meatus but is narrower than in Gomphotaria. The 
zygomatic portion of the squamosal is long and slender and forms a 
splintlike suture with the jugal. 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



93 







wB P^. 



 



v 




B , 



I 




^ 




Figure 1 0. Dusignathus seftoni, new species. A, B, SDSNH 38342, holotype skull. A, ventral view (stereophotographs); B, dorsal view. C, D, E, SDSNH 
20801, referred right dentary. C, lateral view (stereophotographs); D. medial view; E, occlusal view (stereophotographs). Scale bar, 5 cm. 



94 



Thomas A. Demere 









\ 






Figure II. Dusignathus seftoni, new species, SDSNH 43873, referred left humerus 
(stereophotographs); C, posterior view. Scale bar. 5 cm. 



A, anterior view (stereophotographs); B, lateral view 



The palate is broad and not obviously arched. A pair of large 
incisive foramina are positioned 15 mm posterior to the lateral 
incisors (Fig. 10A). The palatine foramina lie 30 mm anterior to the 
palatine/maxilla suture. The posterior border of the palate is not 
preserved. 

The anterior margins of the frontals along the midline are el- 
evated slightly above the level of the maxillae and nasals. This 
elevation is even more pronounced in Gomphotaria. The anterior 
portion of the interfrontal suture is slightly depressed and becomes 
obscured posteriorly in a fine median sulcus that continues posteri- 
orly into the interparietal suture. Farther posteriad the interparietal 
suture is marked by a low but distinct sagittal crest that merges 
posteriorly with the elevated right and left portions of the 
lambdoidal crest. The lambdoidal crest flares posteriorly, over- 
hanging the largely missing occipital shield, which preserves no 
evidence of an occipital crista. The lateral walls of the braincase are 



broadly convex (as in Neotherium, Imagotaria, Alachtherium, and 
Odobenus), not concave (as in Pontolis and Gomphotaria). The 
interorbital constriction is prominent and positioned posteriorly 
against the anterior border of the braincase. 

The orbital wall, on the right side, preserves much of the frontal/ 
maxilla suture as it descends from the antorbital process to meet the 
palatine where the suture bifurcates into the frontal/palatine and 
palatine/maxilla sutures. There is no lacrimal bone, and thus the 
maxilla forms the entire anterior border of the orbit. The frontal/ 
palatine suture is present for a short distance anteriorly but is lost 
posteriorly in a narrowly elongated orbital vacuity. A thin plate of 
palatine separates the orbital vacuity from the maxilla and thus the 
palatine/maxilla suture has a broad exposure on the orbital wall as it 
descends to the sphenopalatine foramen. From here the suture 
continues ventrally to the broken palatal margin. The orbito- 
sphenoid is difficult to interpret because of breakage but seems to 



Two New Species of Fossil Walruses from ihe Upper Pliocene San Diego Formation 



95 



extend posteriorly from the posterior border of the orbital vacuity to 
the alisphenoid. The dorsal border of the orbitosphenoid is marked 
by a distinct ethmoidal foramen. A large arcuate opening marks the 
region of the missing optic foramen and orbital fissure. The dorsal 
portion of the alisphenoid is well preserved and broadly exposed on 
the anterolateral border of the braincase. The parietal/squamosal 
suture is well preserved and runs horizontally from the enlarged 
alisphenoid back to the broken lambdoidal crest. The pterygoids are 
not preserved. 

Within the brain case, the cribriform plate is preserved as a 
teardrop-shaped structure, widest ventral ly. A portion of the bony 
tentorium is preserved on the right side of the braincase. and has no 
expression externally. 

Upper dentition. — Only the intra-alveolar portions of six teeth 
(right and left C\ right and left I 3 , P 1 " 2 ) are preserved in the 
holotype cranium. The empty alveoli of I : , P 1 ^*. and M' - indicate 
only single-rooted teeth. There is no I 1 alveolus. Alveolar diameters 
for the right upper dental arcade I : -M 2 are as follows 
(anteroposterior length/transverse width in mm): 11.8/6.7; 14/20; 
22/29.5; 13/12.5; 11/11; 14/10; 10/9; 8/7.4; 8.4/6.4. 

The lateral incisor has an oval cross-section, with the long axis 
of the cross-section directed somewhat transversely. The right I 1 
preserves a thin ( 1 mm) outer cementum layer surrounding a dense 
orthodentine core. The right C 1 of the new species is greatly en- 
larged relative to P 1 and has a large open pulp cavity (visible in 
computerized tomography images), suggesting either immaturity or 
continuous growth. The cross-sectional shape of this tooth is oval 
proximally (intra-alveolar), becoming roughly triangular distally. 
The distal cross-section approximates a right triangle, with the right 
angle placed medially and the hypotenuse corresponding to the 
labial surface of the tooth (Fig. 10A). In the holotype of D. 
santacruzensis, C is oval throughout and has a closed root. In the 
rostral fragment (UCR 15244) from the Almejas Formation, Baja 
California, referred to D. santacruzensis by Repenning and Tedford 
(1977:46), C 1 has an oval cross-section and is approximately equal 
in size (alveolar diameter) to P 1 . The canines of D. seftoni consist of 
a thin outer cementum layer and an inner massive orthodentine 
core, as in Gomphotaria. There is no evidence of a central globular 
orthodentine column as in the tusked odobenine walruses. No traces 
of enamel were observed on the narrow remnant of crown. The 
anterolateral surface of the canine has a single shallow longitudinal 
groove quite different from the regular longitudinal fluting seen in 
Gomphotaria (see Barnes and Raschke 1991 ) and the three or four 
shallow longitudinal grooves of Odobenus (see Ray 1960). 

There are alveoli for six postcanine teeth. The alveolus for P 3 is 
a deep oval opening with the long axis of the cross-section rotated 
medially 36° to the sagittal plane. One wall of this alveolus is marked 
by a faint vertical ridge, presumably corresponding to an incipient 
bifid root. This alveolus extends at least 26 mm into the maxilla. The 
walls of this and all cheek-tooth alveoli continuously taper to the 
root, quite unlike the bulbous peglike roots of Gomphotaria. The 
alveolus for P* is also relatively deep ( 15 mm) but nearly circular in 
cross-section. Alveoli for M 1 " 2 are shallow (6 and 4 mm, respec- 
tively) and also circular in cross-section. In occusal aspect, the 
postcanine tooth row forms a broad, laterally convex arc aligning 
with I 1 . The canine is positioned slightly outside of this arc. I 3 lies 
somewhat medial to C, not entirely anterior to it as in Neotheriwn, 
Imagotaria, and Pontolis. All of the postcanine alveoli are closely 
appressed to each other, with no intra-alveolar spaces. A conspicu- 
ous diastema 10 mm wide separates C 1 from I 3 . This incisor-canine 
diastema is too narrow to accommodate an enlarged C,, suggesting 
that the lower canine occluded with the upper lateral incisor as 
suggested for D. santacruzensis [see Repenning and Tedford (1977); 
both the holotype and the referred partial rostrum from the Almejas 
Formation, UCR 15244]. 

Dentary.— SDSNH 20801 is a complete right dentary (Table 4) 



with empty alveoli for a reduced incisor (possibly I,), enlarged 
canine, and five postcanine teeth (presumably P,-M,). Although 
the specimen was damaged, a good cast of the specimen, made 
before the damage occurred, is available at the USNM. 

The horizontal ramus is deep dorsoventrally (Fig. 10C) as in D. 
santacruzensis and thick transversely as in Gomphotaria. Two large 
mental foramina are located midway along the lateral surface of the 
dentary, one each below P, and P,. The anterior mental foramen is 
oriented dorsoanteriorly, while the larger and more posterior mental 
foramen is oriented dorsomedially. In D. santacruzensis there are 
also two mental foramina, one each below P, and P,. In medial 
aspect, the mandibular symphyseal surface is a narrow oval (as in 
D. santacruzensis and Pontolis, USNM 335563) and not a broad 
oval (as in Gomphotaria). A portion of the medial wall of the left 
canine alveolus is preserved indicating a fused symphysis. The 
posteroventral portion of the symphysis is marked by a large globu- 
lar and deeply excavated genial tuberosity, which contrasts with the 
more slender, ridgelike tuberosity of D. santacruzensis. The 
coronoid process is large and rises at an angle of about 55° from the 
tooth row. In Gomphotaria and Pontolis (USNM 335563) the 
coronoid rises at a shallower angle (40° and 35°, respectively). The 
masseteric fossa in D. seftoni is more deeply excavated than in any 
living or fossil odobenid. The fossa is divided into upper and lower 
portions by a conspicuous horizontal masseteric ridge. The base of 
the fossa is marked anteriorly by a deep depression that is continu- 
ous posteriorly with a sharply margined shelf that extends as a 
horizontal surface to the mandibular condyle. A similar sharply 
margined masseteric shelf was described for a gigantic proximal 
mandibular fragment (UCR 15245) collected from the Almejas 
Formation and questionably referred to D. santacruzensis by 
Repenning and Tedford (1977:47). The mandibular condyle of 
UCR 15245 measures 107 mm in width; that of SDSNH 20801, 75 
mm. The condyle of D. seftoni is broad, slender, and spindle-like. 

As mentioned, the symphysis is fused, and preserves a portion 
of the medial wall of the left canine alveolus. The region between 
the two lower canines is narrow and in occulsal aspect is shaped 
like an isosceles triangle, with the most acute angle pointing for- 
ward. The medial walls of the right and left canine alveoli come to 
within 20 mm of each other, leaving no area for dorsally placed 
incisors. Although badly damaged, a small alveolus for an incisor is 
closely appressed to the anterior border of the canine. This alveolus, 
best seen on the USNM cast, is approximately 38 mm below the 
canine alveolar margin. The canine alveolus is large and deep and 
extends to a point below the alveolus for P 2 . The canine's root (as 
determined from the alveolus) has an oval cross-section. The alveo- 
lar portion of the horizontal ramus in occlusal aspect is broad and 
laterally convex. The lingual borders of the cheek-tooth alveoli are 
higher than those on the labial border. 

The pterygoid process (angular process) is large and projects 
medially approximately 20 mm as a hooklike flange as far as the 
medial border of the mandibular condyle. Anterior to this process, 
in lateral aspect, the ventral border of the ramus makes an abrupt 
flexure at the level of the distinct marginal process (Davis 1964). 
The marginal process itself is laterally swollen and rugose, marking 
the insertion for the digastricus muscle. The marginal process on 
the type right dentary of D. santacruzensis is slender and not 
enlarged. In lateral aspect, the ventral border of the ramus, between 
the marginal process and the globular genial tuberosity, is broadly 
sigmoidal, as in D. santacruzensis. 

The medial surface of the coronoid process is marked by a 
distinct and curving strutlike ridge that is inset from, but parallel to, 
the anterior coronoid crest. 

Lower dentition. — Although no teeth are present in the referred 
dentary, the alveoli are well preserved. Postcanine alveoli 1^4 
(presumably P,_,) are transversely elongate ovals at the level of the 
tooth row. In Dusignathus santacruzensis P, has a circular alveolus, 



% 



Thomas A. Demere 



while P,-M, are elongated ovals. In D. seftoni the cross-sectional 
alveolar diameters for P,_, (length/width in mm) are as follows: 23/ 
19: 28/19; 29/22: 21/16. The conical alveolus for M, is circular at 
the level of the tooth row ( 13 mm diameter). The canine alveolus 
measures 48 by 34 mm. Alveolar depths for C|-M, are as follows: 
109 mm; 48 mm; 51 mm; 51 mm; 40 mm; 26 mm, respectively. 
Morphological details of the canine alveolus suggest that the root 
was open and had shallow widely spaced longitudinal grooves. 
Preserved on the alveolar walls of P,_, are delicate vertical septa 
suggestive of vestigially bifid roots. In occlusal aspect, the entire 
tooth row (canine through M, ) forms a laterally convex arch. 

Perhaps the most unusual feature of the lower tooth row is the 
orientation of the alveoli in occlusal aspect, specifically the angle 
made between the greatest cross-sectional dimension of the alveo- 
lus and a parasagittal plane. From back to front the roots undergo a 
progressive torsion, which in the right dentary is expressed as a 
successive counterclockwise rotation of the greatest cross-sectional 
dimension relative to a parasagittal plane: P 4 has rotated 52°, P, 
67°, P, 96°, and P, 130° (Fig. 10E). 

Referral of the dentary (SDSNH 20801) to this new species is 
made on the basis of the comparably shortened tooth rows, laterally 
convex arching of the tooth rows, rotation of alveoli, and large size, 
features also observed in the holotype cranium. 

Humerus. — A nearly complete left humerus (SDSNH 43873) 
and a partial left humerus (SDSNH 38256) are here referred to the 
new taxon (Fig. 1 1 ). SDSNH 38256 consists of the diaphysis and 
distal epiphysis, with the proximal epiphysis (including the capitu- 
lum and tuberosities) missing. SDSNH 43873 is complete except 
for damage to the proximal end (lateral one-third of the capitulum 
missing) and distal end (ectepicondyle missing). The following 
description relies primarily on features visible on SDSNH 43873. 

The shaft is slender and similar in form to the humerus (USNM 
23870) referred by Repenning and Tedford (1977) to Imagotaria 
downsi. The slenderness of the shaft and its large size (Table 5) 
suggest similarity to Gomphotaria pugnax (see Barnes and Raschke 
1991 ). In posterior aspect, the lateral outline of the shaft is nearly 
straight and the medial outline is broadly concave. This contrasts 
with the more acutely concave medial profile of the humerus 
(UCMP 65318) questionably referred to Dusignathus santa- 
cruzensis by Repenning and Tedford (1977). Gross comparisons 
between UCMP 65318. USNM 23870, and SDSNH 43873 suggest 
that the latter two (Imagotaria downsi and Dusignathus seftoni) are 
more similar to each other than either is to UCMP 65318 (cf. 
Dusignathus santacntzensis). It should be noted that the holotype 
of D. santacruzensis does not include a humerus and that UCMP 
65318 was only tentatively referred to this taxon. Thus, the actual 
morphology of the humerus of D. santacruzensis remains uncer- 
tain. However, this is not the case for Gomphotaria, another 
dusignathine walrus. 

As in Gomphotaria. the proximal end of the humerus of D. 
seftoni has a rounded capitulum (head) positioned distinctly below 
the slender greater tuberosity. In Valenictus, Pliopedia (USNM 
187328), and cf. D. santacruzensis (UCMP 65318) the greater 
tuberosity and humeral head are nearly at the same level. The lesser 
tuberosity of D. seftoni is knoblike and positioned distinctly below 
the capitulum. in contrast to the condition in Imagotaria (USNM 
23870), in which the lesser tuberosity is nearly at the same level as 
the head. In medial aspect, the lesser tuberosity is broadened dis- 
tally, while in anterior aspect the tuberosity is narrower than in 
Valenictus. Pliopedia. and cf. D. santacruzensis. The bicipital 
groove is broad and U-shaped. 

The insertion for the deltoideus muscle is elongate and posi- 
tioned on the pectoral crest as in Imagotaria. cf. D. santacruzensis, 
Gomphotaria, and Aivukus and differs from the posterolateral^ 
displaced and isolated deltoid tubercles of Odohenus, Valenictus. 
and Pliopedia (USNM 187328). The pectoral crest itself is a slender 



and elongate keeled ridge that descends posteriorly, with some 
flexion, to join the distal surface of the shaft, as in Imagotaria 
(USNM 23870). This flexed posterior segment of the crest is inter- 
mediate in form between the gradually tapered crests of Odohenus, 
Valenictus, and Pliopedia (USNM 187328) and the abruptly flexed 
crests of Aivukus and Gomphotaria. As in all odobenids the distal 
portion of the pectoral crest is directed toward the medial lip of the 
trochlea, which is considerably broader than the radial capitulum. 
Distally, the trochlear surface makes an acute angle of about 76° 
with the shaft axis. 

The entepicondyle is small relative to that of Valenictus. It is 
shaped much as in Imagotaria (USNM 23870) (i.e., a medially 
flattened knob in anterior aspect), not being triangular as in 
Pliopedia (USNM 187328) and cf. D. santacruzensis (UCMP 
65318). Internally, the shaft of the humerus of D. seftoni is com- 
posed of cancellous bone, not osteosclerotic bone as in Valenictus. 
At 346 mm, SDSNH 43873 is longer (greater tuberosity to radial 
capitulum) than either Pliopedia pacifica (USNM 187328, 306 
mm) orcf. Dusignathus santacruzensis (UCMP 653 18. 271.6 mm) 
(Repenning and Tedford 1977). 

Assignment of SDSNH 43873 and SDSNH 38256 to Dusi- 
gnathus seftoni is based in part on the largeness of the former, 
which is compatible in size with the large mandible also referred to 
this species (SDSNH 20801). In addition, the elevated greater 
tuberosity of D. seftoni and the flexed pectoral crest are distinctive 
features shared with another dusignathine, Gomphotaria. And fi- 
nally, the overall generalized morphology of the referred humeri 
clearly separates them from humeri of the only other odobenid 
known from the San Diego Formation. Valenictus chulavistensis. 

Phylogenetic relationships. — Dusignathus seftoni is a dusigna- 
thine closely related to the late Miocene walruses D. santacruzensis 
and Gomphotaria pugnax. Synapomorphies supporting this rela- 
tionship (Fig. 7) include (3) posteriorly directed V-shaped nasal/ 
frontal suture, (32) upper and lower canines enlarged as tusks, and 
(45) dentary with sinuous ventral border (numbers refer to charac- 
ters as discussed by Demere 1994, this volume). 

Referral of D. seftoni to Dusignathus is based largely on fea- 
tures of the lower jaw. These include the sharply divergent man- 
dibular arch and presumed shortened rostrum, as well as the ex- 
tremely deep horizontal ramus and unreduced closely appressed 
lower canines. In both species of Dusignathus. rostral shortening 
did not result in loss of cheek teeth. However, in D. seftoni the 
accommodation of the cheek teeth into a shortened tooth row 
involved rotation and close appression of the roots of individual 
teeth. In D. santacruzensis the lower postcanine teeth lack any 
indication of rotated roots. Although the possibility exists that root 
rotation is an ontogenetic feature, D. seftoni is distinguished by 
other characters, including C 1 with triangular cross-section, C, 
enlarged as a tusk, laterally convex upper and lower cheek-tooth 
rows, and larger size. 

In Dusignathus seftoni as in Gomphotaria pugnax. plesio- 
morphic features such as a distinct sagittal crest, robust coronoid 
process, and large masseteric fossa imply a powerful jaw depressor 
musculature. 

Discussion. — The genus Dusignathus is now known from two 
species, one from the late Miocene of central California and possibly 
Baja California and a second from the late Pliocene of southern 
California. Dusignathus seftoni. the geologically youngest known 
dusignathine walrus, clearly shows that members of this clade sur- 
vived into late Pliocene time along the eastern North Pacific margin. 

The holotype skull of D. seftoni is from a subadult individual, 
possibly a male, while the referred dentary and large humerus are 
from adult animals, almost certainly males. The dimensions of the 
dentary suggest that the new species was large, approaching mod- 
ern Odobenus. Repenning and Tedford (1977) suggested that the 
type of D. santacruzensis was probably a female, so sexual dimor- 



Two New Species of Fossil Walruses from the Upper Pliocene San Diego Formation 



97 



phism may account for some of the size discrepancy between the 
two species. 

The shortened rostrum with steeply inclined tusks of D. seftoni 
is convergent with the condition in Odobenus. The size and cross- 
sectional shape of the upper and lower canines differ sufficiently 
from those of D. santacruiensis (the holotype and referred rostrum 
of Repenning and Tedford 1977:46. UCR 15244) to suggest that 
evolution of the San Diego species involved tusk development. The 
generalized morphology of the referred humeri suggests that this 
species, like Imagotaria, might have been more like otariids in its 
swimming habits than modern Odobenus. This is especially evident 
in the enlarged greater tuberosity, high and elongate pectoral crest, 
and relatively unenlarged entepicondyle. Repenning's ( 1976) sug- 
gestion that Imagotaria was a generalist neritic carnivore might 
apply equally to Dusignathus seftoni. 

SUMMARY 

The two new species of walruses described here increase our 
knowledge of odobenid evolution in many ways. Valenictus 
chulavistensis is possibly the most completely known fossil 
odobenine, represented by essentially every major skeletal element. 
This taxon preserves a reduced dentition previously unknown for 
marine carnivorans, emphasizing the morphological extremes pos- 
sible in the marine realm. Valenictus chulavistensis clearly shows 
that possession of ever-growing tusklike upper canines is an inher- 
ited feature shared with the fossil Alachtherium and modern 
Odobenus. This realization, coupled with the observation that wal- 
ruses do not use their tusks directly in benthic foraging, lends 
support to the hypothesis that walrus tusks evolved as social display 
structures, in a sense similar to the antlers of cervids. In addition, 
many other morphological features shared by Odobenus. Alach- 
therium. and Valenictus provide direct evidence for primitive char- 
acter states near the base of the tusked odobenine clade. 

Assignment of the new species to Valenictus clarifies the taxo- 
nomic and phylogenetic aspects of this formerly problematic taxon 
and provides a sense of the true taxonomic diversity of odobenine 
genera. 

Dusignathus seftoni is the third described dusignathine species 
and illustrates the range of taxonomic diversity in this clade of 
double-tusked walruses. The shortened rostrum, with its condensed 
but complete dentition, parallels the condition in tusked odobenines 
but is associated with an unreduced temporal musculature. 

Valenictus chulavistensis and Dusignathus seftoni were sympa- 
tric along the eastern North Pacific margin during the late Pliocene, 
approximately 2-3 Ma. illustrating the odobenids' past diversity. 
V. chulavistensis and D. seftoni may have avoided direct competi- 
tion through resource partitioning, with the former specializing on 
benthic invertebrates and the latter remaining a generalist neritic 
fish and squid eater. 

Extinction of the entire dusignathine clade must be viewed as a 
Pleistocene event, with only Odobenus of the odobenine clade 
surviving to the Recent. Recognition of Valenictus chulavistensis in 
Califomian Tertiary deposits illustrates that tusked odobenines re- 
mained a part of the North Pacific pinniped fauna at least into the 
late Pliocene. 



ACKNOWLEDGMENTS 

Most of the fossils described in this report were salvaged from 
construction sites in the Rancho Del Rev housing development, city 
of Chula Vista. Fossils were collected by field crews from 
PaleoServices, Inc., of San Diego. Special thanks are extended to 
Richard A. Cerutti, Matthew W. Colbert, Bradford O. Riney, Donald 
R. Swanson, and Stephen L. Walsh of PaleoServices, Ed Elliott. 



Ken Screeton. and Mark Carpenter of McMillin Communities, and 
Doug Reid of the city of Chula Vista Planning Department. 
Children's Hospital of San Diego and especially Glenn Daleo of 
that institution generously provided technical expertise and use of 
their computerized axial tomography facility. For permission to 
study fossils under their care I thank also Lawrence G. Barnes 
(LACM) and Clayton E. Ray (USNM). Fritz Hertel assisted with 
measurements. Matthew W. Colbert. Blaire Van Valkenburgh. and 
Francis H. Fay provided critical review of the manuscript. 



LITERATURE CITED 

Addicott, W. O. 1972. Provincial middle and late Tertiary molluscan 
stages. Temblor Range, California. Pp. 1-26 in E. H. Stinemeyer 
(ed.). Proceedings of the Pacific Coast Miocene Biostratigraphic 
Symposium. Society of Economic Paleontologists and Mineralo- 
gists, Pacific Section. 

Barnes, L. G. 1973. Pliocene cetaceans of the San Diego Formation. Pp. 
37-43 in R. Arnold and R. J. Dowlen (eds.). Studies on the Geology 
and Geologic Hazards of the Greater San Diego Area, California. 
San Diego Association of Geologists. San Diego, California. 

. 1979. Fossil enaliarctine pinnipeds (Mammalia: Otanidae) 

from Pyramid Hill, Kern County, California. Natural History Mu- 
seum of Los Angeles County Contributions in Science 318. 

1989. A new enaliarctine pinniped from the Astoria Formation. 



Oregon, and a classification of the Otanidae (Mammalia: Car- 
nivora). Natural History Museum of Los Angeles County Contri- 
butions in Science 403. 
-, and R. E. Raschke. 1991. Gomphotaria pugnax, a new genus 



and species of late Miocene dusignathine otariid pinniped 
(Mammalia: Camivora) from California. Natural History Museum 
of Los Angeles County Contributions in Science 426. 

Beneden, P. J. van. 1877. Description des ossements fossiles des envi- 
rons d'Anvers. Premiere partie. Pinnipedes ou amphitheriens. 
Annales du Musee Royal d'Historie Naturelle de Belgique 1:1-88. 

Berry. E. W.. and W. K. Gregory. 1906. Prorosmarus aileni, a new genus 
and species of walrus from the upper Miocene of Yorktown. Vir- 
ginia. American Journal of Science 244:60-62. 

Berta, A., and T. A. Demere. 1986. Callorhinus gilmorei n. sp. (Car- 
nivora: Otanidae) 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 of 
Vertebrate Paleontology 10:141-157. 

Chandler. R. M. 1990. Fossil birds of the San Diego Formation, late 
Pliocene. Blancan. San Diego County. California. Ornithological 
Monographs 44:77-161. 

Davis. D. D. 1964. The giant panda. A morphological study of evolu- 
tionary mechanism. Fieldiana: Zoology Memoirs 3:1-334. 

Demere. T. A. 1983. The Neogene San Diego Basin: A review of the 
marine Pliocene San Diego Formation of southern California. Pp. 
187-195 in D. K. Larue and R. J. Steel (eds.). Cenozoic Marine 
Sedimentation. Pacific Margin, U.S.A. Society of Economic Pale- 
ontologists and Mineralogists, Pacific Section. 

. 1986. The fossil whale Balaenoptera davidsonii (Cope 1872), 

with a review of other Neogene species of Balaenoptera (Cetacea: 
Mysticeti). Marine Mammal Science 2:277-298. 

. 1994. The family Odobenidae: A phylogenetic analysis of fossil 

and living taxa, in A. Berta andT. A. Demere (eds.). Contnbutions 
in marine mammal paleontology honoring Frank C. Whitmore, Jr. 
Proceedings of the San Diego Society of Natural History 29:99- 
124. 

Domning. D. P.. and T. A. Demere. 1984. New material of Hydrodamalis 
cuestae (Mammalia: Dugongidae) from the Miocene and Pliocene 
of San Diego County, California. San Diego Society of Natural 
History Transactions 20:169-188. 

English, A. W. 1980. Structural correlates of forelimb function in fur 
seals and sea lions. Journal of Morphology 151:325-352. 

Erdbnnk, D. P. B„ and P. J. H. van Bree. 1990. Further observations on 
fossil and subfossil odobenid material (Mammalia: Camivora) 



98 



Thomas A. Demere 



from the North Sea. Beaufortia 40:85-101. 

Fay. F. H. 1982. Ecology and biology of the Pacific walrus, Odobenus 
rosmarus divergent Illiger. North American Fauna 74. 

Giffin. E. B. 1992. Functional implications of neural canal anatomy in 
recent and fossil marine carnivores. Journal of Morphology 
2 1 4:357-374. 

Gordon, K. R. 1981. Locomotor behavior of the walrus [Odobenus). 
Journal of Zoology (London) 195:349-367. 

Hanna. G. D. 1926. Paleontology of Coyote Mountains, Imperial 
County, California. Proceedings of the California Academy of 
Sciences, series 4, 14:51-186. 

Hasse, G. 1910. Les morses du Pliocene Poederlien a Anvers. Bulletin 
de la Societe Beige de Geologic de Paleontologie et d'Hydrologie, 
Brussels, Memoire 23:293-322. 

Hennig, W. 1966. Phylogenetic Systematics. University of Illinois Press. 
Urbana, Illinois. 

Hertlein. L. G., and U. S. Grant, IV. 1960. The geology and paleontology 
of the marine Pliocene of San Diego. California. Part 2a. Paleontol- 
ogy. San Diego Society of Natural History Memoir 2a:73-l 33. 

, and — — . 1972. The geology and paleontology of the marine 

Pliocene of San Diego. California, Part 2b. Paleontology. San 
Diego Society of Natural History Memoir 2b: 143— 409. 

Howard, H. 1949. New avian records for the Pliocene of California. 
Carnegie Institution of Washington Publication 584:177-199. 

. 1958. Miocene sulids of southern California. Los Angeles 

County Museum Contributions in Science 25. 

Howell. A. B. 1929. Contribution to the comparative anatomy of the 
eared and earless seals (genera Zalophus and Phoca). Proceedings 
of the United States National Museum 73 (15): 1—142. 

Kellogg. R. 1927. Fossil pinnipeds from California. Carnegie Institution 
of Washington Publication 346:27-37. 

. 1931. Pelagic mammals from the Temblor Formation of the 

Kern River region. California. Proceedings of the California Acad- 
emy of Sciences, series 4, 19:217-397. 

Kew, W. S. W. 1914. Tertiary echinoids of the Carrizo Creek region in 



the Colorado Desert. University of California Publications in Geo- 
logical Sciences Bulletin 8:39-60. 
Kidwell. S. M. 1988. Taphonomic comparison of passive and active 

continental margins: Neogene shell beds of the Atlantic coastal 

plain and northern Gulf of California. Palaeogeography, Palaeo- 

climatology, Palaeoecology 63:201-223. 
Mammerickx. J., and K. D. Klitgord. 1982. Northern East Pacific Rise: 

Evolution from 25 m.y. B.P. to the present. Journal of Geophysical 

Research 87:6751-6759. 
Mitchell, E. D. 1961. A new walrus from the Imperial Pliocene of 

southern California: with notes on odobenid and otariid humeri. 

Los Angeles County Museum Contributions in Science 44. 
. 1968. The Mio-Pliocene pinniped Imagotaria. Journal of the 

Fisheries Research Board of Canada 25:1843-1900. 
Ray, C. E. 1960. Trichecodon huxleyi (Mammalia: Odohenidae) in the 

Pleistocene of southeastern United States. Bulletin of the Museum 

of Comparative Zoology 122:129-142. 
Repenning, C. A. 1976. Adaptive evolution of sea lions and walruses. 

Systematic Zoology 25:375-390. 
, and R. H. Tedford. 1977. Otarioid seals of the Neogene. tinned 

States Geological Survey. Professional Paper 992. 
Robinette, H. R., and H. J. Stains. 1970. Comparative study of the 

calcanea of the Pinnipedia. Journal of Mammalogy 51:527-541. 
Rutten. L. 1907. On fossil trichechids from Zealand and Belgium. K. 

Akademie van wetenschappen, Amsterdam 10:2-14. 
Schremp. L. A. 1981. Archaeogastropoda from the Pliocene Imperial 

Formation of California. Journal of Paleontology 55:1 123-1136. 
Sivertson, E. 1954. A survey of the eared seals (family Otariidae) with 

remarks on the Antarctic seals collected by N/K "Norvegia" in 

1928-1929. Det Norske Videnskaps-Akademii Oslo 36:1-76. 
Vaughan. T. W. 1917. The reef-coral fauna of Carrizo Creek, Imperial 

County, and its significance. United States Geological Survey 

Professional Paper 98:355-386. 
Wyss, A. R.. 1987. The walrus auditory region and the monophyly of 

pinnipeds. American Museum Novitates 287. 



The Family Odobenidae: 
A Phylogenetic Analysis of Fossil and Living Taxa 

Thomas A. Demere 

Department of Paleontology, San Diego Natural History Museum. P. O. Box 1390, San Diego. California 92112. and Department 

of Biology, University of California. Los Angeles. California 90024 

ABSTRACT. — The modem walrus. Odobenus rosmarus, is a relict species, the lone survivor of a formerly diverse group of odohenid pinnipeds. 
Walruses had evolved in the North Pacific by at least the middle Miocene and were moderately diverse (five sympatric species) by the late Miocene. 
Even as recently as 3 million years ago there were at least three contemporaneous species of odobenids, two in the North Pacific and at least one in 
the North Atlantic. The Pleistocene record for walruses documents tusked odobenine walruses in both northern ocean basins as late as 500.000 years 
ago. 

A phylogenetic (cladistic) analysis of all well-documented fossil and living walruses supports the monophyly of the Odobenidae. There are two 
major odobenid clades. the Dusignathinae {Pontolis, Gomphotaria, and Dusignathus) and the Odobeninae. The latter group contains the archaic 
odobenine Aivukus and a well-supported clade of tusked odobenines here named the Odobenini (Pliopedia, Alachtherium, Prorosmarus, Valenictus, 
and Odobenus). Neotherium and Imagotaria are generalized early odobenids at the base of the walrus radiation. 



INTRODUCTION 

Walruses (family Odobenidae) are represented today by a single 
living species, the holarctic Odobenus rosmarus. Perhaps the most 
characteristic anatomical feature of Odobenus is its pair of elon- 
gated ever-growing upper canine teeth (tusks) found in adults of 
both sexes. A rapidly improving fossil record reveals that these 
unique structures evolved in only a single lineage of walruses, the 
Odobenini, and that walruses enjoyed at least two major radiations. 
As noted by Repenning and Tedford ( 1977). the fossil record thus 
shows that "tusks do not a walrus make." 

With the exception of an unconfirmed report of a walrus skull 
from the Mio-Pliocene Pisco Formation of coastal Peru, all fossil 
walruses are currently known only from Neogene deposits of the 
northern hemisphere. The oldest records are from the North Pacific, 
from middle Miocene deposits of central California correlative with 
the Barstovian North American Land Mammal Age (NALMA). 
Later Miocene (Clarendonian and Hemphillian NALMA correla- 
tive) records include fossils from the western United States (Or- 
egon and California), Mexico (Baja California), and Japan. Plio- 
cene fossils are known from the eastern and western United States 
(Virginia, North Carolina, South Carolina, Florida, Oregon, and 
California), Great Britain, Belgium, and Japan. Pleistocene wal- 
ruses are known from the eastern and western United States (Maine, 
Massachusetts, New Jersey, Virginia, North Carolina. South Caro- 
lina, California, and Alaska), Canada, Great Britain, the Nether- 
lands. France, and Japan. Thus the odobenids' fossil record in- 
cludes specimens collected from Neogene deposits on both shores 
of the North Pacific and North Atlantic oceans. The ranges of fossil 
walruses extend into modern temperate and even subtropical lati- 
tudes (Ray 1960; Repenning 1976: Repenning etal. 1979). suggest- 
ing that the arctic lifestyle of modern Odobenus rosmarus is the 
result of rather recent dispersal and adaptation to boreal conditions. 

Kellogg (1922:46-58) offered the first detailed discussion and 
review of walrus taxonomy and classification. He placed the fossil 
taxa Alachtherium, Prorosmarus, and Trichecodon, along with 
modern Odobenus. in the family Odobenidae and utilized a mor- 
phological series to present a phylogeny for the group. Because 
Kellogg did not recognize the odobenid affinities of such fossil 
pinnipeds as Dusignathus, Pliopedia, and Pontolis his concept of 
the Odobenidae corresponds to what modern workers recognize as 
the more exclusive subfamily Odobeninae. a clade containing pri- 
marily tusked walruses. 

Repenning and Tedford (1977) summarized the state of 
odobenid paleontology and systematics as it was known at the time 
and recognized two subfamilies (Odobeninae and Dusignathinae) 
within an inclusive family Odobenidae (Table 1 ). In the Odobeninae 



they included the fossil taxa Aivukus, Alachtherium, and Proros- 
marus as well as modern Odobenus; in the wholly extinct 
Dusignathinae they included the fossil taxa Neotherium, 
Imagotaria. Dusignathus, Pliopedia. Valenictus. and Pontolis. 

Barnes (1989) proposed a classification (Table 1) without an 
inclusive family Odobenidae, instead recognizing the three sub- 
families Odobeninae [= Odobeninae of Repenning and Tedford 
(1977)]. Imagotariinae (Neotherium, Pelagiarctos, Imagotaria. 
Pontolis). and a restricted Dusignathinae (Dusignathus. Pliopedia, 
Valenictus). He grouped these three subfamilies with desmato- 
phocids, otariids, and "enaliarctids" in an all-inclusive family 
Otariidae. 

Barnes and Raschke (1991) followed the classification of 
Barnes (1989) and emphasized that only members of the 
Odobeninae were walruses, calling "imagotariine" and dusigna- 
thine taxa "walrus-like." Although this distinction may seem merely 
semantic, it suggests the authors' failure to recognize the common 
ancestry of "imagotariine," dusignathine, and odobenine taxa. I 
present evidence for this common ancestry below and so use the 
term "walrus" for all fossil and living odobenids. 

The monophyly (sensu Hennig 1966) of the Odobenidae has not 
been explicitly demonstrated. Although previous workers (e.g., 
Kellogg 1922; Repenning and Tedford 1977; Barnes 1989) have 
presented characters useful for differentiating odobenids from other 



Table 1 . Previous classifications of fossil and living 
odobenids. 



Repenning and Tedford (1977) 



Barnes (1989) 



Family Odobenidae 
Subfamily Odobeninae 

Aivukus 

Alachtherium 

Prorosmarus 

Odobenus 
Subfamily Dusignathinae 

Neotherium 

Imagotaria 

Pontolis 

Dusignathus 

Pliopedia 

Valenictus 



Family Otariidae 

Subfamily Odobeninae 

Aivukus 

Alachtherium 

Prorosmarus 

Odobenus 
Subfamily Dusignathinae 

Dusignathus 

Pliopedia 

Valenictus 
Subfamily Imagotariinae 

Neotherium 

Pelagiarctos 

Imagotaria 

Pontolis 



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

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

Proc. San Diego Soc. Nat. Hist. 29:99-123, 1994 



100 



Thomas A. Demere 



pinnipeds, these authors made no clear distinction between primi- 
tive and derived character states. Only shared derived character 
states (synapomorphiesl serve as evidence of a group's monophyly. 
In this report 1 provide the first phylogenetic (cladistic) analysis 
of odobenids and discuss highlights of walrus evolution as illumi- 
nated by the resulting phylogenetic framework. In doing so 1 also 
review the fossil record of odobenids (especially specimens and 
work published since 1977). emphasizing those features most rel- 
evant to phylogenetic reconstructions (i.e., synapomorphies). 

SOURCES OF DATA 

Fossil odobenids typically are represented by fragmentary ma- 
terial, most often isolated postcrania. Partial and complete skulls, as 
well as lower jaws, are also known. Unfortunately, many skeletal 
elements are not widely represented among named taxa, hampering 
comparisons. Important exceptions include the nearly complete 
holotype skeletons of Gomphotaria pugnax (see Barnes and 
Raschke 1991) and Valenictus chulavistensis (see Demere 1994, 
this volume), as well as type and referred material of Aivukus 
cedrosensis and Imagotaria dowsii (see Repenning and Tedford 
1977). 

Morphological data presented in this report are in many cases 
based on direct examination of fossil specimens in museum collec- 
tions. I examined undescribed material of Desmatophoca and 
Pinnarctidion currently understudy by A. Berta. undescribed mate- 
rial of Neotherium under study by L. G. Barnes, undescribed mate- 
rial of Prorosmarus under study by C. E. Ray. and new material of 
Pontolis. 

Morphological data were also obtained from published reports, 
for Aivukus from Repenning and Tedford ( 1977); for Alachtherium 
from Hasse (1910), Rutten (1907), Berry and Gregory (1906). 
Kellogg (1922), and Erdbrink and van Bree (1990); for 
Desmatophoca from Barnes (1972, 1987) and Repenning and 
Tedford (1977); for Dusignathus from Kellogg (1927) and 
Repenning and Tedford ( 1977); for Enaliarctos from Mitchell and 
Tedford (1973), Berta and Ray (1990), and Berta (1991); for 
Gomphotaria from Barnes and Raschke (1991); for Imagotaria 
from Mitchell ( 1968). Repenning and Tedford ( 1977), and Barnes 
(1989); for Neotherium from Kellogg (1931) and Barnes (1988); 
for Odobenus from Ray ( 1960) and Fay ( 1982); for Pliopedia from 
Kellogg (1921) and Repenning and Tedford (1977); for 
Pinnarctidion from Barnes ( 1979); for Pontolis from True ( 1909) 
and Repenning and Tedford (1977); for Prorosmarus from Berry 
and Gregory (1906) and Repenning and Tedford (1977); for 
Pteronarctos from Barnes (1989, 1990); and for Valenictus from 
Mitchell ( 1 96 1 ), Repenning and Tedford ( 1 977). and Demere (1994, 
this volume). 

Recent workers (Rowe 1988; de Queiroz and Gauthier 1990) 
have called for rigorous use of the terms "definition" and "diagno- 
sis" in systematic biology. In their usage, definition of a taxon is 
based upon ancestry and taxonomic membership, and diagnosis is a 
listing of shared derived homologous characters (synapomorphies) 
and the level of generality at which they occur. I use a third term. 
characterization, to refer to a listing of distinguishing characters, 
both shared derived and shared primitive homologous characters, 
and their taxonomic distribution. 

Geologic ages and biostratigraphic correlations of fossil 
odobenids discussed here are modified from Ray ( 1 976), Repenning 
and Tedford (1977), and Barnes (1989). Postcanine tooth is abbre- 
viated Pc. Institutional acronyms for cited fossil specimens include 
BMNH, British Museum of Natural History, London, England; 
GMAU, Geological Museum of Amsterdam University, 
Amsterdam, Netherlands; IGCU, Instituto de Geologia, Ciudad 
Universitaria, Universidad Nacional Autonoma de Mexico, Mexico 



City, Mexico; IRSNB, Institut Royal des Sciences Naturelles de 
Belgique. Antwerp. Belgium; LACM. Section of Vertebrate Pale- 
ontology, Natural History Museum of Los Angeles County, Los 
Angeles. California; MCZ. Museum of Comparative Zoology, 
Harvard University. Cambridge, Massachusetts: NSM-PV, Verte- 
brate Paleontology, National Science Musuem, Tokyo, Japan; 
SBNHM. Department of Geology, Santa Barbara Natural History 
Museum. Santa Barbara. California; SDSNH, Department of Pale- 
ontology, San Diego Natural History Museum, San Diego. Califor- 
nia; UCMP. Museum of Paleontology, University of California, 
Berkeley. California; UCR, Department of Geological Sciences, 
University of California. Riverside, California; USNM. Depart- 
ment of Paleobiology, National Museum of Natural History, 
Smithsonian Institution, Washington, D.C. 

PHYLOGENETIC ANALYSIS 

Fifty-three skeletal characters (25 binary and 28 multistate) 
were scored for nine species/genera of fossil and extant odobenids. 
In addition, characters were scored for six outgroup taxa. The 
character-taxon matrix is presented in Table 2. As Rowe ( 1988) has 
suggested, I omitted incomplete and/or poorly known fossil taxa 
from the computer analysis but later added them to the phylogeny 
manually, using synapomorphies observed in the incomplete mate- 
rial. Taxa falling into this category include Dusignathus santa- 
cruzensis, Pliopedia pacifica, and Prorosmarus alleni. Kam- 
tschatarctos, Pelagiarctos, and Prototaria, included by other work- 
ers in the Odobenidae. were excluded from this analysis because of 
their extreme incompleteness or their inaccessibility to me. 

Character polarity within the Odobenidae is based on compari- 
son with a series of successive outgroups (Maddison et al. 1984). 
These polarity decisions were used to construct a hypothetical 
ancestor. To test this procedure. I also had computer alogrithms 
decide global polarities for the ingroup (Swofford 1993). Both 
techniques yielded the same results. Outgroup taxa used were 
Enaliarctos, Pteronarctos, Otariidae (sensu Repenning and Tedford 
1977). Pinnarctidion. Desmatophocidae (sensu Repenning and 
Tedford 1977), and Phocidae. Assumption of either of the compet- 
ing hypotheses of pinniped relationships, monophyly (Wyss and 
Flynn 1993; Berta and Wyss 1994, this volume) or diphyly (Barnes 
1989), did not affect assessment of polarities within the 
Odobenidae. 

Enaliarctos, the earliest and least divergent pinniped, retains 
many primitive arctoid features (Barnes 1989; Berta 1991). 
Pteronarctos. believed to be the sister taxon to all other pinnipeds, 
shares the derived orbital-wall morphology (i.e., maxilla forming 
anterior margin of orbit) of this group (Berta 1991 ). Pinnarctidion 
is closely related to the desmatophocids (Barnes 1989; Berta and 
Wyss 1994, this volume), considered to be the sister taxon of 
odobenids (Repenning and Tedford 1977; Berta and Wyss 1994. 
this volume). 

The data were analyzed on a Macintosh SE computer using 
PAUP( version 3.1.1; Swofford 1993) with all but characters 33 and 
37 (see below under Dentition) treated as unordered. Unweighted 
and weighted treatments of the data were explored. In the latter 
case, all characters were assigned weights scaled to the number of 
possible state changes ( 12 for two states. 6 for three, 4 for four, and 
3 for five). Computer runs were made with a hypothetical ancestor 
used to root the trees. In another series of computer runs, global 
polarity decisions made by the computer algorithm were used to 
root the trees. Global polarity decisions at the ingroup node were 
the same whether all outgroup taxa were used or whether only 
Enaliarctos and the Desmatophocidae were used. Use of only two 
outgroup taxa had the advantage of producing shorter trees with 
fewer homoplastic character arrangements. 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



101 



Table 2. Character-taxon matrix showing the distribution of skull, dental, mandibular, and postcranial characters among fossil and 
modern odobenids. Characters scored represent the ancestral state; characters scored 1—4 represent derived states. Missing data are 
scored as ?. 







1 


2 


3 


4 


5 


6 


7 


8 


9 


1 


1 1 


1 2 


1 3 


1 4 


1 5 


1 6 


1 7 


1 8 


1 9 


2 


2 1 


2 2 
















































1 


Enaliarctos spp. 

















7 


















































2 


Pteronarctos sp. 




































































3 


Otariidae 





0&1 


2 


2 


1 


2 


2 


0&1 


0&1 


0&2 








0&1 

















S 2 











4 


Pinnarctidion spp. 

















? 
































1 

















5 


Desmatophocidae 







1 








3 
































1 





08,1 


2 








6 


Phocidae 







1 





OS t 


2&3 




0&1 














2 

















1 


2 








7 


Neotherium mirum 

















1 

















































8 


Imagotaria spp. 










2 




1 




1 













1 




























9 


Pontolis magnus 







1 


2 




1 


















1 




1 





2 





2 











1 


Gomphotaria pugnax 








1 


2 




1 

















? 


7 








? 





2 











1 1 


Dusignathus spp. 







1 


2 




1 











? 


7 


? 


7 








2 

















1 2 


Aivukus cedrosensis 







7 


1 




1 











? 


7 


? 


7 








2 


? 


7 


1 


1 





1 3 


Alachlherium cretsii 


1 




7 


1 




1 






2 


1 


1 


1 


2 




1 


1 


2 


1 


1 


1 


? 


1 


1 4 


Valenictus chulavistensis 


1 







1 




1 






2 


1 


1 


1 


2 




1 


1 


2 


1 


1 


1 


1 


1 


1 5 


Odobenus spp 


2 


2 





1 




1 






2 


1 


1 


1 


2 




1 


1 


2 


1 


1 


1 


1 


1 







2 3 


2 4 


2 5 


2 6 


2 7 


2 8 


2 9 


3 


3 1 


3 2 


3 3 


3 4 


3 5 


3 6 


3 7 


3 8 


3 9 


4 


4 1 


4 2 


4 3 


4 4 
















































1 


Enaliarctos spp. 




































































2 


Pteronarctos sp. 












































1 





7 


? 


? 





? 


? 


3 


Otariidae 





0S1 


























OS 1 


0&1 


2 


2 


1 &2 























4 


Pinnarctidion spp 

















? 





? 

















1 








? 


? 


? 





7 


? 


5 


Desmatophocidae 






































1 &2 


2 


1 S2 























6 


Phocidae 





0&1 


0&1 























1 


0&1 


5 


2 


1 























7 


Neotherium mirum 



































1 


1 


1 


1 























8 


Imagotaria spp. 





7 


n& 1 


























1 


1 


1 


2 























9 


Pontolis magnus 





? 

















1 











1 


? 


? 


2 


1 














1 


1 


1 


Gomphotaria pugnax 





1 


2 














1 











1 


2 


2 


2 


1 




















1 1 


Dusignathus spp. 





1 


1 


1 








1 


1 











2 


2 


2 


2 


1 


1 








1 


1 


1 


1 2 


Aivukus cedrosensis 


7 





1 








1 








? 





1 


1 


3 


2 


3 


1 

















? 


1 3 


Alachtherium cretsii 


1 


7 


1 


1 


1 


2 








1 


1 


2 


2 


3 


2 


3 


2 








1 


1 





1 


1 4 


Valenictus chulavistensis 


1 


2 


3 


3 


1 


3 








2 


2 


3 


3 


4 


3 


4 


3 


1 





1 


1 





1 


1 5 


Odobenus spp. 


1 


2 


2 


2 


1 


2 


1 





1 


1 


2 


2 


3 


2 


3 


2 


1 


1 


1 


1 













4 5 


4 6 


4 7 


4 8 


4 9 


5 


5 1 


5 2 


5 3 


















1 


Enaliarctos spp. 


























? 


2 


Pteronarctos sp. 


7 


7 


? 


7 


? 


? 


? 


7 


? 


3 


Otariidae 











1 

















4 


Pinnarctidion spp. 











? 


? 


? 





? 


? 


5 


Desmatophocidae 











1 








1 


1 


1 


6 


Phocidae 




















2 








7 


Neotherium mirum 










2 


1 


1 


1 


1 


? 


8 


Imagotaria spp. 










2 


1 


1 


1 


1 


? 


9 


Pontolis magnus 










? 


? 


1 


1 


1 


1 


1 


Gomphotaria pugnax 










2 


1 


1 


1 


1 


1 


1 1 


Dusignathus spp. 










2 


? 


? 


? 


? 


? 


1 2 


Aivukus cedrosensis 










2 


2 


1 


? 


? 


? 


1 3 


Alachtherium cretsii 


1 







2 


2 


1 


? 


7 


? 


1 4 


Valenictus chulavistensis 


2 


1 




2 


2 


1 


1 


1 


1 


1 5 


Odobenus spp. 


2 







2 


2 


1 


1 


1 


1 



Employing the branch-and-bound option in PAUPand varying 
two factors (i.e., outgroup content and character weight) produced 
the following results: 

Hypothetical ancestor/unweighted data — Nine most parsimoni- 
ous trees of 101 steps and consistency index of 0.861. 

Hypothetical ancestor/weighted data — Six most parsimonious 
trees of 746 steps and consistency index of 0.834. 



Two outgroup taxa/unweighted data — Three most parsimoni- 
ous trees of 108 steps and consistency index of 0.830. 

Two outgroup taxa/weighted data — Two most parsimonious 
trees of 786 steps and consistency index of 0.830. 

The same major groups were recognized in all trees, with differ- 
ences in topology involving arrangement of terminal taxa within 
the Dusignathinae and Odobenini. A strict (Nelson) consensus tree 



102 



Thomas A. Demere 



summarizing these topologies is presented in Figure 1 . Figure 2 is a 
composite ciadogram combining the 50% majority-rule consensus 
trees with the manually plotted incomplete fossil taxa. 

The results of the PAUP analysis were exported to MacClade 
version 3.0 (Maddison and Maddison 1992) and examined for 
patterns of character evolution within the most parsimonious to- 
pologies. Various alternate hypotheses were tested and compared 
with the PAUP-generated hypotheses. This allowed empirical 
evaluation of the robustness of the various phylogenetic hypothesis. 

Odobenid Monophyly 

The result of this computer-assisted phylogenetic analysis is a 
well-supported hypothesis of odobenid monophyly based on cra- 
nial, dental, and postcranial synapomorphies. Within the 
Odobenidae. a series of monophyletic groups can be recognized. 

Odobenidae. — Diagnosis of this family is based upon five un- 
equivocal synapomorphies (numbers refer to characters as dis- 
cussed under Character Evidence below): (6) antorbital processes 
constructed from both frontal and maxilla, (47) distal trochlea of 
humerus with diameter of medial lip greater than diameter of ca- 
pitulum, (48) distal portion of radius with enlarged radial process, 
(49) insertion for pollicle extensor muscle on first metacarpal de- 
veloped as a pit or rugosity, and (50) scapholunar with distinct pit 
for the magnum. 

The Odobenidae are defined as the monophyletic group con- 
taining the most recent common ancestor of Neotherium and 
Odobenus and all of its descendants. The family includes Aivukus, 
Alachtherium, Dusignathus, Gomphotaria, Imagotaria, Neo- 
therium, Odobenus. Pliopedia. Pontolis, Prorosmarus, and Valenic- 
tus (Table 2). 

So defined, the Odobenidae are the same as recognized by 



Repenning and Tedford ( 1977), who characterized this taxon by a 
number of features, many of which I have used also. Repenning and 
Tedford, however, used other characters that are not applicable at 
the level of the Odobenidae but are derived at a more general level 
within the Pinnipedia (e.g., skull without prominent supraorbital 
processes, also seen in desmatophocids), are synapomorphies of 
taxa within the Odobenidae (e.g., postcanine tooth roots simple and 
peglike, seen only in dusignathines and odobenines; I, medial to C,, 
seen only in odobenines; femoral head distinctly higher than greater 
trochanter, seen only in Odobenus), or are characters whose distri- 
bution within the Pinnipedia is not well known (e.g., occipital 
condyles widely flaring, bony eustachian canal relatively large, and 
ratio of areas of tympanic membrane and oval window). 

Barnes ( 1989) offered a phylogenetic hypothesis for "otarioids" 
that also included a listing of features characteristic of specific taxa. 
Although Barnes did not formally recognize an inclusive 
Odobenidae, his branching diagram depicts such a grouping (Barnes 
1989: fig. 9, node 12). Some of Barnes' characters I have used as 
well, but Barnes also used characters symplesiomorphic at the level 
of the Odobenidae (e.g., large lesser trochanter of femur), characters 
homoplastic within the Pinnipedia [e.g.. trochanteric fossa of femur 
lost (independently lost in several pinniped groups), optic foramen 
in posteroventral position (also seen in otariids)], and characters that 
are difficult to define (e.g.. radius and ulna shortened). 

I recognize three characters initially considered to represent 
odobenid synapomorphies but now seen to be more generally dis- 
tributed and supportive of a sister-group relationship between 
odobenids and desmatophocids. These characters include (51) as- 
tragalus with calcaneal tuber, (52) calcaneal tuber of calcaneum 
with prominent tuberosity, and (53) entocuneiform strongly over- 
riding the medial articular facet of the mesocuneiform. In all cases 
the derived state occurs also in Allodesmus. 




Odobenidae 



Odobeninl 



Odobeninae 



Figure 1. Strict-consensus tree of proposed relationships among fossil and living odobenids (characters discussed in text). 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



103 



^ 






/ 
j> «.° v *v 

c& ,e> o^ 



fS? 






^ 



^ 



^ 



c?- cnP ,<y <nP' 

^ X *' ^ ^ / „/ ./ > 



^ • ^ </* </* f / ^ ^ / y y y 



s» j*e * & 






# ^ 2* 






Odobenini 




Odobeninae 



Odobenidae 



Figure 2. Fifty-percent majority-rule consensus tree with incomplete odobenid taxa added (dashed lines). 



"Imagotariinae" . — This taxon is paraphyletic, its members, 
Neotherium and Imagotaria, possessing the basal synapomorphies 
of the Odobenidae but lacking the synapomorphies of the 
dusignathine and odobenine walruses. Characterization of these 
"primitive" walruses is based primarily on characters they lack (i.e., 
the derived states of characters 38, 39, 40, and 46). 

Imagotaria + Dusignathinae + Odobenini. — This group is sup- 
ported by five unequivocal synapomorphies: (4) frontal/maxilla 
suture straight and divergent, (5) antorbital process large, (9) palate 
arched transversely, (13) pterygoid strut broad, and ( 14) basioccipi- 
tal broad and pentagonal. Two equivocal synapomorphies also po- 
tentially diagnose this clade: (8) infraorbital foramen enlarged (re- 
versed in Pontolis) and (37) P 4 with bilobed roots (reversed in 
Pontolis). 

Dusignathinae + Odobeninae . — A monophyletic group con- 
taining odobenine and dusignathine walruses is supported by two 
unequivocal synapomorphies: (17) paramastoid process flattened 
and (38) postcanine tooth enamel thin or lost. Two equivocal syna- 
pomorphies are potentially diagnostic of this clade: (36) P 4 
protocone shelf strongly reduced or lost (also seen in phocoids) and 
(37) postcanine teeth single-rooted (also seen in phocoids and some 
otariids). 

Dusignathinae . — Diagnosis of this taxon is based upon one 
unequivocal synapomorphy: (30) upper and lower canines enlarged 
as tusks. In addition three equivocal synapomorphies may diagnose 
this taxon: (3) nasal/frontal suture V-shaped and posteriorly di- 
rected (a more acute "V" occurs in desmatophocids and phocids), 
(19) sagittal crest enlarged (reversed in Dusignathus; also seen in 
certain otariids), and (24) orbital vacuity anteriorly placed (also 
occurs in otariids). 



The Dusignathinae are defined as the monophyletic group con- 
taining the most recent common ancestor of Pontolis and 
Dusignathus and all of its descendants and include Gomphotaria 
also. 

The subfamily Dusignathinae so defined is more exclusive than 
that of Repenning andTedford ( 1977) and except for the addition of 
Pontolis nearly the same as that of Barnes and Raschke ( 1 99 1 ), who 
considered Pontolis an "imagotariine" walrus from the limited 
features preserved on the holotype braincase. Recognition and re- 
ferral of additional cranial as well as postcranial material provides a 
more complete understanding of this taxon. 

Odobeninae. — Diagnosis of this taxon is based upon seven 
unequivocal synapomorphies: (4) frontal/maxillary suture trans- 
versely directed, (21) postorbital process of jugal dorsoventrally 
expanded, (28) C, less than 75% the size of C\ (33) postcanine 
teeth reduced to five, (35) P 3 and P 4 with simple peglike crowns, 
(37) P 4 with single circular root, and (49) first metacarpal with 
insertion for pollicle extensor muscle developed as a rugosity. In 
addition, one equivocal synapomorphy. (31) lower canine 
premolariform, potentially diagnoses this taxon. The crown of the 
lower canine of Aivukus is unknown and may or may not have been 
caniniform. Thus premolariform lower canines may have evolved 
at a lower level of universality and represent a synapomorphy of the 
Odobenini. 

The Odobeninae are here defined as the monophyletic group 
containing the most recent common ancestor of Aivukus and 
Odobenus and all of its descendants. The subfamily thus includes 
Aivukus, Alachtherium, Odobenus, Pliopedia, Prorosmarus, and 
Valenictus. 

Under this definition, the subfamily Odobeninae is nearly the 



104 



Thomas A. Demere 



same as that of Repenning and Tedford ( 1977) and Barnes ( 1989) 
except for the addition of Valenictus and Pliopedia. 

Odobenini (new taxon). — Diagnosis of this new taxon is based 
upon 14 unequivocal synapomorphies: ( 1 ) external narial opening 
elevated above incisive margin. (9) palate arched transversely and 
longitudinally, ( 10) hard palate elongated (also occurs in the otariid 
Otaria), (16) mastoid processes as widest part of cranium, (20) 
zygomatic portion of squamosal blunt and robust. (22) temporal 
fossa shortened. (24) orbital vacuity posteriorly placed. (27) C 1 
with well-developed globular dentine column, (28) C, less than 
409r the size of C. (32) P 1 medial to C, (33) three or four upper 
postcanine teeth, (38) adult postcanine tooth crowns with cementum 
only (no enamel), (41) mandibular terminus vascular, and (45) 
deltoid tubercle of humerus on extreme lateral side of pectoral crest 
or separated from crest. Five equivocal synapomorphies also may 
diagnose this clade: (15) mastoid enlarged (also occurs in Pontolis), 
(26) I' medial to C (also occurs in Dusignathus), (31) C, 
premolariform. (34) tooth row between P 1 and M 1 laterally convex 
(also in Dusignathus), and (42) mandibular arch sharply divergent 
(also occurs in Dusignathus). 

In addition, six characters interpreted by the PAUP analysis as 
derived at the level of the Odobeninae appear a posteriori to repre- 
sent synapomorphies of the Odobenini. These include (11) palatine 
telescoped beneath alisphenoid, (12) hamular process broad, (13) 
pterygoid strut lost, (18) lambdoidal crest with distinct flattened 
traction surface. (19) sagittal crest lost (also variably seen in 
phocoids), and (23) optic foramen funnel-shaped. Although the 
condition of these characters is yet unknown in Aivukus (i.e.. they 
were coded as missing), because most of the six are correlated with 
cranial telescoping (exemplified by Odohenus and Valenictus), the 
absence of telescoping in Aivukus suggests that they are actually 
derived at the level of the Odobenini. 

The Odobenini are defined as the monophyletic group contain- 
ing the most recent common ancestor of Alachtherium and 
Odohenus and all of its descendants. Membership includes 
Alachtherium, Odohenus, Pliopedia, Prorosmarus, and Valenictus. 
Prorosmarus is assigned to this taxon because it possesses derived 
characters 28. 3 1 . and 4 1 . Pliopedia is considered a member of the 
Odobenini on the basis of derived characters 18, 19, and 45. Previ- 
ously, Pliopedia and Valenictus were considered members of either 
the Dusignathinae (Repenning and Tedford 1977) or the 
"Imagotariinae" (Barnes 1989). The type genus of this new tribe is 
Odohenus. 

As defined here, the Odobenini contain all of the odobenine 
walruses with enlarged upper canines of tripartite construction and 
telescoped crania. In this light, the basal odobenine Aivukus is a 
metaspecies (Donaghue 1985) and possible ancestor of the 
Odobenini (i.e., it possesses the odobenine synapomorphies but 
lacks the many synapomorphies of the tusked odobenines). 

Character Evolution 

Dentition. — Odobenids underwent a general evolutionary trend 
toward homodonty that entailed a reduction in the number of roots 
from three to two to one, simplification of postcanine tooth crowns 
from three to two to one cusps, and loss of enamel. A similar 
(except for loss of enamel), convergent pattern of tooth simplifica- 
tion also occurred in otariids (Berta and Demere 1986), desmato- 
phocids (Barnes 1987), and phocids (C. A. Repenning, pers. 
comm.). 

Tusks evolved independently in the Dusignathinae and the 
Odobenini. In the former, enlargement of the upper canines was 
accompanied by enlargement of the lower canines (Repenning and 
Tedford 1977). In the Odobenini, only the upper canines were 
enlarged. This enlargement, and the associated development of a 
central globular dentine column, first evolved in the most recent 



common ancestor of Alachtherium, Valenictus, and Odohenus and 
is strong evidence that modern Odohenus inherited its tusks. In this 
historical light these unique dental structures clearly do not repre- 
sent adaptations to the present arctic range of Odohenus but rather 
are structures transported into this boreal habitat by the temperate 
and/or subtropical ancestor of Odohenus. As discussed by Demere 
( 1994, this volume) the tusks most likely evolved as structures for 
social display under the pressures of sexual selection. 

Other dental evolutionary trends in the Odobenini include re- 
duction in number of postcanine teeth with successive loss of M 1 
and M ; , reduction in number of upper incisors from three to two to 
one (I 1-2 are lost), displacement of I 3 posteriad to a position medial 
to C 1 and in line with the postcanine teeth, and migration of Pc' 
(probably P : ) to a position medial to C 1 . Valenictus chulavistensis 
exhibits the most derived complex of dental characters with loss of 
all teeth, save the upper tusks. 

Cranium. — The most recent common ancestor of Alachtherium, 
Valenictus, and Odohenus had a skull different from that of other 
odobenids. The external nares had moved from a ventral position 
almost level with the tooth row to an elevated position well above 
the tooth row. a modification associated with palatal vaulting and 
the evolution of suction feeding. Elongation of the palate was also 
part of this adaptation to oral suction and involved telescoping of 
the rostrum back and under the anterior portion of the braincase, 
with the posterior border of the palatine reaching a position in line 
with the postglenoid fossa and the orbitosphenoid becoming com- 
pressed in the orbital wall and changing from a horizontally elon- 
gated bone to one that is steeply inclined and marked by a funnel- 
shaped optic foramen. The pterygoid strut was lost with the con- 
comitant shift in the origin for the internal pterygoid muscle to the 
orbital wall of the palatine bone. The hamular process of the ptery- 
goid moved to a more medial position and attained a broad, hori- 
zontally directed form. The temporal fossa was shortened and the 
sagittal crest was lost as the temporalis musculature became re- 
duced. The mastoid bone was greatly enlarged both horizontally 
and ventrally as the neck musculature developed to buttress the 
massive head with its enormous canines. 

Skulls of dusignathine walruses are less specialized than those of 
the Odobenini. but nonetheless have diverged from those of 
Imagotaria and Neotherium. Greatly enlarged sagittal crests in 
Gomphotaria and Pontolis imply a well-developed temporalis mus- 
culature, which together with retention of six postcanine teeth and a 
relatively flat palate suggests a more generalist marine predator than 
the Odobenini. which feed by benthic suction. Dusignathus evolved 
several features convergent with the Odobenini (shortened rostrum, 
convex postcanine tooth row, and fused mandibular symphysis) but 
also retained enlarged lower canines (to go with the large uppers) 
and a strong temporalis musculature. 

Lower jaw. — The lower jaws of odobenids reveal a variety of 
modifications, undoubtedly strongly correlated with the changes 
seen in the skulls. In odobenines the lower jaw evolved a strongly 
upturned symphyseal region that curves medially around the 
enlarged upper canines in the Odobenini. The anterior border of 
the mandible developed into an expanded roughened and pitted 
surface for the mobile lips. This character complex evolved in the 
common ancestor of Alachtherium and Odohenus. Alachtherium 
and Prorosmarus retain two lower incisors and four postcanines. 
In Odohenus the incisors are lost and the postcanine series is 
reduced to only three teeth. As with its upper dentition. Valenictus 
is the most dentally derived of the Odobenini, with loss of all 
lower teeth. Fusion of the mandibular symphysis evolved in the 
common ancestor of Valenictus and Odohenus. In Valenictus the 
symphyseal region is delicate and widest dorsally. In Odohenus 
mandanoensis from the Pleistocene of Japan (as in O. rosmarus), 
the symphysis is buttressed by the addition of bone so that it is 
widest ventrally. Evidently this strongly buttressed symphysis 



The Family Odobemdae: A Phylogenetic Analysis of Fossil and Living Taxa 



105 



evolved in the common ancestor of O. mandanoensis and O. 
rosmarus. 

CHARACTER EVIDENCE 

The following section presents the cranial, dental, and postcra- 
nial characters employed in the cladistic analysis. For each charac- 
ter, alternate character states, taxonomic distribution of states, and 
polarity assessment are discussed. Outgroup taxa include species of 
Enaliarctos, Pteronaretos, Otariidae, Pinnarctidion, Desmatopho- 
cidae (Desmatophoca and Allodesmus), and Phocidae. Some dis- 
cussions also include a posteriori assessments of character evolu- 
tion based upon the distribution of character states in the proposed 
phylogenetic hypothesis (Fig. 1). 

Cranium 

1. External narial opening. = low. 1 = intermediate. 2 = high. 
The external narial opening is low and almost level with the tooth 
row in most pinnipeds (primitive condition). In Odobenus rosmarus 
the external nares are elevated well above the tooth row (Repenning 
and Tedford 1977) (derived condition 2). The nares' position in 
Alachtherium (see Erdbrink and van Bree 1990:pl. IB) and 
Valenictus is intermediate (derived condition 1 ). Elevation of the 
narial opening may be related to the extreme palatal vaulting char- 
acteristic of the suction-feeding Odobenini. Enlargement of the 
upper canines and broadening of the muzzle may also be correlated 
with this elevated position of the external narial opening. 

2. Ascending process of premaxilla, overlap with nasal. = 
long, 1 = short. 2 = none. In Allodesmus (Barnes 1972), 
Desmatophoca (Barnes 1987), and phocids (Wyss 1987) the as- 
cending process of the premaxilla has a very short overlapping 
contact externally with the nasal. In Odobenus rosmarus the two 
bones overlap only within the nasal opening, with the maxilla 
contacting the nasal along its entire external lateral border. In many 
neonates of O. rosmarus an irregular band of premaxilla (ascending 
process) is visible externally, sandwiched between the maxilla and 
nasal. This is roofed over in adults. In fossil odobenids an ascending 
process is always visible externally, where preserved. The length of 
overlap, however, is variable; it is relatively long (premaxilla over- 
laps more than 50^ of nasal) in Gomphotaria and Neotherium, 
short (overlap <50 c /c) in Imagotaria, Dusignathus, Aivukus, and 
Valenictus. Wyss' ( 1987) and Barnes' ( 1992) suggestion that a long 
overlap is the primitive condition is supported by a long overlap in 
Enaliarctos emlongi (see Berta 1991 ). Pteronaretos goedertae (see 
Barnes 1989). and Pinnarctidion (A. Berta, pers. comm.). I postu- 
late that the short overlap evolved independently at least twice, 
once in the desmatophocids and again in the odobenids. The condi- 
tion in Gomphotaria represents a reversal. 

3. Nasal/frontal suture. = transverse. I = V-shaped, 2 = W- 
shaped. The posterior border of the nasal is blunt and nearly trans- 
verse in most outgroup taxa and in Neotherium, Imagotaria, and 
Odobenus (Figs. 3A. E) (primitive condition). The condition in 
Aivukus and Alachtherium is unclear. Although two derived states 
are recognized, the homology of the first, a V-shaped nasal/frontal 
suture (point of V directed posteriorly between frontals. Figs. 3B- 
D), is questionable. Two variations of this condition are seen with 
Pontolis (USNM 314300). Gomphotaria, and Dusignathus having 
a broad V-shaped suture (Fig. 3D) and Desmatophoca, Allodesmus 
(Fig. 3B), and phocids (Fig. 3C) having an acute V-shaped nasal/ 
frontal suture. This distribution suggests the independent evolution 
of character state 1 in phocoids and in dusignathine walruses. A 
second derived condition occurs in living and fossil otariids (e.g., 
Zalophus, Eumelopias, Olaria, and Thalassoleon), in which the 
suture (Fig. 3F) is W-shaped (i.e., the frontals extend anteriorly 
between the nasals; King 1983). 




m 







m 



f 



Figure 3. Rostral suture patterns of selected fossil and living pinnipeds. 
A, Enaliarctos, Pteronaretos, and Neotherium: B, Desmatophoca and 
Allodesmus; C, Phoca, Leptonychotes, and Halichoerus; D, Pontolis, 
Gomphotaria, and Dusignathus; E, Imagotaria and Odobenus; F, 
Eumelopias, Olaria, and Zalophus. Skeletal elements: f. frontal; m, maxilla; 
n, nasal. Not to scale. 



4. Frontal/maxilla suture. = V-shaped, 1 = straight, transverse, 
2 = straight, divergent. In all of the fossil outgroup taxa, as well as 
in Neotherium (LACM 131950), the frontal/maxilla suture is char- 
acterized by a narrow anteriorly directed V-shaped segment of the 
frontal that invades the maxilla immediately adjacent to the nasals 
(Fig. 3A) (primitive condition). Loss of this V-shaped frontal seg- 
ment is derived, and two states can be recognized. Lateral to the 
nasals the suture is either transverse (state 1, Fig. 3E) or 
anterolateral^ divergent (state 2, Fig. 3D) relative to the sagittal 
plane. In Imagotaria and the dusignathines Pontolis. Gomphotaria, 
and Dusignathus, the suture, as it leaves the nasals, is straight and 
sharply divergent relative to the sagittal plane (approximately 62°, 
55°, and 58°. respectively). In the odobenines Aivukus, 
Alachtherium, Valenictus. and Odobenus, the suture is straight and 
nearly transverse (approximately 85°, 80°, 74°. and 85°, respec- 
tively). 

5. Antorbital process. = small/absent. I = large. The primitive 
condition is either a small, weakly developed antorbital process, as 
in Enaliarctos. Pteronaretos. Pinnarctidion, and Neotherium. or no 
process, as in Allodesmus and Desmatophoca. The derived condi- 
tion of a large process occurs in otariids. certain lobodontine 
phocids. and the later-diverging odobenids Imagotaria. Pontolis 
(seen on USNM 335554). Gomphotaria. Dusignathus (seen on 



1()6 



Thomas A. Demere 



SDSNH 38342), Alachtherium (seen on GMAU K-8052), 
Valenictus (seen on SDSNH 38228), and Odobenus. The distribu- 
tion of this character suggests that the large antorbital process 
evolved independently in lobodontine phocids, otariids, and 
odobenids (except Neotherium). 

6. Antorbital process. = constructed from frontal, 1 = con- 
structed from frontal and maxilla, 2 = constructed from maxilla 
only, 3 = absent. The small antorbital process of Pteronarctos 
appears to lie entirely within the frontal bone (Barnes 1990) (primi- 
tive condition). In all fossil and modern odobenids the frontal/ 
maxilla suture splits the antorbital process, which is thus formed 
from both bones. A second, derived condition evolved indepen- 
dently in otariids and lobodontine phocids, in which the antorbital 
process is anterior to the frontal/maxilla suture and entirely within 
the maxilla. The absence of the process in desmatophocids and 
most phocids is assigned state 3. 

7. Supraorbital processes offrontals. = weak, 1 = absent. 2 = 
strong. In all of the fossil outgroup taxa the supraorbital process is 
present but weakly developed (primitive condition). Two derived 
character states are recognized. The supraorbital process is absent 
in fossil and living odobenids and in phocids (derived state 1 ). A 
strongly developed process (derived state 2) does not occur among 
the ingroup but is present in living and fossil otariids (see 
Repenning andTedford 1977; King 1983). 

8. Infraorbital foramen. = small, 1 = large. The infraorbital 
foramen in outgroup taxa and in Neotherium (LACM 131950) is 
small (primitive condition) relative to that seen in Odobenus. En- 
larged foramina are found in all other fossil odobenids. In 
Dusignathus, Alachtherium, Valenictus. and Odobenus the enlarge- 
ment is associated with a shortened rostrum. The shortened rostrum 
of Dusignathus is convergent on the condition seen in the 
Odobenini. In Imagotaria. Gomphotaria, and Aivukus. however, no 
rostral shortening is associated with the large infraorbital foramen. 
The enlarged foramen in Odobenus rosmarus is correlated with 
increased innervation and blood flow to the muzzle with its mus- 
tache. The phocid Erignathus has a well-developed rostral mus- 
tache and a correspondingly large infraorbital foramen. Reversal to 
the primitive condition is hypothesized for Pontolis, with its elon- 
gated rostrum and relatively small infraorbital foramen. 

9. Palate. = flat, 1 = arched transversely, 2 = arched longitudi- 
nally and transversely. The primitive condition of a relatively flat 
palate is invariant in the outgroup taxa and also occurs in 
Neotherium. The palate of Imagotaria is transversely but not longi- 
tudinally arched (derived condition I ). The Odobenini have highly 
vaulted/arched palates (derived condition 2). This vaulting occurs 
in both the transverse and longitudinal planes, with the degree of 
vaulting greatest between the anterior incisors and the end of the 
postcanine tooth row. The vaulted palate is associated with strong 
oral suction, the mode of feeding of the living walrus (Fay 1982). 
The suction-feeding otariid Otaria byronia also has a vaulted/ 
arched palate but the greatest degree of its arching is more posteri- 
orly positioned, between the last postcanine tooth and the internal 
narial opening, suggesting that the tusked odobenines and Otaria 
independently acquired functionally similar but nonhomologous 
arched palates. The distribution of this character suggests that a 
transversely arched palate evolved early in odobenid history and 
that the vaulted palate of the Odobenini is a synapomorphy of this 
group of specialized walruses. 

10. Pidate. = short, 1 = long, with long maxilla and short 
palatine, 2 = long, with long maxilla and long palatine. The primi- 
tive condition of a relatively short palate (Figs. 4A, B) occurs in 
outgroup taxa and in Neotherium. Imagotaria, Gomphotaria, and 
Aivukus. The derived condition of an elongated palate evolved 
independently in two groups of pinnipeds. In the Odobenini (Figs. 
4E-G) only the maxilla is elongated and the rostrum is telescoped 
against and beneath the braincase (derived condition I ). In Otaria 



the palate is formed from both elongated palatines and maxillae and 
is not telescoped beneath the braincase (derived condition 2). 

The length of the palate is evaluated by determining the position 
of the anterior border of the internal narial opening relative to the 
position of the postglenoid fossa. The narial border of elongated 
palates reaches the level of the postglenoid fossa. Elongation of the 
palate is correlated with an increased efficiency of oral suction (Fay 
1982). 

11. Palatine. = abutting alisphenoid, 1 = underlying ali- 
sphenoid. In pinnipeds, the palatine bone generally forms a squa- 
mous suture with the pterygoid bone (which it overlies) and a plane 
suture with the alisphenoid (primitive condition). This condition 
(Fig. 5A) occurs in all outgroup taxa and most odobenids (e.g., 
Neotherium. Imagotaria. Pontolis). The derived condition occurs in 
the tusked odobenines, whose palatine and pterygoid share a plane 
suture and palatine and alisphenoid share a squamous suture. In this 
configuration (Fig. 5B) the alisphenoid almost entirely overlies the 
palatine, with the result that the pterygoid lies entirely posterior to 
the palatine (externally). The condition in Aivukus is unknown, but 
other aspects of the skull (e.g., short palate) suggest that it had the 
primitive condition. 

1 2. Hamular process of pterygoid. = narrow, 1 = broad. Most 
pinnipeds have a hamular process that is transversely compressed, 
anteroposteriorly elongated, and hooked posteroventrally (primi- 
tive condition). In contrast, the enlarged hamular process of the 
Odobenini is transversely broadened, dorsoventrally compressed, 
and flared and hooked posterolaterally. A special condition occurs 
in Otaria. in which the hamular process is transversely compressed 
and projects ventrally. The condition in Aivukus is unknown, but 
other aspects of its skull (e.g., short palate) suggest that it had the 
primitive condition. Character state 1 represents a synapomorphy 
of the Odobenini. 

13. Pterygoid strut. = narrow. 1 = broad, 2 = absent. The 
pterygoid strut (Barnes 1990), defined as the horizontally posi- 
tioned expanse of palatine, alisphenoid, and pterygoid lateral to the 
internal narial opening and hamular process, is the site of origin for 
the internal pterygoid muscle. The primitive condition as seen in 
Enaliarctos, Pteronarctos, and Neotherium is a distinct but narrow 
pterygoid strut. Two derived conditions can be recognized. A broad 
pterygoid strut with a large ventral exposure of the alisphenoid and 
pterygoid occurs in Imagotaria (Fig. 4B) and Pontolis (USNM 
314300, Fig. 4C) (derived condition 1). The pterygoid strut is 
absent in the Odobenini (Figs. 4E-G) (derived condition 2). In the 
latter group, the palatine underlies the alisphenoid (character 11) 
and the muscle attachment is confined to the orbital wall. Loss of 
the pterygoid strut is also seen in phocids and the otariid Otaria. In 
the latter case, the alisphenoid lies posterior to the palatine and the 
hamular process forms a vertical surface continuous with the orbital 
wall. In both cases the palate is extended posteriorly, an apparent 
adaptation for strong oral suction. 

14. Basioccipital. = narrow and parallel-sided, 1 = broad and 
pentagonal. The primitive condition occurs in the outgroup taxa as 
well as in Neotherium (LACM 131950). A basioccipital that is 
relatively short, broad, and pentagonal represents the derived con- 
dition and occurs in many odobenids. e.g., Imagotaria (USNM 
335594), Pontolis (USNM 3792), Valenictus (SDSNH 38227), and 
Odobenus. Expressing the width of the basioccipital as a percentage 
of its length provides a means for evaluating this character. In 
Odobenus the width of the basioccipital is about 125% of its length, 
in Pontolis I0()9f. in Neotherium about 80%, and in Enaliarctos 
mealsi about 83%. The derived condition is defined as a ratio 
greater than 90%. 

15. Mastoid process. = small, 1 = large. The mastoid process is 
primitively small in the outgroup taxa and in Neotherium, 
Imagotaria, Gomphotaria, and Aivukus. The derived condition of a 
greatly enlarged mastoid process constructed internally of cancel- 





s 



^ 











Figure 4. Ventral aspect of skulls of selected fossil and living pinnipeds. A. Enaliarctos mealsi (Jewett Sand, early Miocene, after Tedford 1976); B, 
lmagotarki downsi (Santa Margarita Formation, late middle Miocene, after Repenning and Tedford 1977); C. Pontolis magma (Empire Formation, late late 
Miocene); D, Aivukus cedrosensis (Almejas Formation, late late Miocene, after Repenning and Tedford 1977); E, Alachtherium cretsii (Scaldisian sands, 
early Pliocene, after Hasse 1910); F, Valenictus chulavistensis (San Diego Formation, late Pliocene); G, Odobenus rosmarus (Recent). Scale bar, 5 cm. 



108 



Thomas A. Demere 





Figure 5. PterygoidValisphenoid suture patterns in odobenids. A, taxa 
outside the Odobenini; B, Odobenini. Skeletal elements: al, alisphenoid; pt, 
pterygoid; pa. palatine. Not to scale. 



lous bone occurs in the tusked odobenines Odobenus, Valenictus, 
and Alachtherium. In Pontolis (USNM 314300) the mastoid pro- 
cess is enlarged also but does not descend to the same level as that 
of the Odobenini and presumably enlarged independently. 

Enlargement of the mastoid process is related to an increase in 
mass of the neck musculature of Odobenus and other tusked 
odobenines. 

16. Widest part of skull. = zygomatic arch. 1 = mastoid 
processes. In Enaliarctos, Pinnarctidion, Desmatophoca, and early 
odobenids (e.g.. Neotherium and Imagotaria) the skull is widest at 
the level of the zygomatic arch (primitive condition). In the 
Odobenini the mastoid region is the widest part of the skull (Figs. 
4E-G). This synapomorphy is probably correlated with enlarge- 
ment of the mastoid process and shortening of the skull. As men- 
tioned. Pontolis has an enlarged mastoid, but its skull is not short- 
ened and consequently the widest part of the cranium is at the 
zygomatic arch. 

17. Paramastoid process. = small, knoblike, 1 = elongated, 
posteriorly directed, 2 = flattened. The paramastoid process of 
Enaliarctos, Pteronarctos, otariids, and phocids as well as 
Neotherium and Imagotaria is primitively small and knoblike 
(Berta 1991). In Pinnarctidion. Desmatophoca, and Allodesmus 
(see Barnes 1987: fig. 9). the process is elongated and posteriorly 
directed (derived condition 1). In the dusignathine and odobenine 
walruses the process is flattened and platelike (derived condition 2). 

18. Lambdoidal crest. = crestlike, 1 = flattened. The 
lambdoidal crest (occipital crest of Repenning and Tedford 
1977:52) in most pinnipeds is a sharp, narrow ridge following the 
parietal/occipital suture from the cranial vertex to the mastoid 
process. Primitively, this crest is narrow and overhangs the occipital 
shield as a posterodorsally directed projection (e.g., in outgroup 
taxa and Neotherium and Pontolis). In Alachtherium (see Erdbrink 
and van Bree 1990: pi. 1), Pliopedia (see Repenning and Tedford 
I977:pl. 24, fig. 6), Valenictus (see Demere 1994, this volume:figs. 
1A, 2A,B), and Odobenus the vertex of the lambdoidal crest is 
marked by a large flattened crescentic traction surface presumably 
for insertion of the splenius musculature (an important neck exten- 
sor). In their reconstruction of the cranium of Aivukus, Repenning 
and Tedford (1977: fig. 1 ) included an incipient flattened traction 
surface on the cranial vertex. However, this region is not well 
preserved on the holotype cranium, and its presence in Aivukus 
cannot be accurately determined. Other aspects of the skull (e.g., 
relatively small mastoid process) suggest that Aivukus had the 
primitive condition. 

19. Sagittal crest. = small, 1 = absent, 2 = large. Outgroup taxa 
as well as most early odobenids (e.g., Neotherium and Imagotaria) 
have a distinct but low sagittal crest (primitive condition). Two 
derived character states are recognized. In the Odobenini the sagit- 
tal crest is completely lost (derived condition 1). In the dusigna- 
thines Pontolis (USNM 395567) and Gomphotaria, the sagittal 
crest is greatly enlarged (derived condition 2). The small sagittal 
crest on the holotype of Dusignathus seftoni (SDSNH 38342) may 



be related to the specimen's not being mature, and it is possible that 
adult males had the enlarged crests seen in Pontolis and 
Gomphotaria. Repenning and Tedford ( 1977) reported that Aivukus 
lacks a sagittal crest. However, the holotype cranium is damaged in 
this area and thus cannot be accurately evaluated. The unreduced 
temporal fossa (character 22) of Aivukus, suggesting well-devel- 
oped temporalis musculature, indicates that this taxon may have 
possessed a sagittal crest. The sagittal crest is variable in several 
species of extant pinnipeds (King 1983). in most cases because of 
sexual dimorphism, males typically having stronger crests than 
females (e.g., Zalophus). The crest in undescribed specimens of 
Desmatophoca varies between small and absent (A. Berta, pers. 
comm.). 

20. Zygomatic process of squamosal. = long/slender, 1 = short 
and robust, 2 = expanded. The zygomatic portion of the squamosal 
is primitively long and slender in outgroup taxa and Neotherium, 
Imagotaria, Gomphotaria. Dusignathus. and Aivukus. Two derived 
states are recognized. The Odobenini possess a shortened and ro- 
bust process that has a plane suture with the jugal (derived state 1). 
Desmatophoca. Allodesmus (see Bames 1972). and phocids (see 
King 1983) have an expanded process with a mortised jugal/squa- 
mosal suture (derived state 2). 

21. Postorbital process of jugal. = small, 1 = dorsoventrally 
expanded. The postorbital process of the jugal is relatively small in 
the outgroup taxa and Neotherium (primitive condition). The dorso- 
ventrally expanded and robust process of Aivukus, Valenictus, and 
Odobenus is a synapomorphy of the Odobeninae. Although the 
zygomatic arch is not preserved in Alachtherium the shortening of 
its temporal fossa suggests that this taxon also had the derived 
postorbital process. 

22. Temporal fossa. = elongate. 1 = shortened. The primitive 
condition of a relatively long temporal fossa occurs in the outgroup 
taxa and many early odobenids. The Odobenini possess the derived 
condition of an anteroposteriorly shortened temporal fossa. Reduc- 
tion in the size of the temporal fossa probably correlates with 
telescoping of the cranium and a decrease in the strength of the 
temporalis muscle. Functionally, this change is associated with 
emphasis in the Odobenini on suction feeding rather than biting. 

23. Optic foramen and orbitosphenoid. = platelike. 1 = funnel- 
shaped. In Enaliarctos (Mitchell and Tedford 1973) and 
Pteronarctos (Barnes 1990) the optic foramen is situated anteriorly 
in the orbitosphenoid, which continues as a plate well anterior to the 
foramen. The primitive condition is also seen in Pinnarctidion. 
Neotherium (LACM 131950), and Imagotaria (USNM 335594), as 
well as in otariids. In the Odobenini. the optic foramen is funnel- 
shaped and lies almost directly above the orbital fissure. In addi- 
tion, the orbitosphenoid is not produced as a plate anterior to the 
foramen. The condition in Aivukus is unknown, but other aspects of 
the skull (e.g., lack of cranial telescoping) suggest that it had the 
primitive condition. 

24. Orbital vacuity: = absent, 1 = present/anteriorly posi- 
tioned. 2 = present/posteriorly positioned. The orbital wall of the 
outgroup taxa, as well as of Neotherium, Imagotaria, and Aivukus is 
a continuous, unbroken surface (primitive condition). Extant pinni- 
peds (i.e., otariids, phocids, and Odobenus). however, as well as 
some fossil taxa (e.g.. Gomphotaria, Dusignathus. and Valenictus) 
possess unossified areas or vacuities in the orbital wall. The occur- 
rence of vacuities is considered derived (Wyss 1987), and two states 
are recognized. In otariids. phocids. and dusignathine walruses the 
vacuities are anteriorly placed, with the maxilla forming the ante- 
rior border of the vacuity (derived state 1 ). In contrast, the orbital 
vacuity in Odobenus and Valenictus is posteriorly placed relative to 
derived state 1 and is bordered anteriorly by a thin plate of the 
palatine (derived state 2). Valenictus possesses the same condition 
seen in Odobenus. The nature of the anterior border of the orbital 
vacuity in Alachtherium is unknown, but is most likely condition 2. 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



109 



The distribution of character state I suggests that anteriorly placed 
vacuities evolved independently in phocids. otariids, and 
dusignathines. 

Dentition 

25. Upper incisors. = I'~ 3 , 1 = I 2 " 3 only, 2 = L only. 3 = incisors 
lost. Outgroup taxa, as well as early odobenids (e.g., Neotherium), 
have the primitive condition of three upper incisors per side (Figs. 
4A-C). Three derived character states are identified. Reduction to 
two upper incisors (i.e., loss of I') occurs in Dusignathus, Aivukus. 
and Alachtherium (derived condition I ). Imagotaria displays a 
polymorphism with three incisors (Fig. 4B| in the holotype 
(SBNHM 342) and a referred female (USNM 23858: Repenning 
and Tedford 1977) but only two incisors (I 1 lost) in a juvenile male 
(Repenning and Tedford 1977:pl. 8, fig. 1) and an undescribed 
cranium from the Empire Formation (USNM 335599). 

Reduction to only one incisor (I 1, : lost) in Gomphotaria and 
adult Odobenus represents a second derived state (Fig. 4G). A 
polymorphism also crops up in Odobenus, in which I 2 is sometimes 
present (Fay 1982). Loss of all incisors in Valenictus (Fig. 4F) is an 
autapomorphy of this taxon (derived condition 3). 

The distribution of this character indicates independent loss of 

1 1 in "monachine" seals and later-diverging odobenids (e.g., 
Imagotaria + dusignathines + odobenines) and independent loss of 

1 2 in Odobenus and Gomphotaria. The condition in Pontolis is 
viewed as a reversal. In the Odobenini loss of incisors is correlated 
with an emphasis on suction feeding. 

26. Position of P. = anterior to C, 1 = medial to C (anterior to 
incisive foramen). 2 = medial to C 1 (lateral to incisive foramen). 3 = 
I' absent. In the outgroup taxa and early odobenids (e.g., Neo- 
therium and Imagotaria) the upper incisors, especially I 1 : , are 
positioned at the anterior border of the premaxilla (Figs. 4A. B) 
anterior to the canine (primitive condition). In Dusignathus 
(SDSNH 38342) and Alachtherium (GMAU K-8052) the "incisors 
are medial to the anterior half of the canine (Fig. 4E) but anterior to 
the incisive foramen (derived condition 1 ). In adult Odobenus, I 1,2 
are typically lost and I 3 is more posteriorly located on the medial 
side of the enlarged canine (Fig. 4G). lateral to the incisive foramen 
(derived condition 2). Loss of all upper incisors (Fig. 4F) is an 
autapomorphy of Valenictus. 

The distribution of these character states suggests that derived 
condition 1 evolved independently in Dusignathus and the 
Odobenini. 

27. Globular dentine in C 1 . = absent, 1 = present. Ray ( 1960) 
described the unique structure of the ever-growing tusks of 
Odobenus (O. rosmarus and O. hu.xleyi), noting the central column 
of globular dentine surrounded first by a thick ring of orthodentine 
and then by a thin outer layer of cementum. Enamel is lacking in 
adult Odobenus tusks (Fay 1982). This unique dental structure is 
also seen in Valenictus (see Demere 1994, this volume) and is 
inferred for Alachtherium on the basis of isolated tusks from the 
Antwerp Pliocene and for Prorosmarus on the basis of isolated 
tusks from the Yorktown Formation (C. E. Ray, pers. comm.). 
Although Gomphotaria also has enlarged upper canines, computed 
tomography images reveal a thin outer layer of cementum overly- 
ing a thick core of orthodentine. with no evidence of a central 
globular dentine column. Barnes and Raschke (1991) also noted 
patches of enamel on the tusks of Gomphotaria. The canines of 
Aivukus lack globular dentine (Repenning and Tedford 1977). Pos- 
session of a central column of globular dentine is a synapomorphy 
of the Odobenini. 

28. Size ofC, relative to C'. = nearly equal (100-80%). I = 
reduced (75-45%), 2 = very reduced (40-20% ), 3 = C, absent. The 
anteroposterior diameter of the lower canines can be expressed as a 
percentage of the anteroposterior diameter of the upper canines. In 



Enaliarctos emlongi this measure is 104%, in Imagotaria downsi 
84-86%. in Dusignathus santacruzensis 85%, in Gomphotaria 
pugnax 86%, in Pontolis magnus 89%, in Aivukus cedrosensis 65%, 

and in Odobenus rosmarus 29-31%. Nearly equal (100-80%) up- 
per and lower canines represent the primitive condition. Two de- 
rived states may be recognized. Lower canines reduced to between 
75% and 45% of the upper canines (e.g., Aivukus) represent derived 
condition 1; further reduction (40-20%) (e.g., Odobenus) repre- 
sents derived condition 2. Although no associated upper and lower 
dentitions are known for Alachtherium, it is clear that the lower 
canines of the holotype dentary of A. crelsii are considerably smaller 
than the greatly enlarged upper canines inferred from empty alveoli 
in the known crania. The same is almost certainly true for 
Prorosmarus, in which the lower canine is small and the lateral 
concavity of the dentary (in occlusal aspect) suggests enlarged 
upper canines. Character state 1 is an odobenine synapomorphy: 
character state 2 is a synapomorphy of the Odobenini. Loss of C, is 
an autapomorphy of Valenictus. 

29. Upper canine. = procumbent, 1 = steeply inclined. In the 
outgroup taxa and Neotherium, Imagotaria, Gomphotaria, 
Alachtherium, and Valenictus the upper canine is strongly to moder- 
ately procumbent (primitive condition). In Odobenus and 
Dusignathus seftoni (see Demere 1994. this volume) the canines are 
more vertically oriented (derived condition). This derived state 
evidently evolved independently in the two taxa. Such convergence 
between Dusignathus seftoni and Odobenus rosmarus is also seen 
in characters 34, 39. and 42. 

30. C' and C, enlarged as tusks. = no, 1 = yes. In the outgroup 
taxa and Neotherium and Imagotaria the upper and lower canines 
are not enlarged relative to the adjacent postcanine teeth (primitive 
condition). Enlargement of both upper and lower canines is a 
synapomorphy of the dusignathines. As mentioned (character 27), 
the upper canines of these taxa are constructed differently from 
those of the tusked odobenines. Although Dusignathus 
santacruzensis has relatively elongated upper and lower canines of 
equal size (Repenning and Tedford 1977). they are not enlarged as 
tusks. In D. seftoni, however, upper and lower tusks are developed. 
Evaluation of this character is based on the size difference between 
the diameters of the canines and those of adjacent postcanine al- 
veoli. 

31. Lower canine. = caniniform, 1 = premolariform, 2 = 
absent. Outgroup taxa and Neotherium, Imagotaria. and 
Dusignathus primitively possess caniniform lower canines. 
Premolarization, in Odobenus, Alachtherium, and Prorosmarus (see 
Berry and Gregory 1906). involves loss of the enamel crown and 
carina, transverse expansion of the root, and shortening and simpli- 
fication of the crown. Because the condition in Aivukus is currently 
unknow n. premolarization of the canine may represent a synapo- 
morphy of the Odobeninae or more exclusively of the Odobenini. 
Loss of the lower canines is an autapomorphy of Valenictus. 

32. Position of Pc 1 . = posterior to C 1 , 1 = medial to C 1 , 2 = 
absent. Primitively, the position of the first upper postcanine tooth 
(Pc 1 ) in carnivorans is posterior to the upper canine (Figs. 4A-C). 
In Alachtherium (GMAU K-8052 and the Hasse 1910 specimen) 
and Odobenus, Pc 1 has moved to a position medial to the canine 
(Figs. 4E. G). The distribution of this character suggests that condi- 
tion 1 is a Odobenini synapomorphy. with the loss of all lower 
postcanine teeth being an autapomorphy of Valenictus (Fig. 4F). 

33. Upper postcanine teeth. = six, I = five. 2 = three or four. 3 
= zero. Outgroup taxa and Neotherium. Pontolis, and Dusignathus 
have the primitive pinniped dental formula of six postcanine teeth 
(Barnes 1989:Berta 1991). Reduction to five teeth (i.e.. loss of M : ) 
as in Aivukus (derived condition I ). to three or four teeth as in 
Alachtherium and Odobenus (derived condition 2), or zero as in 
Valenictus (condition 3) represents three derived states. Although 
the type of Gomphotaria pugnax has only five postcanine teeth, a 



110 



Thomas A. Demere 



referred rostrum (JMTC 907-170) has alveoli for six teeth, suggest- 
ing that the condition in this species is variable. 

34. P'-M'. postcanine tooth row: = laterally concave. 1 = 
straight. 2 = laterally convex, 3 = postcanine teeth absent. In the 
outgroup taxa. the upper postcanine tooth row forms a sigmoidal 
curve with the P'-M 1 portion laterally concave in occlusal view 
(Barnes 1989). In Neotherium, Imagotaria (Fig. 4B), Aivukus. 
Gomphotaria, and Pontolis (Fig. 4C, D) the tooth row is nearly 
straight. The tooth row is also variably straight in phocids and 
otariids. In Alachtherium (Fig. 4E), Odobenus (Fig. 4G), and 
Dusignathus (see Demere 1994, this volume) the tooth row is 
laterally convex. Loss of all postcanine teeth is an autapomorphy of 
Valenictus. 

This distribution suggests that a straight tooth row evolved 
several times, once in the common ancestor of Neotherium and all 
other odobenids, and again in certain otariids. A convex tooth row 
evolved twice, once in the Odobenini and once in Dusignathus. 
This convergence towards a laterally convex tooth row is probably 
a function of the rostral shortening seen in both Dusignathus and 
the tusked odobenines. 

35. P 3 and P 4 , crowns. = three cusps, 1 = two cusps, paracone 
emphasized, 2 = one cusp only, 3 = simple peglike crown, 4 = teeth 
absent, 5 = complex labial cusps. The crowns of P' and P 4 of 
Enaliarctos mealsi, E. emlongi, and E. barnesi are characterized by 
a well-developed paracone. a distinct metacone. and a strong 
protocone shelf that is anteromedially placed (Barnes 1989; Berta 
1991). In Pinnarctidion (USNM 314325), P 4 has a well-developed 
paracone. a strong metacone, and a well-developed posteromedial 
protocone shelf (A. Berta. pers. comm.). In Desmatophoca, P 4 has a 
well-developed central cusp (= paracone), a small posterolateral 
cuspule (metacone?), and a narrow cingulum rimmed by tiny lin- 
gual cuspules. In Neotherium, P 1 has a well-developed paracone, a 
distinct but reduced metacone. and a posteromedial shelf with a 
well-developed lingual cuspule (L. G. Barnes, pers. comm.). In 
Imagotaria downsi the crowns of P' and P 4 possess a strong 
anterolateral cusp (= paracone) but a very weakly developed pos- 
terolateral cuspule (metacone?). The strong protocone shelf is 
posteromedially placed and has a well-developed lingual cuspule 
(Repenning and Tedford 1977). In a referred rostrum of Gompho- 
taria pugnax (JMTC 907- 1 70), the crown of P 4 has a single lateral 
cusp (= paracone) with a distinct lingual cingulum expanded 
slightly at its posteromedial corner. The posterior premolars in 
Dusignathus santacruzensis are unknown. The simple single- 
cusped conical crown of P : in the type, however, suggests that the 
condition of the crown of P 4 was like that described for 
Gomphotaria. In the Odobeninae, P' and P 4 have simple peglike 
crowns (Repenning and Tedford 1977: Fay 1982; Erdbrink and van 
Bree 1990), an odobenine synapomorphy. Loss of all postcanine 
teeth is an autapomorphy of Valenictus. The condition in Pontolis is 
unknown. 

36. P 4 , protocone shelf. = strong and anteromedially placed, 1 
= strong and posteromedially placed with small cuspules, 2 = 
reduced or absent, 3 = P 4 absent. Enaliarctos has a functional 
carnassial (Berta 1991) with a strong protocone shelf that is 
anteromedially placed. Pinnarctidion and Neotherium have a P 4 
with a posteromedially placed protocone shelf. This modification is 
associated with loss of the embrasure pit between P 4 and M 1 and is 
correlated with reduction in occlusal shear (i.e., P 4 is no longer a 
functional carnassial). In Neotherium, the protocone shelf bears two 
small cuspules. In Imagotaria downsi the protocone shelf is re- 
duced (relative to Neotherium) and is variable in the number and 
size of cuspules. In other odobenids the shelf is greatly reduced or 
absent. Loss of P 4 is an autapomorphy of Valenictus. 

The distribution of this character suggests that a reduced proto- 
cone shelf evolved at least twice, once in the common ancestor of 
the dusignathines and odobenines and once in the desmatophocids. 



37. P 4 . number of roots. = three; 1 =two; 2 = one bilobedroot; 
3 = one root. 4 = P 4 absent. As noted by Barnes ( 1989) and Berta 
( 1 99 1 ). the P 4 of Enaliarctos and Pinnarctidion has three separate 
roots, one above each of the principal cusps. Three derived charac- 
ter states are recognized. Derived state 1 is characterized by a 
reduction in the number of roots to two as in Pteronarctos (Barnes 
1990). Desmatophoca (A. Berta. pers. comm.), and Neotherium (L. 
G. Barnes, pers. comm.). with the single anterior root well sepa- 
rated from the posterior bilobed root and formed from coalesced 
metacone and protocone roots. Derived state 2 is represented by a 
further reduction to only a single vestigially bilobed root through 
fusion of the anterior root to the posterior root as seen in Imagotaria. 
Pontolis (USNM 314300). Dusignathus (SDSNH 38342), and 
Gomphotaria. In the latter two taxa the roots are relatively swollen 
and have a diameter greater than the tooth crowns'. Derived state 3 
consists of reduction to a single nearly circular root and is an 
odobenine synapomorphy. This reduction in number of postcanine 
tooth roots is correlated with development of homodonty (Barnes 
1989). Loss of all postcanine teeth is an autapomorphy of 
Valenictus. 

The distribution of this character indicates that reduction to two 
roots evolved independently in the outgroup taxa Pteronarctos and 
Desmatophoca and in the odobenids. Similar trends towards root 
reduction and homodonty evolved independently in otariids 
(Repenning and Tedford 1977; Berta and Demere 1986; Barnes 
1989). 

38. Postcanine tooth enamel. = well-developed enamel layer; 
1 = thin and/or patchy enamel layer; 2 = cementumonly (no enamel 
on adult teeth), 3 = postcanine teeth lost. Well-developed enamel is 
primitive for carnivorans. Two derived characters are recognized. 
The thin enamel crowns of Dusignathus (see Repenning and 
Tedford 1977) and the patchy remnants of enamel on adult teeth of 
Gomphotaria (see Barnes and Raschke 1990) and Aivukus (see 
Repenning and Tedford 1977) represent derived state 1. Loss of an 
enamel layer is seen in adult Odobenini and represents derived state 
2. Loss of all postcanine teeth is an autapomorphy of Valenictus. 

Mandible 

39. Mandibular symphysis. = unfused. 1 = fused. Pinnipeds 
primitively have an unfused mandibular symphysis. The derived 
condition of a fused symphysis occurs in Valenictus and Odobenus. 
There is some sexual dimorphism in Odobenus rosmarus. whose 
adult females may have unfused symphyses (C. E. Ray, pers. 
comm.). The mandibular symphysis is also fused in Dusignathus 
seftoni. 

The distribution of this character suggests that a fused symphy- 
sis evolved twice, once in the common ancestor of Valenictus and 
Odobenus and again in Dusignathus. A less parsimonious hypoth- 
esis is that fusion evolved several times within the Odobenini. 

40. Mandibular symphysis with extra bone. = no, 1 = yes. 
Odobenus rosmarus is characterized by a very heavy and swollen 
mandibular symphysis that is widest ventrally. Odobenus man- 
danoensis also has a very heavy symphyseal region, but in it the widest 
part is dorsally placed (Tbmida 1 989: figs. 4A-C). In other odobenine 
walruses, as well as pinnipeds in general, there is no increase in the 
mass of the symphysis (primitive condition). The derived condition is 
a synapomorphy uniting the species of Odobenus. Increased mass of 
the symphysis reduces kinesis between the left and right dentaries and 
may be an adaptation to strong oral suction. 

41. Mandibular terminus. = smooth, compact, 1 = vascular. 
The distal terminus of the lower jaw in tusked odobenines is rough- 
ened and pitted, in contrast with the smooth surfaces seen in other 
pinnipeds (primitive condition). Fay ( 1982) discussed the oral open- 
ing in Odobenus, noting that the terminus of the lower jaw is 
covered by a tough, cornified surface that he suspected functioned 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



111 



to hold prey securely. The roughened surface of the odobenine 
mandibles may be related to increased vascularization for this 
cornified lower lip. Although the condition in Aivukus is unknown, 
that taxon's lack of a laterally convex tooth row (character 34) 
suggests that Aivukus retained the primitive condition. 

42. Mandibular arch. = nearly parallel, 1 = sharply divergent. 
The mandible of Enaliarctos emlongi (Berta 1991 ) forms an elon- 
gated arch in occlusal aspect, with the right and left rami meeting at 
a sharply acute angle of about 33° (primitive condition). As deduced 
from the structure of the upper jaw, other species of Enaliarctos, 
Pteronarctos, and Pinnarctidion had a similar mandibular architec- 
ture. In Neotherium this angle is about 20°, in Imagotaria about 36°. 
and in Gomphotaria about 42°. The derived condition of a sharply 
divergent mandibular arch occurs in Dusignathus (angle of 58°- 
60°), and the Odobenini (angle of 58°-60°). A sharply divergent 
mandibular arch is correlated with rostral shortening. 

The distribution of the derived condition in these taxa suggests 
convergent evolution of a shortened rostrum in Dusignathus and the 
Odobenini. 



43. Dentary, ventral harder. = straight, 1 = sinuous. In lateral 
view, the ventral border of the horizontal ramus (between the 
symphysis and pterygoid process) is nearly straight in the outgroup 
taxa, Neotherium (Fig. 6A), and Imagotaria (Fig. 6B) (primitive 
condition). The condition in the Odobenini is complicated by the 
upturning of the symphysis, but in general the ventral margin 
preserves the primitive condition (Figs. 6D-F). As noted by 
Repenning and Tedford (1977), the type dentary of Dusignathus 
santacruzensis has a markedly sinuous ventral margin (Fig. 6C). 
This condition also occurs in Dusignathus seftoni (SDSNH 20801 ) 
and Pontolis (USNM 335563). 

The distribution of this character suggests that a sinuous ventral 
border may have evolved only once in the common ancestor of 
Pontolis and Dusignathus. The condition in Gomphotaria would 
thus represent a reversal. An alternate hypothesis is parallel evolu- 
tion of a sinuous ventral border in Dusignathus and Pontolis. 

44. Dentary, marginal process. = weakly developed. 1 = 
strongly developed. The marginal process (Davis 1964:61) is vari- 
ably developed in pinnipeds. This process is the main area for 








Figure 6. Lateral aspect of right dentaries of selected fossil and living odobenids. A, Neotherium minim (Round Mountain Silt, middle Miocene): B. 
Imagotaria downsi (Santa Margarita Formation, late middle Miocene, after Repenning and Tedford 1977); C, Dusignathus santacruzensis (Punsima 
Formation, late late Miocene); D. Alachtherium eretsii (Saldisian sands, early Pliocene); E, Valenktus chulavistensis (San Diego Formation, late Pliocene); 
F, Odobenus rosmarus (Recent). Scale bar. 5 cm. 



112 



Thomas A. Demere 



insertion of the digastric muscle, the principal jaw depressor. The 
primitive condition of a weakly developed process (or no process at 
all) occurs in Neotherium and Imagotaria. The derived condition of 
an enlarged marginal process occurs in Pontolis, Dusignathus, 
Valenictus,Alachtherium, and Prorosmarus. In Odobenus rosmarus 
the marginal process is secondarily reduced, possibly in relation to 
the increased mass of the horizontal ramus. The reduced marginal 
process of Gomphotaria is also viewed as a reversal. 

Postcrania 

45. Humerus, deltoid tubercle. = on pectoral crest. 1 = on 
lateral edge of crest, 2 = off crest. The scar for insertion of the 
deltoideus muscle is primitively located on the pectoral crest (Fig. 
7B) of the humerus (Repenning and Tedford 1977). This condition 
is seen in the outgroup taxa and in many odobenids [Neotherium. 
Imagotaria, Pontolis, and Aivukus). In Odobenus, Valenictus (see 
Demere 1994: fig. 6, this volume), and Pliopedia (see Repenning 
and Tedford 1977:pl. 17, fig. 3) the scar is separate from and 
posterior to the crest (Fig. 7A). In Alachtherium (IRSNB M.170) 
the tuberosity occupies an intermediate position (i.e., posterior to 
the crest but still joined to it). A left humerus (MCZ 7713) referred 
to Prorosmarus alleni by Repenning and Tedford ( 1977; citing C. 
E. Ray) also preserves the intermediate condition. 

The distribution of the derived character states suggests a trans- 
formation series from a laterally placed deltoid insertion to eventual 
separation of the insertion from the crest. Character state 1 appears 
to represent a synapomorphy of the Odobenini. while state 2 repre- 
sents a synapomorphy uniting Pliopedia, Valenictus. and Odobenus. 

46. Humerus, medial entepicondyle. = small, 1 = enlarged. 
Primitively, the entepicondyle of the pinniped humerus is small 
(Fig. 7) relative to the greatly enlarged entepicondyle observed in 




— dt 



Figure 7. Anterior aspect of left odobenid humeri. A, Odobenus 
rosmams; B. Aivukus cedrosensis (after Repenning and Tedford 1977). 
Skeletal elements: dt. deltoid tubercle; gt, greater tuberosity; pc, pectoral 
crest. 



Valenictus (see Repenning and Tedford 1977; Demere 1994, this 
volume). Enlargement of the entepicondyle is probably related to 
an increased mass of the intrinsic flexor musculature of the fore- 
limb (Howell 1929) and is an autapomorphy of Valenictus. 

47. Humerus, diameter of distal trochlea. = medial lip same 
(or smaller) diameter as distal capitulum. 1 = medial lip diameter 
greater than distal capitulum. Among pinnipeds, the anteroposterior 
diameter of the medial lip of the distal humeral trochlea is smaller 
than or equal to the diameter of the distal capitulum (primitive 
condition). The derived condition of a distinctly larger medial lip is 
a synapomorphy for odobenids (Repenning and Tedford 1977:8). 

48. Radius, distal end. = unexpanded. 1 = expanded, with 
small radial process. 2 = expanded, with large radial process. As 
noted by Berta and Ray (1990), the distal end of the radius of 
Enaliarctos mealsi is unexpanded (primitive condition). Two de- 
rived conditions are recognized, with only one applying to the 
ingroup. An expanded distal end with a relatively small radial 
process (Repenning and Tedford 1977:pl. 2) occurs in otariids and 
Allodesmus (derived condition 1). An expanded distal end with a 
relatively large and distally projecting radial process is found in 
odobenids and represents a synapomorphy for the entire group. 

49. Metacarpal I, insertion ofpollicle extensor. = smooth. 1 = 
pit. 2 = rugosity. The insertion for the pollicle extensor muscle on 
the dorsoproximal surface of metacarpal I is variably expressed. 
Primitively, the dorsoproximal surface is smooth. In Imagotaria, 
Pliopedia, and Gomphotaria there is a conspicuous pit (Barnes 
1989) (derived condition 1). while in Aivukus, Alachtherium. 
Valenictus. and Odobenus a rugosity marks the insertion 
(Repenning and Tedford 1977:21 ) (derived condition 2). Presence 
of a pit or rugosity is an odobenid synapomorphy. 

50. Scapholunar. = no pit for magnum, 1 = well-formed pit. In 
odobenids, the magnum articulates with the scapholunar in a con- 
spicuous and often deep pocket or pit (Repenning and Tedford 
1977:36). Although the derived condition is invariant in the 
ingroup, it provides support for the monophyly of the group. The 
primitive condition is a flat articular surface for the magnum as 
preserved in Allodesmus. phocids, and otariids. 

5 1 . Astragalus, calcaneal process. = absent, 1 = present, 2 = 
elongated. As discussed by Berta and Ray ( 1 990: 1 5 1 ), the astraga- 
lus of Enaliarctos mealsi lacks a posteromedial calcaneal process. 
This represents the primitive condition for pinnipeds. Two derived 
character states are recognized, with only one found in the ingroup. 
In Allodesmus (see Kellogg 1931: fig. 53) and odobenids the calca- 
neal process is distinct but variably developed (derived condition 
1 ), either strong and posteroventrally directed as in Imagotaria 
(Repenning and Tedford 1977:pl. 14, figs. 25, 26) or more weakly 
developed as in the Odobenini. The second derived state occurs in 
the phocids. whose astragalus has a strong caudally directed calca- 
neal process that is nearly as long as the calcaneal tuber ( King 1 983; 
Berta and Ray 1990). 

The distribution of character state I suggests that a distinct but 
unelongated process evolved in the common ancestor of odobenids 
and the Desmatophocidae. 

52. Calcaneum. calcaneal tuber. = straight. I = prominent 
medial tuberosity. The calcaneal tuber is primitively straight-sided 
in pinnipeds (Repenning and Tedford 1977; Berta and Wyss 1994, 
this volume). The derived condition of a prominent medial tuberos- 
ity on the proximal end of the calcaneal tuber occurs in Allodesmus 
and odobenids, especially Imagotaria (Repenning and Tedford 
1977: pi. 15, figs. 3-6). Although this character does not vary within 
the ingroup it does provide evidence for the sister-group relation- 
ship of odobenids and desmatophocids. 

53. Entocuneiform. mesocuneiform articulation. = abutting, 1 
= overlapping. Primitively, the entocuneiform of pinnipeds articu- 
lates with the mesocuneiform along a straight butt joint. In addition, 
the distal articular facet for articulation with the first metatarsal is 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



113 



short and quadrate. The primitive condition is preserved in extant 
otariids. The derived condition is an entocuneiform that strongly 
overlaps the mesocuneiform such that the eoncave articular facet 
for the mesocuneiform is positioned subparallel to the facet for the 
navicular articulation, not at right angles to it. In addition, the distal 
condyle of the entocuneiform is relatively elongated and rounded. 
This derived character complex occurs in all the studied odobenids 
in which the tarsals were preserved (Pontolis, Gomphotaria, 
Valenictus, and Odobenus). Kellogg ( 1925: 107-108: fig. 14) briefly 
discussed this character complex in his description of a fossil 
odobenid ankle from the Towsley Formation (UCMP24070-24082) 
that he tentatively referred to Pontolis magnus. Repenning and 
Tedford ( 1977:24) transferred this specimen to Imagotaria downsi. 
In a later report Kellogg ( 193 1 :292-294; figs. 60-63) described the 
entocuneiform of Allodesmus, noting similarities with the same 
bone in Odobenus. The condition in Enaliarctos is unknown. 

The distribution of this character in the outgroup taxa supports 
two alternative hypotheses, that the derived state evolved indepen- 
dently in desmatophocids and odobenids or, more parsimoniously, 
that the derived state evolved only once in the common ancestor of 
these two taxa. 

SYSTEMATICS 

This review is not meant to serve as a formal and exhaustive 
systematic treatment but rather as a summary of recent literature 
and new specimens collected since the monograph of Repenning 
and Tedford (1977). For certain taxa, the discussions are more 
extensive than for others. This difference in treatment is due either 
to recovery of new and previously undescribed material referable to 
nominal taxa or to taxonomic and/or nomenclatural complexity that 
requires elaboration. 

Because the primary emphasis of this report is to explore phylo- 
genetic relationships within the Odobenidae, the taxon discussions 
focus on synapomorphies rather than on symplesiomorphies. 
Symplesiomorphies are important in characterizing taxa, but be- 
cause they do not provide evidence of recent common ancestry they 
are not as useful for unraveling phylogeny. 

Class Mammalia Linnaeus, 1758 

Order Carnivora Bowdich, 1 82 1 

Infraorder Pinnipedia Illiger. 1811 

Family Odobenidae Allen, 1880 

Definition. — The monophyletic group containing the most re- 
cent common ancestor of Neolherium and Odobenus and all of its 
descendants. 

Diagnosis. — Pinnipeds with antorbital process constructed from 
both frontal and maxilla, supraorbital process of frontal absent; 
diameter of medial lip of distal trochlea of humerus greater than 
diameter of capitulum, distal portion of radius with enlarged radial 
process, insertion for pollicle extensor muscle on first metacarpal 
developed as a pit or rugosity, and scapholunar with distinct pit for 
magnum (odobenid synapomorphies), 

Neotherium Kellogg. 1931 

Type species. — Neotherium mirum Kellogg, 193 1 . 

Distribution. — Middle Miocene of the eastern North Pacific. 

Included species. — Type species only. 

Emended diagnosis. — A small odobenid with adventitious root 
on P 2 (possible autapomorphy, Barnes 1989:14), antorbital process 
bifurcated by maxilla/frontal suture, supraorbital process of the 



frontal lost, medial distal trochlea of humerus broader than radial 
capitulum (odobenid synapomorphies); embrasure pit on palate 
between P 4 and M 1 lost, medial tuberosity of calcaneus prominent, 
and calcaneal process of astragalus distinct (synapomorphies at 
level of desmatophocid and odobenid common ancestry). 

Neotherium mirum Kellogg, 1931 

Neotherium mirum Kellogg 1931 ; Mitchell and Tedford 1973; Mitchell 
1961; Barnes 1988. 1989. 

Lectotype. — USNM 11542, a right calcaneum (Mitchell and 
Tedford 1973:266). 

Type locality. — Sharktooth Hill, Kern County, California. 

Horizon and age. — Sharktooth Hill Bonebed. Round Mountain 
Silt, middle Miocene (Barstovian NALMA correlative, ca. 13-14 
Ma). 

Diagnosis. — As for the genus. 

Referred material. — Kellogg (1931) included an astragalus 
(USNM 11543), cuboid (USNM 11552), and navicular (USNM 
11548) in the original hypodigm and referred several additional 
postcranial elements to this taxon. Mitchell and Tedford (1973) 
referred additional material, including a humerus (LACM 4319), to 
it. Barnes (1988) referred a partial dentary (LACM 12300) to N. 
mirum and mentioned a large collection of topotypic material under 
study. 

Discussion. — Repenning and Tedford (1977) recognized the 
odobenid affinities of the type material of Neotherium mirum, 
noting the prominent medial tuberosity of the calcaneum and the 
distinct calcaneal process of the astragalus. Humeri referred to this 
taxon preserve the distal articular synapomorphies of the 
Odobenidae. Barnes (1988:fig. 4) referred a partial left dentary to 
N. mirum and described the anterior lower dentition. P,-M, are 
double-rooted and have transversely compressed crowns with a 
prominent central cusp flanked by reduced anterior and posterior 
cuspules. In most aspects, Neotherium is a generalized odobenid, 
possessing the basal synapomorphies of the family but lacking the 
derived features of later-diverging members. 

Repenning and Tedford (1977) and Barnes (1988) noted a dis- 
tinct bimodality in the size of skeletal elements and suggested that 
N. mirum was sexually dimorphic. 

A large collection of topotypic cranial (including a complete 
skull), dental, and postcranial material is currently under study by 
L. G. Barnes, who generously allowed me to examine critical 
specimens. I discuss this material only in the context of the phylo- 
genetic analysis (Table 2). 

Imagotaria Mitchell, 1968 

Type species. — Imagotaria downsi Mitchell. 1968. 

Distribution. — Late middle to late late Miocene of the eastern 
North Pacific. 

Included species. — Imagotaria downsi Mitchell. 1968 and 
Imagotaria sp. cf. /. downsi. 

Emended diagnosis. — Medium-sized odobenids with strongly 
developed and posteroventrally directed calcaneal process of the 
astragalus (possible autapomorphy of Imagotaria), upper postcanine 
teeth with single, vestigially bilobed roots (also seen in certain 
dusignathines), palate transversely arched, frontal/maxilla suture 
straight and transverse, premaxilla with overlap of nasal short, 
antorbital process enlarged, infraorbital foramen large, pterygoid 
strut broad, and basioccipital broad and pentagonal (synapomor- 
phies at level of dusignathine and odobenine common ancestry). 

Imagotaria downsi Mitchell, 1968 

Imagotaria downsi Mitchell, 1968; Repenning and Tedford 1977. 
Imagotaria sp. Barnes 1971. 



114 



Thomas A. Demere 



Holotype. — SBNHM 342. parts of the skull, partial right and 
left dentaries, isolated upper and lower postcanine teeth, partial 
hyoid, partial atlas, partial thoracic vertebra, glenoid regions of 
both scapulae, and partial right and left humeri. 

Type locality. — Great Lakes Carbon Company quarry, Lompoc. 
Santa Barbara County. California. 

Horizon and age. — Sisquoc Formation (diatomite); late middle 
to early late Miocene (Clarendonian NALMA correlative, ca. 9-12 
Ma). 

Diagnosis. — As for the genus. 

Referred material.— Mitchell (1968:1865-1866) referred a 
horizontally split and crushed skull (USNM 13487) from the 
Sisquoc Formation to /. downsi. Repenning and Tedford ( 1977:22- 
24) referred cranial and postcranial material collected from the 
Santa Margarita Formation (early late Miocene. Clarendonian 
NALMA correlative) as exposed in Santa Cruz County, California, 
to /. downsi. Most of this referred material came from a single fossil 
accumulation, perhaps a rookery. 

Discussion. — Imagotaria downsi exhibits several symplesio- 
morphies useful in distinguishing it from other members of the 
clade Imagotaria + Dusignathinae + Odobeninae, including low 
sagittal crest, inflated tympanic bulla, and P 3 - 4 with strong 
anterolateral cusp (= paracone), weakly developed posterolateral 
cuspule (metacone?), and strong posteromedially placed protocone 
shelf with well-developed lingual cuspule. Thus, Imagotaria 
downsi is more readily distinguished by characters that it lacks than 
by autapomorphies, i.e.. it possesses the synapomorphies of the 
clade Imagotaria + Dusignathinae + Odobeninae but lacks the 
features derived in the odobenines and dusignathines). Its only 
probable autapomorphy is the strongly developed calcaneal process 
of the astragalus (Repenning and Tedford 1977:39-40). The hu- 
merus of Imagotaria downsi is generalized, with an elongated and 
keeled pectoral crest that descends with a marked flexion to the 
distal surface of the relatively slender shaft. 

With the additional specimens referred to Imagotaria downsi by 
Repenning and Tedford ( 1977), this taxon is one of the most com- 
pletely known fossil odobenids. The sample size is large enough to 
reveal both sexual dimorphism and polymorphism in certain dental 
features (e.g.. either two or three upper incisors). 

Dentition: 31. 1C.4P. 2M x 2 = 36(38) 
21, 1C.4P. 1(2)M 

Imagotaria sp. cf. /. downsi 

Material. — In the Emlong Collection at the USNM (Ray 1976) 
are several complete skulls, including USNM 335599 and 335594, 
isolated jaws, and postcrania referable to Imagotaria. This material 
may represent more than one taxa and will be described in a 
subsequent report. 

Locality. — Coos Bay, Coos County, Oregon. 

Horizon and age. — Empire Formation; late late Miocene 
(Hemphillian NALMA correlative, ca. 5-8.5 Ma). 

Discussion. — Referral of these fossils to Imagotaria is based on 
their possession of derived characters not seen in Neotherium but 
reported for /. downsi and on their lack of dusignathine and 
odobenine synapomorphies. The Oregon specimens differ from the 
type material in their mastoid-paramastoid crest being aligned 
subparallel to the sagittal plane (not divergent as in the holotype) and 
consistent loss of I'. The palate is arched transversely and bordered 
by straight postcanine tooth rows possessing four premolars and two 
molars. I' is lost, and I 2 - 3 are positioned anterior to the unenlarged 
canine. The pterygoid strut is extremely broad. The sagittal crest is 
low and extends posteriorly to a sharply edged and overhanging 
lambdoidal crest. The frontal lacks a supraorbital process, but the 
antorbital process is large and split by the frontal/maxilla suture. 



This material was included in this report because it allows 
determination of character states not preserved in the type and 
referred material of /. downsi (e.g., maxilla/frontal suture and 
antorbital processes). 

Subfamily Dusignathinae (sensu Repenning and Tedford. 1977) 

Definition. — The monophyletic group containing the most re- 
cent common ancestor of Pontolis and Dusignathus and all of its 
descendants. 

Diagnosis. — Odobenids with upper and lower canines enlarged 
as tusks; nasal/frontal suture V-shaped and posteriorly directed (a 
more acute "V" occurs in desmatophocids and phocids); sagittal 
crest enlarged (reversed in Dusignathus; also seen in certain 
otariids); and orbital vacuity anteriorly placed (also occurs in 
otariids). 

Dusignathus Kellogg. 1927 

Type species. — Dusignathus santacruzensis Kellogg, 1 927 

Distribution. — Late late Miocene to late Pliocene of the eastern 
North Pacific. 

Included species. — Dusignathus santacruzensis Kellogg, 1927, 
and Dusignathus seftoni Demere. 1994 (this volume). 

Definition. — The monophyletic group containing the most re- 
sent common ancestor of D. santacruzensis and D. seftoni and all of 
its descendants. 

Emended diagnosis. — Medium-sized dusignathine walruses 
with mandibular rami deep, left and right dentaries meeting at acute 
angle of 60° at symphysis to form a narrow V-shaped "chin," lower 
canines closely appressed to each other, and rostrum relatively 
shortened (autapomorphies of Dusignathus). Equivocal 
dusignathine synapomorphies that support alternative arrangements 
of members of this clade include upper and lower canines slightly to 
greatly enlarged, squamosal fossa relatively large above the external 
auditory meatus (apomorphies shared with Gomphotaria), and ven- 
tral border of dentary sinuous (apomorphy shared with Pontolis). 

Dusignathus santacruzensis Kellogg, 1927 

Dusignathus santacruzensis Kellogg, 1927; Mitchell 1968; Repenning 
and Tedford 1977; Barnes and Raschke 1991 

Holotype. — UCMP 27121, associated left and right dentaries, 
part of right maxilla with C and P 1 -, isolated teeth (including two 
upper incisors), cranial fragment with lambdoidal crest and portion 
of occiput, and partial right temporal bone (mastoid, petrosal, and 
base of zygomatic process of squamosal). 

Type locality. — UCMP V-2701 . Santa Cruz County, California. 

Horizon and age. — Purisima Formation, late late Miocene 
(Hemphillian NALMA correlative, ca. 5-8.5 Ma). 

Emended diagnosis. — A species of Dusignathus with tooth 
wear in-line on anterior and posterior margins of postcanine tooth 
crowns (possible autapomorphy), auditory bulla slightly inflated, 
sagittal crest low. occipital crest low. canines not ever-growing, and 
zygomatic arch slender (plesiomorphies shared with Enaliarctos. 
Neotherium, and Imagotaria). 

Referred material. — Repenning and Tedford (1977:43—44) re- 
ferred fore- and hindlimb elements collected from the Purisima 
Formation to D. santacruzensis. These authors also referred a ros- 
tral fragment (UCR 15244) with I 3 , C 1 , and P 1 " 3 collected from the 
Almejas Formation (late late Miocene, Hemphillian NALMA cor- 
relative of Isla Cedros, Baja California. Mexico) to this taxon. A 
complete right humerus ( UCMP 653 1 8 ) collected from the Purisima 
Formation was questionably referred to D. santacruzensis by 
Repenning and Tedford ( 1977). 

Discussion. — The maxillary fragment of the holotype contains 



The Family Odohenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



115 



C' and P K 2 , with the premolars single-rooted with smooth, thin 
enamel crowns and slight lingual cingula. In occlusal aspect, the 
canine and premolars are in line (i.e., C' is not set outside the tooth 
row as in odobenines). The maxilla is not swollen around the base 
of the procumbent and deeply rooted C (root extending to a posi- 
tion above P : ). The crown of C' is caniniform and only slightly 
worn (in marked contrast to the extreme wear shown by the lower 
canine that suggests occlusion with a large I 1 , as noted by 
Repenning and Tedford ( 1977)]. C 1 is not greatly enlarged relative 
to the postcanine teeth but apparently is positioned lateral to the 
lower dentition when the jaws occlude. 

The lower jaw has an unfused and elongated bony symphysis 
with a slender but distinct genial tuberosity (Fig. 6C). The horizon- 
tal rami are deep and narrow with sinuous ventral borders. C, is 
unreduced relative to C and has a closed root. The roots of all the 
postcanine teeth are short, swollen, oval, single-rooted pegs, in 
which the greatest diameter exceeds that of the crown. Gomphotaria 
has similar lower postcanine teeth except that the roots are more 
circular in cross-section (Barnes and Raschke 1991). In 
Dusignathus, wear on the upper and lower postcanine teeth is 
concentrated on the anterior and posterior edges of the thin enamel 
crowns. 

Cranial fragments preserve portions of a low sagittal crest, a 
distinct lambdoidal crest, and an occipital crest. The type's right 
squamosal fragment preserves a slender zygomatic process and 
large open external auditory meatus. The mastoid process makes an 
acute angle with the long axis of the skull. 

Except for the wear pattern, dental features alone do not distin- 
guish Dusignathus santacruzensis from Gomphotaria pugnax. 
Dusignathus differs however, in being much smaller and having 
widely divergent, relatively shortened, deep mandibular rami. This 
distinctive mandible suggests that Dusignathus had a shorter ros- 
trum than did Gomphotaria. In addition, Gomphotaria had enlarged 
and fluted upper canines, Dusignathus much smaller and smooth- 
surfaced canines. 

The humerus ( UCMP 653 1 8) questionably referred to this taxon 
by Repenning and Tedford (1977) is quite generalized and pre- 
serves the elongated and sharply keeled pectoral crest (with con- 
joined deltoid insertion) seen in Neotherium, Imagotaria, and 
Gomphotaria. 

Repenning and Tedford (1977) discussed the holotype and re- 
ferred material of D. santacruzensis thoroughly. Unfortunately, no 
new material has been collected since. 

Dentition: 1(2)1. 1C.4P. 1(?)M x 2 = 26(28) 
01, 1C.4P, 1M 

Dusignathus seftoni Demere, 1994 

Dusignathus seftoni Demere, 1994 (this volume). 

Holotype. — SDSNH 38342, skull lacking the basicranium. 

Type locality.— SDSNH locality 3468, Chula Vista, San Diego 
County. California. 

Horizon and age. — San Diego Formation, late Pliocene 
(Blancan NALMA correlative, ca. 2-3 Ma). 

Diagnosis. — A species of Dusignathus with upper and lower 
postcanine teeth forming a laterally convex arch (in occlusal as- 
pect), postcanine teeth in upper and lower jaws with transverse axes 
of roots medially rotated, roots of all postcanine teeth closely 
appressed, dentary with masseteric fossa deeply excavated and 
symphysis fused (autapomorphies of D. seftoni), upper and lower 
canines enlarged, nasal/frontal suture V-shaped posteriorly 
(dusignathine synapomorphies), sagittal crest low, six postcanine 
teeth, and I 1 lost (symplesiomorphies at level of the Odobenidae). 

Referred material. — SDSNH 20801, right dentary with portions 
of the symphyseal region of the left dentary; SDSNH 43873, com- 



plete left humerus; SDSNH 38256. damaged left humerus. All 
collected from the San Diego Formation. 

Discussion. — This species is assigned to Dusignathus because 
it shares with the type species a shortened rostrum, narrow symphy- 
seal "chin," and deep mandibular rami with a sinuous ventral 
border. D. seftoni differs from the type species primarily in overall 
size (larger), configuration of the tooth row (laterally convex in 
occlusal aspect), orientation of the roots of the postcanine teeth (in 
transverse cross-section the long axes of the roots are rotated medi- 
ally progressively from the back of the tooth row to the front), and 
size of the upper and lower canines (both enlarged). The lower jaw 
of D. seftoni is very robust with a deeply excavated masseteric fossa 
and well-developed marginal process. The large genial tuberosity is 
swollen and excavated. The holotype skull of D. seftoni represents 
an immature (probably male) individual and has the distinctive 
nasal/frontal suture seen in Gomphotaria and referred material of 
Pontolis (see below) in which the nasals extend posteriorly as a 
wedge between the frontals (Fig. 3D). 

The humerus of this species of Dusignathus is generalized 
compared to that of the Odobenini. The shaft is slender, the greater 
tuberosity extends above the proximal capitulum, the deltoid inser- 
tion is positioned on the elongated pectoral crest, and this crest 
descends with a marked flexion to the distal portion of the shaft. 

Dentition: 21. 1C. 4P. 2M x 2 = 32 
II. 1C, 4P, 1M 

Gomphotaria Barnes and Raschke, 1 99 1 

Tvpe species. — Gomphotaria pugnax Barnes and Raschke, 
199L 

Distribution. — Late late Miocene of the eastern North Pacific. 

Included species. — Type species only. 

Emended diagnosis. — A large dusignathine walrus with roots of 
enlarged upper and lower canines finely fluted and covered with 
thick cementum, mastoid-paramastoid crest compressed antero- 
posteriorly and expanded dorsoventrally (autapomorphies of 
Gomphotaria), upper and lower canines greatly enlarged, squamo- 
sal fossa above the external auditory meatus relatively large (syna- 
pomorphies shared with Dusignathus), body large, and sagittal 
crest highly elevated and arched (synapomorphies shared with 
Pontolis, USNM 335567); postcanine teeth with bulbous crowns, 
thin enamel, and smooth narrow lingual cingula (synapomorphies 
at the level of the common ancestry of the Odobeninae and 
Dusignathinae); I 1 - 2 lost (convergent with Odohenus): marginal 
process of dentary weakly developed, overlap of nasal by ascending 
process of premaxilla relatively long, and ventral border of man- 
dible straight (reversals to primitive condition). 

Gomphotaria pugnax Barnes and Raschke, 1991 

Gomphotaria pugnax Barnes and Raschke, 1991 . 

Holotvpe. — LACM 121508, nearly complete skeleton includ- 
ing skull, mandible, and most postcranial elements. 

Tvpe locality. — LACM locality 4631, Marblehead, San 
Clemente, Orange County, California. 

Horizon and age. — Capistrano Formation; late late Miocene 
(Hemphillian NALMA correlative, ca. 5-8.5 Ma). 

Diagnosis. — As for the genus. 

Referred material. — JMTC 907-170, damaged rostrum with 
palate intact and teeth and/or alveoli for I 3 , C\ P'" 4 , M'~ 2 . This 
specimen was collected from the Oso Sand Member of the 
Capistrano Formation and is currently assigned only a field number 
in the catalog system of John Minch and Associates. Mission Viejo, 
California. Eventually this specimen will be donated to the Orange 
County Natural History Foundation. 



116 



Thomas A. Demere 



Discussion. — The holotype includes almost every skeletal ele- 
ment, making Gomphotaria pugnax the most completely known 
dusignathine. A baculum indicates a male. Many of the postcranial 
elements are characterized by pathological exostoses suggesting 
arthritis. The condylobasal length of 466 mm indicates an enormous 
pinniped, second in size only to Pontolis magnus (e.g., USNM 
335567. condylobasal length of 600 mm). With its large and procum- 
bent upper lateral incisors, enlarged and procumbent upper and 
lower canines with deeply fluted roots, vertically oriented palatine 
foramina, highly elevated and arched sagittal crest, and relatively 
small orbit, Gomphotaria pugnax is distinctive dentally and crani- 
ally. Dusignathine features include the broadly V-shaped invasion of 
the frontals by the nasals, enlarged upper and lower canines, and 
upper postcanine teeth with single vestigially bilobed roots. That the 
loss of I 1,2 and retention of a large I 1 is not a result of the senility of 
the holotype is confirmed by the referred rostrum. In this specimen 
(JMTC 907-170), which lacks any of the exostoses of the type, the 
right I 3 is preserved in its alveolus and is large with a conical and 
well-worn crown. Wear is presumably the result of occlusion with 
the enlarged C,. The premolars preserved with this specimen (P 2 ^ 1 ) 
all have nearly circular roots that are swollen and greater in diameter 
than the somewhat oval crowns. An alveolus for M 2 suggests that the 
reduced dentition of the type skull is pathological. The enamel on the 
postcanine teeth is thin as in Dusignatkus. However, occlusal wear is 
apical and does not extend onto the anterior and posterior margins of 
the crown, as is seen in the type of D. santacruzensis. A distinct 
lingual cingulum is preserved on P 3, 4 and is marked by a slight 
posterolingual expansion. There are no accessory cusps orcuspules. 

The humerus of Gomphotaria is generalized and. like that of 
Dusignathus seftoni, has a greater tuberosity that rises distinctly 
above the proximal capitulum. The humerus has a large and elon- 
gated pectoral crest, which descends abruptly to the distal portion 
of the humeral shaft. The deltoid tubercle is positioned on the 
lateral margin of the pectoral crest. 

Dentition: II. IC.4P. 1(2)M x 2 = 26(28) 
01, 1C, 4P, IM 

Pontolis True 1905 

Type species. — Pontolis magnus (True, 1905) 

Distribution. — Late late Miocene of the eastern North Pacific. 

Included species. — Type species only. 

Emended diagnosis. — A large dusignathine walrus with elon- 
gated rostrum and marked interorbital constriction ( possible autapo- 
morphies of Pontolis), infraorbital foramen unenlarged, I 1 " 3 all 
present. M'~ 2 double rooted, P,_, double-rooted, M,_ : double rooted 
(reversals to conditions at base of the Odobenidae), sagittal crest 
greatly enlarged (apomorphy shared with Gomphotaria), mandible 
with sinuous ventral border (apomorphy shared with Dusignathus). 
mastoid process enlarged (homoplasy shared with the Odobenini), 
nasal/frontal suture V-shaped, and C 1 and C, large (dusignathine 
synapomorphies). 

Pontolis magnus (True, 1905) 

Pontoleon magnus True, 1905. 

Pontolis magnus (True. 1905); True 1909; Kellogg 1922; Repenning 
andTedford 1977: Barnes and Raschke 1991. 

Holotype. — USNM 3792; complete basicranium and partial 
occiput of a damaged skull. 

Type locality. — Sea cliffs near Empire. Coos County, Oregon. 

Horizon and age. — Empire Formation, late late Miocene 
(Hemphillian NALMA correlative; ca. 5-7 Ma). 

Diagnosis. — As for the genus. 

Referred material. — The Emlong Collection at the USNM con- 
tains undescribed material here referred to Pontolis magnus. This 



material includes a skull and partial skeleton (USNM 335567), a 
nearly complete skull (USNM 314300), a rostrum (USNM 335554), 
a right dentary (USNM 335563), and isolated postcrania. 

Description of referred material . — The following description is 
based primarily upon USNM 314300, a complete skull, lacking 
only the dorsal portion of the braincase including the sagittal crest 
and occipital crest. There are alveoli for I 1 \ C 1 . P 1 ^, and M'~ 2 . 
P 1_1 are all single-rooted with oval alveoli. P M alveoli have lateral 
septa (vestigially bifid roots). M 1 has two roots (posterior root 
bilobed), andM : has two separate roots of equal sizes. C 1 andC, are 
large relative to adjacent teeth but not tusklike. The rostrum is 
elongated and associated with a small braincase (relative to overall 
length of skull) and narrow interorbital constriction. The anterior 
outline of the braincase is triangular in dorsal aspect. The nasals are 
relatively elongated and extend as a V-shaped wedge between the 
frontals. The infraorbital foramen is not especially enlarged and is 
hidden by the swollen maxilla when viewed in anterior aspect. The 
frontal bone at the frontal/nasal suture is not raised above the dorsal 
level of the maxilla and nasal as in Gomphotaria pugnax and 
Dusignathus seftoni. The antorbital process is strongly developed 
and split by the maxilla/frontal suture (USNM 335554). The frontal 
lacks a supraorbital process. The temporal fossa is long and narrow. 
The zygomatic arch forms the widest part of the skull. The postero- 
lateral border of the zygomatic process of the squamosal slopes 
ventrolaterally in the transverse plane, while anteriorly the process 
is long and slender and forms a simple overlapping contact with the 
jugal. The postorbital process of the jugal is relatively small. The 
sagittal crest is sharply keeled and has a sagittal groove running 
posteriorly to the cranial vertex. In USNM 335567. the sagittal crest 
is strongly developed (as in Gomphotaria) but does not extend 
above the general dorsal outline of the skull as viewed in lateral 
aspect. Instead, the roof of the braincase is depressed (in contrast to 
its more elevated position in Gomphotaria) and broadly convex at 
the level of the external auditory meatus. In USNM 335567, the 
lambdoidal crest flares posteriorly and overhangs the occiput, 
which is vertically oriented and marked by a strong occipital crest. 
The occipital condyles are small (relative to the overall size of the 
skull) whereas the mastoids are enlarged, resembling somewhat 
those of Odobenus. The palate is broad and slightly arched trans- 
versely. The hamular process of the pterygoid is delicate and ex- 
tends posteriorly to the level of the posterior border of the glenoid 
fossa. The pterygoid strut is narrow, and the anterior border of the 
internal nares is smoothly curved. The auditory bulla is flattened 
with smooth external surfaces, and the basisphenoid is arched 
transversely. 

Skull measurements of USNM 314300 (in mm): condylobasal 
length 528, braincase length 310, rostrum length 218. zygomatic 
width 255. mastoid width 237, interorbital width 47. Measurements 
of empty alveoli of USNM 3 1 4300 (anteroposterior diameter/trans- 
verse diameter, in mm): I 3 28/25, C 43/34. P 1 21/16, P : 23/16. P 3 
21/13. P 4 21/13. M 1 15/14, M 2 21/8. 

The lower jaws here referred to Pontolis preserve a large 
caniniform canine, at least one incisor (I,), P,_,. and M, ,. In 
USNM 335567 M,_, are missing. The crowns of P, 3 , as preserved 
in USNM 335563, are characterized by a single central cusp (= 
paraconid) and a strong lingual cingulum that extends around to the 
posterior margin of the tooth. The anterior portion of each tooth is 
worn to the extent that the paraconid is nearly lost. Whether or not 
there were any anterior cusps or cuspules is unclear. Overall, the 
dentary is elongated, with a distinct genial tuberosity, unfused and 
narrowly oval symphysis (in medial aspect), shallow horizontal 
ramus with a sinuous ventral border, well-developed marginal pro- 
cess, and large coronoid process without a deeply excavated masse- 
teric fossa. 

Referral of this material to Pontolis is based on the following 
features shared with the holotype: overall large size, narrow ptery- 



The Family Odohenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



117 



goid strut, transversely convex braincase, smooth and flattened 
bulla, bulla smoothly joining with mastoid (i.e., a prominent groove 
between the stylomastoid foramen and hyoid fossa is lacking), large 
and pentagonal basioccipital, and lateral border of zygomatic pro- 
cess of squamosal sloping ventrolaterally in transverse plane. Both 
the holotype braincase and the Emlong fossils were collected from 
the Empire Formation at Coos Bay, Oregon. 

Discussion. — Pontolis magnus was the first fossil pinniped to 
be named from the Pacific coast. The holotype, USNM 3792, is a 
large incomplete braincase, described and illustrated by Repenning 
and Tedford (1977:42-43; pi. 18, fig. 5). The important new mate- 
rial described here documents a truly tremendous animal — one 
skull ( USNM 335567) measures 600 mm in condylobasal length. In 
contrast, skulls of modern adult males of Odobenus rosmarus 
divergens range between 380 and 430 mm long (Fay 1985). 

Pontolis magnus is considered a dusignathine on the basis of its 
large upper and lower canines and V-shaped nasal/frontal suture. 
The retention of numerous dental plesiomorphies like separate 
roots on M'-\ P,^, and M, •, and three upper incisors underlines the 
taxon's mosaic nature. 

Dentition: 31. 1C. 4P.2M x 2= 32(36) 
II, 1C. 4P. 2(())M 

Subfamily Odobeninae (sensu Repenning and Tedford, 1977) 

Definition. — The monophyletic group containing the most re- 
cent common ancestor of Aivukus and Odobenus and all of its 
descendants. 

Diagnosis. — Odobenids with frontal/maxillary suture trans- 
versely directed, postorbital process of jugal dorsoventrally ex- 
panded, C, less than 75% the size of C postcanine teeth reduced to 
five, P 3 and P 4 with simple peglike crowns, P 4 with single circular 
root, first metacarpal with insertion for pollicle extensor muscle 
developed as a rugosity, and lower canine premolariform 
(odobenine synapomorphies). 

Aivukus Repenning and Tedford, 1977 

Type species. — A. cedrosensis Repenning and Tedford, 1977. 

Distribution. — Late late Miocene of the eastern North Pacific. 

Included species. — Type species only. 

Characterization. — A medium-sized odobenid with frontal/ 
maxillary suture directed transversely, postorbital process of jugal 
dorsoventrally expanded, C, less than 75% the size of C, P 3 - 4 with 
simple peglike crowns, first metacarpal with insertion for pollicle 
extensor muscle developed as a rugosity (odobenine synapomor- 
phies), I' lost (apomorphy shared with Dusignathus and 
Alachtherium),C ] not enlarged as tusk and lacking a central column 
of globular dentine, humerus with elongated and sharply keeled 
pectoral crest, and deltoid tubercle joined with pectoral crest 
(symplesiomorphies at the level of the Pinnipedia). 

Aivukus cedrosensis Repenning and Tedford, 1977 

Aivukus cedrosensis Repenning and Tedford, 1977; Barnes and Raschke 
1991. 

Holotype. — IGCU 901, a partial skull, partial left dentary, and 
partial front limb. 

Type locality.— UCR locality RV 7309, Isla Cedros, Baja Cali- 
fornia Sur, Mexico. 

Horizon and age. — Almejas Formation (lower portion), late 
late Miocene (Hemphillian NALMA correlative, ca. 5-8.5 Ma). 

Characterization. — As for the genus. 

Referred material— Repenning and Tedford (1977:14) referred 
several postcranial elements, including a complete humerus (UCR 
15243) and isolated carpal bones, to A. cedrosensis. 



Discussion. — Repenning and Tedford ( 1977) described the type 
and referred material of Aivukus cedrosensis thoroughly. As noted 
by these authors and confirmed here, Aivukus is clearly an 
odobenine with its dorsoventrally elongated postorbital process of 
the jugal. frontal/maxilla suture transverse and nearly perpendicular 
to sagittal plane, lower canine reduced in size relative to the upper 
canine, and postcanine teeth peglike with single, nearly circular 
roots and heavy "jacket" of cementum. The mosaic nature of this 
taxon is evident in its retention of numerous plesiomorphies (rela- 
tive to the Odobenini). including an elongated and slender rostrum, 
upper canine not ever-growing and lacking a central column ot 
globular dentine, lower canine larger than Pc,, mandibular symphy- 
sis unfused, and deltoid tubercle incorporated into the pectoral crest 
of the humerus (Fig. 7B). 

Although Repenning and Tedford (1977:fig. 1) presented a 
complete restoration of the skull of Aivukus, portions of the skull 
are not preserved on the holotype. For example, the cranial roof is 
not preserved at the midline, and thus it is impossible to determine 
whether there was a sagittal crest (as in Neotherium, Imagotaria, 
Pontolis, Gomphotaria, and Dusignathus) or a flattened cranial 
vertex (as in Alachtherium, Pliopedia, Valenictus, and Odobenus). 
Also, the nasal/frontal suture is not discernible in the holotype, and 
the occipital condyles, pterygoids, palatines, auditory bullae, basi- 
occipital. and most of the alisphenoid and nasals are missing. This 
incompleteness calls into question some of the odobenine features 
discussed by Repenning and Tedford (1977), such as the domed and 
crestless braincase, transverse nasal/frontal suture, broad pterygoid 
hamuli, and pentagonal basioccipital. My analysis indicates that 
Aivukus represents the least divergent odobenine walrus and is a 
possible ancestor of the tusked odobenines. 

Dentition: 21. 1C.4P. !M x2 = 28? 
?I, 1C, 4P, 1M 

Tribe Odobenini new name 

Definition. — The monophyletic group containing the most re- 
cent common ancestor of Alachtherium and Odobenus and all of its 
descendants. 

Diagnosis. — Odobenines with the external narial opening el- 
evated above the incisive margin, palate arched transversely and 
longitudinally, hard palate elongated (also occurs in the otariid 
Otaria), mastoid processes as widest part of cranium, zygomatic 
portion of squamosal blunt and robust, temporal fossa shortened, 
orbital vacuity posteriorly placed, upper canine with well-developed 
globular dentine column. C, less than 40% the size of C P' medial 
to upper canine, upper postcani ne teeth three or four, adult postcanine 
tooth crowns with cementum only ( no enamel ), mandibular terminus 
vascular, deltoid tubercle of humerus on extreme lateral side of 
pectoral crest or off crest, mastoid enlarged (also occurs in Pontolis), 
I 3 medial to upper canine (also occurs in Dusignathus), lower canine 
premolariform, tooth row between P 1 and M 1 laterally convex (also 
in Dusignathus), mandibular arch sharply divergent (also occurs in 
Dusignathus), palatine telescoped beneath alisphenoid, hamular pro- 
cess broad, pterygoid strut lost, lambdoidal crest with distinct flat- 
tened traction surface, sagittal crest lost (also variably seen in 
phocoids), and optic foramen funnel-shaped. 

Pliopedia Kellogg, 1921 

Type species. — Pliopedia pacifica Kellogg, 1921. 

Distribution. — Late late Miocene of the eastern North Pacific. 

Included species. — Type species only. 

Emended diagnosis. — A medium-sized member of the 
Odobenini with lesser tuberosity of the humerus expanded medially 
(possible autapomorphy of Pliopedia), braincase broadly convex, 
sagittal crest absent, lambdoidal crest with flattened traction sur- 



118 



Thomas A. Demere 



face, occipital shield hemispherical, humerus with low pectoral 
crest descending gradually to distal end. and deltoid tubercle well 
separated from pectoral crest (synapomorphies of the Odobenini). 

Pliopedia pacifica Kellogg, 1921 

Pliopedia pacifica Kellogg, 1 92 1 ; Repenning and Tedford 1 977; Barnes 
andRaschke 1991. 

Holotype. — USNM 13627, associated left humerus, radius, 
ulna, metacarpals, metatarsals, and phalanges. 

Type locality. — Santa Margarita. San Luis Obispo County, Cali- 
fornia. 

Horizon and age. — Paso Robles Formation, late late Miocene 
(Hemphillian NALMA correlative, ca. 5-6 Ma, Repenning and 
Tedford 1977:49). 

Characterization. — As for the genus. 

Referred material. — Repenning and Tedford ( 1 977:49) referred 
a partial skeleton (USNM 187328, including a braincase, rib and 
portions of the right and left forelimbs collected from the Etchegoin 
Formation, late late Miocene, Hemphillian NALMA correlative of 
central California) to P. pacifica. This referral was made on the 
basis of the morphology of the humerus. Although Barnes and 
Raschke (1991:13) subsequently removed this specimen from P. 
pacifica, citing work in progress, I retain it in this (axon. 

Discussion. — Repenning and Tedford (1977:49-53) discussed 
the type and referred material of Pliopedia pacifica thoroughly. The 
holotype itself consists of fragmentary material preserving little 
diagnostic morphology except a distal humeral articulation with the 
medial lip of the trochlea broader than the radial capitulum and a 
metacarpal I with a conspicuous pit for insertion of the pollicle 
extensor muscle (both odobenid synapomorphies). More important 
is the partial skeleton (USNM 187328) from the Etchegoin Forma- 
tion. The lesser tuberosity of the right humerus of this specimen 
(Repenning and Tedford 1977:pl. 17) is unique in being medially 
expanded and positioned distinctly below the proximal capitulum. 
The greater tuberosity is at the same level as the capitulum. The 
humerus also exhibits several features characteristic of the 
Odobenini, including a pectoral crest that gradually joins the shaft 
distally and a deltoid tubercle well separated from the pectoral 
crest. The partial skull of USNM 187328 (Repenning and Tedford 
1977:pl. 24, fig. 6) also exhibits several important features of the 
Odobenini. including a low and broadly rounded braincase, absence 
of a sagittal crest and its replacement by a sagittal sulcus, and a 
broad flattened and crescentic traction surface on the lambdoidal 
crest. The occipital shield is hemispherical in posterior aspect. 

Pliopedia pacifica differs fromAlachtherium in lacking a rectan- 
gular occipital shield and a deltoid tubercle joined with the pectoral 
crest of the humerus. It differs from Prorosmarus in lacking a deltoid 
tubercle joined with the pectoral crest of the humerus and from 
Valeniclus in its humerus' lacking an enlarged entepicondyle. It 
differs from Odobenus in the lesser tuberosity of the humerus being 
medially expanded. Repenning and Tedford (1977:52) noted the 
strong resemblance between the referred braincase and that of 
Odobenus. yet they chose to place Pliopedia with the dusignathine 
walruses. It is clear from the phylogenetic analysis that P. pacifica is 
a member of the Odobenini and the oldest known member of this 
group. This assignment suggests that Pliopedia pacifica also pos- 
sessed the elongated specialized tusks of the Odobenini. 

Alachtherium du Bus, 1867 

Type species. — A. cretsii du Bus, 1867. 
Distribution. — Pliocene of the eastern North Atlantic Ocean. 
Included species. — Type species only. 

Emended diagnosis. — A large member of the Odobenini with 
humerus approximately 15% larger than that of Odobenus 



rosmarus, posterior outline of occipital shield rectangular, lower 
postcanine teeth more widely spaced than those of Odobenus 
(autapomorphies of Alachtherium); external narial opening el- 
evated, palate elongated and vaulted, C enlarged and tusklike. C 
with central column of globular dentine, upper and lower postcanine 
teeth with single circular roots, mandibular symphysis elongated, 
mastoid process enlarged (synapomorphies of the Odobenini), man- 
dibular symphysis unfused, I 1 transversely in line with I : at anterior 
end of the premaxilla (not displaced posteriorly, adjacent to P 1 and 
medial of C 1 . as in Odobenus). external auditory meatus large and 
open in ventral aspect between postglenoid process and mastoid 
process (in Odobenus the two processes are closely appressed to 
one another), C 1 (tusk) more procumbent than in Odobenus, and 
coronoid process of mandible elongated (plesiomorphies relative to 
Odobenus). 

Alachtherium cretsii du Bus, 1867 

Alachtherium cretsii du Bus, 1867; van Beneden 1877; Kellogg 1922. 
Trichechus antverpiensis Rutten, 1907. 
Alachtherium antwerpiensis Hasse. 1910. 

Odobenus antx'erpiensis van der Feen 1968; Erdbrink and van Bree 
1986, 1990. 

Holotype. — Complete right dentary (IRSNB M.168) illustrated 
by van Beneden (1877) and Berry and Gregory (1906). 

Type locality. — Wyneghem. Fort I, Antwerp Basin, Belgium. 

Horizon and age. — Scaldisian sands; early Pliocene. 

Diagnosis. — As for the genus. 

Referred material. — Van Beneden (1877) referred several 
Scaldisian walrus specimens, including a partial cranium (IRSNB 
M. 169) and a complete humerus (IRSNB M.170), to this taxon. 

Taxonomic history. — Rutten (1907) examined the partial cra- 
nium of Beneden (IRSNB M.169), concluding that it was incom- 
patible with the holotype dentary. Rutten was concerned primarily 
with presumed size differences between the skull and dentary and 
with the large coronoid process of A. cretsii (Fig. 6D), which he 
suggested would be too massive to fit between the convex braincase 
and zygomatic arch. Rutten ( 1907) removed the Beneden cranium 
from his concept of A. cretsii and instead designated it the type of a 
new taxon, Trichechus antverpiensis. 

Hasse (1910) described cranial (Fig. 4E) and postcranial mate- 
rial belonging to at least four individuals, both adults and juveniles, 
from the younger Merxemian (= Poederlian, upper Pliocene) sands 
north of Antwerp (van der Feen 1968). Like Rutten (1907), Hasse 
concluded that the type jaw of A. cretsii was incompatible with his 
new walrus material and so erected yet another new species, 
Alachtherium antwerpiensis. 

Van der Feen (1968) illustrated and described a posterior cranial 
fragment collected from the mouth of the Scheldt River, Nether- 
lands, referring to it as Odobenus antverpiensis (Rutten) but not 
discussing why it was not referable to the genus Alachtherium or 
the species A. cretsii. 

Erdbrink and van Bree ( 1990) described and illustrated a nearly 
complete cranium (GMAU K-8052) from offshore Zeeland, Neth- 
erlands, referring it to Odobenus antverpiensis (Rutten) on the basis 
of its depressed (not ridgelike) sagittal suture, broad and high 
lambdoidal crest, weakly developed occipital crest, large size (rela- 
tive to O. rosmarus), anterior position of I 3 (anterior to C 1 . not 
posterior as in O. rosmarus). low (dorsoventral) position of the 
external narial opening (relative to the elevated position seen in O. 
rosmarus). and small peglike postcanine teeth with sharp edges. 
From illustrations of GMAU K-8052 it appears that this specimen 
is conspecific with the skull of A. antverpiensis figured by Hasse 
(1910) and the cranial fragment used by Rutten (1907) as the 
holotype of T. ant\<erpiensis. Similarities include the rectangular 
posterior outline of the occipital shield, broad and vaulted palate. 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



119 



large diastema between I 3 and C, sinuous postcanine tooth row 
(when the diastema is included as part of the tooth row's curvature), 
anteriorly placed incisors, and procumbent C Differences between 
the skulls include five postcanine teeth in Hasse's skull vs. four in 
GMAU K-8052, retention of I 1 in GMAU K-8052 vs. loss of this 
tooth in Hasse's skull, and more anterior placement of Pc 1 (medial 
to C) in Hasse's skull vs. posterior placement of this tooth in 
GMAU K-8052. Erdbrink and van Bree (1990) concluded that O. 
antverpiensis (Rutten) is a senior synonym of A antwerpiensis 
Hasse. Surprisingly, these authors failed even to mention A. cretsii 
and offered no reasons for not synonymizing O. antverpiensis 
(Rutten) with A. cretsii du Bus. 

Reasons for lumping all taxa into Alachtherium cretsii include 
the largeness of both the type dentary (Fig. 6D) and type and 
referred crania (Fig. 4E) of T. anh'erpiensis Rutten, A. antwerpi- 
ensis Hasse, and O. atmerpiensis (Rutten) (in van der Feen 1968; 
Erdbrink and van Bree 1990), sinuous lower and upper postcanine 
tooth rows (a tracing of the sinuous occlusal outline of the type jaw 
conforms to the sinuous occlusal outline of the upper jaw of both 
the Hasse skull and GMAU K-8052). and possession of two lower 
incisors in the type dentary of A. cretsii and two upper incisors in 
the skull of Hasse (Fig. 4E). The skull of Hasse also has a more 
elongated snout region than does that of Odobenus, which is com- 
patible with the rather elongated symphyseal region in the type 
dentary of A. cretsii. 

Possible reasons for retaining two species of Alachtherium in- 
clude the early Pliocene age (Scaldisian) of the holotype dentary of 
A. cretsii vs. the late Pliocene age (Poederlian) of the Hasse (1910) 
material (A. antwerpiensis) and retention of five postcanine teeth 
vs. four in GMAU K-8052. 

Discussion. — The features of Alachtherium cretsii linking it to 
the Odobenini include its elongated and ever-growing upper canine 
(with central column of globular dentine), C, reduced to size of the 
premolars, mastoid enlarged and descending ventrad to level of the 
hamular process of the pterygoid, sagittal crest lost and replaced by 
a sagittal sulcus marking the interparietal suture, palate elongated 
and longitudinally arched, pterygoid strut lost, lambdoidal crest 
flattened, and Pc' medial to C 1 . 

In general, Alachtherium preserves intermediate character states 
in the transition from the primitive dental condition of Aivukus to 
the derived condition of Odobenus. In adults of O. rosmarus, V 2 
are typically lost (Fay 1982), I 3 is displaced posteriorly to the 
medial side of C 1 and in line with the postcanine teeth, and Pc 1 is 
placed anterior and medial of C (Fig. 4G). In Alachtherium, I 1 is 
variably present, I 2 is always present, I 3 is consistently positioned 
anterior to C 1 , and Pc 1 is placed posterior to C (Fig. 4E). Major 
differences are seen in the lower jaws, where Odobenus has a 
strongly fused and massive symphysis and a reduced coronoid 
process (Fig. 6F). in contrast to the unfused and slender symphysis 
and large coronoid process of Alachtherium (Fig. 6D). 

The humerus (IRSNB M.170) of Alachtherium is longer than 
that of Odobenus (435 vs. 390 mm) and similarly slender, more 
slender than that of Valenictus. The pectoral crest is elongated but 
not sharply keeled. The deltoid tubercle is laterally displaced but 
still associated with the pectoral crest. The biciptal groove is 
broadly open, and the lesser tuberosity is distinctly below the 
proximal capitulum. The entepicondyle is not enlarged and is more 
quadrate than the triangular entepicondyle of Odobenus. 

I retain Alachtherium as a genus distinct from Odobenus be- 
cause of the differences noted in the postcanine teeth, the more 
procumbent canines, the more elongated rostrum, the broader pal- 
ate, the less elevated external narial opening, the unfused and 
unswollen mandibular symphysis, and features of the humerus. 

Dentition: 2(3)1. !C.4(5)Pc x 2 = 28(32) 
21, 1C, 4Pc 



Prorosmarus Berry and Gregory, 1906 

Type species. — Prorosmarus alleni Berry and Gregory, 1906 

Distribution. — Early Pliocene of the western North Atlantic. 

Included species. — Type species only. 

Characterization. — A medium-sized member of the Odobenini 
with lower postcanine teeth closely appressed to each other 
(apomorphy shared with Odobenus): distal tip of lower jaw rough- 
ened and pitted, mandibular symphysis elongate and sloping, lower 
canine equal in size to Pc, (i.e.. C, reduced), horizontal ramus in 
occlusal aspect laterally concave anteriorly, becoming convex in 
region of postcanines (synapomorphies of the Odobenini), 
postcanine teeth with single nearly circular roots and simple crowns 
(odobenine synapomorphies), and unfused mandibular symphysis 
(plesiomorphy shared with most pinnipeds). 

Prorosmarus alleni Berry and Gregory, 1906 

Prorosmarus alleni Berry and Gregory, 1 906; Kellogg 1 922; Repenning 
andTedford 1977; Barnes and Raschke 1991. 

Holotype— USNM 9343, partial left dentary. 

Type locality. — Beach at Yorktown, Virginia. 

Horizon and age. — Yorktown Formation, early Pliocene (latest 
Hemphillian NALMA correlative, ca. 3.5-5 Ma). 

Characterization. — As for the genus. 

Referred material— Left humerus (MCZ 7713; C. E. Ray in 
Repenning and Tedford 1977:13); numerous isolated skeletal ele- 
ments in the USNM collections from the Lee Creek Mine (C. E. 
Ray, pers. comm.). All referred material collected from the 
Yorktown Formation. 

Discussion. — The holotype dentary is missing the mandibular 
condyle, coronoid process, pterygoid process, and masseteric fossa. 
The horizontal ramus is deep dorsoventrally and moderately thick 
transversely. The unfused symphysis extends below the anterior 
border of Pc,. There are many similarities with the holotype dentary 
of Alachtherium cretsii. including reduced and premolariform ca- 
nine, possession of alveoli for four postcanine teeth (presumably 
P,^,, Berry and Gregory 1906) and two incisors, single-rooted and 
circular postcanine alveoli, distinct but unswollen genial tuberosity, 
well-developed marginal process, large and deeply set mental fora- 
men beneath intra-alveolar septum separating Pc, and Pc 2 , and 
dorsally expanded and pitted incisive margin. 

Important differences between the type dentaries of P. alleni and 
A. cretsii include smaller overall size, less upturned symphysis, 
position of mental foramen below Pc , rather than Pc 2 , and postcanine 
alveoli closely appressed to each other and positioned squarely on 
alveolar rather than medial margin of dentary. C. E. Ray (pers. 
comm.) believes these and other features suggest a close relation- 
ship, perhaps conspecificity, between the two taxa. Prorosmarus 
may be the female of Alachtherium. Sexual dimorphism may be 
responsible for the greater intra-alveolar distance, larger overall size, 
and greater degree of symphyseal upturning seen in A. cretsii. 
Resolution of this issue must await discovery of diagnostic cranial 
material of Prorosmarus from the Yorktown Formation. 

As discussed by Repenning and Tedford (1977:13), the left 
humerus (MCZ 7713) referred to this taxon falls within the size 
range of modern Odobenus humeri. It differs, however, in the 
position of the deltoid insertion, which is still located on the pectoral 
crest rather than being a separate tubercle on the lateral side of the 
shaft. Other differences include a wider bicipital groove, more 
robust distal extremity, and less triangular entepicondyle. Impor- 
tantly, the pectoral crest is more keeled and descends with some 
flexure to the distal portion of the shaft. This flexion, however, is not 
as pronounced as in Imagotaria, Dusignathus, Gomphotaria, or 
Aivukus. 



120 



Thomas A. Demere 



On the basis of these features, P. alleni is a member of the 
Odobenini close to the common ancestry of Alachtherium and 
Odobenus. 

Dentition: ?I. 1C. ?P. ?M 
21, 1C. 4P, 0M 

Valenictus Mitchell, 1961 

Type species. — Valenictus imperialensis Mitchell, 1961. 

Distribution. — Late late Miocene and late Pliocene of the east- 
ern North and Central Pacific. 

Included species. — Valenictus imperialensis Mitchell, 1961, 
and Valenictus clutlavistensis Demere, 1994 (this volume). 

Definition. — The monophyletic group containing the most re- 
cent common ancestor of V. imperialensis and V. clutlavistensis and 
all of its descendants. 

Emended diagnosis. — A member of the Odobenini with humeri 
characterized by pectoral crest broad, not a sharply keeled ridge, 
greater tuberosity thickened, entepicondyle greatly enlarged, shaft 
short and robust, and bicipital groove narrow (synapomorphies of 
Valenictus). 

Valenictus imperialensis Mitchell, 1961 

Valenictus imperialensis Mitchell, 1961 ; Mitchell 1968; Repenning and 
Tedford 1977. 

Holotype.— LACM (C1T) 3926, nearly complete left humerus. 

Type locality: — LACM (C1T) locality 472, Coyote Mountains, 
Imperial County, California. 

Horizon and age. — Imperial Formation (Coyote Mountain 
Clays), late late Miocene to early Pliocene (Hemphillian NALMA 
correlative, 4-6 Ma). 

Emended diagnosis. — A species of Valenictus with entepicon- 
dyle of humerus rounded, knoblike, and positioned distally and 
posterior outline of humeral shaft nearly straight (apparent autapo- 
morphies of V. imperialensis). 

Referred material. — Repenning and Tedford (1977:53) referred 
USNM 13643, the distal end of a right humerus collected from the 
San Joaquin Formation (latest Miocene to early Pliocene, probable 
Hemphillian NALMA correlative of inland central California) to 
V. imperialensis. As discussed below, close examination of this 
specimen suggests it is associated more closely with Valenictus 
clutlavistensis. 

Discussion. — Features shared with the modern walrus, 
Odobenus, include a greater tuberosity thickened and only slightly 
elevated above the proximal capitulum, a pectoral crest broad and 
not developed as a sharply keeled ridge, and a deltoid tubercle 
separated from the pectoral crest and located on the lateral margin 
of the shaft. The humerus is relatively stocky compared to those of 
Odobenus, Alachtherium, and Imagotaria. 

Previous workers have assigned Valenictus to the Dusigna- 
thinae, perhaps because of its many unique derived features. Its 
possession of a deltoid tubercle located posterolateral to a low 
pectoral crest and a greater tuberosity elevated only slightly above 
the proximal capitulum, however, clearly demonstrate its affinities 
with the Odobenini. 

The single specimen of Valenictus imperialensis is from a ma- 
rine rock unit known for its distinctive assemblages of warm-water 
molluscan taxa of Caribbean and tropical Pacific affinities (Demere 
1993). 

Valenictus chulavistensis Demere, 1994 

Valenictus chulavistensis Demere, 1994 (this volume). 

Type material. — Holotype, SDSNH 36786, a partial skeleton 
preserving the left side of the skull and mandible as well as nearly 



every postcranial element. Paratype, SDSNH 38227. a nearly com- 
plete skull with both canines. 

Holotype and paratype locality. — SDSNH locality 3551, Chula 
Vista. San Diego County. California. 

Horizon and age. — San Diego Formation, late Pliocene 
(Blancan NALMA correlative, ca. 2-3 Ma). 

Diagnosis. — A large species of Valenictus with edentulous 
dentary, edentulous premaxilla and postcanine portion of maxilla, 
osteosclerotic long bones, astragalus with broad sulcus calcanei. 
very reduced collum tali, and coalesced navicular and sustentacular 
facet (autapomorphies of V. chulavistensis). Distinguished from V. 
imperialensis by the following features of the humerus: larger size, 
more sigmoidal posterior profile, sharply keeled supinator ridge, 
more robust and rectangular entepicondyle, and more obtuse angle 
between the shaft and the axis of the distal trochlea. 

Referred material. — Numerous cranial, dental, and postcranial 
remains from the San Diego Formation (see Demere 1994. this 
volume). The distal fragment of a right humerus (USNM 13643) 
from the San Joaquin Formation (latest Miocene to early Pliocene, 
probable Hemphillian NALMA correlative of inland central Cali- 
fornia), originally referred to V. imperialensis by Repenning and 
Tedford (1977:53). 

Discussion. — Valenictus clutlavistensis is known from almost 
every skeletal element (see Demere 1994, this volume). Dentally, it 
is the most divergent odobenid, having lost all teeth except the 
enlarged upper canines (Fig. 4F). This taxon has the following 
synapomorphies of the Odobenini: upper canines enlarged and 
ever-growing, with three dental layers (globular dentine, ortho- 
dentine, and cementum), palate elongated and arched longitudi- 
nally as well as transversely, mastoid process as widest part of skull, 
temporal fossa shortened with blunt and robust zygomatic arch, 
hamular process of pterygoid broad, pterygoid strut lost, lambdoidal 
crest flattened, sagittal crest lost, and optic foramen funnel-shaped. 

The mandible of V. chulavistensis is completely edentulous and 
delicate with a narrow but fused and strongly upturned symphysis 
(Fig. 6E). The alveolar margin is sharply keeled, and the occlusal 
outline is sinuous, accommodating the enlarged canines of the 
upper jaw. 

The humerus of V chulavistensis has the stocky shaft and 
enlarged entepicondyle characteristic of the genus. The distal end of 
a humerus from the San Joaquin Formation (USNM 13643) re- 
ferred to V imperialensis by Repenning and Tedford (1977) is 
better referred to V chulavistensis because its entepicondyle is 
more cubic, the widest portion of its entepicondyle is positioned 
more proximally, and the long axis of its entepicondyle is rotated 
anterodorsally. 

Dentition: 01. 1C.0P. 0M x 2 = 2 
01, 0C, OP, 0M 



Odobenus (Brisson, 1762) 

Type species. — Odobenus rosmarus (Linnaeus, 1758). 

Distribution. — Pleistocene and Holocene of the North Atlantic, 
North Pacific, and Arctic oceans. 

Included species. — Odobenus rosmarus (Linnaeus, 1758); 
Odobenus mandanoensis Tomida. 1989; Odobenus huxleyi 
(Lankester, 1865); Odobenus koninckii (van Beneden, 1871). 

Definition. — The monophyletic group containing the most re- 
cent common ancestor of O. huxleyi and O. rosmarus and all of its 
descendants. 

Emended diagnosis. — A member of the Odobenini with man- 
dibular symphysis fused and greatly reinforced, lower canine incor- 
porated into postcanine tooth row, external nares well elevated 
above the incisive margin of the premaxilla, external overlap of 
premaxilla and nasal lost, C enlarged as a tusk and oriented nearly 



The Family Odohenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



121 



vertically, I 1 : and 1, , lost (sometimes present as an atavism). I 3 
posteriorly placed (relative to C 1 ). external auditory meatus con- 
stricted anteroposteriorly (autapomorphies of Odobenus), C a tusk 
and composed of three layers including a central column of globu- 
lar dentine, palate elongated and vaulted, palate telescoped beneath 
braincase, hamular process of pterygoid enlarged and laterally di- 
rected, lambdoidal crest flattened, sagittal crest lost and replaced by 
a sulcus, temporal fossa shortened, and external narial opening 
elevated (synapomorphies of the Odobenini). 

Odobenus rosmarus (Linnaeus, 1758) 

Phoca Rosmarus Linnaeus, 1758 

Odobenus rosmarus (Linnaeus. 1758): Harington 1984; Harington and 
Beard 1992; Harington et al. 1993; Fay 1982, 1985; Erdbrink and van Bree 
1986, 1990 

Triehechus virginianus De Kay, 1 842 

Trichechus huxleyi (in part): Rutten 1907 

Holotype. — None designated. 

Type locality. — None designated. 

Horizon and age. — Pleistocene to Recent. 

Diagnosis. — As for the genus. 

Discussion. — Fay (1982) recognized two subspecies of 
Odobenus rosmarus. O. r. rosmarus from the North Atlantic and O. 
r. divergens from the North Pacific. Fossil and/or subfossil remains 
closely resembling modern O. rosmarus (= Trichechus virginianus) 
have been recovered from coastal deposits and the inner continental 
shelf of eastern North America (Ray 1975; Harington 1977; 
Harington et al. 1993; Parris 1983). and from the English Channel 
and North Sea (Rutten 1907; van der Feen 1968; Erdbrink and van 
Bree 1986). Included in this material are several complete and nearly 
complete crania as well as isolated tusks, mandibles, and assorted 
postcrania. Late Wisconsin remains of O. rosmarus have also been 
reported from British Columbia (Harington and Beard 1992) and 
California (Harington 1984). Erdbrink and van Bree (1986) re- 
viewed English and Dutch Pleistocene walrus material and offered 
anatomical criteria for assigning it to O. rosmarus. Undescribed 
Pleistocene crania from the western Atlantic seaboard in USNM are 
conspecific with O. rosmarus (Ray 1992). Odobenus rosmarus is 
currently the only nominal species of Odobenus for which the skull 
(Fig. 4G) and mandible (Fig. 6F) are confidently known. 

Dentition: II. 1C. 3(4)Pc x 2 = 18(19) 
01, 1C. 3Pc 

Fay (1982) reported that some Pacific walruses have three 
incisors, four premolars, and two molars as an atavism. 

Odobenus huxleyi (Lankester, 1 865) 

Trichechodon huxleyi Lankester, 1865. 

Trichecus (sic) huxleyi Lankester 1 880. 

not Trichechus huxleyi Rutten 1907. 

Trichechodon huxleyi (in part) Kellogg 1922. 

Odobenus huxleyi van der Feen 1968; Erdbrink and van Bree 1986. 

Holotype. — Partial upper canine, BMNH 46000. 

Type locality. — Sutton, Suffolk County. England. 

Horizon and age. — Red Crag (Waltonian). early Pleistocene 
(Erdbrink and van Bree, 1986). 

Emended diagnosis. — A species of Odobenus known confi- 
dently only from tusks, which have cementum and outer 
orthodentine layers thinner than those of O. rosmarus (possible 
autapomorphies of O. huxleyi). 

Discussion. — Lankester (1865, 1880) stated that Trichechodon 
huxleyi was distinguishable from O. rosmarus by the greater curva- 
ture and diameter of its tusks. Erdbrink and van Bree ( 1986) found 
that the supposed greater curvature and diameter of Lankester's 
tusks fall within the range of variation of modern Odobenus 



rosmarus and therefore are unreliable for distinguishing the two 
taxa. They noted instead that the cementum and outer orthodentine 
layers of the type and referred tusks of O. huxleyi are much thinner 
than those of O. rosmarus and so used this character complex as the 
sole diagnostic feature of O. huxleyi. 

Since Lankester"s original description, additional specimens 
have been referred to O. huxleyi, most notably a nearly complete 
cranium (with tusks) illustrated by Rutten (1907). In their review of 
North Sea fossil odobenids Erdbink and van Bree ( 1986) restricted 
the concept of O. huxleyi to the type dental material and several 
additional tusks dredged from the North Sea. Under their view, 
Rutten's and several other cranial specimens referred to O. huxleyi 
are instead assignable to O. rosmarus. It should be pointed out, 
however, that Erdbrink and van Bree (1986) also assigned speci- 
mens here referred to Alachtherium (i.e., A. antverpiensis = A. 
cretsii) to Odobenus (i.e., O. antverpiensis). These authors included 
in Odobenus all odobenids with tusks possessing a central column 
of globular dentine. Erdbrink and van Bree ( 1986) suggested that O. 
huxlevi may prove to be a senior synonym of A. antverpiensis if and 
when tusks having thin cementum and outer orthodentine layers 
and so positively assignable to the latter species are found. Since 
thin dental layers could actually have arisen in the common ances- 
tor of O. huxleyi and A. cretsii. this feature alone may not be 
sufficient to diagnose a taxon, and O. huxleyi may not be diagnos- 
able at all (i.e.. a nomen nudum). In this report, the concept of O. 
huxleyi is that proposed by Erdbrink and van Bree ( 1986). 

Odobenus koninckii sensu lato (van Beneden, 1 87 1 ) 

Trichechodon koninckii van Beneden, 1871 
Trichechodon koninckii (in part) van Beneden 1877 
Trichechodon huxleyi (in part) Kellogg 1922 

Holotype. — Partial upper canine (original lost, but cast sur- 
vives; IRSNB cast 2892; Rutten 1907). 

Type locality. — Antwerp, Belgium. 

Horizon and age. — Scaldisian sands, early Pliocene. 

Emended characterization. — A species of Odobenus with four 
equally large lower postcanine teeth behind a canine of larger 
circumference. 

Referred material. — Partial left dentary (van Beneden 1877; no 
catalog number). 

Discussion. — The cast now serving as the basis for Trichecho- 
don koninckii is not by itself diagnostic except in its preservation of 
the dental-layer synapomorphies of the Odobenini. Therefore 
Rutten ( 1907) and van der Feen (1968) dismissed T. koninckii as a 
nomen nudum. 

This designation would be easy to accept had van Beneden 
(1877) not referred additional fossils to this taxon. Most important 
is a partial left dentary preserving alveoli for C,-Pc 4 (van Beneden 
1877: pi. 6. figs. 5-7). This specimen appears to have had a broad 
and fused symphysis (C. E. Ray, pers. comm.). a synapomorphy of 
Odobenus. Its other important features include nearly circular and 
equal-sized alveoli for Pc,^,, very thin septa between the alveoli, 
straight tooth row with the alveolus for C, closely appressed to that 
of Pc,, diameter of C, greater than those of Pc M , and sigmoidal 
lateral outline of horizontal ramus in occlusal aspect. This character 
complex defines a taxon close to modern Odobenus and indicates 
that the genus may have evolved during the Pliocene. The taxo- 
nomic relationship of the referred dentary to the type of O. koninckii 
is unclear and may never be known unless more complete material 
is found. Thus I refer to this taxon as O. koninckii sensu lato. 

Odobenus mandanoensis Tomida. 1989 

Odobenus mandanoensis Tomida. 1989. 

Holotype. — NSM-PV 18911, partial mandible preserving sym- 



122 



Thomas A. Demere 



physeal portions of left and right rami with roots of left C,, P 2 , and 
right C,. 

Type locality. — Sand and gravel mine near village of Mandano. 
city of Kisarazu, Chiba Prefecture, Japan. 

Horizon and age. — Mandano Formation, upper Pleistocene (ca. 
0.5 Ma). 

Emended diagnosis. — A species of Odobenus with symphysis 
widest above the mental foramen rather than below, portion of 
dentary lateral to alveoli for Pc, thicker than in O. rosmarus, dorsal 
longitudinal margin of symphysis gently sloping, and diastemata 
between C, and Pc, wider than in O. rosmarus (autapomorphies of 
O. mandanoensis). 

Discussion. — This species' most notable feature is its massive 
and strongly fused mandibular symphysis, a synapomorphy of 
Odobenus. Tomida (1989) was not certain that his taxon differed 
from O. huxleyi but distinguished it from O. koninckii sensu lato 
(i.e., van Beneden's partial mandible) by its smaller symphyseal 
area and more steeply sloping dorsal longitudinal symphyseal mar- 
gin. 

Features of the Odobenini preserved in the holotype include the 
following: lower canine incorporated into postcanine tooth row, 
symphysis strongly fused and massive, incisive margin of symphy- 
sis with precanine constriction, lower incisors lost, right and left 
tooth rows forming angle of 40° to 42° at the symphysis, lower 
canine premolariform, and adult teeth lacking enamel. This taxon 
demonstrates that Odobenus was in the western North Pacific as 
early as 500,000 years ago. 

Dentition: ?I. ?C. ?P. ?M x 2 
01. 1C. 1+P. ?M 

SUMMARY 

Computer-assisted phylogenetic analysis of fossil and living 
odobenids supports the monophyly of the group and recognizes two 
principal clades. the Dusignathinae and the Odobeninae. This phy- 
logenetic framework suggests that odobenids evolved during the 
middle Miocene in the North Pacific Ocean and diversified during 
the later Miocene, dispersing to the North Atlantic Ocean by the 
early Pliocene. The earliest odobenids lacked enlarged upper ca- 
nines, confirming that "tusks do not a walrus make." Tusks evolved 
independently in the dusignathine and odobenine lineages. 
Dusignathine walruses developed enlarged lower as well as upper 
canines, while odobenines evolved the greatly enlarged tusks seen 
in modern Odobenus. The evolution for social display of these 
enormous structures is associated with many other modifications of 
the skull and mandible. This high degree of divergence is recog- 
nized in the naming of a new taxon, the Odobenini, which contains 
the fossil genera Alachtherium, Pliopedia. Prorosmarus, and 
Valenictus as well as modern Odobenus. 

ACKNOWLEDGMENTS 

I gratefully acknowledge Clayton E. Ray (USNM), Lawrence 
G. Barnes (LACM), and John A. Minch (John Minch & Associates) 
for allowing me to study specimens in their care. This report 
benefited from discussions with Annalisa Berta, Lawrence G. 
Barnes, Matthew W. Colbert, and Clayton E. Ray. Matthew W. 
Colbert, Clayton E. Ray. Charles A. Repenning. and Blaire Van 
Valkenburgh provided critical reviews of the manuscript. 

LITERATURE CITED 

Barnes, L. G. 1971. Imagotaria (Mammalia: Otariidae) from the late 
Miocene Santa Margarita Formation near Santa Cruz, California. 



PaleoBios 10. 

. 1972. Miocene Desmatophocinae (Mammalia: Carnivora) from 

California. University of California Publications in Geological 
Sciences 89. 

— . 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: Otariidae) from Washington. Natural History Mu- 
seum of Los Angeles County Contributions in Science 382. 

. 1988. A new fossil pinniped (Mammalia: Otariidae) from the 

middle Miocene Sharktooth Hill Bonebed California. Natural His- 
tory Museum of Los Angeles County Contributions in Science 396. 

. 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. 

. 1990. A new enaliarctine pinniped of the genus Pteronarctos 

(Mammalia: Carnivora) from the Astoria Formation, Oregon. Natu- 
ral History Museum of Los Angeles County Contributions in Sci- 
ence 422. 

— . 1992. A new genus and species of middle Miocene enaliarctine 
pinniped (Mammalia: Carnivora) from the Astoria Formation in 
coastal Oregon. Natural History Museum of Los Angeles County 
Contributions in Science 431. 

, 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. 

Beneden, P. J. van. 1877. Description des ossements fossiles des envi- 
rons d'Anvers. Premiere partie. Pinnipedes ou amphitheriens. 
Annales du Musee Royal d'Historie Naturelle de Belgique 1:1-88. 

Berry. E. W., and W. K. Gregory. 1 906. Prorosmarus alleni, a new genus 
and species of walrus from the upper Miocene of Yorktown, Vir- 
ginia. American Journal of Science 244:60-62. 

Berta, A. 1991. New £>M//flr(7(«(Pinnipedimorpha)from theOligocene 
and Miocene of Oregon and the role of "enaliarctids" in pinniped 
phylogeny. Smithsonian Contributions to Paleobiology 69:1-33. 

, 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 :1 11-126. 

. and C. E. Ray. 1990. Skeletal morphology and locomotor 

capabilities of the archaic pinniped Enaliarctos mealsi. Journal of 
Vertebrate Paleontology 10:141-157. 

, and A. R. Wyss. 1994. Pinniped phylogeny. 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 29:33-56. 

Davis, D. D. 1964. The giant panda. A morphological study of evolu- 
tionary mechanism. Fieldiana: Zoology Memoirs 3:1-334. 

Demere, T. A. 1993. Fossil mammals from the Imperial Formation 
(upper Miocene-lower Pliocene), Coyote Mountains, Imperial 
County, California. Pp. 82-85 in R. E. Reynolds and J. Reynolds 
(eds.). Ashes, Faults, and Basins. San Bernardino County Museum 
Association Special Publication 93-1. 

. 1994. Two new species of fossil walruses (Pinnipedia: 

Odobenidae ) from the upper Pliocene San Diego Formation. Cali- 
fornia. In A. Berta and T. A. Demere (eds.). Contributions in marine 
mammal paleontology honoring Frank C. Whitmore, Jr. Proceed- 
ings of the San Diego Society of Natural History 29:77-98. 

Donoghue, M. J. 1985. A critique of the biological species concept and 
recommendations for a phylogenetic alternative. Bryologist 
88:172-181. 

Erdbrink. D. P. B.. and P. J. H. van Bree. 1986. Fossil Odobenidae in the 
Dutch collections (Mammalia. Carnivora). Beaufortia 36:13-33. 

. and — . 1990. Further observations on fossil and subfossil 

odobenid material (Mammalia: Carnivora) from the North Sea. 
Beaufortia 40:85-101. 

Fay. F. H. 1982. Ecology and biology of the Pacific walrus, Odobenus 
rosmarus diverge/is Illiger. North American Fauna 74. 

. 1985. Odobenus rosmarus. Mammalian Species 238. 



The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa 



123 



Feen. P. J. van der. 1968. A fossil skull fragment of a walrus from the 
mouth of the river Scheldt (Netherlands). Bijdragen tot de 
Dierkunde 38:23-30. 

Harington, C. R. 1977. Marine mammals in the Champlain Sea and the 
Great Lakes. Annals of the New York Academy of Sciences 
288:508-537. 

. 1984. Quaternary marine and land mammals and their 

paleoenvironmental implications — some examples from northern 
North America. Carnegie Museum of Natural History Special Pub- 
lication 8:51 1-525. 

, and G. Beard. 1992. The Qualicum walrus: A late Pleistocene 

walrus (Odobenus rosmarus) skeleton from Vancouver Island. Brit- 
ish Columbia, Canada. Annales Zoologici Fennici 28:31 1-319. 
— , T. W. Anderson, and C. G. Rodrigues. 1993. Pleistocene walrus 
(Odobenus rosmarus) from Forteau, Labrador. Geographie Phy- 
sique et Quaternaire 47: 1 1 1 - 1 1 8. 

Hasse, G. 1910. Les morses du Pliocene Poederlien a Anvers. Bulletin 
de la Societe Beige de Geologic de Paleontologie et d'Hydrologie, 
Memoire 23:293-322. 

Hennig.W. 1966. Phylogenetic Systematics. University of Illinois Press. 
Urbana, Illinois. 

Kellogg. R. 1921. A new pinniped from the upper Pliocene of Califor- 
nia. Journal of Mammalogy 2:212-226. 

. 1922. Pinnipeds from Miocene and Pleistocene deposits of 

California. University of California Publications in Geological 
Sciences 13:23-132.' 

. 1925. Structure of the flipper of a Pliocene pinniped from San 

Diego County, California. Camegie Institution of Washington Pub- 
lication 348:97-116. 

. 1927. Fossil pinnipeds from California. Carnegie Institution of 

Washington Publication 346:27-37. 

1931. Pelagic mammals from the Temblor Formation of the 



Kern River region. California. Proceedings of the California Acad- 
emy of Sciences (4th series) 19:217-397. 

King. J. E. 1983. Seals of the World. Second edition. Comstock, Ithaca, 
New York. 

Lankester, E. R. 1865. On the sources of the mammalian fossils of the 
Red Crag, and on the discovery of a new mammal in that deposit, 
allied to the walrus. Quarterly Journal of the Geological Society of 
London 2 1(3):22 1-232. 

. 1880. On the tusks of the fossil walrus, found in the Red Crag of 

Suffolk. Linnaean Society of London Transactions. Series 2, Zool- 
ogy 2:213-221. 

Maddison. W. P.. and D. R. Maddison. 1992. MacClade: Analysis of 
Phylogeny and Character Evolution. Version 3.0. Sinauer, 
Sunderland. Massachusetts. 

Maddison, W. P., M. J. Donoghue. and D. R. Maddison. 1984. Outgroup 
analysis and parsimony. Systematic Zoology 33:83-103. 

Mitchell, E. D. 1961. A new walrus from the Imperial Pliocene of 
southern California: With notes on odobenid and otariid humeri. 
Los Angeles County Museum Contributions in Science 44. 



— . 1968. The Mio-Pliocene pinniped Imagotaria. Journal of the 

Fisheries Research Board of Canada 25:1843-1900. 

— , and R. H. Tedford. 1973. The Enaliarctinae. A new group of 

extinct aquatic Carnivora and a consideration of the origin of the 

Otariidae. Bulletin of the American Museum of Natural History 

151:201-284. 
Pams, D. C. 1983. New and revised records of Pleistocene mammals of 

New Jersey. Mosasaur 1:1-21. 
Queiroz. K. de, and J. Gauthier. 1990. Phylogeny as a central principle 

in taxonomy: Phylogenetic definitions of taxon names. Systematic 

Zoology 39:307-322. 
Ray. C. E. I960. Trichecodon huxleyi (Mammalia: Odobenidae) in the 

Pleistocene of southeastern United Stales. Bulletin of the Museum 

of Comparative Zoology 122: 129-142. 

— . 1975. The relationships of Hemicaulodon effodiens Cope 1 869 

(Mammalia: Odobenidae). Proceedings of the Biological Society 

of Washington 88:281-304. 

— . 1976. Fossil marine mammals of Oregon. Systematic Zoology 

25:420-436. 

— . 1992. The time has come, the walrus said. Virginia Explorer 

8(l):2-7. 
Repenning, C. A. 1976. Adaptive evolution of sea lions and walruses. 

Systematic Zoology 25:375-390. 
, 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 the Changing Environment. Oregon 

State University Press, Corvallis, Oregon. 
Rowe. T. 1988. Definition, diagnosis, and origin of Mammalia. Journal 

of Vertebrate Paleontology 8:241-264. 
Rutten, L. 1907. On fossil trichechids from Zealand and Belgium. 

Proceedings of the Section of Sciences. Koninklijke Akademie van 

Wetenschappen te Amsterdam 10:2-14. 
Swofford. D. L. 1993. PAUP: Phylogenetic Analysis Using Parsimony, 

Version 3.1. Computer program distributed by the Illinois Natural 

History Survey, Champaign, Illinois. 
Tomida, Y. 1989. A new walrus (Carnivora, Odobenidae) from the 

middle Pleistocene of the Boso Peninsula. Japan, and its implica- 
tions for odobenid paleobiogeography. Bulletin of the National 

Science Museum. Tokyo, Series C, 15:109-119. 
True. F. W. 1909. A further account of the fossil sea lion. Pontolis 

magntts, from the Miocene of Oregon. Pp. 143-148 in W. H. Dall 

(ed.). The Miocene of Astoria and Coos Bay, Oregon. United States 

Geological Survey Professional Paper 59. 
Wyss, A. R.. 1987. The walrus auditory region and the monophyly of 

pinnipeds. American Museum Novitates 287. 
, 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 Phylogeny, Placentals. Springer- Verlag, 

New York. New York. 



Phylogenetic Relationships of Platanistoid River Dolphins 
(Odontoceti, Cetacea): Assessing the Significance of Fossil Taxa 

Sharon L. Messenger 

Department of Biology, San Diego State University. San Diego, California 92182 

ABSTRACT. — The superfamily Platanistoidea (sensu Simpson 1945) includes four extant monotypic genera of mostly freshwater dolphins 
(Inia geoffrensis . Pontoporia blainvillei, Lipotes vexillifer, and Platanista gangetica) and approximately 20 fossil species. Character states 
diagnosing the Platanistoidea are almost entirely primitive, thus uninformative in revealing phylogenetic relationships. Recent phylogenetic 
analyses question the monophyly of the group and suggest that some of the taxa are more closely related to members of the Delphinoidea (i.e., extant 
and fossil dolphins, porpoises, narwhals, and belugas). Studies of soft-anatomical characters, including nasal passage anatomy and facial 
musculature, have elucidated relationships within the extant Odontoceti but have not resolved the status of the Platanistoidea. Although soft- 
anatomical characters often cannot be inferred from fossils, fossil taxa improve resolution, especially within the Platanistoidea. for the following 
reasons: morphological diversity seen in these fossils provides insight into the variability and distribution of some osteological characters, some 
fossil families (e.g., the Squalodontidae and Eurhinodelphidae) have been proposed as the nearest relatives of at least some of the extant 
Platanistoidea, and some of these fossil taxa represent groups temporally close to the ancestral node, allowing more accurate resolution of the 
ancestral condition at the internal nodes of the cladogram. If these fossil families are closely related to the Platanistoidea. their exclusion from 
phylogenetic studies could lead to incorrect polarity assessment, incomplete views of character evolution, and specious conclusions of relationships. 
Fossil taxa sometimes have been used, however, when their monophyly or phylogenetic position within the Odontoceti were in question. 
Recognizing nonmonophyletic groups may effectively exclude taxa from the analysis, again decreasing the probability of recovering the true 
phylogeny. The best inference of phylogenetic relationships will ultimately come from consideration of all available data, including fossil taxa. 
molecular data, and soft-anatomical characters, analyzed with rigorous phylogenetic methods. 



INTRODUCTION 

Platanistoid (sensu Simpson 1945) river dolphins include four 
extant monotypic genera of mostly freshwater dolphins found only 
in the Amazon (Inia geoffrensis), Yangtze (Lipotes vexillifer), and 
Ganges and Indus (Platanista gangetica) river systems and a re- 
stricted area of the southwest Atlantic Ocean (Pontoporia 
blainvillei). Additionally, approximately 20 fossil species, exclud- 
ing fragmentary material, have been regarded as closely related to 
river dolphins (Muizon 1987:13, 1988a:162). Currently, the river 
dolphins are among the most endangered of all cetaceans (Brownell 
et al. 1989), yet their basic biology, including their systematic 
relationships, remains poorly known. 

The taxonomy of the river dolphins has fluctuated for more than 
100 years. Some researchers (Flower 1867; Winge 1921; Slijper 
1936; Simpson 1945) have proposed a monophyletic origin for river 
dolphins, placing the genera either into one family, the Platanistidae, 
or into separate families within the same superfamily, the 
Platanistoidea, the latter arrangement emphasizing their great mor- 
phological differences. Others (Gray 1863, 1866; Miller 1918, 1923; 
Kellogg 1928) have regarded the extant river dolphins as polyphyl- 
etic, generally placing Pontoporia within the Delphinidae. During 
the second half of this century the river dolphins' monophyly has 
been widely accepted (Hershkovitz 1966;Kasuya 1973; Pilleri et al. 
1982; Zhou 1982; Barnes 1985; Barnes etal. 1985;Gaskin 1985; for 
opposing views see Rice 1 977; Fordyce 1 983 ), despite the characters 
diagnosing the group, such as a long, narrow rostrum and elongate 
mandibular symphysis, being demonstrably primitive or equivocal 
at the level of the Platanistoidea. Thus the monophyly of river 
dolphins has not been established on the basis of shared derived 
features. Recent phylogenetic analyses question it (Muizon 1984, 
1987, 1988a, 1991;Heyning 1989) and suggest that some genera are 
more closely related to members of the Delphinoidea, which include 
the dolphins, porpoises, narwhals and belugas. Yet none of these 
analyses has attempted to incorporate all available data (i.e., some 
analyses have not included fossils as terminal taxa, while others have 
excluded soft-anatomical characters). 



'Present Address: Department of Zoology. University of Texas. 
Austin, Texas 78712-1064. 



Both Heyning (1989) and Muizon (1984, 1987, 1988a, 1991) 
have attempted to reconstruct the phylogenetic relationships of 
odontocete whales by using cladistic methodology, yet each used 
quite different approaches. Heyning (1989) analyzed the relation- 
ships of extant families of odontocetes by using a large number of 
soft-tissue characters, while Muizon (1984, 1987. 1988a. 1991), 
using osteological characters, focused on fossil taxa. These studies 
have resolved some odontocete relationships, but some of their 
hypotheses conflict. It is not my objective in this paper to compare 
these hypotheses to detect the effects of fossil taxa in phylogenetic 
studies, as the studies differ not only in the inclusion or exclusion of 
fossils but also in the choice of characters included, method of 
polarity assessment, and use of computer-assisted programs to 
generate most parsimonious trees. These studies simply represent 
the current state of knowledge of the relationships of odontocete 
whales, within the context of which I investigate the effect of the 
exclusion of fossils in resolving river dolphin relationships. 

I have taken data on fossil taxa from Muizon (1984, 1987, 
1988a, 1991), although his inclusion of nonmonophyletic fossil 
taxa and use of fossil taxa with unresolved relationships may under- 
mine his hypotheses, as will be seen below. 

PREVIOUS CLADISTIC STUDIES 

With the addition of fossil taxa into a phylogenetic analysis of 
the Odontoceti. Muizon (1984) concluded that the river dolphins 
are paraphyletic (i.e.. not including all of the descendants of their 
most recent common ancestor). The fossil families included in his 
studies, such as the Squalodontidae, Squalodelphidae. and Eurhino- 
delphidae. are important in their being more diverse osteologically 
than any extant odontocete family. When included in an analysis 
with extant odontocetes. their unique combination of primitive and 
derived character states introduced a greater degree of character 
conflict and imposed topological changes in the phylogenetic hy- 
potheses. Among the extant river dolphins, Muizon (1988a, 1991) 
retained only Platanista in the Platanistoidea (Fig. la, Platanisti- 
dae). He placed Pontoporia and Inia in the Inioidea, the sister taxon 
to the Delphinoidea, Lipotes in the Lipotoidea. the sister taxon to 
the clade including both the Inioidea and Delphinoidea. 

Soft-tissue characters of the nasal passage complex, used by 



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

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

Proc. San Diego Soc. Nat. Hist. 29:125-133. 1994 



126 



Sharon L. Messenger 




Muizon, 1988a, 1991 



(b) 











DELPHINOIDEA 








1 


Ul 




UJ 

< 

Q 




Ul 

< 
a 


1 


< 

a 

i- 


Ul ' 
< Ul 
Q < 


QC 
UJ 

t- 
Ul 

tn 

> 

X 


Ul 

< 

a 

X 

a 


I— 

V) 

z 
< 

< 


Ul 

< 
o 

z 


z 
o 

Q 
O 

z 
o 


HOCOENI 
ELPHINID 


Q- 


N 


0. 


s 


a- q 



^^6 Heyning, 1989 

Figure 1 . Alternative hypotheses of relationships of the Odontoceti. (a) 
Cladogram based on both extant and fossil taxa, redrawn from Mui/on 
(1988a, 1991). The Inioidea include Pomoporia and Inia. (b) Cladogram 
based on 42 (including 18 soft-anatomical) characters of extant taxa only 
(from Heyning 1989). The Iniidae include Inia. Lipotes, and Pomoporia 
Numbers next to bars indicate the number of synapomorphies supporting 
that clade. 



Heyning (1989) in his analysis of extant odontocetes. also have 
resolved some relationships among extant odontocete families (Fig 
lb). For example, Heyning (1989) cited the development of a 
vestibular sac as one of the synapomorphies (i.e.. shared, derived 
character states) linking the Iniidae (including Inia, Pontoporia, 
and Lipotes) with the Delphinoidea to the exclusion of Platanista, 
also implying that the Platanistoidea (sensu Simpson 1945) are 
paraphyletic or polyphyletic. He did not address relationships 
within the Iniidae. Although both studies concluded that 
platamstoids are not monophyletic and separated Platanista from 
the remaining river dolphins. Heyning (1989) did recognize the 
other three river dolphins as a monophyletic taxon, the Iniidae, 
whereas Muizon (1988a) indicated that this grouping is itself 
paraphyletic. Nonetheless, Heyning (1989) stated that platanistoid 
relationships have not been resolved conclusively and emphasized 
the need for all platanistoid species to be reanalyzed. 

Another and perhaps more significant difference in the two 
proposed hypotheses is in the relationship of ziphiids to physeterids. 
Four characters of the nasal passage (confluence of nasal passages. 



presence of a blowhole ligament, presence of premaxillary sacs, 
and development of the proximal sac into an inferior vestibule/ 
nasofrontal sac/posterior nasal sac complex) were used by Heyning 
(1989) as synapomorphies uniting the Ziphiidae (beaked whales) 
with the clade including the Platanistidae (Platanista only), Iniidae, 
and Delphinoidea and excluding the Physeteridae. Soft-anatomical 
features were also among the character states he used to unite 
Physeter and Kogia into a monophyletic group, the Physeteridae 
(presence of a spermaceti organ and frontal and distal sacs), and to 
establish the monophyly of the Ziphiidae (presence of throat 
grooves). Yet Muizon (1984, 1991) recognized the Physeteridae 
and Ziphiidae as a monophyletic group. On the basis of features 
evident in fossil taxa, especially Squaloziphius emlongi, which he 
considered a ziphiid, Muizon (1984, 1991 ) determined that charac- 
ters previously thought to be primitive for odontocetes, such as the 
absence of the lateral plates of the pterygoids, were derived in 
parallel in the clade including physeterids and ziphiids and the 
clade including the remaining odontocetes. 

These examples demonstrate the need for a re-evaluation of the 
Platanistoidea, as well as odontocetes in general. Although neither 
soft-anatomical characters nor fossils resolved platanistoid rela- 
tionships, the value of both has been clearly demonstrated. 



EFFECTS OF EXCLUDING TAXA 
ON PHYLOGENY RECONSTRUCTION 

Many have debated the usefulness of fossil taxa in phylogenetic 
analyses (Simpson 1961; Hennig 1966; Patterson 1981; Doyle and 
Donoghue 1987; Gauthier et al. 1988; Donoghue et al. 1989; 
Huelsenbeck 1991; Novacek 1992). While some (e.g., Simpson 
1961 ) have advocated the special qualities of fossils, emphasizing 
ancestor-descendant relationships, others (e.g., Patterson 1981) 
have contended that fossils offer no additional information and 
should not affect the topology of a cladogram based solely on extant 
taxa. Yet Doyle and Donoghue ( 1987). in their phylogenetic analy- 
sis of angiosperms, and Gauthier et al. ( 1988), in their re-evaluation 
of amniote relationships, have demonstrated that the consideration 
of fossil taxa can affect hypothesized relationships dramatically. 
Huelsenbeck ( 1 99 1 ), through the use of computer simulations, has 
proposed conditions under which fossils might provide both more 
and less resolution than extant taxa alone. According to Gauthier et 
al. (1988), "fossils should be most important in phylogenetic infer- 
ence when the group of interest is old and only a few, highly 
modified, terminal taxa are extant." This statement agrees with 
Felsenstein's (1978) prediction that parsimony methods can be 
positively misleading (i.e., the method will not converge on the real 
phylogeny despite the addition of more data) in lineages in which 
the scaled lengths of branches leading to terminal taxa are much 
longer than those of internal branches. This situation is directly 
applicable to the river dolphins. Each of the four monotypic extant 
genera exhibits a unique combination of primitive and derived 
character states. In my own analyses, I have found for the river 
dolphins many more autapomorphies than characters elucidating 
relationships among them. These four extant species constitute less 
than 20% of the total number of known river dolphin species, even 
if only well-preserved fossil taxa are considered. Also, several 
families within the river-dolphin group, as defined by Muizon 
( 1984) (e.g.. Squalodontidae, Squalodelphidae), as well as in ceta- 
ceans in general [e.g., Archaeoceti, Eurhinodelphidae (= Rhab- 
dosteidae), Cetotheriidae], are represented exclusively by fossil 
members, evidence that cetacean history conceals far more diver- 
sity than the order shows today. This lost diversity represents lost 
information. 

Fossil taxa are important in systematics for the following rea- 
sons: first, fossil taxa may represent outgroups (i.e., taxa closely 



Phylogenetic Relationships of Platanistoid River Dolphins (Odontoceti. Cetacea): Assessing the Significance of Fossil Taxa 



127 



related to the group under study that are used to determine the 
direction of character evolution! phylogenetically closer to the 
ingroup than are extant forms. Similarly, fossil taxa, especially 
those temporally close to the ancestor, should be more representa- 
tive of the condition at the ancestral node. If condition at the nodes 
are better known, the resulting phylogeny will better approximate 
the true phylogeny (Huelsenbeck 1991). Second, fossil taxa may 
provide information on intermediate character states, showing that 
some characters vary continuously, although they appear discon- 
tinuous in extant taxa. Without these fossil taxa such character 
states may be mistakenly interpreted as nonhomologous. Third, a 
fossil taxon that is a sister taxon of a living form may retain many 
plesiomorphic character states and may render alternative hypoth- 
eses of relationships more parsimonious (Doyle and Donoghue 
1987; Gauthier et al. 1988; Donoghue et al. 1989). Potential prob- 
lems resulting from the exclusion of fossil taxa can be illustrated by 
examples in platanistoid systematics. 

Fossils as Outgroup Taxa 

Outgroup taxa are used in phylogenetic analyses to determine 
the direction of character transformations, i.e., polarity of character 
states. If fossil taxa represent outgroups phylogenetically closer to 
the ingroup than any extant taxon, addition of these fossil taxa 
could change polarity assignments at the outgroup node. Because 
previous investigators have proposed that some river dolphins are 
more closely related to members of the Delphinoidea, the ingroup 
in investigations of the relationships of extant platanistoids must 
include the Delphinoidea. Therefore, the first outgroup should be 
the Ziphiidae, followed by the Physeteridae and, if necessary, the 
Mysticeti and terrestrial mammals (Heyning 1989). In Muizon's 
(1984. 1987, 1988a, 1991) studies including fossil taxa. the 
Agorophiidae (sensu Fordyce 1981), Squalodontidae. Squalodel- 
phidae. and Eurhinodelphidae represent fossil groups more closely 
related to the ingroup than are some of the extant outgroups. The 
effect that these additional fossils can have on polarity assessment 
is illustrated by a particularly interesting and complex structure in 
cetaceans, the pterygoid bone. 

Cetaceans possess a pterygoid that, in some members, is di- 
vided into medial and lateral lamina (Fig. 2). The condition of the 
lateral lamina of the pterygoid, extending posteriorly beyond the 
level of the pterygoid hamulus, varies widely in the Odontoceti, 
especially among some of the extant river dolphins, and homolo- 
gies are unclear (Cozzuol 1989a). For this example, however, I will 
assume that all lateral lamina are homologous. The presence of the 
lateral lamina of the pterygoid has been interpreted as both 
plesiomorphic (Fraser and Purves 1960; Muizon 1984; Fordyce 
1985) and apomorphic (Barnes 1985; Cozzuol 1989a). This charac- 
ter can be polarized differently depending on whether or not fossil 
taxa are considered (Figs. 3a, b). Among extant taxa, the lateral 
plate is present in mysticetes (Fraser and Purves 1960), Platanista 
gangetica, Pontoporia hlainvillei, some species of the Phocoenidae 
(e.g., Phocoenoides dalli), and some individuals of Lagenor- 
hynchus albirostris (Cozzuol 1989a). The pterygoids of the earliest- 
diverging extant odontocetes, the Physeteridae and Ziphiidae, lack 
a lateral lamina. The lateral lamina of mysticetes, creating a shallow 
fossa in the posterior margin of the pterygoid (Fraser and Purves 
1960), differs greatly from that of any extant odontocete and may 
not be homologous. Therefore, by the outgroup method of 
Maddison et al. (1984), the lateral lamina of extant odontocetes is 
derived (Fig. 3a). Among fossil taxa, the pterygoid bears a lateral 
lamina in archaeocetes, agorophiids, ziphiids (Squaloziphius 
emlongi), squalodontids, squalodelphids, platanistids (Zarhachis 
and Pomatodelphis), and eurhinodelphids. If the structures are ho- 
mologous and some of the fossil taxa are more closely related to the 
ingroup than to any extant outgroup taxon, as Muizon ( 1991 ) has 



suggested, the fossil taxa imply that the lateral lamina of the ptery- 
goid could be primitive in the clade including the river dolphins and 
Delphinoidea (Fig. 3b). 

Similarly, the size of the posterior process of the tympanic bulla 
is a character whose polarity can be interpreted differently when 
fossil taxa are included in or excluded from phylogenetic analysis. 
The tympanies of the Physeteridae and Ziphiidae (and Mysticeti) 
exhibit a large posterior process that becomes incorporated into the 
cranium between the squamosal and the exoccipital suture and is 
visible on the exterior of the skull. All other extant odontocetes 
except Platanista exhibit a much smaller posterior process that is 
no longer visible on the exterior of the cranium; Platanista has a 
posterior process somewhat intermediate in size. Outgroup com- 
parison of extant taxa only implies that the large posterior process 
of the tympanic of physeterids, ziphiids, and mysticetes is primitive 
and the small posterior process is derived. Muizon ( 1984), however, 
found that the posterior process of Platanista resembles that of 
agorophiids and considered this moderately small posterior process 
as the plesiomorphic condition in odontocetes. Therefore, he con- 
sidered the enlarged posterior process of physeterids and ziphiids 
derived, constituting a synapomorphy uniting the two families and 
and their fossil relatives into a monophyletic group. He considered 
the much smaller process of the Lipotoidea, Inioidea, and Delphi- 
noidea to be a derived condition representing a synapomorphy of 
that clade. 

A character traditionally used to unite the river dolphins is their 
elongated mandibular symphysis. Indeed, all of them possess a 
mandibular symphysis measuring over one-half of the total length 
of the mandible. Heyning ( 1989), however, found that agorophiids, 
eurhinodelphids, and Steno (a delphinid) also possess elongated 
mandibular symphyses. Because the origin and taxonomic distribu- 
tion of an elongated mandibular symphysis was unclear, Heyning 
gave it less weight, though he considered this character derived, 
having evolved independently three times, in Physeteridae. Pla- 
tanistidae. and Iniidae. If the relationships of fossil and extant 
odontocetes proposed by Muizon ( 1988a. 1991 ) are correct and the 
elongated mandibular symphysis is derived, the character must 
have evolved independently seven times, in agorophiids, 
physeterids, eurhinodelphids, platanistids. Lipotes, iniids, and 
Steno. If the elongated symphysis is primitive for toothed whales, 
however, its independent loss in Kogia, ziphiids, and delphinioids 
and reappearance in Steno requires only five steps. With the addi- 
tion of fossil taxa it is no longer more parsimonious to use the 
presence of an elongated mandibular symphysis to unite any of the 
river dolphin species. 

Fossil Taxa and Increased Diversity of Character States 

Fossil taxa can also affect phylogenetic inferences because 
additional information on intermediate states of characters seen in 
some fossils may be used to link taxa that had not been considered 
closely related. Extant taxa may be highly derived, with homolo- 
gous features lost or difficult to detect. Fossil taxa may illustrate the 
variability of some characters, aiding in determining their homolo- 
gies. For example, Platanista and its fossil relatives exhibit an 
articular process on the periotic bone. This process is associated 
with a fossa in the squamosal bone and, in some taxa (e.g., the 
Platanistidae), fits so tightly into the fossa that the periotic cannot 
be removed without breaking the process. A similar process seen in 
another fossil family, the Eurhinodelphidae, appears to be homolo- 
gous. Zarhachis, a fossil platanistid. however, exhibits both the 
articular process and the process seen in the Eurhinodelphidae, 
indicating that these processes may not be homologous (Muizon 
1987). 

Some fossil taxa, such as the Squalodontidae, exhibit intermedi- 
ate or additional character states not seen in any extant taxon. Two 






Llpal. 



Figure 2. Ventral view of skulls showing different morphologies of the pterygoid and palatine bones in several species of cetaceans (modified from 
Muizon 19841. (a), Archaeocete (Zygorhiza kochii); (b). mysticete (Balaenoptera musculus); (c), eurhinodelphid (Eurliinodelphis bossi); (d), Ponloporia 
blainvillei: (e). Inia geoffrensis; (fl, ziphiid (Mesoplodon bidens); (g), delphinid {Lissodelpbis peroni). Lip, lateral lamina of the pterygoid; Llpal, lateral 
lamina of the palatine; Lmp, medial lamina of the pterygoid; Pal. palatine; Prf, falciform process; Prh, hamular process; Pt, pterygoid process; Sp, pterygoid 



Phylogenetic Relationships of Platanistoid River Dolphins (Odontoceti, Cetacea): Assessing the Significance of Fossil Taxa 



129 



(a) 





111 

< 

Q 




Ill 

< 

Q 




Ul 

< 


Ill 

< 


m 






C/) 




1- 


Q 


< 


1- 


DC 
111 
1- 
111 

> 


Ill 




z 
o 
o 
o 

z 


z 


a 


LU 
O 

1- 
V) 


< 

a 

i 


z 
< 
1- 
< 


LU 

< 


111 

o 
o 
o 


z 

X 
Q. 

_l 


> 


I 


a. 


_1 


z 


o 


X 


Ul 


s 


Q. 


N 


Q. 


s 


a. 


a 


+ 


- 


- 


+ 


+.- 


- 


+ 


+ 




( ») = Lateral plate present 
(-) = Lateral plate absent 




Figure 3. Distribution of states (in parentheses) of the lateral plate of the 
pterygoid in representative cetaceans. +, plate present; — , plate absent, (a), 
Cladogram based on extant taxa only (from Heyning 1989). At the outgroup 
node (bar) the plate is absent, (b) Cladogram based on both both fossil and 
extant taxa (from Muizon 1991 ). At the outgroup node presence or absence 
of the plate is equivocal. 



characters, a subcircular fossa in the squamosal bone and an articu- 
lar process of the periotic, are unique to the Platanistoidea (sensu 
Muizon 1987. 1991). The deep subcircular fossa is positioned 
posteromedial to the postglenoid process of the squamosal and 
dorsal to the periotic. It may be a result of the expansion of the 
peribullary sinus, a basicranial air sinus that surrounds the periotic 
and tympanic bones (Muizon 1987). The articular process, dis- 
cussed above, is found on the lateral surface of the periotic at the 
junction between the posterior process and the body of the periotic. 
This process articulates with a fossa in the squamosal bone at the 
base of the postmeatal process. These characters are well developed 
in Platanista, fossil platanistids {Zarhachis and Pomatodelphis), 
and the Squalodelphidae. According to Muizon ( 1987). they occur 
in some members of the Squalodontidae (e.g., Squalodon and 
Eosqualodori) but are much less developed. Nonetheless, these 



characters have been used as synapomorphies diagnosing the 
Platanistoidea, as defined by Muizon (1987. 1991). In a phyloge- 
netic analysis of extant taxa only, these characters would be consid- 
ered autapomorphies of Platanista, thus offering no information 
about the phylogenetic relationships of Platanista within the 
Odontoceti. One important phylogenetic implication of this inclu- 
sion of fossil taxa (Muizon 1984,1987) is that it is no longer most 
parsimonious to retain Platanista. with its presumed fossil relatives 
{Zarhachis, Pomatodelphis, Squalodelphidae, and Squalodontidae), 
in the clade including the remaining river dolphins. 

Fossils as Sister Taxa Retaining Plesiomorphic Characters 

Fossils may affect the topology of a cladogram if they represent 
sister taxa retaining plesiomorphies. As discussed above, the evi- 
dence of fossils led Muizon ( 1987, 1991 ) to unite Platanista with 
Zarhachis. Pomatodelphis, the Squalodelphidae. and Squalodonti- 
dae and separate it from the remaining river dolphins, the Inioidea 
and Lipotoidea. He hypothesized that the Squalodontidae and 
Squalodelphidae are the sister taxa of Platanista, Zarhachis, and 
Pomatodelphis (Figs. 4a, b). Muizon (1987,1991) proposed this 
relationship on the basis of derived characters (e.g., subcircular 
fossa of the squamosal bone, articular process of the periotic), yet 
the Squalodontidae are otherwise primitive. To include Platanista 
and its fossil relatives in a clade with the remaining river dolphins 
implies a great number of reversals in the fossil taxa (Fig. 4a). For 
example, 12 characters of the Squalodontidae, such as heterodont 
dentition and unfused lacrimal and jugal bones, would have to be 
considered reversals. As a consequence. Platanista, with its fossil 
relatives, has been placed as the sister taxon to the clade including 
the Eurhinodelphidae. Lipotidae. Iniidae. and Delphinoidea (Fig. 
4b). This arrangement implies that the characters shared by the 
platanistids, Lipotes, and Inia are convergences or plesiomorphies. 

These examples illustrate that fossil taxa can indeed have a 
significant impact on the topology of a cladogram and should be 
considered in cladistic analyses. 

ALTERNATE METHODS OF PHYLOGENETIC 
RECONSTRUCTION 

Application of correct phylogenetic methodology (Hennig 
1966; Eldredge and Cracraft 1980; Wiley 1980) is necessary to 
avoid erroneous inferences of relationships. Proper cladistic meth- 
odology includes the use of monophyletic groups as operational 
taxonomic units, polarization of characters on the basis of compari- 
son with at least two outgroups that consist of the taxa most closely 
related to the ingroup (Watrous and Wheeler 1981; Maddison et al. 
1984), and the use of computer-assisted algorithms (e.g., PAUP; 
Swofford 1990), especially when data sets are large or characters 
are inconsistent. To date, only one phylogenetic study (Heyning 
1989) addressing platanistoid relationships has employed a com- 
puter program (PAUP. version 2.4.1), and it did not present a 
published matrix of character-state assignments. Any attempt to 
reproduce the results of such an analysis requires that such a matrix 
be reconstructed on the basis of character descriptions in the text 
that are not always complete. Very few studies sufficiently describe 
character states to the species level or describe intraspecific poly- 
morphism, both of which are necessary for accurately reconstruct- 
ing character matrices. Other studies (Muizon 1984, 1987) have 
included nonmonophyletic taxa (e.g.. the Squalodontidae) or have 
used alternative, less reliable methods to polarize characters, such 
as assuming earlier taxa are more primitive. The following ex- 
amples illustrate these problems in platanistoid systematics. 

A significant problem in recognizing a nonmonophyletic taxon 
is that some members of that taxon may be more closely related to 



130 



Sharon L. Messenger 



(a) 



in 

o _ 

O I- 

LU UJ 

< o 
x p 

O tf) 

cc >- 

< 5 



UJ 

< 
9 

x 

Q. 
O 

rr 
O 
C3 

< 



5 Q 

E i 

* a 

« _i 

3 UJ 

< •£• ° 
Cog? 

SS f 1 s 

X £ g- 3 

a. n V) uj 



< 
o 
rx 



UJ 

uj < 

M 



o 

Q 
O 

< 
3 



UJ 

a 



•2 

.2 -a< 



o a -2 -8 < 

O O 3 3 §9: § 
w w a. rq <x_i 5 



< 

UJ 
CI 



X 

a 




(b) 



F 
u 
o 
o 

UJ 

< 

X 

o 

rr 

< 



o 

F 
en 

> 
S 



UJ 

< 
9 

x 
a. 
O 
rr 
O 
O 
< 







& 

c 
o 


LU 


LU 
< 








LU 
< 


UJ 

< 
a 




1 

Q> 
</> 

.a 

N 
O 


< 
Q 

F 

z 


Q 

I 

a 






■2 
5 


Q 

X 

a. 


rx 

UJ 

i- 

LU 


UJ 

< 
9 


o 

Q 

o 

< 


LU 
Q 

O 

_i 
< 


2 

C 


u 

(0 


■8 

o 

*- 

to 


LU 

a 
O 

z 

X 


>- 


X 


to 

3 


3 


D 


2 


-c 


E 
o 


rr 


X 


a 


o 


o 


03 


to 


z> 


a 


N 


<J5 


c/> 


(/) CL 


r3 


Q. 


LU 



< 

9 uj 

O Q 

a = 



< 

UJ 

9 

o 

z 

X 
Q. 

_l 
UJ 

a 




Figure 4. Alternative phylogenetic positions of Platanista and its fossil 
relatives (modified from Muizon 1988a, 1991). Muizon (1984. 1988a, 
1991 ) has proposed that fossil taxa Squalodontidae and Squalodelphidae are 
most closely related to Platanistidae {Platanista, Zarhachis, and 
Pomatodelphis). Numbers, number of synapomorphies; R, reversals, (a). 
River dolphins constituting a monophyletic group, implying 1 2 reversals in 
the Squalodontidae. (b) It is more parsimonious to remove Platanista and its 
fossil relatives from the remaining river dolphins and place them as the 
sister taxon to the clade including the Eurhinodelphidae. Lipotidae, Inndae, 
and Delphinoidea. 



the ingroup than others. Not recognizing those members separately 
could have the same effect as excluding them from the analysis. 
Also, since a nonmonophyletic taxon can contain members of more 
than one monophyletic group, if the taxon is polyphyletic, such taxa 
may appear misleadingly diverse. As mentioned earlier, the in- 
creased diversity of character states seen in fossils can be useful in 
establishing homologies or uniting taxa. If these groups are 
nonmonophyletic, however, they could confound rather than re- 
solve phylogenetic relationships. Alternatively, a paraphyletic 
taxon, by definition not including all descendants of a common 
ancestor, may appear misleadingly uniform. Since most phyloge- 



netic studies of river dolphins (Muizon 1984, 1987, 1988a, 1991) 
have considered nonmonophyletic taxa, these problems need to be 
addressed. 

Groups such as the Agorophiidae, Squalodontidae, and Eurhi- 
nodelphidae have not been demonstrated to be monophyletic but 
rather have been defined by plesiomorphic character states. Addi- 
tionally, Squaloziphius emlongi, considered by Muizon ( 1 99 1 ) to be 
an important early diverging ziphiid. is considered by others not to 
be closely related to the Ziphiidae (Heyning pers. comm.). Several 
of these taxa are considered by some researchers (Fordyce 1985) to 
be grades, and they are very possibly paraphyletic. The Agoro- 
phiidae and Squalodontidae, often described as primitive odon- 
tocetes, include stratigraphically early fossil taxa united largely by 
plesiomorphies such as heterodont dentition and incompletely tele- 
scoped skulls. Several of the taxa included in these families are 
represented by only fragmentary material. To date, no diagnosis of 
the Agorophiidae on the basis of derived character states has been 
attempted, and the group is in much need of study. Nevertheless, it 
has been used as an outgroup taxon in studies of platanistoid 
relationships (Muizon 1984, 1991; Heyning 1989). 

Although Muizon (1987) stated that the Squalodontidae could 
be nonmonophyletic, he included that family in his redefinition of 
the Platanistoidea as the sister taxon of the Squalodelphidae and 
Platanistidae. The Squalodontoidea, as defined by Winge (1921), 
Rice (1967). Rothausen (1968), and Barnes (1985), include the 
Agorophiidae. Fordyce ( 1985) stated not only that agorophiids did 
not share a most recent common ancestor with squalodontids, but 
also that some genera within the Squalodontidae are more closely 
related to other taxa. such as the Squalodelphidae and Platanistidae. 
Cozzuol (1989b) believed the Squalodontidae to be polyphyletic 
and. in an attempt to resolve this problem, removed Prosqualodon 
from the family while including the eurhinodelphids. Later, Muizon 
(1991) proposed that a subset of the genera he had previously 
placed in the family (Muizon 1987) form a clade. The status of the 
Squalodontidae is still not completely resolved. 

The monophyly of the Eurhinodelphidae is also in question and 
requires further study. Although Fordyce (1985) stated that this 
family has not been diagnosed on the basis of derived character 
states, Muizon ( 1991 ) listed one synapomorphy for it, lengthening 
of the premaxillary portion of the rostrum such that the rostrum 
extends farther anteriorly than the mandible. Another problematic 
family, the Acrodelphidae {sensu Abel 1905), contains species that 
have been placed in the Eurhinodelphidae or as the sister taxon to 
the Eurhinodelphidae (Muizon 1988b). 

This also brings into question the monophyly of the 
Acrodelphidae. Barnes (1985) defined the family as including 
Schizodelphis, Pomatodelphis, and probably Zarhachis but recom- 
mended re-evaluation of it. Muizon ( 1988b) stated that the family 
had traditionally included Acrodelphis, Schizodelphis, Eoplata- 
nista, Champsodelphis, and, according to some researchers, 
Pomatodelphis and Zarhachis. In his revision, he broke up the 
family Acrodelphidae, restricting it to the type specimen of 
Acrodelphis and leaving it as incertae sedis. He placed Schizo- 
delphis sulcatus into the Eurhinodelphidae, stated that Acrodelphis 
is a junior synonym of Champsodelphis, placed Acrodelphis 
ombonii into a new genus. Dalpiazina [subsequently proposing it as 
a possible sister taxon to Squalodontidae (Muizon 1991)], placed 
Champsodelphis tetragorhinus into a new genus. Medocinia, in- 
cluded in the Squalodelphidae, and placed Pomatodelphis and 
Zarhachis into the Platanistidae. This example underscores the 
need for a re-evaluation at all levels. Under such circumstances 
where the taxonomy appears to be very unstable, it is best to 
disregard the current classification and regard each species, or 
specimen, as a separate operational taxonomic unit. 

Not only is the monophyly of several taxa in question, so are 
their phylogenetic positions within the Odontoceti. This can cause 



Phylogenetic Relationships of Platanisloid River Dolphins (Odonloceti, Celacea): Assessing the Significance of Fossil Taxa 



131 



problems in determining appropriate outgroups and reconstructing 
character states at ancestral nodes. Some workers (Barnes 1985; 
Fordyce 1985; Cozzuol 1989b) have stated that at least some 
squalodontids represent an early-diverging lineage within the 
Odonloceti. At least three alternative branching sequences of the 
Squalodontidae have been suggested (Fig. 5): ( 1 ) as the sister taxon 
to the clade including the Platanistidae and Squalodelphidae 
(Muizon 1987, 1991); (2) as the sister taxon to the Ziphiidae 
(Fordyce 1985); (3) as one of the earliest diverging lineages within 
the Odontoceti (Barnes 1985; Cozzuol 1989b; Heyning 1989). If at 
least some members of the Squalodontidae are demonstrated to 
have diverged before the Physeteridae and/or Ziphiidae, this again 
could change polarity assignments for lineages branching off sub- 
sequently and ultimately may affect the topology of the cladogram. 
Similarly, the Eurhinodelphidae (Fig. 6) have been suggested as 

( 1 ) the sister taxon to the Delphinida (sensu Muizon 1988a). which 
include the Iniidae, Lipotidae. and Delphinoidea (Muizon 1988a); 

(2) an early-diverging lineage that may have originated within the 
Squalodontidae (Barnes 1985; Cozzuol 1989b); or (3) members of 
the family Delphinidae (Kellogg 1928). Fordyce (1983) mentioned 
similarities between eurhinodelphids and platanistids but concluded 
that further study is required to determine if these similarities are 
synapomorphies. These radically different hypotheses of relation- 
ships emphasize the need for more study of this group. Misplace- 
ment of the Eurhinodelphidae or its recognition as a nonmono- 
phyletic family could lead to incorrect polarity assignments. 

As has been demonstrated earlier, appropriate choice of the 
outgroups serving as the basis for character polarity is vital to 
inferring phylogenetic relationships. The outgroup-comparison 
method has been demonstrated to be the most objective method for 
determining character-state polarity (Watrous and Wheeler 1981). 
When possible, more than one outgroup should be used and the 
branching sequence of outgroups should be determined on the basis 
of shared, derived features. Yet several cladistic studies have failed 
to polarize character states on the basis of more than one outgroup 
(e.g., Barnes 1985). Others often have resorted to the stratigraphic 
record, generally looking at the stratigraphically earliest members 
of the ingroup to assign polarities (Muizon 1984, 1987, 1988a. 




Figure 6. Alternative phylogenetic positions of the Eurhinodelphidae. as 
proposed by various researchers. The family represents ( 1 ) the sister taxon 
of the Delphinida (Muizon 1988a, 1991), (2) an early-diverging lineage 
originating within the Squalodontidae (Barnes 1985; Cozzuol 1989b). or (3) 
a subset of the Delphinidae (Kellogg 1928). 



1991 ). When fossil taxa within the ingroup are used, characters may 
be polarized incorrectly and the resulting phylogenetic relation- 
ships may be based on shared primitive characters. 

Finally, computer-assisted programs (e.g., PAUP, Swofford 
1990) should be used to analyze phylogenetic relationships. The 
assumptions (e.g., whether or not characters were ordered or 
weighted) made during the computer analyses should be described. 
The matrix of character states used in the computer analysis should 
also be published. If character-state matrices cannot be reproduced 
accurately from the descriptions given in the text of a published 
phylogenetic analysis, the results of the analysis are not reproduc- 
ible. 



UJ 

o 

O 

UJ 

< 

X 

o 

DC 

< 



< 

o 
i 

Q. 

o 

DC 

o 
o 

< 



UJ 

< 



o 

D 

o 

_l 

< 
a 



< 
ui 9 
< t 



H 
UJ 
H 
UJ 

c/> 

>- 

X 

rx 



if 

o 

I 

(A 

3 

I x 

d? n 



UJ 

< 
9 

x 

Q. 



O 

Q 

o 

_l 

< 
a 



UJ 

< 
9 

x 

0) D. 

id 



§ a .2 •§ o < 
-j .a -s I ? 9 

< c g tax h- < 

3 a « eccoq 

(A V) Q. iq Q. UJ 3 ± 



< 

UJ 

9 
o 



X 
Q. 




Figure 5. Alternative phylogenetic positions of the Squalodontidae. as 
proposed by various researchers. The family represents ( 1 ) the sister taxon 
of the Squalodelphidae and Platanistidae ( Muizon 1 987. 1 99 1 ),( 2 ) the sister 
taxon of the Ziphiidae (Fordyce 1985). or (3) an early-diverging lineage of 
odontocetes (Barnes 1985; Cozzuol 1989b; Heyning 1989). 



DISCUSSION 

Clearly, much work still needs to be done before the phyloge- 
netic relationships of many odontocete taxa are sufficiently under- 
stood. The problems regarding the phylogenetic position and/or 
monophyly of some fossil taxa, however, do not negate their impor- 
tance in phytogeny. As the phylogenetic relationships of the earliest 
diverging lineages become further resolved and monophyletic 
groups are identified, assessments of character polarities and hy- 
potheses of character evolution will change. This is especially 
relevant for cetaceans and river dolphins in particular, of which a 
large proportion of the species are extinct. It is important not to 
attribute special qualities to fossils or to overlook the inherent 
biases of the fossil record. The fossil record of cetaceans is skewed, 
since most fossil taxa are found in deposits originating in shallow 
seas or estuaries and very few pelagic species are known. The 
selective preservation of certain bony elements, such as periotic 
bones or teeth, is another source of bias. Fossils inherently lack 
certain characters available in extant taxa. such as soft tissue and 
DNA. As Heyning (1989) showed, such characters also provide 
important information for resolution of phylogenetic relationships 
and should be included in data sets even though they are lacking 
from fossil material. Lack of certain characters is not restricted to 
fossil taxa. Extant taxa may be effectively incomplete if some of 
their characters are so highly derived that homologies cannot be 
determined (e.g.. nasal sacs of physeterids versus other 



132 



Sharon L. Messenger 



odontocetes). The addition of fossil taxa will generally increase the 
number of missing characters in the data matrix. Missing character 
data will increase the number of equally parsimonious trees but 
should not give misleading trees. The increase in the number of 
equally parsimonious trees may be disconcerting; however, the 
quality of a phylogeny should not be based on its recovering a 
single most parsimonious tree, since that can be accomplished with 
relatively high reliability with randomized data, at least with mo- 
lecular data (Hillis 1991; Hillis and Huelsenbeck 1992). The best 
approximation of phylogenetic relationships should consider all 
available data, including fossil taxa and soft-tissue characters, ana- 
lyzed with rigorous and testable cladistic methodology. 

ACKNOWLEDGMENTS 

I especially thank A. Berta, D. Archibald, J. McGuire, L. 
Grismer, P. Mabee, R. Etheridge, and M. Simpson for thoughtful 
discussions and for helping me develop my ideas in the preparation 
of the manuscript. An earlier draft of this paper was improved 
through the comments of A. Berta, T Demere, J. McGuire, L. 
Grismer. and three anonymous reviewers. 1 thank J. Mead, C. 
Potter. T. Demere, P. Unitt, S. Bailey, T. Daeschler, W. Fuchs, for 
lending or providing access to specimens used in this study. This 
work was funded in part by grants from the Society of Sigma Xi 
(Grant-in-Aid of Research), the Smithsonian Institution (Graduate 
Student Fellowship), the American Museum of Natural History 
Lerner-Gray Fund for Marine Research, and the Department of 
Biology at San Diego State University. 

LITERATURE CITED 

Abel, O. 1905. Les odontocetes du Boldenen (Miocene superieur) 
d'Anvers. Memoires du Musee Royal d'Histoire Naturelle de 
Belgique 3:1-155. 

Barnes, L. G. 1985. Fossil pontoporiid dolphins (Mammalia: Cetacea) 
from the Pacific coast of North America. Natural History Museum 
of Los Angeles County Contributions in Science 363. 

. D. P. Domning. and C. E. Ray. 1985. Status of studies on fossil 

marine mammals. Marine Mammal Science 1:15-53. 

Brownell, R. L., Jr., K. Ralls, and W. F. Perrin. 1989. The plight of the 
"forgotten" whales. Oceanus, pp. 5-11. 

Cozzuol, M. 1989a. An alternative interpretation of the evolutionary 
significance of the bony lateral lamina of the pterygoid and its 
implication for odontocete systematics. Abstract of Posters and 
Papers, Fifth International Thenological Congress. Rome, 22-29 
August 1989, 1:481. 

. 1989b. On the systematic position of the genus Prosqualodon 

Lydekker, 1893, and some comments on the family Squalodonti- 
dae. Abstract of Posters and Papers, Fifth International 
Thenological Congress, Rome, 22-29 August 1989, 1:483. 

Donoghue, M. J., J. A. Doyle, J. Gauthier, A. G. Kluge. and T. Rowe. 
1989. The importance of fossils in phylogeny reconstruction. An- 
nual Review of Ecology and Systematics 20:431^160. 

Doyle, J. A., and M. J. Donoghue. 1987. The importance of fossils in 
elucidating seed plant phylogeny and macroevolution. Review of 
Paleobotany and Palynology 50:63-95. 

Eldredge, N., and J. Cracraft. 1980. Phylogenetic Patterns and the 
Evolutionary Process: Method and Theory in Comparative Biol- 
ogy. Columbia University Press. New York, New York. 

Felsenstein, J. 1978. Cases in which parsimony or compatibility meth- 
ods will be positively misleading. Systematic Zoology 27:401- 
410. 

Flower, W. H. 1867. Description of the skeleton of I iiia geoffrensis and 
of the skull of Pontoporia blainvillii, with remarks on the system- 
atic position of these animals in the order Cetacea. Transactions of 
the Zoological Society of London 6:87-1 16. 

Fordyce, R. E. 1981. Systematics of the odontocete whale Agorophius 
pvgmaeus and the family Agorophiidae (Mammalia: Cetacea). 
Journal of Paleontology 55:1028-1045. 



. 1983. Rhabdosteid dolphins (Mammalia: Cetacea) from the 

Middle Miocene. Lake Frome area. South Australia. Alcheringa 
7:27^+0. 

. 1985. The history of whales in the Southern Hemisphere. Pp. 

79-104 in J K. Ling and M. M. Bryden (eds.). Studies of Sea 
Mammals in South Latitudes. Proceedings of a Symposium of the 
52nd ANZAAS Congress, Sydney. 

1 992. Cetacean evolution and Eocene/Oligocene environments. 



Pp. 368-381 in D. Prothero and W. Berggren (eds.). Eocene- 
Ohgocene Climatic and Biotic Evolution. Princeton University 
Press. Princeton, New Jersey. 

Fraser, F. C. and P. E. Purves. 1960. Hearing in cetaceans. Bulletin of 
the British Museum (Natural History). Zoology 7:1-140. 

Gaskin, D. E. 1985. The Ecology of Whales and Dolphins. Heinemann 
Educational Books, London, England. 

Gauthier, J., A. G. Kluge, and T. Rowe. 1988. Amniote phylogeny and 
the importance of fossils. Cladistics 4:105-209. 

Gray, J. E. 1863. On the arrangement of the cetaceans. Proceedings of 
the Zoological Society of London, pp. 197-202. 

. 1866. Catalogue of seals and whales in the British Museum. 

2nd edition. British Museum (Natural History), London. 

Hennig, W. 1966. Phylogenetic Systematics. University of Illinois Press, 
Urbana, Illinois. 

Hershkovitz, P. 1 966. Catalog of Living Whales. United States National 
Museum Bulletin 246:1-259. 

Heyning, J. 1989. Comparative facial anatomy of beaked whales 
(Ziphiidae) and a systematic revision among the families of extant 
Odontoceti Natural History Museum of Los Angeles County Con- 
tributions in Science 405. 

Hillis, D. M. 1991. Discriminating between phylogenetic signal and 
random noise in DNA sequences. Pp. 278-294 in M. M. Miyamoto 
and J. Cracraft (eds.). Phylogenetic Analysis of DNA Sequences. 
Oxford University Press, Oxford. England. 

, and J. P. Huelsenbeck. 1992. Signal, noise, and reliability in 

molecular phylogenetic analyses. Journal of Heredity 83:189-195. 

Huelsenbeck, J. P. 1991. When are fossils better than extant taxa in 
phylogenetic analysis? Systematic Zoology 40:458^469. 

Kasuya, T. 1973. Systematic consideration of recent toothed whales 
based on the morphology of tympano-periotic bone. Scientific 
Reports of the Whales Research Institute, Tokyo 25:1-103. 

Kellogg. R. 1928. The history of whales, their adaptation to life in the 
water Quarterly Review of Biology 3:29-76, 174—208. 

Maddison. W. P., M. J. Donoghue, and D. R. Maddison. 1984. Outgroup 
analysis and parsimony. Systematic Zoology 33:83-103. 

Miller, G. S., Jr. 1918. A new river dolphin from China. Smithsonian 
Miscellaneous Collections 68:1-12. 

. 1923. The telescoping of the cetacean skull. Smithsonian Mis- 
cellaneous Collections 76: 1 -7 1 . 

Muizon, C. de. 1984. Les vertebres fossiles de la Formation Pisco 
(Perou). Deuxieme partie: Les odontocetes (Cetacea, Mammalia) 
du Pliocene Inferieur de Sud-Sacaco. Editions Recherche sur les 
Civilisations Memoire 50. 

. 1987. The affinities of Notocetus vanbenedeni, an early Mio- 
cene platanistoid (Cetacea, Mammalia) from Patagonia, southern 
Argentina. American Museum Novitates 2904. 

. 1988a. Les relations phylogenetiques des Delphinida (Cetacea, 

Mammalia). Annales de Paleontologie 74: 159-227. 

. 1988b. Le polyphyletisme des Acrodelphidae. Odontocetes 

longirostres du Miocene europeen. Bulletin du Museum National 
d'Histoire Naturelle, Section C, 4eme serie, 10:31-88. 

1991. A new Ziphiidae (Cetacea) from the early Miocene of 



Washington state (USA) and phylogenetic analysis of the major 
groups of odontocetes. Bulletin du Museum National d'Histoire 
Naturelle, Section C, 4eme sene, 12:279-326. 

Novacek. M. J. 1 992. Fossils, topologies, missing data, and the higher level 
phylogeny of eutherian mammals. Systematic Biology 41:58-73. 

Patterson, C. 1981. Significance of fossils in determining evolutionary 
relationships. Annual Review of Ecology and Systematics 12:195- 
223. 

Pilleri, G., G. Marcuzzi, and O. Pilleri. 1982. Speciation in the 
Platanistoidea, systematic, zoogeographical and ecological obser- 
vations on recent species. Investigations on Cetacea 14:15-46. 

Rice, D. W. 1967. Cetaceans. Pp. 291-324 in S. Anderson and J. Knox- 



Phylogenetic Relationships of Plalanistoid River Dolphins (Odontoceti, Cetacea): Assessing (he Significance of Fossil Taxa 



133 



Jones (eds.). Recent Mammals of the World. Ronald Press, New 
York. New York. 
. 1977. A list of marine mammals of the world. NOA A Technical 



Report NMFSSSRF-71 1. 
Rothausen, K. I%8. Die systematische Stellung der europaeischen 

Squalodontidae (Odontoceti: Mammalia). Palaontologische 

Zeitschrift 42:83-104. 
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. 
. 1961. Principles of Animal Taxonomy. Columbia University 

Press. New York, New York. 
Slijper. E. 1936. Die cetacean: Vergleichend-anatomisch und 

systematisch. Capita Zoologica 7:1-590. 



Swofford. D. L. 1990. PAUP: Phylogenetic Analysis Using Parsimony, 
version 3.0. Computer program distributed by the Illinois Natural 
History Survey, Champaign, Illinois. 

Watrous, L. E., and Q. D. Wheeler. 1981. The out-group comparison 
method of character analysis. Systematic Zoology 30:1-11. 

Wiley. E. O. 1980. Phylogenetics: The Theory and Practice of Phyloge- 
netic Systematics. Wiley, New York, New York. 

Winge, H. 1921. A review of the interrelationships of the Cetacea. 
Smithsonian Miscellaneous Collections 72:1-81. 

Zhou, K. 1982. Classification and phylogeny of the superfamily 
Platanistoidea. with notes on evidence of the monophyly of the 
Cetacea. Scientific Reports of the Whales Research Institute. To- 
kyo 34:93-108. 



Are the Squalodonts Related to the Platanistoids? 

Christian de Muizon 
Institut Frangaisd' Etudes Andines, URA 12 CNRS, Casilla 18-1217, Lima 18, Peru 

ABSTRACT. — Traditionally, the superfamily Platanistoidea (Odontoceti, Cetacea) includes the four families of living river dolphins: the 
Platanistidae, Iniidae. Pontoporiidae. and Lipotidae. New studies regard the Platanistoidea as polyphyletic and classify the Iniidae. Pontoponidae, 
and Lipotidae in the infraorder Delphinida, which also includes the Delphinoidea. Previously, I suggested that the Platanistidae, Squalodelphidae, 
Squalodontidae, and Prosqualodon are closely related and together form a monophyletic superfamily Platanistoidea. Here. I test the platanistoid 
affinities of the squalodonts by examining the possibility of close relationships with other groups of odontocetes such as the Delphinida. the 
Physeterida (more precisely the Ziphioidea), and the Eurhinodelphidae. For each grouping, several features regarded as key characters in odontocete 
phytogeny are considered and an attempt is made to establish synapomorphies with the Squalodontidae. However, none of the possible 
synapomorphies that could contradict the placement of the Squalodontidae within the Platanistoidea is satisfactory because either they are 
symplesiomorphies or the structures compared are nonhomologous. Furthermore, none of the synapomorphies that relate the Squalodontidae to the 
Platanistoidea have been observed in the other three groups. Consequently, no argument is found contradicting the platanistoid affinities of the 
Squalodontidae without considerably increasing the number of convergences. Several archaic odontocetes. including Agomphius, Archaeodelphis, 
Xenorophus, Patriocetus, and Microzeuglodon have not been included in the Platanistoidea mainly because diagnostic platanistoid features are not 
observable, either because of incompleteness or inadequate specimen preparation. It is probable that these archaic odontocetes do not belong to the 
Platanistoidea but represent early branches in the evolution of odontocetes. Further studies and discoveries are needed to clarify this point. 



INTRODUCTION 

The family Squalodontidae (sensu Simpson 1945) is a group of 
odontocete cetaceans that lived worldwide during the Oligoeene 
and Miocene. Remains of squalodonts are especially abundant in 
upper Oligoeene and lower to middle Miocene rocks of Europe 
(Jourdan 1861; Paquier 1894; Capellini 1903; Dal Piaz 1904, 1916; 
Gemmelaro 1920; Rothausen 1968), Asia (Mchedlidze 1984), 
North and South America (Lydekker 1894; Kellogg 1923; Cabrera 
1926), Australia and New Zealand (Hall 19ll;Benham 1935;Flynn 
1948). Although abundant, squalodont remains are seldom well 
preserved. Complete or nearly complete skulls and/or skeletons are 
uncommon, and several generic names assigned to squalodonts are 
based on species whose type specimens are too incomplete to allow 
a meaningful comparison and, consequently, a reliable determina- 
tion. Many species of squalodonts are based upon isolated teeth or 
jaw fragments that are inadequate for accurate determination given 
the high individual and interspecific variability of cetaceans. Al- 
though skull elements or partial skulls may be sufficient in other 
groups of cetaceans, because of the scarcity of derived features in 
squalodonts, very complete skulls are necessary in this group to 
provide an adequate basis for comparison. Furthermore, since sev- 
eral key features in odontocete phylogeny are related to the auditory 
region, that region should be known in association with the skull in 
the specimens used for phylogenetic reconstructions. For those 
reasons only five genera (Squalodon, Eosqualodon, Kelloggia, 
Phoberodon, and Prosqualodon) of the family Squalodontidae were 
considered in a previous phylogenetic analysis (Muizon 1991). 
These genera are represented by complete or nearly complete skulls 
and, in part, by nearly complete skeletons. 

Furthermore, as noted above, the family Squalodontidae is a 
conservative group that retains several primitive features of the 
skull and postcranial skeleton. Derived characters appear to be rare, 
and the synapomorphies used to diagnose this family (Muizon 
1991 ) could probably be interpreted as independent acquisitions in 
the various taxa and, if considered alone, are not very satisfactory. 

These difficulties, related to the preservation of the specimens 
or inherent in the group, have made the understanding of squalodont 
relationships complex and contentious. Because of their lack of 
obviously derived characters, squalodonts have been placed with 
various groups of odontocetes. Abel (1914) regarded the 
Squalodontidae as descendants of the Agorophiidae and ancestors 
of the Physeteridae. Ziphiidae, Eurhinodelphidae, and Platanistidae 
(= Platanistoidea sensu Simpson 1945). Slijper ( 1936), contrary to 



Abel, regarded the Squalodontidae as ancestral to the Delphinidae 
(= Delphinoidea sensu Muizon 1988a) as well as to the Eurhino- 
delphidae and Platanistidae, and Rothausen (1968) admitted that 
the Delphinoidea and Platanistoidea had their origin in the Squal- 
odontidae. In the phylogeny proposed by Thenius ( 1969), relation- 
ships among taxa are unclear, although he indicated that the 
Delphinoidea, Eurhinodelphidae, and Platanistidae are more closely 
related to the Squalodontidae than to the Physeteridae and 
Ziphiidae. Barnes et al. (1985 (directly associated the Delphinoidea, 
Squalodelphidae, and Eurhinodelphidae with the Squalodontidae, 
whereas they did not recognize close relationships with the 
Platanistoidea, Ziphiidae, and Physeteroidea. Barnes ( 1990:20) re- 
garded the Squalodontidae as the sister group of the 
Squalodelphidae. and he considered the Squalodontidae, 
Squalodelphidae. and Eurhinodelphidae (= Rhabdosteidae sensu 
Barnes 1990) as a monophyletic group that is the sister group of the 
Delphinoidea. The cladogram presented by Barnes (1990:20) can 
be a better basis for discussion than the phylogenetic tree of p. 10 of 
the same paper, as the author lists numerous characters to justify his 
position. He presented no character analysis, however, and the 
previous works (Kasuya 1973; Zhou 1982; Muizon 1984, 1985, 
1987, 1988a,b; Heyning 1989) on that topic are not taken into 
account, although some characters listed by Barnes were described 
and mentioned as synapomorphies in some of these works. 

The Agorophiidae have often been considered as either the 
primitive sister group of the Squalodontidae (Slijper 1936; Thenius 
1969; Barnes et al. 1985) or ancestral to them (Abel 1914; Dal Piaz 
1977). Winge (1921) included Agomphius within the Squalo- 
dontidae. and Rothausen ( 1968) regarded Agomphius as a member 
of the superfamily Squalodontoidea along with Patriocetus, 
Microzeuglodon, and Agriocetus. Patriocetus was erroneously re- 
garded as a mysticete by Abel (1914) and Winge ( 1921 ). Whitmore 
and Sanders (1977) and Fordyce (1981) reviewed these primitive 
odontocetes and did not include them in the Squalodontidae. The 
family Agorophiidae has been restricted by Fordyce (1981) to 
include only Agomphius pygmaeus; however, several undescribed 
taxa that probably belong to this family are known. 

Recently, 1 have discussed odontocete relationships (Muizon 
1984, 1985, 1987, 1988a, 1991) and have interpreted the Plata- 
nistoidea (sensu Simpson 1945) as a polyphyletic group. The Plata- 
nistoidea (sensu Muizon 1984, 1987, 1991; Heyning 1989) are a 
monophyletic group represented by one living and several fossil 
taxa (see below). In previous works I have classified the 
Squalodontidae in the Platanistoidea (Muizon 1984. 1987, 1991). 



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

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

Proc. San Diego Soc. Nat. Hist. 29:135-146. 1994 



136 



Christian de Muizon 



The purpose of this paper is to test this hypothesis by comparing it 
to alternative hypotheses of relationships with the Delphinida. the 
Physeterida, and the Eurhinodelphidae. In the following sections 
the terms Platanistoidea, Platanistidae, Squalodontidae, Squalo- 
delphidae. Eurhinodelphidae, Physeterida, Physeteroidea, Physe- 
teridae, and Ziphiidae are used sensu Muizon (1988b, 1991). My 
interpretation of platanistoid relationships and the character analy- 
sis that justifies it have already been presented elsewhere (Muizon 
1984, 1985, 1987, 1988a, 1991). The cladogram presented in Fig- 
ure 1 summarizes the main relationships among odontocetes I 
proposed previously. 

Abbreviations. — AMNH, American Museum of Natural His- 
tory, New York, USA; IGUP, Geological Institute of Padua Univer- 
sity, Italy; MHNG. Museum d'Histoire naturelle de Grenoble, 
France; MHNL. Museum d'Histoire naturelle de Lyon, France; 
MNHN, Museum national d'Histoire naturelle, Paris, France; MLP, 
Museo de la Plata, Argentina; USNM, United States National Mu- 
seum of Natural History, Washington. D.C., USA. 

SQUALODONTIDAE AND PLATANISTOIDEA 

I have previously included the Squalodontidae within the 
Platanistoidea (Muizon 1987, 1991 (.Traditionally, this superfamily 
was regarded as a monophyletic group including the four living 
genera of river dolphins, Platanista, Inia, Pontoporia, and Lipotes 
(Winge 1921; Kellogg 1928; Shjper 1936; Simpson 1945; Hersh- 
kovitz 1966; Kasuya 1973). The authors of even recent works 
(Zhou 1982; Barnes 1985; Barnes etal. 1985: Barnes 1990), though 
disputing the position of genera within the superfamily. all accept 
the monophyly of the Platanistoidea {sensu Simpson 1945). 

In a recent revision of fossil and living odontocetes, I concluded 
that this was not the case and that the only extant taxon that should 
be included in the family Platanistidae is Platanista (Muizon 1984, 
1987). Inia, Pontoporia, and Lipotes should be classified within the 
Delphinida. mainly on the basis of several synapomorphies of the 
auditory and pterygoid regions (Muizon 1988a). It is noteworthy 
that Heyning (1989:33. 56-57), using different characters (includ- 
ing some of the soft parts), proposed a nearly identical interpreta- 
tion, providing an independent test of my earlier hypothesis 
(Muizon 1984, 1985. 1987, 1988a). 

The Platanistoidea (sensu Muizon 1984. 1987, 1991) include 
three or possibly four families: the Platanistidae, Squalodelphidae, 
Squalodontidae, and possibly Dalpiazinidae (Muizon 1987. 1988b. 
1991). The Platanistidae include Platanista, Zarhachis, and 
Pomatodelphis. The Squalodelphidae include Squalodelphis, 
Notocetus, Phocageneus, and Medocinia (Muizon 1987, 1988b). 
Four genera, namely, Squalodon, Kelloggia, Eosqualodon, and 
Phoberodon, are preserved well enough to be recognized as belong- 
ing to the Squalodontidae (Muizon 1991 ). 

The genus Prosqualodon, which lacks the synapomorphies of 
the squalodontid-squalodelphid-platanistid clade but possesses 
platanistoid synapomorphies (see Muizon 1991 and below), has 
been excluded from the Squalodontidae and is regarded as the sister 
group of all other Platanistoidea (Fig. 1 ). 

Neosqualodon is probably a derived squalodont. but since none 
of the platanistoid features could be observed on the specimens 
referred to that genus (Dal Piaz 1904: Gemmellaro 1920), I do not 
include it in the family. Moreover, it is also possible that 
Neosqualodon does not belong in the Platanistoidea. 

Sulakocetus, from the late Oligocene of the Caucasus, does not 
have the platanistoid synapomorphies of the scapula (see below) 
and consequently cannot be classified in this group. Judged from 
the illustrations provided by Mchedlidze ( 1984:pl. XIII) and Pilleri 
(1986:pl. V. 2), the premaxillae of Sulakocetus have a thickened 



X? 



** 

»> 






y/y / y y / yy 

LJi3 bjio L? 




Figure 1. Phylogenetic relationships of the major groups of odontocetes. 
I. Maxillae covering partially or totally the supraorbital processes of the 
frontals; large premaxillary foramina; posterior extension of the premaxillae 
contacting the frontals. 2, The designation "other odontocetes" refers to the 
early-diverging members of the order, including Agorophius, an undescribed 
agorophiid (USNM 256517), Archaeodelplus. Microzeuglodon, and 
Xenorophus, which have a less-telescoped skull and almost certainly do not 
represent a monophyletic taxon. 3, Posterior extension of maxillae and 
frontals. which almost contact occipital posteriorly, almost totally exclude 
the panetals from the dorsal surface of the skull (they do not articulate on the 
dorsal surface of the skull and often do not articulate at all), cover the 
temporal fossa, form a continuous crest (temporal crest) from the postorbital 
process to the nuchal crest, and demarcate a small square or trapezoidal 
frontal window on the vertex. 4. For definition of the Physeterida 
(Physeteroidea + Ziphoidea) see Muizon ( 1991 ). 5, Reduction of posterior 
process of tympanic, which is not exposed on exterior of skull in lateral 
view; development of posterior sinus; development of anterior spine on 
tympanic with salient anterolateral convexity separated from spine by a 
marked notch. 6. Reduction or loss of coracoid process on scapula; acromion 
located on anterior edge of scapula; diappearance of supraspinatus fossa on 
the lateral side of the scapula. 7. Palatine bone covered in middle by maxilla 
and pterygoids and divided into small anteroventromedial part and large 
posterodorsolateral area; presence of shallow subcircular fossa medial to 
spiny process of squamosal; presence of a foramen spinosum. 8. Lengthen- 
ing of rostrum and dilation of its apex; increase in width of vomenan 
window on ventral side of rostrum; low, wide and regularly convex or flat 
dorsal process of periotic; opening of mandibular canal larger than in the 
Platanistidae. Squalodelphidae. and Eurhinodelphoidea; increase in size of 
anterior incisors, which lie horizontally: reduction of lateral lamina of 
pterygoid hamulus. 9. Tendency toward thickening of supraorbital process; 
development of supplementary articular mechanism with squamosal on 
lateral edge of periotic; appearance of deep subcircular fossa dorsal to spiny 
process of squamosal by deepening of shallow fossa observed in the 
Squalodontidae; loss of double roots on cheek teeth. 10 and 1 1 . for defini- 
tion of these taxa, see Muizon (1987). 12. Loss of articulation of tympanic 
with squamosal; palatine hollowed out by pterygoid sinuses but no lateral 
lamina of palatine present: expansion of pterygoid sinus outside pterygoid 
fossa in orbit and temporal fossa. 13, For definition of the 
Eurhinodelphoidea (Eoplatanistidae + Eurhinodelphidae). see Muizon 
(1991). 14, For definition of the Delphinida (Lipotidae + Inioidea + 
Delphinoidea) see Muizon (1988a). (For character analyses, see Muizon 
1987, 1988a, and 1991). 



Are the Squalodonts Related to the Platanistoids? 



137 



posterior extremity that could be homologous with the premaxillary 
crest (defined by Moore 1968) observed in the Ziphiidae. If this is 
the case, Sulakocetus should be classified within the Ziphiidae 
since that structure has been regarded as the key character of the 
family (Moore 1968; Muizon 1991). 

The genus Eoplatanista (middle Miocene of Italy) has previ- 
ously been proposed as belonging in the Platanistidae (Pilleri 1985. 
1989). However, as mentioned by Barnes et al. (1985) and as 
discussed by Muizon (1988b). Eoplatanista is not a platanistid. 
Pilleri's interpretation is poorly supported for the following rea- 
sons: ( 1 ) there is no discussion of the synapomorphies that diagnose 
the Platanistidae and Platanistoidea, (2) the synapomorphies of the 
Platanistidae and Platanistoidea as diagnosed by Muizon (1987. 
1991, this paper) are not present in Eoplatanista, (3) Eoplatanista 
(family Eoplatanistidae) shares synapomorphies with the 
Hurhinodelphidae to form the superfamily Eurhinodelphoidea, 
which represents the sister group of the Delphinida (see Muizon 
1988b, 1991 ). and (4) the tooth similarities between Platanista and 
the type specimen of Eoplatanista italica mentioned by Pilleri are 
the result of the old age of the individual and in some respects are 
also found in the specimens referred by Pilleri (1989) to a new 
genus. Zignodelphis. I regard Zignodelphis as a junior synonym of 
Eoplatanista (see Muizon 1988b). 

The principal synapomoiphies supporting the diagnosis of the 
Platanistoidea used here are based upon three anatomical regions: 
the scapula, the auditory region, and the palatine bone (see Muizon 
1984, 1987, 1988a, 1991 for detailed discussion and character 
analysis). 

Scapula. — The superfamily Platanistoidea is diagnosed by im- 
portant morphological modifications of the scapula (Muizon 1984, 
1987, 1991 ). The coracoid process is a small rounded protuberance 
(sometimes almost absent), and the acromion process is located on 
the anterior edge of the scapula in such a way that the supraspinatus 
fossa is very reduced or absent (it is possibly shifted medially in 
Platanista) (Fig. 2). These features are probably related to a change 
in use of the forelimb, which needs to be analyzed functionally. The 
loss of the supraspinatus fossa does not equate with the loss of the 
supraspinatus muscle since Pilleri et al. (1976) observed a supra- 
spinatus muscle in Platanista. Therefore, the scapular synapo- 
morphies of the Platanistoidea can be defined as ( 1 ) great reduction 
of the coracoid process and (2) great reduction or loss of the 
supraspinatus fossa, with the acromion process located on the ante- 
rior edge of the bone. 

Among cetaceans, the plesiomorphic condition is found in the 
archaeocetes (the earliest known cetaceans), which have a scapula 
whose coracoid process is clearly fingerlike and straight (Kellogg 
1936 and personal observations), an acromion located on the lateral 
side, and a well-developed supraspinatus fossa. However, the mor- 
phology of the coracoid process is derived in comparison to that of 
the outgroup (i.e., mesonychid condylarths, regarded as the closest 
relatives of cetaceans, Prothero et al. 1988). The scapula in the 
outgroup bears a small, rounded, and blunt coracoid process that is 
recurved medially (see USNM 299745, a partial skeleton referred 
to cf. Mesonyx sp. from the Bridger Basin, Wyoming). 

In summary, cetaceans are diagnosed by the development of a 
straight fingerlike coracoid process of the scapula and by a ten- 
dency to reduce the supraspinatus fossa. Within cetaceans, the 
Delphinida (sensu Muizon 1988) are even more derived in having a 
coracoid process that is long, flattened, and often expanded at its 
apex. The absence of an elongated coracoid process in some ceta- 
ceans is also regarded as a derived feature. Among the odontocetes, 
reduction of the coracoid process, positioning of the acromion on 
the anterior edge of the scapula, and loss of the supraspinatus fossa 
occur in Platanista, Notocetus, Squalodon, Kelloggia, and 



Prosqualodon. Cozzuol and Humbert-Lan (1989:484) stated, 
contra Muizon (1987, 1991). that the type specimen of Phoberodon 

arctirostris has a scapula with a "conspicuous supraspinatus fossa 
which is larger than in the modern Ziphiidae" and that its acromion 
"is not on the anterior edge of the scapula"; furthermore, "the 
coracoid process (of the type specimen of P. arctirostris) is broken 
close to the base and its section permits us to infer that it was well 
developed." Cabrera ( 1926:387), however, noted that the coracoid 
process is represented merely by a small extension of the anterior 
border of the glenoid cavity. Furthermore, if the coracoid process is 
broken, I question its degree of development. The illustration 
(Cabrera 1926:fig. 8) suggests that the acromion is not located on 
the anterior edge of the bone. If Cozzuol and Hutnbert-Lan's obser- 
vation is correct (unfortunately no illustration of the specimen was 
provided), then Phoberodon should be excluded from the 
Platanistoidea and, therefore, from the Squalodontidae. Like 
Sulakocetus, Phoberodon is a genus that needs revision and the 
discovery of better-preserved skeletons. 

In several genera classified in the Platanistoidea. including 
Zarhachis, Pomatodelphis, Squalodelphis, Phocageneus, Medo- 
cinia, and Eosqualodon, the scapula is unknown. However, these 
genera possess other derived features of their auditory and ptery- 
goid regions allowing their inclusion in this group (Muizon 1987, 
1988b; see below). 

The scapulae of Inia, Pontoporia, and Lipotes have a large 
coracoid process and an acromion located on the lateral side of the 
bone. Thus these genera lack the derived character states found in 
the Platanistoidea. On the contrary, their morphology is similar to 
that of the other Delphinida (Muizon 1988a). 

A reduced coracoid process and an acromion located at the 
anterior edge of the scapula are features also present in the 
mysticetes Balaena and Megaptera. In these two taxa. however, the 
acromion is also very reduced (Balaena) or absent (Megaptera), 
while the acromion of the Platanistoidea is very large, as in other 
odontocetes. Consequently, the scapula of the Platanistoidea differs 
from that of Balaena and Megaptera. and it is probable that these 
differences represent different functional modifications. Reduction 
of the coracoid process is therefore interpreted as a convergence 
between mysticetes and odontocetes. 

Auditory region. — The Platanistidae and the Squalodelphidae 
(sensu Muizon 1987) have been regarded as sister taxa mainly 
because both share a deep subcircular fossa located dorsal to the 
spiny process of the squamosal (Muizon 1987:fig. 3 and p. 5, 
Muizon 1991 :fig. 11). The lack a subcircular fossa, the 
plesiomorphic condition, characterizes other cetaceans. As stated 
elsewhere (Muizon 1987, 1991), this structure could represent a 
simple extension of the peribullary sinus, which surrounds the 
periotic and part of the tympanic of odontocetes. It is noteworthy, 
however, that the subcircular fossa almost always possesses numer- 
ous foramina, implying an important function in blood supply, as 
hypothesized by Fordyce (this volume). A subcircular fossa is 
observed in Platanista, Pomatodelphis, Zarhachis, Notocetus, 
Medocinia, and Squalodelphis. A shallow subcircular fossa is also 
present in Squalodon bariensis, S. calvertensis, S. tiedemani, and 
Eosqualodon latirostris. In those species the foramina are present 
but larger and less numerous than in the Platanistidae and 
Squalodelphidae. Furthermore, in Squalodon and Eosqualodon, the 
subcircular fossa opens anteromedially in a groove (oriented 
anteromedially-posterolaterally) that leads into a small foramen 
located between the fossa and the foramen ovale. In Squalodon 
calvertensis (USNM 328343 and 214644) apparently this passage 
opens in the foramen ovale. The same foramen has been described 
in Waipatia by Fordyce (1994, this volume) as the foramen 
spinosum. In Zarhachis and Pomatodelphis a foramen spinosum is 



138 



Christian de Mui/.on 




Figure 2. Lateral view of the scapula in some odontocetes. a, Squalodon (from USNM 22902); b, Prosqualodon (from AMNH 29022); c, Notocetus 
(from AMNH 29060); d. Platanista (from MNHN 1870-79); e, Eurhinodelplm (from USNM 1 1867); f, Pontoporia (from MNHN 1934-375); g, Phocoena 
(from MNHN 1982-137). All figures are schematic and not drawn to scale. AC. acromion; CP. coracoid process. 



positioned between the subcircular fossa and the foramen ovale. In 
those genera, however, the groove is sometimes absent or reduced 
but never as deep as in Squalodon. In Zarhachis flagellator ( USNM 
10991 ), the ventral edge of the foramen is formed by the junction of 
two lips, indicating that a canal has been formed by the closure of 
the lips of a "gutter". Furthermore, in USNM 13768 {Pomato- 
delphis determined as Zarhachis by Muizon 1987), in which the 
internal side of the braincase is visible, the foramen apparently 
opens at least partially within the cranial cavity in the parietal, 
while in USNM 10911 (Zarhachis) the foramen spinosum clearly 
opens in the foramen ovale. In Platanista the foramen has disap- 



peared because a large cranial hiatus has developed. Therefore, 
there seems to be some interspecific (or intraspecific) variation in 
the cerebral opening of the foramen spinosum. Nevertheless, I 
consider the shallow subcircular fossa of Squalodon and Eosqua- 
lodon to be homologous with that observed in the Platanistidae and 
Squalodelphidae because ( I ) it is located in the same position, 
dorsal to the spiny process (its lateral border being exactly dorsal, 
sometimes dorsolateral, to the base of the spiny process), (2) it 
bears a similar vascularization, suggesting that both functioned in 
blood supply, and (3) it is associated anteromedially with a similar 
foramen spinosum. 



Arc the Squalndonts Related to the Platanistoids'.' 



139 



A vascularized subcircular fossa with a foramen spinosum has 
not been found in the Physetenda or in the clade consisting of the 
Eurhinodelphoidea and Delphinida. In Kentriodon and in some 
Recent Delphinidae a shallow depression is sometimes observed 
(as an individual variation) in a position close to that in Squalodon, 
but it is generally more medial. However, there is no foramen 
spinosum, and the position of the depression, when present, is 
variable. Therefore, it is likely that the condition sometimes ob- 
served in the Delphinoidea is not homologous but instead related to 
the presence of a cranial hiatus and to the important development in 
the Delphinoidea of the penbullary sinus, which tends to excavate 
the bones surrounding the periotic and the tympanic. In contrast, the 
subcircular fossa of the Platanistoidea seems to be associated with 
the circulatory system (Fordyce 1994, this volume). A subcircular 
fossa was not seen in Eosqualodon langeweischei, possibly because 
the tympanic and periotic of the specimen illustrated by Rothausen 
( 1968) have not been removed from the skull. The same is true for 
Kelloggia barbara Mchedlidze (1984); however, the morphology 
of Kelloggia is very similar to that of Squalodon, and I suspect the 
former to be a junior synonym of the latter. A subcircular fossa is 
not present in Prosqualodon australis (AMNH 29022 and MLP 5-8 
and 5-9). It is also noteworthy that Phoberodon arctirostris has no 
subcircular fossa (M. A. Cozzuol. pers. comm.), contrary to my 
earlier statement (Muizon 1991) that would confirm its exclusion 
from the Platanistoidea. The presence of a foramen spinosum in the 
Squalodelphidae could not be determined since no specimens were 
available for this study. 

The families Platanistidae and Squalodelphidae also share a 
supplementary articular mechanism with the squamosal on the 
lateral edge of the periotic (Fig. 3). The plesiomorphic condition, 
absence of this articular mechanism, characterizes the Archaeoceti, 
Ziphiidae, Physeteridae, Eurhinodelphoidea (sensu Muizon 1988b). 
and Delphinida. This stucture is a hooklike articular process in the 



IAPP 




ARP 



ARP 




Figure 3. Lateral view of the periotic of some odontocetes. a, Squalodon 
(from MHNL Dr 15); b. Notocetus (from AMNH 29060); c, Phocageneus 
(from USNM 21039); d. Pomatodelphis (from USNM 1X7414); e, 
Platanista (from MNHN 1870-79). All figures are schematic and not drawn 
to scale. ARP, articular rim of the periotic; HAPP, hooklike articular process 
of the periotic; IAPP, incipient articular process of the periotic. 



Platanistidae (Platanista, Zarhachis, and Pomatodelphis). In 
Platanista it is not possible to remove the periotic from the skull 
without breaking cither the process or the corresponding fossa on 
the squamosal. In the Squalodelphidae, there is no true hooklike 
process but a less derived articular rim (as in Notocetus and Phoca- 
geneus) or a straight somewhat conical process (Squalodelphis). 
This feature has not been found in the Squalodonadae, but the 
periotic of Squalodon bariensis shows a preliminary stage in the 
development of the condition found in the Platanistidae and the 
Squalodelphidae. On the anterior border of the articular facet for the 
tympanic, the periotic of S. bariensis displays a groove extending 
from the fossa crus breve incudis to the apex of the process. This 
groove receives the spiny process of the squamosal, with which it 
articulates tightly (Muizon 1991 :fig. II, 12). The anterior crest that 
delimits the articular groove is elevated, and its posterior extremity 
forms a peg (Fig. 3A) that is in the same position as the articular rim 
and the hooklike process of the Squalodelphidae and Platanistidae. 
In fact, the supplementary articular mechanism observed in the 
Platanistidae and Squalodelphidae is the result of the important 
increase in size and thickness of the anterior crest of the articular 
groove for the spiny process of the squamosal. That interpretation is 
clear in a comparison of the periotics of Squalodon calvertensis 
(USNM 187315) and Zarhachis flagellator (USNM 26274). The 
condition observed in Squalodon bariensis is well marked in the 
holotype of Squalodon calvertensis, and I observed it (sometimes 
marked much better, with an incipient articular rim) in most of the 
specimens of the large collection of Squalodon periotics in the U.S. 
National Museum. That feature, although very common, is not 
constant in the genus Squalodon. Furthermore, it was not observed 
in the other squalodontid genera, since the periotic is either un- 
known (Phoberodon arctirostris) or it has not been removed from 
the skull (Eosqualodon langeweischei and Kelloggia barbara). The 
periotic of Prosqualodon davidis has been illustrated by Flynn 
( 1948). but the figure does not show enough detail, and the holotype 
(and only known specimen) has been lost. The periotic referred by 
True ( 1909) to Prosqualodon australis does not have the articular 
modification observed in Squalodon. This periotic is not associated 
with a skull. However, several skulls of Prosqualodon australis 
(with associated periotics) recently collected in Patagonia confirm 
True's interpretation (Cozzuol, pers. comm.). 

Palatine bone. — Previously, the Platanistidae were diagnosed 
by the structure of their palatine bones (Muizon 1987). In 
Pomatodelphis and Zarhachis, the palatines are not articulated 
ventrally on the palate as in most odontocetes; they clearly have 
migrated dorsolateral^ and are surrounded by the maxilla and the 
pterygoid (which partially overlap them). In Platanista the condi- 
tion is even more specialized, as the pterygoid totally overlaps the 
palatine and the posterior part of the maxilla (Kellogg 1924). The 
plesiomorphic condition in which the maxillae are totally separated 
from the pterygoids by the palatines is found in the Archaeoceti, in 
the Agorophiidae (new genus from the Oligocene of Oregon under 
study by E. Fordyce, USNM 256517), and in primitive Ziphiidae 
(e.g., Squaloziphius emlongi). 

An interesting observation has been made on the further pre- 
pared skull of the holotype of Squalodon bariensis (Muizon 1991 ). 
The palatines of this specimen are clearly divided into a small 
ventromedial portion contacting the other palatine and a large pos- 
terolateral portion in a position similar to that in Pomatodelphis and 
Zarhachis (Fig. 4). The condition in Squalodon appears to be the 
consequence of a posterior extension of the maxilla, which partially 
overlaps the palatine, dividing it into two areas on the surface of the 
skull and contacting the pterygoid in its anterior region. Further- 
more, as clearly seen in the lateral view of the skull of 5. bariensis 
(Muizon 1991;fig. 10), the pterygoid also overlaps the ventral limit 
of the palatine. This arrangement, in which the palatine is split in 



140 



Christian de Muizon 







Figure 4. Lateral view of the skull in some odontocetes. a, Squalodon (from MHNLDr 15); b, Pomatodelphis (from USNM 187414); c. Eurhinodelphis 

(reconstruction from various USNM specimens); d, Delphinus (from private specimen). All figures are schematic and not drawn to scale. DPP, dorsal part 
of the palatine; FPS, falciform process of the squamosal; Fr. frontal; LLP. lateral lamina of the palatine; Mx. maxilla; Pal, palatine; Pt, pterygoid; PTH, 
pterygoid hamulus; Sq, squamosal; VPP, ventral part of the palatine. 



two parts by a posterior extension of the maxilla, is not common 
among odontocetes. It is regarded here as a first step toward the 
platanistid specialization in which the ventromedial part of the bone 
has disappeared. This feature was not observed in the Squalo- 
delphidae, the sister group of the Platanistidae, probably as a conse- 
quence of the poor preservation of that suture on the four well- 
preserved skulls of the Squalodelphidae examined [Notocetus 
vanbenedeni (two skulls) and Squalodelphis fabianii (one skull), 
and Medocinia tetragorhina (one skull)]. It is probable that this 
character was present in the Squalodelphidae since, at the anterior 
region of the fossa for the hamular lobe of the pterygoid sinus, no 
trace of the palatine can be seen medially, a condition indicating 
that the bone has probably been displaced laterally as in the 
Platanistidae or covered by the maxilla or pterygoid in its medial 
part. The same observation can be made on the badly crushed skull 
of the holotype of Medocinia tetragorhina (Muizon 1988b:80). 
That feature was likely present in Squalodon calvertensis (holo- 
type, USNM 10484, and USNM 328343) and is obvious in 
Squalodon tiedemani (USNM 183023 and 424070); however, it 
could not be observed in Squalodon bellunensis and Phoberodon 
arctirostris because of poor preservation of the skulls. Furthermore, 
the skulls of the holotypes of Eosqualodon langewieschei and 
Kelloggia barbara require further preparation to allow better obser- 
vation of the palatine bones. In Prosqualodon davidis and P. austra- 



lis the relationships of the palatine with the maxilla and the ptery- 
goid are similar to those in nonplatanistoid odontocetes, and the 
maxilla does not overlap the palatine to contact the pterygoid. The 
reconstruction of the palatine suture of Squaloziphius emlongi 
(Muizon 199l:fig. 4). which is shown divided by a pterygoid- 
maxilla contact, is an editorial error; as indicated in the text (Muizon 
1991:290). the maxilla-palatine suture shows the plesiomorphic 
condition of a "regular parabola opening posteriorly and whose 
branches run obliquely from the median suture to the anterior edge 
of the ventral opening of the infraorbital foramen"; in that species 
there is no contact between the maxilla and the pterygoid. Such a 
contact is seen in the Physeterida as individual variation in some 
specimens of Kogia (Fraser and Purves 1960:pl. 16), Hyperoodon 
(Fraser and Purves 1960:pl. 8), Berardius (Fraser and Purves 
1960;pl. 9) Ziphius (Fraser and Purves 1960:pl. 10), and 
Mesoplodon. The condition seen in the Recent Ziphiidae (it is 
absent in all the fossil forms in which that region of the skull is 
preserved) is a result of the enlargement of the pterygoid, which 
overlaps the palatine anteriorly. It is also absent in the fossil kogiid 
Scaphokogia. In Squalodon bariensis and 5. tiedemani (USNM 
424070) the separation of both parts of the palatine is partially 
achieved by a posterior extension of the maxilla, which overlaps it 
and contacts the pterygoid. Consequently, the conditions observed 
in Squalodon and in the Physeterida are not homologous. 



Are the Squalodonts Related to the Platanistoids? 



141 



Muizon (1991 :302) mentioned that the speeimen from St. Paul- 
Trois-Chateaux (Department of Drome, Franee; MHNG 5000) 
referred to Squalodon bariensis has a W-shaped suture of the pala- 
tine with the maxillae and. therefore, no contact between the maxil- 
lae and the pterygoids. This arrangement is due to the specimen's 
being a young animal in which the posterior extension of the 
maxillae had not yet fully developed. 

RELATIONSHIPS OF THE DALPIAZINIDAE 

The family Dalpiazinidae Muizon, 1988, is monotypic and rep- 
resented so far by a single species. Dalpiazina ombonii. The diag- 
nosis of this taxon and the description of the specimens referred to it 
have been presented previously along with discussions of its tax- 
onomy and taphonomy (Muizon 1988b). 

Before addressing the taxonomy and phylogenetic relationships 
of Dalpiazina ombonii. I must recount the great confusion existing 
in the associations of the specimens initially described by Longhi 
( 1898) as belonging to Champsodelphis ombonii. Considering this, 
I have formally designated a lectotype forD. ombonii. IGUP 26405, 
a partial mandible associated with a fragment of maxilla (Muizon 
1988b:64). The other specimens described by Longhi cannot be 
unequivocally referred to Dalpiazina ombonii and should not be 
taken into account for taxonomy or phylogeny (Muizon 1988b:62- 
66). I have also stated that the individual "B" described by Dal Piaz 
(1977:30) was composed of elements that, very probably, do not 
belong to the same taxon, and I have recommended that they be 
removed from the hypodigm (Muizon 1988b:66-67). The 
hypodigm of Dalpiazina ombonii should therefore be restricted to 
three specimens: (1) the lectotype of Longhi (1898), (2) the indi- 
vidual "A" of Dal Piaz (1977:27 and pi. I), a partial cranium and 
rostrum, a left periotic, and a cervical vertebra (Muizon 1988b:67- 
76), and (3) the individual "C" of Dal Piaz (1977:33 and pis. II, 19 
and 20). 

Specimens "A" and "C" of Dal Piaz show some features of the 
skull and periotic (which could be regarded as synapomorphies) 
that could indicate a relationship of Dalpiazina ombonii with the 
Squalodontidae (Muizon 1988b. 1991). This interpretation dis- 
agrees with that of Pilleri (1985, 1989), who classified D. ombonii 
within the Delphinoidea. I am reluctant to accept Pilleri's assess- 
ment for two reasons: ( 1 ) he has not explained his concept of the 
Delphinoidea and the synapomorphies he employed to diagnose 
that superfamily, and (2) the character states used to diagnose the 
Delphinoidea (Muizon 1988a; Heyning 1989; Barnes 1990) either 
are not observed (except for single-rooted teeth with conical 
crowns) in specimens confidently referred to Dalpiazina ombonii 
(see above) or are not preserved. Because of the high variability of 
cetacean teeth, they should not be used the sole character diagnos- 
ing a taxon (according to Pilleri's concept of delphinoid teeth, the 
majority of the teeth of young adult eurhinodelphids should be 
classified within the Delphinoidea). Consequently, I disagree with 
Pilleri's classification of D. ombonii in the Delphinoidea. 
"Dalpiazella" is an error by Pilleri et al. (1989:223-224) for 
Dalpiazina. 

The Squalodon-like morphology of the periotic of Dalpiazina 
ombonii has been discussed by Muizon (1988b) and accepted by 
Fordyce ( 1994, this volume). However, it is uncertain whether the 
similarities of the periotics of Dalpiazina and Squalodon represent 
synapomophies, with the probable exception of the morphology of 
the dorsal process (see Muizon 1988b, 1991 ). Four character states 
that are possible synapomorphies relating Dalpiazina to the Squalo- 
dontidae (Muizon 1991) are ( 1 ) longer rostrum with enlarged apex, 
(2) wider vomerian window on the ventral side of the rostrum, (3) 
low, wide and regularly convex dorsal process of the periotic, and 
(4) opening of the mandibular canal larger than in the Platanistidae, 



Squalodelphidae, and Eurhinodelphoidea. Pilleri (1989:385) re- 
jected the first synapomorphy, arguing that it is also present in the 
Delphinidae. The condition observed in some short-snouted 
delplunids (Globicephalinae) is different from that of the Squal- 
odontidae and Dalpiazina. In the Globicephalinae, the rostral por- 
tions of the maxilla and premaxilla are equally long and both reach 
the apex of the shortened rostrum; on the long rostrum of the 
Squalodontidae and Dalpiazina the maxilla is shorter than the 
premaxilla and never reaches the apex. The anterior widening of the 
premaxillae on the rostrum of the Globicephalinae is due to the 
strong development of the melon, which characterizes that subfam- 
ily. Consequently, the condition in the Globicephalinae is not ho- 
mologous with that of the Squalodontidae and Dalpiazina. 

The hypothesized close relationship between the Dalpiazinidae 
and the Squalodontidae suggested in earlier works (Muizon 1988b, 
1991) remains poorly supported since none of the platanistoid 
synapomorphies mentioned above were observed on the available 
specimens because of their poor preservation. Consequently, the 
Dalpiazinidae are a possible sister group of the Squalodontidae, 
although this hypothesis has yet to be confirmed. 

SQUALODONTS AND OTHER ODONTOCETES 

In the following section I compare the Platanistoidea with the 
three other major groups of odontocetes: the Delphinida. the 
Physterida, and the Eurhinodelphoidea. 

Delphinida. — The Delphinida have been diagnosed by Muizon 
( 1988a: 164) on the basis of nine synapomorphies: ( 1 ) acquisition of 
a lateral lamina on the palatine, (2) virtual loss of the posterior 
region of the lateral lamina of the pterygoid. (3) excavation of the 
posterodorsal portion of involucrum of the tympanic, (4) reduction 
of the anterior process of the periotic. (5) development of a ventral 
swelling and tubercule on the periotic, (6) increase in size of the 
processus muscularis of the malleus, (7) enlargement of the trans- 
verse apophyses of the lumbar vertebrae as triangular blades, (8) 
anterior inflexion of the anteroventral angle of the sigmoid process 
of the tympanic, and (9) frontals narrower than (or as wide as) nasal 
on vertex. A continuous lateral lamina of the pterygoid is observed 
in Pontoporia; however, this condition has been regarded as a 
reversal in some Pontoporiidae (Barnes 1985; Muizon 1988a). 

The oldest known fossil Delphinida belong to the family 
Kentriodontidae Barnes, 1978, a taxon whose definition and rela- 
tionships with the superfamily Delphinoidea are complex (Muizon 
1988a). Some taxa known by well-preserved specimens come from 
the middle Miocene of North and South America and include the 
genera Kentriodon, Delphinodon, Liolithax, Lophocetus, and 
Atocetus. Oligodelphis is from the late Oligocene of the Caucasus; 
however, in the specimens referred to this genus none of the 
synapomorphies of the Delphinida are preserved. Other fossil 
Delphinida from the Miocene represented by well-preserved speci- 
mens are the lipotid Parapontoporia and the iniid Ischyrorhynchus. 

In their phylogenetic tree, Barnes etal. (1985: fig. 1) considered 
the Squalodontidae and the Delphinoidea closely related. However, 
none of the synapomorphies of the Delphinoidea (Muizon 1988a) 
have been found in the Squalodontidae and consequently their 
inclusion in the Delphinida cannot be supported. If the Squalo- 
dontidae were regarded as the sister group of the Delphinida, the 
platanistoid synapomorphies (e.g.. reduction of coracoid process 
and supraspinatus fossa on the scapula, maxilla covering the pala- 
tine bone, and presence of a shallow subcircular fossa) would have 
to be considered convergences. No synapomorphies have been 
found to relate the Squalodontidae to the Delphinida or even to the 
Delphinoidea. Such a hypothesis would require too many parallel- 
isms and reversals to be acceptable. Only a few authors have 
suggested a close relationship between the Delphinoidea and the 



142 



Christian de Muizon 



Squalodontidae. Winge ( 1921 ) stated that the Delphinidae had their 
origin in the Platanistidae, a family in which he included the four 
living genera of river dolphins. Winge's interpretation ( 1921:46) is 
based principally on his concept of the taxon Platanistidae. If one 
considers that three of the four extant genera that he included (i.e., 
Inia, Pontoporia, and Lipotes) belong to the Delphinida and that the 
Inioidea (Iniidae + Pontoporiidae) are the sister group of the 
Delphinoidea, his interpretation is understandable. Under my inter- 
pretation of the Platanistoidea. which excludes the Iniidae. 
Pontoporiidae, and Lipotidae, no close relationship between the 
Delphinoidea and the Platanistoidea, and thus the Squalodontidae. 
can be established. 

Physeterida. — This infraorder includes two superfamilies, the 
Physteroidea and the Ziphioidea, and is regarded here as a mono- 
phyletic taxon. However, Heyning ( 1989) and Heyning and Mead 
(1990). from analysis of the morphology of the air sacs and nasal 
tracts as well as osteological data, did not recognize a close rela- 
tionship between these superfamilies and regarded the 
Physeteroidea as the sister group of the remaining extant families of 
odontocetes. According to this interpretation the Physeterida are 
paraphyletic. These authors noted that the physeteroid morphology 
of the nasal tracts, although highly specialized, retains several 
primitive features. This interpretation may be correct; however, 
because of the spectacular modification due to scaphidiomorphy 
(development of a large supracranial basin) of this region of the 
physeteroid skull, it is also possible that the plesiomorphic features 
of the nasal tracts recognized by Heyning and Mead actually repre- 
sent apomorphic features (reversals) imposed by its hyperspeciali- 
zation (for instance, the morphology of the premaxi 11a of the middle 
Miocene Orycterocetus crocodilinus indicates that the bone could 
very well have contained a small premaxillary sac). Furthermore. 
Heyning's interpretation implies convergence of the five synapo- 
morphies used to define the Physeterida (Muizon 1991 :fig. 5). The 
monophyletic alternative is accepted here, although it is clear that a 
careful comparison of the synapomorphies supporting each inter- 
pretation is needed. 

It is noteworthy that Heyning's (1989) cladogram considers 
only living odontocetes. The exclusion of fossil taxa from phyloge- 
netic reconstructions is a subjective choice that certainly introduces 
errors that cannot be compensated for by computer-generated cla- 
dograms, since it may result in a more parsimonious cladogram 
markedly different from one that includes both living and fossil 
taxa. The significance of fossils in the reconstruction of phyloge- 
netic relationships has been demonstrated by several authors 
(Gauthieret al. 1988; Donoghue et al. 1989; Novacek 1992), and it 
has been recommend that no phylogeny should ignore the fossil 
record. The introduction of fossil taxa in phylogeny allows better 
definition of character states and homoplasies and consequently 
provides more information for phylogenetic reconstruction. 

Of the three characters presented by Heyning ( 1989:fig 39) to 
diagnose the nonphyseterid odontocetes, one (character 16, the 
presence of premaxillary sacs) is present in the Agorophiidae. 
which still retain joint parietals on the vertex and nasals overhang- 
ing the bony nares but have the maxilla covering the supraorbital 
process of the frontal, the key synapomorphy of the odontocetes. 
The two other characters (blowhole ligament and nasal passages 
confluent) are soft-anatomical characters and cannot be evaluated 
among fossil odontocetes. The feature "temporal fossa roofed over 
by expansions of the maxillae," retained by Heyning ( 1989:fig 39. 
character 22) as a synapomorphy of the Ziphiidae and Delphi- 
noidea, is also found in the middle Miocene physeterid Oryctero- 
cetus crocodilinus and platanistid Zarhachis flagellator, while it is 
very poorly developed in the early Miocene ziphiid Squaloziphius 
emlongi and middle Miocene kentriodontid delphinoid 
Kampholophos semdus. Furthermore, the feature "facial asymme- 



try" (Heyning 1989:fig. 39, character 4) cannot be retained as an 
odontocete synapomorphy since it is absent in several fossil 
odontocetes [e.g., Agorophius, an undescribed agorophiid (USNM 
256517). Patriocetus, and Arcliaeodelphis]. Consequently, the in- 
troduction of the Agorophiidae or of fossil taxa of extant families 
into Heyning's cladogram also introduces character conflicts that 
can be resolved by including the Physeteroidea and the Ziphiidae in 
the same monophyletic taxon. 

Abel (1914) proposed a phylogeny of cetaceans that suggests a 
close relationship between the Squalodontidae and the Ziphiidae. It 
is true that the Ziphiidae show several similarities with the 
Squalodontidae, such as the anterior extension of the pterygoid 
contacting the maxilla (as an individual variation), overlapping the 
palatine, and dividing the palatine on the palate (see above). 

In Squalodon bariensis the ventrolateral portion of the ptery- 
goid hamulus is reduced. This can be regarded as the first step 
toward the condition observed in the living ziphiids, in which the 
pterygoid hamulus has lost its lateral lamina (the polarity of that 
character state has been discussed by Muizon 1984). Study of the 
fossil ziphiids Ninozipliius and Squaloziphius indicates that the 
pterygoid condition observed in living ziphiids is the result of 
reduction of the lateral lamina that is present but vestigial on the 
edges of the pterygoid hamulus of Ninozipliius platyrostris (early 
Pliocene of Peru) and partially closes the hamular fossa laterally in 
the older Squaloziphius emlongi (early Miocene of Washington 
State. USA). Furthermore, the condition in S. emlongi shows that in 
ziphiids the reduction of the lateral lamina of the pterygoid is first 
achieved by reduction of its posterior part, lateral to the 
basioccipital. The second step is observed in Ninozipliius 
platyrostris, in which the anterior portion of the lateral lamina of the 
pterygoid (i.e., the lateral lamina of the pterygoid hamulus) has also 
disappeared. However, the condition of the pterygoid in Squalodon 
bariensis indicates that the reduction of the lateral lamina would be 
initiated by loss of its hamular portion, which is contradicted by 
data provided by the fossil ziphiids. The condition in Squalodon 
must therefore be regarded as independent of that in ziphiids. 

The premaxi Uae of the Squalodontidae and Prosqualodon con- 
tact the frontals posteriorly on the vertex (Fig. 5). The same feature 
is present in the Ziphiidae, Squaloziphius, Ziphirostrum. 
Choneziphius. Ziphius. Hyperoodon, and almost always in 
Mesoplodon (at least on the left side of the skull). A premaxilla- 
frontal contact is also seen in the Eurhinodelphidae. Eoplata- 
nistidae. Platanistidae, Squalodelphidae, and Dalpiazinidae. This 
character is absent in the Delphinida (although it is observed occa- 
sionally as a result of individual variation), in the Berardiini 
[Berardius, Tasmacetus, Ninozipliius (undescribed specimen)], and 
in the Physetendae. It is also present in some mysticetes. Appar- 
ently a premaxilla-frontal contact is absent in archaeocetes and 
thus is derived in cetaceans. However, it is noteworthy that all early 
odontocetes show a premaxilla-frontal suture (Agorophius, 
Patriocetus, Xenorophus. and Archeodelphis). Furthermore, the 
oldest known ziphiid, Squaloziphius emlongi, possesses this fea- 
ture. Consequently, it seems reasonable to infer that, within the 
odontocetes, the plesiomorphic state is a contact between the pre- 
maxilla and frontal. The apomorphic condition in odontocetes is the 
loss of the articulation, which is interpreted as having evolved 
convergently in several groups. In the Delphinida. the apomorphic 
condition is the consequence of reduction of the posterior apex of 
the premaxillae. which is extreme in the Phocoenidae. In the 
Berardiini it is probably related to the increase in size of the nasals, 
and in the Physeteridae it is probably related to the development of 
a large supracranial basin (scaphidiomorphy). 

Consequently, the two features mentioned above (i. e., reduc- 
tion of the lateral lamina of the pterygoid and presence of a premax- 
illa-frontal suture) that could be regarded as synapomorphies of the 



Are the Squalodonts Related to (he Platanistoids? 



L43 





Figure 5. Dorsal view of the vertex in some odontocetes. a, Squalodon (from MHNL Dr 15); b, Prosqualodon (from Kellogg 1928); c. Eurhinodelphis 
(from USNM 8X42); d. Zarkachis (from USNM 1091 1 ); e,Agowphius (from Fordyce 1981 ); f. Patrioceius (from K. Rothausen, unpublished thesis. 1965); 
g,Xenoroplms (from Kellogg 1923); h.Archaeodelphis (from Kellogg 1928); i, Squaloziphius (from USNM 181528); j, Ziphirostrum (from Kellogg 1928). 
All figures are schematic and not drawn to scale. Fr, frontal; Mx. maxilla; Na. nasal; Oc. occipital; Pa. parietal; Pal. palatine; Pmx. premaxilla. 



Squalodontidae and the Ziphiidae cannot be accepted. Furthermore, 
the absence in each family of synapomorphies of the other rein- 
forces the hypothesis that there is no close relationship between the 
Squalodontidae and Ziphiidae. 

Fordyce (1985) suggested a close, possibly sister-group, rela- 
tionship of the Squalodontidae and Ziphiidae, basing his assess- 
ment on four derived features: ( 1 ) relatively deep rostrum. (2) teeth 
inserted on the lateral flanks of the rostrum, (3) robust zygomatic 
process, and (4) twisted transversely inflated anterior process of the 
periotic. However, the distribution of these character states among 
other odontocetes suggests that they cannot be retained as 
synapomorphies. As defined by Fordyce. the feature "deep ros- 
trum" is imprecise; it should mention at which level of its length the 
rostrum is deep. If the depth of the rostrum is measured at its base, 
the feature cannot be used since it also occurs in archaeocetes, in 
several primitive odontocetes (e. g., Xenorophus), in an undescribed 
agorophiid (USNM 256517), and in most eurhinodelphids. If the 
character is defined as "rostrum deep in its anterior half," the 
synapomorphy cannot be used since it occurs in some Physeteroidea 
(Scaphokogia) and because it is not constant in the Ziphiidae 
{Squaloziphius emlongi, Ninoziphius platyrostris, and Berardius 
have a fairly flat rostrum). Furthermore, because of the wide varia- 
tion among odontocetes in the morphology of the rostrum, charac- 
ter states based upon its shape or length should not be used at higher 
taxonomic levels. The second character proposed by Fordyce is 
unclear since teeth are inserted on the lateral flanks of the rostrum 
in all odontocetes. The third character, the strong zygomatic pro- 
cess, is found not only in the Ziphiidae but also in all Platanistidae, 
Squalodelphidae, Prosqualodon, and, to a certain extent, in the 
Agorophiidae. On the contrary, the zygomatic process' being ro- 
bust, long, and slightly recurved ventrally could represent a 
synapomorphy of the Platanistoidea (sensu Muizon 1987, 1991). In 
the Ziphiidae this feature is much less marked (except in 



Squaloziphius, where its tremendous size is more a consequence of 
enlargement of the postglenoid process and represents an 
autapomorphy), and the process is never as long and always 
strongly recurved ventrally. It is also possible that the robustness of 
the process observed in the Squalodontidae represents a 
symplesiomorphy since a large zygomatic process is also found in 
some primitive odontocetes such as Agorophius, an undescribed 
agorophiid (USNM 256517), and Patriocetus. However, this hy- 
pothesis seems less probable as in these genera the process is 
generally less robust than in the Platanistoidea. The fourth proposed 
derived feature, the morphology of the anterior process of the 
periotic, is also found in the Platanistidae and Squalodelphidae, 
which have the same transversely inflated anterior process. It is true 
that the ventrally twisted anterior process is not found in those two 
families, but it is always observed in the Squalodontidae, 
Eurhinodelphoidea, and Physeterida. In summary, three of the four 
derived features listed by Fordyce (1985) are problematic. 

Eurhinodelphidae. — This extinct family of odontocetes is the 
best candidate, besides the Platanistoidea, for a close relationship 
with the Squalodontidae. The Eurhinodelphidae and the superfam- 
ily Eurhinodelphoidea (Eurhinodelphidae and Eoplatanistidae) rep- 
resent plausible morphological descendants of the Squalodontidae. 
However, the Eurhinodelphidae do not share any convincing 
synapomorphies with the Squalodontidae or with the Platanistoidea 
(sensu Muizon 1987). One possible synapomorphy is the presence 
of a contact between the premaxillae and the frontals. However, as 
discussed above, this is probably a symplesiomorphy among 
odontocetes, and the delphinoid condition, in which only the nasals 
contact the premaxillae. is an apomorphy. 

Furthermore, the Squalodontidae, Platanistidae, Squalo- 
delphidae, and Prosqualodon have a strong anterior spine on the 
tympanic, a feature also present in the Eurhinodelphoidea and 
Lipotidae. This character is absent in the Archaeoceti, an 



144 



Christian de Muizon 



undescribed agorophiid (USNM 256517, under study by E. 
Fordyce), the Physeterida. and the Delphinoidea. Absence of the 
anterior spine in earlier-diverging groups (Archaeoceti and 
Agorophiidae) indicates that that condition is plesiomorphie, and 
thus the presence of the spine is apomorphic. However, the absence 
of an anterior spine on the tympanic in the Delphinoidea would 
contradict that interpretation. In fact, among the Delphinida (sensu 
Muizon 1988a). the Lipotidae, the earliest-diverging lineage, have a 
strong anterior spine: the Inioidea, the next lineage to diverge (sister 
group of the Delphinoidea), have a very reduced spine, and the 
Delphinoidea lack spines (the projections sometimes observed in 
Tursiops and Globicephala are regarded as hyperossification due to 
old age, since they are totally absent in young adults). Conse- 
quently, it seems that the tendency in the Delphinida is reduction of 
the anterior spine. If one admits that the absence of a spine is the 
plesiomorphie state, one must admit that the condition in the 
Delphinida is a reversal. The acquisition of a strong anterior spine 
on the tympanic has been regarded as a synapomorphy of the group 
Platanistoidea + Eurhinodelphoidea + Delphinida (Muizon 1991 ). 
Consequently, the common occurrence of an anterior spine in the 
Squalodontidae and in the Eurhinodelphidae is a symplesiomorphy 
within that group. 

The Eurhinodelphidae. as well as the Eoplatanistidae. do not 
show any of the synapomorphies of the Platanistoidea (Muizon 
1987; 1991 ): they have a scapula with a large coracoid process and 
a well defined supraspinatus fossa (Fig. 2e). Furthermore, the pala- 
tine morphology, the subcircular fossa, and the modification of the 
posterior process of the periotic that relate the Squalodontidae to 
the platanistid-squalodelphid clade are absent in the Eurhino- 
delphoidea. In other respects, the Eurhinodelphoidea have been 
regarded (Muizon 1991 ) as the sister group of the Delphinoidea on 
the basis of three synapomorphies: ( 1 ) loss of articulation of the 
tympanic with the squamosal. (2) palatine hollowed out by the 
pterygoid sinus with no lateral lamina of the palatine, and (3) 
expansion of the pterygoid sinus outside of the pterygoid fossa and 
into the orbit and temporal fossa. 

This interpretation, however, is contradicted by the morphology 
of the involucrum of the tympanic, which is very similar in the 
Eurhinodelphoidea and Ziphioidea. In both groups the dorsal face 
of the involucrum shows a well-developed indentation that is never 
present to this extent in other odontocetes (Fig. 6). This feature has 
not been found in the Physeteroidea, indicating that the Eurhino- 
delphoidea could represent the sister group of the Ziphiidae, not of 
the Physeterida. It is also possible that the modified bulla of the 
Physeteroidea has altered the expression of that feature. In fact, the 
Eurhinodelphoidea do not show any of the five synapomorphies of 
the Physeterida (Muizon 1991), and, therefore, if one admits the 
monophyly of the Physeterida. the morphology of the involucrum 
in the two groups must be regarded as convergent. The involucrum 
is olive-shaped to conical in the Platanistoidea and Squalodontidae 
and has neither the indentation observed in the Eurhinodelphoidea 
and Ziphioidea nor the sigmoid shape of the Delphinida (Fig. 6). 

DISCUSSION AND CONCLUSIONS 

Three alternate hypotheses that relate the Squalodontidae to 
nonplatanistoid groups have been tested; none has proven satisfac- 
tory since their acceptance would increase the number of 
convergences. Furthermore, the characters used to support a rela- 
tionship of squalodonts to nonplatanistoid taxa have been demon- 
strated to be symplesiomorphic, homoplastic, or not homologous. 
Consequently, the hypothesis that squalodontids are platanistoids 
(Muizon 1987) is more parsimonious. 

Contrary to what is generally accepted. I have excluded 
Prosqualodon from the family Squalodontidae mainly because of 








Figure 6. Dorsomedial view of the involucrum of the tympanic of some 
odontocetes. a, Squalodon (from MHNL Dr 15); b. Platanista (from MNHN 
1870-79); c. Eurhinodelphis (from MNHN AMN 69); d, Lipotes (from 
AMNH 5333); e, Ziphius (from MNHN 1962-152); f, Pontoporia (from 
MNHN 1934-375); g. Delphinus (from private collection). All figures are 
schematic and not drawn to scale. 



the lack of a partial covering of the palatine by the maxilla and lack 
of a subcircular fossa in the roof of the middle ear cavity (Muizon 
1991). This genus, which has the typical platanistoid scapula 
(Muizon 1987). must, however, be included in the superfamily and 
represents the sister group of all other Platanistoidea. Cozzuol and 
Humbert-Lan (1989) have related the genus Prosqualodon to the 
Delphinida mainly on the basis of what they have observed to be a 
lateral lamina on the palatine of the holotype of Prosqualodon 
australis and some unpublished specimens. Such a duplication is 
also mentioned by Flynn (1948). However. I consider that 
Prosqualodon lacks a true lateral lamina. Instead, both palatines 
present a small lateral crest, a consequence of the posteroventral 
excavation of the palatines (due to the pterygoid bone's being 
hollowed by the pterygoid sinus). The pterygoids of Prosqualodon 
are not opened anteriorly and the pterygoid sinus does not contact 
the palatine as it does in the Delphinida. A similar lateral crest of the 
palatine is observed in the skull of the type of Squaloziphius 
emlongi (see Muizon 1991). in the Squalodontidae, and in some 
specimens of the Eurhinodelphidae (Schizodelphis barnesi. USNM 
187130). Consequently, the condition in Prosqualodon may be 



Are the Squalodonts Related to the Platanistoids? 



14? 



convergent with that of the Delphinidu. but I do not regard them as 
homologous and do not accept the classification of Prosqualodon 
among the Delphinida as proposed by Cozzuol and Humbert-Lan 
(1989). Furthermore. Prosqualodon lacks a ventral rim on the 
ventrolateral side of the anterior process of its periotic, the involu- 
crum of its tympanic is not sigmoid as in all the Delphinida. and on 
its vertex the frontals are wider than the nasals (for character 
analysis see Muizon 1988a). 

Nevertheless, this interpretation of squalodont relationships is 
weakened by some of the Squalodontidae not being sufficently well 
known. In fact, the skulls of the type specimens of Eosqualodon 
langewieschei and Kelloggia barbara need further preparation to 
disclose their auditory and pterygoid regions. Furthermore, as stated 
above, the auditory region of Prosqualodon is poorly known, and 
the only perioti.es referred to that genus are those attached to the 
skull of the type off! davidis, which is now lost (Fordyce 1982:49), 
and that of P. australis. figured by True (1909). 

The platanistoid relationships of the squalodonts I advocate 
here are not new, having been proposed previously (Abel 1914; 
Slijper 1936; Rothausen 1968). The last author suggested that the 
Platanistoidea {sensu Simpson 1945) arose in the late Oligocene 
between the Patriocetidae and the Squalodontidae, in which he 
included Prosqualodon. 

In this study, as in previous works (Muizon 1987; 1991 ), I have 
not considered several archaic taxa including Patriocelus, 
Agorophius, Archaeodelphis. Xenowphus, Agriocelus. and Micro- 
zeuglodon. As shown by Fordyce (1981). most of these taxa are 
odomocetes, and their primitive morphology indicates only a "pre- 
squalodontid" grade of evolution. Patriocetus is probably an ar- 
chaic squalodontid or platanistoid. Agriocelus, known by only a 
single poorly preserved skull, could be a squalodontid or plata- 
nistoid but is too little known to be taken into account here. The six 
genera mentioned above must be classified within the odontocetes 
since they possess what I regard as the key synapomorphy of the 
group, the posterodorsal extremity of the maxillae partially or 
totally overlapping the supraorbital process of the frontals and 
extending posteriorly behind the preorbital process. This condition 
does not exist in other cetaceans, i.e., archaeocetes and mysticetes, 
including the Aetiocetidae. 

The taxa mentioned above have been included by Rothausen 
(1968) in the superfamily Squalodontoidea. However, they repre- 
sent several primitive stages in the early evolution of the 
odontocetes, and it is likely that they do not belong to the same 
monophyletic taxon. Agorophius, Archaeodelphis, and Micro- 
zeugolodon differ from the Squalodontidae in having the parietals 
still visible on the vertex dorsally (this observation cannot be made 
on Xenowphus, in which this part of the skull is not preserved) 
(Fig. 5). Consequently, they show a less advanced telescoping of 
the skull than do the other odontocetes, making them resemble the 
archaeocetes more than the Squalodontidae. For this reason, they 
probably represent early branches in odontocete phylogeny, and I 
do not think that they should be classified in the Squalodontidae. 
The vertex of other more typical odontocetes (more derived in this 
respect) is characterized by a more posterior extension of the maxil- 
lae and frontals, which almost exclude the parietals from the dorsal 
surface of the skull (they do not articulate on the dorsal surface of 
the skull and often do not articulate at all). The posteriorly ex- 
panded maxillae and frontals also cover the temporal fossa, form a 
continuous crest (temporal crest) from the the postorbital process to 
the occipital crest, and demarcate a small square or trapezoidal 
frontal window on the vertex. 

Furthermore, some of these odontocetes of "pre-squalodontid" 
grade are represented by incomplete skulls, some of them totally 
lacking the rostrum or braincase, and the pterygoid and auditory 
regions are almost always absent or very damaged. Their poor 



preservation, as well as their scarcity, make phylogenetic analysis 
difficult. As shown by Fordyce (1981). the phylogenetic relation- 
ships of these primitive odontocetes are obscure, and more com- 
plete specimens are needed to clarify their evolutionary history and 
systematics. 

ACKNOWLEDGMENTS 

Special thanks are due to A. Berta. T. A. Demere. and E. 
Fordyce for fruitful discussions and numerous comments that 
helped to improve the manuscript and to S. L. Messenger and J. E. 
Heyning, who reviewed the manuscript and provided useful com- 
ments. I also thank C. E. Ray, who read the manuscript, and A. 
Dagand. who made the illustrations. 

LITERATURE CITED 

Abel, O. 1914. Die Vorfahren der Bartenwale. Denkschriften der 
Mathematisch-Naturwissenschaften 90: 155-244. 

Barnes. L. G. 1985. Fossil pontoporiid dolphins (Mammalia. Cetacea) 
from the Pacific coast of North America. Natural History Museum 
of Los Angeles County Contributions in Science 363. 

. 1990. The fossil record and evolutionary relationships of the 

genus Tursiops. Pp. 3-26 in S. Leatherwood and R. R. Reeves 
(eds.). The Bottlenose Dolphin. Academic Press, San Diego, Cali- 
fornia. 

. D. P. Domning. and C. E. Ray. 1985. Status of studies on fossil 



marine mammals. Marine Mammal Science 1:15-53. 
Benham, W. B. 1935. Fossil Cetacea of New Zealand. Part III. The skull 

and other parts of the skeleton of Prosqualodon hamiltoni n. sp. 

Transactions of the Royal Society of New Zealand 67:8-14. 
Cabrera. A. 1926. Cetaceos fdsiles del Museo de La Plata. Revista del 

Museo de La Plata 29:363-41 1 . 
Capellini, G. 1903. Avanzi di Squalodonte nella arenana di grumi dei 

Frati presso Schio. Memorie della Reale Academia di Scienze del 

Istituto di Bologna, Classe Scienze Fisiche 10:437-445. 
Cozzuol, M. A., and Humbert Lan. G. 1989. On the systematic position 

of the genus Prosqualodon Lydekker, 1893, and some comments 

on the odontocete family Squalodontidae. Abstracts of Posters and 

Papers. Fifth International Theriological Congress. Rome. 22-29 

August 1989. 1:483-484. 
Dal Piaz, G. 1904. N eosqualodon nuovo genere della famigli degli 

Squalodontidi. Memoires de la Societe Paleontologique Suisse 

31:1-19. 
. 1916. Gli Odontoceti del Miocene Bellunese. II. Squalodon. 

Memorie dell' Istituto Geologico della Reale Universita di Padova 

4:1-94. 
. 1977. Gli Odontoceti del Miocene Bellunese. V-X. 

Cyrtodelphis. Acrodelphis, Protodelphinus, Ziphiodelphis, 

Scaldicetus. Memorie dell'Istituto Geologico della R. Universita di 

Padova, Publ. dal Prof. G. Dal Piaz, Allegato al vol. 4 (1916):1- 

128. 
Donoghue. M., J. Doyle. J. Gauthier. A. Kluge, and T. Rowe. 1989. The 

importance of fossils in phylogeny reconstruction. Annual Review 

of Ecology and Systematics 20:431—460. 
Flynn. T T. 1948. Description of Prosqualodon davidii Flynn, a fossil 

cetacean from Tasmania, with a note on the microscopic tooth 

structure by J. Thorton Carter. Transactions of the Zoological Soci- 
ety of London 26:153-197. 
Fordyce, R. E. 1981. Systematics of the odontocete whale Agorophius 

pygmaeus and the family Agorophiidae (Mammalia. Cetacea). 

Journal of Paleontology 55:1028-1045. 
. 1982. A review of Australian fossil Cetacea. Memoirs of the 

National Museum, Victoria 43:43-58. 
. 1985. The history of whales in the southern hemisphere. Pp. 

79-104 in J. K. Link and M. M. Bryden (eds.). Studies of Sea 

Mammals in South Latitudes. South Australian Museum, Sydney, 

Australia. 
. 1994. Waipatia maerewhenua, new genus and new species 

(Waipatiidae, new family), an archaic late Oligocene dolphin from 



146 



Christian de Muizon 



New Zealand. In A. BertaandT. A. Demere (eds.). Contributions in 
marine mammal paleontology honoring Frank C. Whitmore, Jr. 
Proceedings of the San Diego Society of Natural History 29:147- 
178. 

Fraser, F. C, and P. E. Purves. 1960. Hearing in cetaceans. Evolution of 
the accessory air sacs and the structure and function of outer and 
middle ear in recent cetaceans. Bulletin of the British Museum of 
Natural History, Zoology 7:1-140. 

Gauthier, J., A. Kluge. and T. Rowe. 1988. Amniote phylogeny and the 
importance of fossils. Cladistics 4: 105-209. 

Gemmellaro, M. 1920. II Neosqualodon assenzae Forsyth Major sp. del 
Museo della Universita di Palermo. Giomale delle Scienze Naturale 
ed Economiche di Palermo 32:121-153. 

Hall.T. S. 1911. On the systematic position of the species of Squalodon 
and Zeuglodon described from Australia and New Zealand. Pro- 
ceedings of the Royal Society of Victoria 23:257-265. 

Hershkovitz, P. 1966. Catalogue of the living whales. United States 
National Museum Bulletin 246:1-259. 

Heyning, J. E. 1989. Comparative facial anatomy of beaked whales 
(Ziphiidae) and a systematic revision among the families of extant 
Odontoceti. Natural History Museum of Los Angeles County Con- 
tributions in Science 405. 

, and J. G. Mead. 1990. Evolution of the nasal anatomy of 

cetaceans. Pp. 67-79 in J. A. Thomas and R. A. Kastelein (eds.). 
Sensory Abilities of Cetaceans. Laboratory and Field Evidence. 
Plenum, New York, New York. 

Jourdan, C. 1861. Description de restes de deux grands Mammiferes 
constituant deux genres, l'un le genre Rhizoprion de l'ordre des 
Cetaces et du groupe des Delphinoides: I'autre le genre Dinocyon 
de l'ordre des Camassiers et de la famille des Canides. Annales des 
Sciences Naturelles, Serie 4, Zoologie 16:369-372. 

Kasuya, T. 1973. Systematic consideration of the recent toothed whales 
based on the morphology of tympanoperiotic bone. Scientific Re- 
ports of the Whales Research Institute 25:1-103. 

Kellogg, R. 1923. Description of two squalodonts recently discovered in 
the Calvert Cliffs of Maryland; and notes on the shark-toothed 
cetaceans. Proceedings of the United States National Museum 
62:1-69. 

. 1924. A fossil porpoise from the Calvert Formation of Mary- 
land. Proceedings of the United States National Museum 63:1-39. 
1928. The history of whales, their adaptation to life in the water. 



Quarterly Review of Biology 3:29-76. 
. 1936. A review of the Archaeoceti. Carnegie Institution of 



Washington Publication 482. 

Longhi, P. 1898. Sopra i resti di un cranio di Campsodelphis fossile 
scoperto nella molassa miocenica del bellunese. Atti della Societa 
Veneto-trentina di Scienze Naturali. Sena II, 3(2): 1-59. 

Lydekker, R. 1 894. Contribution to the knowledge of the fossil verte- 
brates of Argentina. Part II. Cetaceans skulls from Patagonia. 
Analesdel Museo de La Plata 1983:1-14. 

Mchedlidze, G. A. 1984. General Features of the Paleobiological Evolu- 
tion of Cetacea. Amerind, New Delhi, India (translation from Rus- 
sian). 

Moore, J. C. 1968. Relationships among the living genera of beaked 
whales with classification, diagnoses and keys. Fieldiana. Zoology 
53:209-289. 

Muizon, C. de. 1984. Les vertebres fossiles de la formation Pisco 
(Perou). Deuxieme partie: Les odontocetes (Cetacea, Mammalia) 
du Pliocene inferieur de Sud-Sacaco. Travaux de ITnstitut Fran^ais 
d'EtudesAndines 27:1-188 



— . 1985. Nouvelles donnees sur le diphyletisme des dauphins de 

riviere (Odontoceti, Cetacea, Mammalia). Comptes Rendus Hebdo- 

madaires des Seances de I'Academie des Sciences, Serie II, 

301:359-361. 

— . 1987. The affinities of Notocetus vanbenedeni, an early Mio- 



cene platanistoid (Cetacea, Mammalia) from Patagonia, southern 
Argentina. American Museum Novitates 2904. 

. 1988a. Les relations phylogenetiques des Delphinida. Annales 

de Paleontologie 74:115-183. 

— . 1988b. Le polyphyletisme des Acrodelphidae, odontocetes 
longirostres du Miocene europeen. Bulletin du Museum National 
d'Histoire Naturelle, Section C, 4eme serie, 10:31-88. 

1991. A new Ziphiidae (Cetacea) from the early Miocene of 



Washington State (USA) and phylogenetic analysis of the major 
groups of odontocetes. Bulletin du Museum National d'Histoire 
Naturelle, Section C, 4eme serie. 12:279-326. 

Novacek, M. J. 1992. Fossils as critical data for phylogeny. Pp. 46-88 in 
M. J. Noavacek and Q. D. Wheeler (eds.). Extinction and Phylog- 
eny. Columbia University Press. New York, New York. 

Paquier, V. 1894. Etude sur quelques cetaces du Miocene. Travaux du 
Laboratoire de Geologie de ['University de Grenoble 3:373-397. 

Pilleri, G. 1976. Acomparative study of the skin and general myology of 
Platanista indii and Delphinus delphis in relation to hydrodynam- 
ics and behaviour. Investigations on Cetacea 6:89-127. 
— . 1986. Beobachtungen an den Fossilen Cetaceen des Kaukasus. 
Hirnananatomisches Institut, Ostermundingen, Switzerland. 
— . 1985. The Miocene cetacea of the Belluno sandstones (eastern 



southern Alps). Memorie di Scienze Geologiche 36:1-87. 

1989. Comments on Christian de Muizon's paper: 



Le 
polyphyletisme des Acrodelphidae, odontocetes longirostres du 
Miocene europeen (The polyphyletism of the Acrodelphidae, 
longirostral Odontoceti of the European Miocene). Investigations 
on Cetacea 22:378-392. 

— , M. Ghir, and C. Krause. 1989. Odontoceti (Mammalia, 
Cetacea) from the lower Miocene of Rosignano. Piedmont, north 
Italy. Investigations on Cetacea 22:189-291. 

Prothero, D. R., E. M. Manning, and M. Fischer. 1988. The phylogeny 
of the ungulates. Pp. 201-234 in M.J. Benton (ed.).The Phylogeny 
and Classification of the Tetrapods. vol. 2. Systematics Association 
Special Volume 35B. Clarendon Press, Oxford, England. 

Rothausen. K. 1968. Die systematische Stellungs der europaischen 
Squalodontidae (Odontoceti, Mammalia). Palaontologische 
Zeitschrift 42:83-104. 

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. 

Slijper, E. J. 1936. Die Cetaceen, vergleichend anatomisch und 
Systematisch. Capita Zoologica 7: 1-590. 

Thenius, E. 1969. Stammesgeschichte der Saugetiere (einschliesslich 
der Hominiden). Handbuch der Zoologie 2:369-722. 

True, F. W. 1909. A new genus of fossil cetacean from Santa Cruz 
Territory, Patagonia; and description of a mandible and vertebrae of 
Prosqualodon. Smithsonian Miscellaneous Collections 52:44 1^456. 

Winge, H. 1921. A review of the interrelationships of Cetacea. 
Smithsonian Miscellaneous Collections 72:1-98. 

Whitmore F. C, A. E. Sanders. 1977. Review of the Oligocene Cetacea. 
Systematic Zoology 25:304-320. 

Zhou. K. 1982. Classification of the superfamily Platanistoidea with 
notes on evidences of the monophyly of the Cetacea. Scientific 
Reports of the Whale Research Institute 34:93-108. 



Waipatia maerewhenua^ New Genus and New Species (Waipatiidae, New 

Family), an Archaic Late Oligocene Dolphin (Cetacea: Odontoceti: 

Platanistoidea) from New Zealand 

R. Ewan Fordyce 

Department of Geology, University of Otago, P.O. Box 56. Dunedin, New Zealand 

ABSTRACT. — Waipatia maerewhenua, from the Otekaike Limestone (late Oligocene). Waitaki Valley. New Zealand, is a new genus and 
species in a new family Waipatiidae (Odontoceti: Platanistoidea) near the base of the radiation of platanistoids. Its features include skull about 600 
mm long; rostrum long and narrow; incisors long, procumbent, and gracile; cheek teeth heterodont and polydont: maxillae telescoped back over 
frontals toward supraoccipital; parietal narrowly exposed on vertex; pterygoid sinus fossa restricted to basicranium; and palatine broad and not 
invaded by pterygoid sinus fossa. Features of the tympano-periotic, periotic fossa, and foramen spinosum indicate platanistoid relationships. 
Waipatia maerewhenua is more closely related to the Squalodelphidae and Platanistidae than to the Squalodontidae. Of the similar small dolphins 
previously identified as Squalodontidae. Mierocetus ambiguus ( late Oligocene. Germany ) and Sachalinocetus cholmicus (early or middle Miocene. 
Sakhalin) are possible waipatiids. Mierocetus hectori (earliest Miocene. New Zealand! is a probable squalodelphid. Prosqualodon marplesi (early 
Miocene. New Zealand) is transferred to Notocetus (Squalodelphidae) as Notocetus marplesi (new combination). Sulakocetus dagestanicus (late 
Oligocene, Caucasus) is probably a waipatiid close to W. maerewhenua. These taxa reveal an early radiation of the Platanistoidea by the late 
Oligocene. 



INTRODUCTION 

This article describes a new family, new genus, and new species 
of late Oligocene marine platanistoid dolphin from New Zealand. 
Heterodont dolphins from Oligocene and Miocene rocks world- 
wide have played a key role in interpretations of cetacean evolution 
because they are transitional in grade between archaic Cetacea 
(Archaeoceti) and extant odontocetes. Waipatia maerewhenua 
meets traditional concepts of the Squalodontidae, a family often 
used for heterodont odontocetes. but is more closely related to the 
Squalodelphidae and Platanistidae than to the Squalodontidae. It is 
an early member of the platanistoid radiation that led to diverse 
Miocene taxa and ultimately to the two extant species of "river 
dolphins" of the genus Platanista; the latter represent the last of the 
Platanistidae and, probably, the superfamily Platanistoidea. 
Waipatia maerewhenua thus has implications for odontocete his- 
tory and for defining and delimiting the Squalodontidae. Squalodel- 
phidae, and Platanistoidea. 

The article has three main sections: ( 1 ) a description reviewing 
morphology and commenting on other taxa as needed to help 
interpret homology, (2) a comparison covering broader aspects of 
morphology, homology, and function, and (3) cladistic relation- 
ships. A new combination, Notocetus marplesi (Dickson. 1964) 
(Platanistoidea: Squalodelphidae). is used throughout for the so- 
called Prosqualodon marplesi of New Zealand. 



MATERIAL AND METHODS 

Descriptions are based on the right or left side, whichever is 
more informative, with differences between right and left men- 
tioned only if asymmetry is evident. Unreferenced statements about 
morphology are based on personal observations. The specimen was 
prepared with pneumatic chisels and scrapers. Fine details were 
prepared under a microscope with an ultrasonic dental scaler and an 
air-abrasive unit; some sutures could not be traced fully because the 
cancellous bone is friable and not permineralized. Photographs 
were taken with a 35-mm Asahi Pentax camera with a 50-mm 
macro lens. Illustrations derived from photographs are not cor- 
rected for parallax. 

Acronyms used here are NMNZ Ma. marine mammal catalog in 
the National Museum of New Zealand, Wellington, New Zealand; 
OM C and OM A, catalogs in Otago Museum, Dunedin, New 
Zealand; OU, fossil catalog in Geology Museum, University of 
Otago, Dunedin, New Zealand; USNM, Department of Paleo- 



biology, National Museum of Natural History, Smithsonian Institu- 
tion. Washington, D.C. 

SYSTEMATICS 

Order Cetacea Brisson, 1762 

Suborder Odontoceti Flower. 1867 

Superfamily Platanistoidea Simpson. 1945 

Family Waipatiidae, new 

Type genus. — Waipatia, new genus. 

Included genera. — Waipatia, new genus, only. 

Diagnosis of family. — As for the only included species, 
Waipatia maerewhenua. in the only included genus, Waipatia, be- 
low. 

Comment. — The family probably includes Sulakocetus 
dagestanicus Mchedlidze, 1976 (late Oligocene, Caucasus), and 
may include species of Mierocetus and Sachalinocetus; these are 
discussed below. 

Genus Waipatia, new 

Type species. — Waipatia maerewhenua, new species. 

Included species. — Waipatia maerewhenua. new species, only. 

Diagnosis. — As for the only included species. Waipatia maere- 
whenua, below. 

Etymology. — From the Maori name Waipati. a place near the 
type locality. Probable derivation: wai, water; pati, shallow. Re- 
garded as indeclinable. Pronunciation: wai-pa-ti. with a pronounced 
as in English "far," and /' as in "he." 

Waipatia maerewhenua, new species 

Figs. 2-8. 9b, lOa-k, 11,12, 13a-g 

Material.— Holotype only, OU 22095: a skull with 23 teeth in 
place, both mandibles. 17 loose teeth, left tympanic bulla, right 
periotic. left periotic lacking anterior process, atlas, natural cast of 
anterior of axis, and anterior thoracic vertebra. Collected by R. 
Ewan Fordyce. A. Grebneff, and R. D. Connell. January 1991. 

Type locality. — North-facing cliff near Waipati Creek, 5 km 
west-southwest of Duntroon and 1.2 km north of "The Earth- 
quakes," North Otago (Fig. 1). Grid reference: NZMS [New 



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

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

Proc. San Diego Soc. Nat. Hist. 29:147-176. 1994 



148 



R. Ewan Fordyce 



top not seen 
30 J 



J=fer 



20:: 



10 



rrrj 



iii 



^ 



iii 



i , i , i 



iii 



Wi 



rff 



iii 



i5S 



i i i 



i i i 



rri 



I : Igl  



0^ 



" 3 9 a 3 
9- 9 :9 i 
 3':3: : S:3. 
9:'9g'3.9'3.'9 



massive limestone with 
sparse brachiopods 

poorly exposed 



cliff top 




base not seen 



"The Earthquakes" 



massive limestone with 
occasional macrofossils 



glauconitic massive limestone 
grades up 

bedded greensand with 
abundant macrofossils 

massive bioturbaled muddy 
greensand with reworked clasls 

°! calra reou s mu dstone 

•^—extensively bored 
massive calcareous mudstone 



NZ slag 


e- 








T 


w 


pj 


z 


>-z 


Z 

O 
f— 
00 


.j. < 


_J UJ 


>< 


<o 
w 5 


PJ 


... 9 ..- 




S 


Z 




J 


< 




PJ 


z 




i4 


Q 




< 


OS 


UJ 


X 


H 


UJ 


PJ 


Z 


U 


H 


3 


O 


O 


Q 


D 




/ Z 


J 






< 


O 


„D .... 




O 


P 


Sq 




< 


< 


<z 








o< 


9 




•ji, m 




0-< 




OtX 


- *pC 




^0 


3 £ 


>-UJ 
_JZ 
OS UJ 


J, CO J 


(*<Z 


— PjQi 


Wo< 


<<~J 


c^< 


Jzo 


ujO 


£<2 






"c 




£ 






20. 



NZ 



E J 



10., 



-,:i 



J- 9^-9 

4)9.' I . V S.- 
J 9 9 9 3' 

base not seen 



15+ m of massive 
limestone is exposed 
sporadically above scarp 



scarp top 



■< — dolphin horizon 
shelly massive limestone 

poorly exposed 

scarp top 

concretionary glauconitic 
limestone 

massive bioturbated 
greensand 



Waipati 



. stage 


~i 






T 




ID 






z 






p 


7 


? 


in 






PJ 






£ 












_) 




w 


pj 


z 


z 


^ 


< 


u. 




z 


n 




< 


O 


x 


O 


pj 


a; 


_J 


r_ 


f- 


O 


O 


z 


UJ 


___ _ 


.J 


[— 


DQ 


Q 


< 


?z 




J 


<< 






q uj 






W bj 






O 








1 


17CWE y 




250 




kilometres 


1 


4SOS ./ 


f{ 


C 


^unedin 







.fChristchurch 
Duntroon 



SOUTH ISLAND, NEW ZEALAND 



older terrace 
gravels 



liments 



kilometres 



Figure 1. Locality map, geological map, and stratigraphic sections. Dot marks type locality of Waipmia maerewhenua. The "Earthquakes" section based 
on Homibrook ( 1%6). Fordyce et al. (1985). and Fordyce and M.A. Ayress field data. The Waipati section from Fordyce's field data; detailed microfossil 
dates unavailable for Waipati. Correlations between "Earthquakes" and Waipati sections are lithostratigraphic. Upper parts of both sections, exposed 
discontinuously on sloping hillsides, are measured less reliably than lower parts. Geological map based on Gage (1957). 



Zealand Mapping Series] 260 metric sheet 140 (1987): 222912; 
near latitude 44° 51.5' S, longitude 170° 37.25' E. See Gage (1957: 
Geological Map No. 2). 

Horizon and age. — Massive limestone with sparse macrofossils 
(Maerewhenua Member), 8-9 m above the base of the Otekaike 
Limestone Formation (Fig. 1 ). Fossil record number I40/f284 (New 
Zealand fossil record file, Geological Society of New Zealand). 
Matrix lacks Globoquadrina dehiscens, a planktonic foraminiferal 



index species for the Waitakian Stage; this species appears nearby 
in the upper Otekaike Limestone (Hornibrook et al. 1989). Other 
foraminifera in the sample indicate a Duntroonian to Waitakian age. 
The upper Duntroonian Stage is likely; this is equivalent to late or 
latest Oligocene, about 24—26 Ma (Hornibrook et al. 1989). Nearby, 
at "The Earthquakes." the stratigraphic sequence is established 
better (Fig. 1; Fordyce et al. 1985; Gage 1957; Hornibrook 1966; 
Hornibrook et al. 1989), reinforcing an upper Duntroonian determi- 



Waipatia maerewhenua. New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



149 



nation. Here, the lower Otekaike Limestone represents the 
Duntroonian Stage (late to latest Oligocene), while Waitakian fau- 
nas (earliest Miocene) appear 13-14 m above the base of the 
limestone. 

Diagnosis. — Odontocete with slightly asymmetrical skull of 
medium size (condylobasal length approximately 600 mm), attenu- 
ated rostrum, heterodont polydont teeth, and basicranium of archaic 
grade. Placed in the Platanistoidea because the periotic has an 
incipient articular process, the anterior process is roughly cylindri- 
cal in cross section and deflected ventrally. and the tympanic bulla 
has an incipient anterior spine, anterolateral convexity, and ventral 
groove extending anteriorly as a series of long fissures. Allied with 
the Squalodelphidae and Platanistidae, rather than the Squalodonti- 
dae, because the long asymmetrical posterior apex of the premaxilla 
extends posterior to the nasal to wedge between the elevated edge 
of the maxilla and frontal on vertex, the cheek teeth are small, the 
incisors are relatively delicate and procumbent, the premaxillary 
sac fossa is relatively wide and expanded medially to form a signifi- 
cant prenarial constriction, the pterygoid sinus fossa is in the 
alisphenoid and/or basioccipital dorsolateral to the basioccipital 
crest and posteromedial to the foramen ovale, the lateral groove 
affects the external profile of the periotic. rendering it sigmoidal in 
dorsal view, the dorsal ridge on the anterior process and body of the 
periotic is associated with a depression near the groove for the 
tensor tympani. the profile of the anteroexternal sulcus of the 
periotic is recurved and concave dorsally, and the squamosal carries 
a smoothly excavated periotic fossa associated with an incipient 
subcircular fossa (enlarged foramen spinosum) dorsal to the 
periotic. More derived than described Squalodelphidae, Platanisti- 
dae, and Dalpiazinidae in that the mandibles have a shorter unfused 
symphysis, the sinus fossa in the alisphenoid and/or basioccipital is 
larger, and the anterior process of the periotic is relatively larger and 
more inflated transversely, with a blunter apex reflected more 
abruptly ventrally. 

Etymology. — From the Maori name Maerewhenua, name of a 
river near the type locality. Probable derivation: maere, perhaps 
from maru. shelter, or maero. the original inhabitants; whenua. 
country or land. Regarded as indeclinable. Pronunciation: mae-re- 
whe-nua, with a pronounced as in English "far." e as "ea" in 
English "leather." wh usually as "f" but sometimes as "wh" as in 
"when," u as double "o" in English "moon." 

General description. — The skull is nearly complete; it lacks the 
apex of the rostrum, the pterygoids, and all but the bases of the 
jugals. There is a little shear (structures on the right side lie anterior 
to those on the left) but no major diagenetic distortion; the brain case 
is slightly crushed. The asymmetry of the nasals, frontals, premaxil- 
lae. and base of the rostrum appears real. The skull was found upside 
down; the right mandible lay bent over the rostrum with its body 
perforated by maxillary teeth. The earbones and 17 partial or whole 
teeth were loose in the matrix around the skull. About 1.5 nr was 
excavated without revealing the rest of the skeleton. 

Cranium. — The cranium (that portion of skull posterior to the 
antorbital notches) is about as long as it is wide. In lateral view 
(Fig. 2e), the orbit is little elevated above the base of the rostrum. 
The external nares open from subvertical narial passages about 
level with the postorbital processes of the frontals. At the level of 
the nasals, the face is up to 30-35 mm deep, indicating well- 
developed maxillo-naso-labialis (facial) muscles (Fig. 2). Facial 
muscle origins, formed by the maxilla, are relatively long and 
narrow and not expanded or deepened posterolaterally; the poste- 
rior of the face is shallow. The maxilla and frontal only partly roof 
the relatively large temporal fossa (Figs. 2a, b). A prominent tempo- 
ral crest with a long straight postorbital border bounds all of the 
dorsal edge of the fossa; within the fossa the braincase is not 
obviously inflated. The intertemporal constriction is reduced, with 



a narrow band of parietals exposed dorsally. The supraoccipital lies 
well forward, not encroached upon by facial elements. 

Rostrum. — The rostrum is relatively long, w ide at its base at the 
antorbital notches, and attenuated anteriorly (Figs. 2a, 3a). Each 
antorbital notch, which transmitted the facial nerve, is open anteri- 
orly but is shallow dorsoventrally. A prominent antorbital (preor- 
bital) process extends forward to bound the notch laterally. Anterior 
to the right notch, the rostral margin of the maxilla flares out to form 
a marked flange missing on the left (Figs 2a, 4a. 5a). The right notch 
is deeper and more U-shaped than the left. As viewed laterally (Fig. 
2c). the premaxilla forms all of the dorsal profile; the ventral 
surface of the rostrum, formed by the maxilla, is roughly flat, and 
the rostrum thins only a little apically. In ventral view (Fig. 2c), the 
anterior half of the rostrum is grooved medially, while posteriorly it 
is gently convex. Palatal ridges are indistinct, and there are no 
rostral fossae for pterygoid sinuses. In dorsal view, the open 
mesorostral groove is wide posteriorly but narrow anteriorly. Other 
profiles of the rostrum are shown in Figs. 2 and 4 

Premaxilla. — Anteriorly, the premaxilla is narrowest in dorsal 
view at mid-rostrum, where it bounds and slightly roofs the 
mesorostral groove. Further forward, the premaxilla forms the api- 
cal 55+ mm of the rostrum. The dorsal rostral suture with the 
maxilla is prominent but not deep. The premaxilla forms an 
internarial constriction medially, where the premaxillary sac fossa 
is widest between the level of the nares and premaxillary foramina. 
Anteriorly, the fossa is nearly horizontal in transverse profile; it 
narrows and is elevated behind the prenarial constriction. Each 
premaxillary foramen is single; the right is longer than the left and 
lies more posteriorly, but both open anterior to the antorbital pro- 
cess. The anteromedial and, particularly, the posterolateral premax- 
illary sulci are prominent (Fig. 4a). but the posteromedial sulcus is 
shallow and indistinct. The nasal plug muscle probably originated 
on the narrow shelf of the premaxilla that overhangs the mesorostral 
groove anteromedial to the premaxillary foramen. Much of the 
outer margin of the premaxilla lateral to the premaxillary sulci 
carries a low thick rounded ridge. In dorsal view, the lateral edge of 
the premaxilla is gently convex around the region of the external 
nares. Lateral to each naris and within the premaxilla is a long 
median premaxillary cleft (new term; Figs. 4d, 5b). perhaps a 
vascular feature, which ascends posteriorly toward the junction of 
premaxilla, maxilla, nasal, and frontal at the vertex. The cleft lies 
just internal to the prominent medial facial crest formed by the 
maxilla and premaxilla and does not strictly mark the boundary 
between the posterolateral plate and posteromedial splint of the 
premaxilla. On the left, the premaxillary cleft grades forward into 
the posterolateral sulcus. 

The premaxilla is split or bifurcated posteriorly into a more 
dorsal, posteromedial thin ascending process (splint) and a more 
ventral posterolateral plate (sensu Fordyce 1981). The posterolat- 
eral plate is developed where a thin portion of the premaxilla 
external to the posterolateral sulcus overlaps the maxilla: this plate 
is conspicuous in lateral view (Figs. 4c, 6b) but is indistinct from 
above (Figs. 4d, 5b). The narrow posteromedial splint extends 
behind each nasal to wedge between the maxilla and frontal, thus 
separating the nasal from the maxilla. The left and right splints are 
asymmetrical (Figs. 4d, 5b). 

Maxilla. — Rostral profiles of the maxilla are shown in Figs. 2a, 
b, e and 4a-c. At least one maxillary foramen opens in the shallow 
depression between the maxillary flange and antorbital notch, and 
two or three foramina also open around each notch, but numbers are 
uncertain because the bone surface is damaged. Contacts with the 
frontal and lacrimal can be localized only to within a few millime- 
ters. The right antorbital process, formed by the lacrimal, is not 
covered by the maxilla. Ventrally. the maxilla forms most of the 
surface of the rostrum; it extends back between the subhorizontally 



[50 



R. Ewan Fordyce 




Figure 2. Waipatia maerewhenua, holotype. OU 22095. Skull, coated with sublimed ammonium chloride. All to same scale; scale = 200 mm. A, dorsal; 
B. right posterolateral; C, ventral; D, posterior; E. left lateral of skull and left mandible. 



15! 




Figure 3. Waipatia maerewhenua, holotype, OU 22095. Skull, coated with sublimed ammonium chloride. A. ventral view, posterior of basicranium, 
right side. Scale = 100 mm. B-E all to same scale; ruler divisions are I mm. B. ventromedial view, posterior of basicranium, right side. C, ventral view, 
posterior of basicranium with periotic in place, right side. D. ventral view, posterior of basicranium with periotic in place, left side. E, ventrolateral view, 
posterior of basicranium showing pterygoid sinus fossa posteromedial to foramen ovale, right side. 




Figure 4. Waipatia maerewhenua, holotype, OU 22095. A-D. skull, coated with sublimed ammonium chloride. A-C at same scale; scale = 100 mm. A. 
anterodorsal; B, skull with articulated mandibles, anterior and slightly dorsal view (mandibles are distorted so that symphysis does not articulate properly); 
C, left anterolateral; D. detail of vertex. Scale = 50 mm. E-J. holotype. left tympanic bulla, coated with sublimed ammonium chloride. Scale = 20 mm. E. 
dorsal; F, ventral; G, oblique dorsolateral of medial face. H, medial; 1, posterior; J. lateral. 



Waipaiia maerewhenua, New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 

temporal fossa 



153 




anteromedial sulcus 

premaxillary foramen 

posleromedial sulcus 

premaxillary sac fossa 

posterolateral sulcus 



supraorbital process of maxilla _ 



frontal \_temporal crest 




maxillary foramina 

supraoccipital 

naris 

nasal 

frontal 

posteromedial splint of premaxilla 



crest on maxilla 

posterolateral plate of premaxilla 
. parietal 

frontal 
. premaxillary cleft 



Figure 5. Reconstructions of dorsal view of skull of Waipaiia maerewhenua. A, general profile. Scale = 200 mm. B. detail of vertex. Scale = 50 mm. 



directed infraorbital foramen and the palatine but does not contrib- 
ute to the anterior wall of the orbit (Figs. 2c. 3a, 7a). 

The cranial part of the maxilla (e.g.. Figs. 2a, 4a, c) forms a long 
narrow supraorbital process that covers all of the frontal but for a 
thin lateral band over the orbit and curves in gently behind the 
nasals. Although the maxilla is slightly thickened just behind the 
antorbital process, there is no facial crest. Each supraorbital process 
has two centrally placed posteriorly directed maxillary foramina 
about level with the postorbital process of the frontal; these fo- 
ramina supplied blood vessels and nerves to the facial muscles. The 
maxilla carries anteriorly directed grooves, not obviously vascular, 
anterior to the maxillary foramina. Posteriorly, the rounded apex of 
the maxilla is separated from the supraoccipital by a thin band of 
the frontal and parietal. Though the maxilla is subhorizontal over 
the orbit, it becomes steeper posteromedially, with a markedly 



concave surface. The maxilla rises abruptly at the vertex to form a 
barely elevated maxillary crest that contacts the premaxilla (anteri- 
orly) and frontal (posteriorly) (Figs. 4c, 5b). just behind the bifurca- 
tion of the premaxilla. 

Palatine. — Broadly exposed palatines form the posterior por- 
tion of the palate between the choanae (posterior nares) at about the 
level of the most posterior cheek tooth (Figs. 2c, 3a, 7a). The 
palatines are continuous transversely across the convex palate, not 
narrowed or split by contact of the pterygoids with the maxillae. 
Contacts with the maxilla and frontal are localized to within a few 
millimeters; the sutures appear to be simple. The palatine sulci on 
the maxilla extend back toward the palatines, but the maxillary- 
palatine suture is preserved too poorly to tell whether a palatine 
foramen is present. Medially the palatines contact each other to 
form an indistinct flat palate bounded by faint palatal crests. Each 



154 



R. Ewan Fordyce 




nasal 



premaxilla 




squamosal 



parietal 

frontal 

posteromedial splint of premaxilla 

posterolateral plate of premaxilla 

premaxillary cleft 

maxillary foramen 



Figure 6. Reconstructions of lateral view of skull of Waipatia maerewhenua. A, general profile. Scale = 200 mm. B. detail of face. Scale = 50 mm. 



palatine is prominently excavated posteroventrally, just below the 
choana, with a shallow, crescentic depression at the pterygopalatine 
suture (Fig. 3a). The palatine lacks a lateral (outer) lamina. 

Pterygoid and pterygoid sinus. — Neither pterygoid is preserved. 
The loss of the pterygoids reveals an overlying large channel tor the 
maxillary branch of the trigeminal nerve (V,). which ran from near 
the foramen ovale internally out via the foramen rotundum to the 
orbit (Figs. 3a, 7a). The long lateral margin of the basioccipital. 
basisphenoid. and vomer in front of the basioccipital crest indicates 
that the inner lamina of the pterygoid was long: an anterior facet on 
the basioccipital crest indicates contact with the pterygoid. Since 
the basisphenoid and vomer are wide (Fig. 3a). the inner lamina of 
the pterygoid was probably narrow, not expanded medially. There is 
no evidence of a well-developed bony lateral lamina of the ptery- 
goid associated with the subtemporal crest; the subtemporal crest is 
the abrupt ventrointernal margin of the temporal fossa (here mainly 
formed by alisphenoid) that extends from near the choanae toward 
the squamosal, to separate the basicranium from the temporal fossa 
and orbit. This well-preserved crest lacks a thin bony ridge, which 
would be expected if a pterygoid lateral lamina had extended ven- 
tral to the crest, and lacks a definite suture for the pterygoid. 
Furthermore, the falciform process of the squamosal (Figs. 3a-e, 
8a, b) lacks evidence of contact with the pterygoid. 

A relatively large hemispherical pterygoid sinus fossa is present; 
despite its name, this fossa lies mainly in the alisphenoid. The 
missing pterygoid probably formed the anterior part of the fossa. 



The fossa apparently did not extend anteriorly or dorsally beyond 
the pterygoid, and there is no evidence of a fossa in the palatine 
(Figs. 2c, 3a). Farther dorsally, the palatine and/or frontal just 
below the orbital infundibulum lacks any channel for an orbital 
extension of the pterygoid sinus; the orbit lacks fossae. Behind the 
orbit, the prominent subtemporal crest (Fig. 7a) further indicates 
that the pterygoid sinus was confined to the skull base. Smooth 
bone surfaces posterior to the main pterygoid fossa indicate other 
lobes of the sinus. Probable fossae include the large depression in 
the alisphenoid anterior to the groove for the mandibular nerve (V 3 ) 
and a smooth depression between the foramen ovale and falciform 
process. A fossa, presumably for a large posteromedial lobe of the 
pterygoid sinus, lies posteromedial to the foramen ovale around the 
carotid foramen. Sutures here are fused; the fossa probably involves 
the alisphenoid. basisphenoid, and the dorsal part of the 
basioccipital crest (Figs. 3e. 8a). 

Nasal. — Nodular, anteroposterior^ short, wide nasals are 
crudely rectangular in dorsal view, with a convex anterior margin 
and a biconcave posterior margin (Figs. 2a, 4a-d). In vertical 
profile the anterior edge is rounded. Each nasal extends posterolat- 
erally between the frontal and premaxilla, markedly so on the left. 
The interdigitating internarial suture and, particularly, the 
nasofrontal suture are depressed but not deep or narrow. The nasals 
only slightly overhang the external nares. 

Mesethmoid. — The mesethmoid forms much of the borders of 
the nanal passages below the nasals (Figs. 4d. 5b). Anteromedially. 



Waipatia maerewhenua, New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



155 



post-tympanic process 

external auditory meatus 

subtemporal crest 

postorbital ridge of frontal 

frontal 

infraorbital foramen 




jugal 

lacrimal 

frontal foramina 

channel for maxillary nerve V2 

pterygoid sinus fossa 



basioccipital crest J exoccipitalJ 



D _ cheek tooth 1 



mandibular foramen 




canine 



Figure 7. Reconstructions of Waiparia maerewhenua. Scale = 200 mm. A, ventral view, skull; B. dorsal view, mandible. 



it forms a low rim on the narial passage at the posterior of the 
mesorostral groove, where it is probably fused with the vomer and/ 
or presphenoid. Further posterodorsally, the mesethmoid forms an 
ossified internarial septum about 10 mm wide. The dorsal surface 
here and further ventrally in the mesorostral groove is diffuse and 
probably carried cartilage that formed a septum between the soft 
tissues of the nares and also filled the mesorostral groove. The 
mesethmoid does not significantly support the nasals (Fig. 4b). 
Behind and laterally, a narrow groove (for the olfactory nerve?) 
ascends to a diagonal depression (for the olfactory foramen?). 

Vomer. — This lines the mesorostral groove, with a thin sliver 
exposed ventrally on the palate (Figs. 2c, 7a) between the maxillae. 
Further posteriorly (Fig. 3a), the sagittal part of the vomer separates 



the choanae, where the narial passages turn abruptly dorsally to- 
ward the external nares. The horizontal part of the vomer extends at 
least 65 mm posterior to the palatine, almost level with the foramen 
ovale, to cover the basisphenoid and broadly roof the basicranium. 
The margins of the horizontal part are subparallel posteriorly but 
(Tare out anteriorly as the choanae widen. 

Lacrimal. — The lacrimal is exposed to dorsal view (Figs. 2a, 
4a-c) at the antorbital process, where it forms the lateral margin of 
the antorbital notch. Sutures are ill defined because thin edges on 
the maxilla are preserved poorly. The lacrimal is thin both dorso- 
ventrally and anteroposteriorly. is transversely wide, is directed 
anterolaterally, and has only a small ventral exposure. Ventrally. the 
transversely wide, anteroposteriorly narrow broken base of the 




basioccipital crest 

posterior lacerate toramen 
condyle 
exoccipital 

hypoglossal foramen 
jugular notch 
squamosal 
paroccipital process 
foramen spinosum 
spiny process 
posterior meatal crest 
exoccipital 

external auditory meatus 



tympanosquamosal recess 
posterior part of falciform process 
anterior part of falciform process 



path for mandibular nerve V3 

pterygoid sinus fossa in alisphenoid-basisphenoid 

carotid foramen 

subtemporal crest 



postglenoid process 

tympanosquamosal recess 
suture for bulla 
post-tympanic process 
posterior portion, 
periotic fossa -i 



B 




anterior meatal crest 
posterior meatal crest 
supratubercular ridge 
anterior portion, periotic fossa 
foramen spinosum 
alisphenoid-squamosal suture 
foramen 1 
foramen ovale 



foramen 2 



L- pterygoid sinus fossa in alisphenoid-basisphenoid 



Figure 8. Interpretations of details of basicranium of Waipatia maerewhenua. Scale = 20 mm. A. ventrolateral, right side with ventral uppermost, 
showing position of fossa for pterygoid sinus in alisphenoid and basioccipital posteromedial to foramen ovale. B, ventral and slightly medial aspect, right 
side showing periotic fossa, presumed foramen spinosum, and other structures around periotic. 



Waipatia maerewhenua, New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



157 



parietal.^ 

.jsP _ superior petrosal sulcus 



B 



basioccipital 



superior process 



'■ } M. apex of superior process 

:$< 

squamosal :.-ja 
.. ..V-l 




posterior lacerate foramen-, 
parietal / f 



dorsal 

A 



external 



internal 



squamosal^i^^ 



pars cochlearis 




ventral 



homologue of superior process 

dorsal crest [= vestigial apex 
of superior process] 

internal auditory meatus 
pars cochlearis 



Basilosaurus cetoides 



Waipatia maerewhenua 



Figure 9. Schematic cross section of the basicranium at the level of the periotic in Basilosaurus cetoides (redrawn from Pompeckj 1922: pi. 2) and 
Waipatia maerewhenua, showing changes in relationship of periotic, squamosal, and parietal. Scale = 10 mm. A. Basilosaurus cetoides: B. Waipatia 
maerewhenua. 



jugal lies immediately behind the antorbital notch. There is no clear 
evidence of a lacrimal canal. 

Frontal. — The frontals have a long (ca. 45 mm) rather tabular 
exposure on the vertex behind the nasals (Figs. 4d, 5b). Since other 
bones on the skull are fused to the degree expected in a subadult or 
adult specimen, the distinct interfrontal suture is noteworthy. The 
frontals are markedly asymmetrical, with the left wider and shorter 
than the right. There are no supraorbital foramina (the large hole 
visible in Fig. 4d is a tool mark). The depressed suture with the 
parietals is partly fused; an interpretation appears in Fig. 5b. Farther 
laterally (Fig. 2b, 4c), the frontal is barely exposed dorsal ly along 
the postorbital margin of the face. 

Ventrally, the frontal forms most of the shallow elongate orbit. 
The preorbital ridge is low and indistinct, without an antorbital 
process, and barely separates the orbit from the infraorbital foramen 
(Fig. 3a). The lateral margin of the frontal is thin; farther medially. 
two or three confluent small frontal foramina open laterally in the 
roof of the orbit near the prominent postorbital ridge (Figs. 3a. 7a). 
Posteromedially, the postorbital ridge appears to contact the 
alisphenoid. Behind the ridge and below the posterior of the face, 
the frontal forms a large posteroventrally directed origin for the 
temporal muscle. 

Parietal. — A narrow slightly depressed band of parietal is ex- 
posed across the vertex between the frontal (Figs. 4d. 5b) and the 
nuchal crest of the supraoccipital. No obvious postparietal foramina 
or inteiparietal are present. A short robust temporal crest formed by 
the parietal separates the vertex from the temporal fossa at an 
intertemporal constriction. The parietal forms the slightly inflated 
anterolateral wall of the brain case (= medial wall of the temporal 
fossa), where it is markedly concave dorsally but convex farther 
ventrally toward the subtemporal crest. 

A nodular exposure of parietal is present in the basicranium 
dorsomedial to the periotic and immediately internal to the squamo- 
sal (Figs. 3a-c, 8a, b), where it faces ventromedially. The parietal 
thus lies immediately dorsal and internal to the dorsal crest of the 
periotic (Figs. 8a, b, 9a. b); such juxtaposition is concomitant with 
the lack of a discrete subarcuate (subfloccular) fossa on the periotic 
and a change in the structure or position of the superior petrosal 
sulcus. Here the parietal lacks evidence of large cavities (presumed 
vascular sinuses) of the sort present anterior to the periotic in the 
archaeocete Basilosaurus cetoides (USNM 6087; Kellogg 1936: 



figs. 5, 6). Medially, the parietal contacts an extensive horizontal 
sheet of apparently fused basisphenoid (anteriorly) and basioccipi- 
tal (posteriorly): the contact of these three elements separates the 
foramen ovale anteriorly from the posterior lacerate foramen and 
clearly isolates the periotic from the cranial cavity (Figs. 3b-e, 8b, 
9b). The term cranial hiatus, used by Fraser and Purves (1960) for 
structures here, seems redundant; furthermore, a perusal of litera- 
ture indicates that the term is used in an ambiguous and misleading 
way. Posteromedially, the parietal borders the posterior lacerate 
foramen, while farther laterally it has a long suture with the 
exoccipital that includes the small foramen 2 of Fig. 8b. The long 
suture with the alisphenoid anterolaterally is formed by a narrow 
fissure that runs obliquely from the foramen ovale to open at the 
foramen spinosum (Figs. 3a-c, 8a, b) dorsal to the periotic. 

Squamosal. — In dorsal view, the zygomatic process parallels 
the axis of the skull at the maximum width of the cranium. The 
process reaches forward to about level with the back of the nasals 
but does not reach the level of the postorbital processes of the 
frontals. The crest of the zygomatic process is rounded transversely 
but nearly flat anteroposteriorly (Figs. 4a, c). Posteriorly, above the 
external auditory meatus and post-tympanic process, the zygomatic 
process carries a large fossa (Figs. 2b, d. e) that angles forward 
almost to the level of the postglenoid process; this fossa forms an 
origin for some or all of the sternomastoideus. scalenus ventralis, 
longus capitis, and mastohumeralis muscles (cf. Howell 1927; 
Schulte and Smith 1918). Dorsally. at the apex of the fossa, the 
broad crest of the zygomatic process passes abruptly into a narrow 
lambdoid crest that curves inward and up onto the supraoccipital. 
Between the parietal and zygomatic process, the dorsal surface of 
the squamosal carries a broad shallow depression that forms the 
floor of the temporal fossa. The apex of the zygomatic process is 
rather short and rounded. 

The ventral surface of the squamosal is complex. In ventral 
view (Figs. 3a-e) the zygomatic process has a steep external face 
and gently rounded internal face, with a short narrow facet for the 
jugal at its apex. Posteriorly, a distinct ridge at the outer margin of 
the tympanosquamosal recess for the middle sinus marks the inner 
edge of the glenoid fossa. The ridge and recess extend ventro- 
laterally onto the robust postglenoid process, where the recess is 
widest; here the skull lacks a postglenoid foramen and the anterior 
transverse ridge associated with this foramen. Near the spiny 



158 



R. Ewan Fordyce 



process (sensu Muizon 1987), the surface of the recess carries a few 
shallow striae, presumably vascular, but there are no clear foramina 
(Figs. 3b, d). Anteriorly, the boundaries of the recess are indistinct, 
without any marked dorsal excavation. 

Anteriorly, the squamosal-alisphenoid suture is indistinct (Fig. 
8b). Contacts are not clear beyond foramen 1 of Fig. 8b. There is no 
pterygoid process of the squamosal at the subtemporal crest, but the 
falciform process (Figs. 3a-e, 8a) is well developed. This process 
has a long base and is bifurcated: an anterior portion extends out as 
a thin platelike subhorizontal spike ventral to the path of the man- 
dibular nerve (V,), while a posterior portion curls ventrally and 
inwards along the apex of the anterior process of the penotic. Both 
parts of the falciform process are thin distally and show no sign of 
contact with the pterygoid. 

The smooth squamosal forms a sporadically vascularized and 
spacious periotic fossa (new term) above the periotic and lateral to 
the parietal (Figs. 3a, b, e, 8b, 9b). The periotic fossa extends 
anteroposteriorly about 22 mm from the base of the falciform 
process almost to the exoccipital, and transversely about 20 mm 
from the inner edge of the tympanosquamosal recess to a crest on 
the inner margin of the squamosal. The latter crest apposes the 
ventrally adjacent dorsal crest of the periotic (Figs. 9b. 1 lb), close 
to the parietal. 

The periotic in W. maerewhenua approximates the squamosal at 
the posterior process (which, though finely porous dorsally, is not 
fused), lateral tuberosity, and part of the anterior process (Figs. 3c, 
d). For the most part, the squamosal and periotic are widely sepa- 
rated dorsally, leaving a spacious cavity between the periotic fossa 
and the periotic (Fig. 9b). There are two ventrolateral fissures that 
open into this cavity immediately anterior and immediately posterior 
to the lateral tuberosity, between the external edge of the periotic and 
the inner edge of the tympanosquamosal recess (Fig. 3c). 

The periotic fossa (Figs. 3a, b, 8b, 9b) is split into two portions 
by a roughly vertical supratubercular ridge (new term; Figs. 3b. 8b) 
11-12 mm anterior to the spiny process. A larger anterior portion 
lies dorsal to the anterior process of the periotic. while the smaller 
posterior portion lies dorsal to the body of the periotic. The anterior 
portion of the periotic fossa probably transmitted the middle menin- 
geal artery, which entered the periotic fossa via the fissure immedi- 
ately anterior to the lateral tuberosity (between the anterior process 
and the inner edge of the tympanosquamosal recess; Figs. 3c, d), 
passed above the periotic, and entered the large foramen spinosum. 
Given the voluminous cavity between the periotic and periotic 
fossa, the artery may have given rise to a rete that filled the anterior 
part of the periotic fossa. A near-obliterated fissure that marks the 
path of the foramen spinosum (Figs. 3a, b, 8a, b) runs forward 
across the squamosal and along or near the parietal-alisphenoid 
suture toward the foramen ovale. The anterior part of the periotic 
fossa and the large foramen spinosum are provisionally regarded as 
homologous with the subcircular fossa sensu Muizon (1987) of 
Notocetus vanbenedeni Moreno. 1892. 

In the smaller posterior portion of the periotic fossa, behind the 
supratubercular ridge, the wall of the squamosal is excavated dorsal 
to the spiny process. The excavation may represent an incipient 
cavity for the articular process of the periotic (as seen in Zarhachis 
flagellator, e.g., Muizon 1987). The posterior portion of the periotic 
fossa could have housed a rete, a lobe of the middle sinus extending 



dorsally from near the spiny process, or apart of the posterior sinus. 
It is not clear how the posterior part of the periotic fossa relates to 
the large posterior sinus fossa shown by Muizon ( 1987: fig. 3a) for 
Notocetus vanbenedeni. 

The external auditory meatus (Figs. 3a-e) is narrow, widens 
laterally and ventrally. and deepens externally; it has a steep ante- 
rior wall. Medially, the meatus is separated from the tympano- 
squamosal recess by a sharp anterior meatal crest (new term; Fig. 
8b) that extends from the postglenoid process to the spiny process; 
in the Archaeoceti, a topographically identical and presumably 
homologous crest lies behind the vestigial postglenoid foramen. 
The posterior wall of the meatus slopes gently back to a low 
posterior meatal crest (new term; Figs. 8a, b), behind which lies the 
post-tympanic process (sensu Pompeckj 1922: pi. 2; = post-meatal 
process of Muizon 1987) of the squamosal (Fig. 8). The anterior 
edge of the posterior process of the bulla overlaps the posterior 
meatal ridge to form part of the meatus. Also, an anterodorsal 
projection from the posterior process of the bulla overlaps the spiny 
process. Three fissures in the post-tympanic process receive ridges 
on the posterior process of the bulla and the posterior (mastoid) 
process of the periotic (Figs. 3. 8a. b). Waipatia maerewhenua is 
amastoid. with a posterior process of the periotic that lies 9-10 mm 
internal to the skull wall, covered ventrally by bulla and hidden 
from lateral view. Behind the articulated periotic is a narrow cleft, 
open ventrally, by which the facial nerve perhaps left the skull. 

Periotic. — The incomplete periotics together provide a clear 
idea of their structure. As this element seems one of the most 
diagnostic single bones among the Cetacea. I consider it here in 
detail. Morphological terms here largely follow Barnes (1978). 
Fordyce (1983). Kasuya (1973). Kellogg (e.g., 1923a). and 
Pompeckj (1922). 

Distinctive features of the periotic (Figs. lOa-k, 1 la-d) include, 
in summary, the large, robust, inflated anterior process with a 
subcircular cross section, an indistinct anterior keel, prominent 
anterointernal sulci (new term; see below), and a blunt apex. The 
lateral tuberosity and fossa incudis are prominent. The deep, later- 
ally compressed, pyriform internal auditory meatus has a rather 
small posterior tractus. a very narrow anterior portion, and a supple- 
mentary opening (for the greater petrosal nerve?) off the facial 
canal (= Fallopian aqueduct of Kellogg) anterior to the internal 
auditory meatus. The smooth subspherical pars cochlearis is rela- 
tively large and dorsoventrally deep. The subcircular dorsal aper- 
ture for the cochlear aqueduct is small and thick-lipped, while the 
aperture for the endolymphatic duct is primitively slitlike. A dorsal 
crest (new term; Fig. lib) forms the vertex of the dorsal surface. 
There is a narrow, smooth facet on the attenuated posterior process. 
Though each periotic is incomplete, it is likely that the axis (as 
viewed dorsally with the ventral face sitting on a flat plane) is 
sigmoidal, as is seen in Notocetus marplesi. Overall profiles are 
shown in Figs. lOa-k and Figs. 1 la-d; details follow. 

Of the two horizontal anterointernal sulci (Fig. lid) on the 
internal face of the anterior process, the dorsal sulcus ends posteri- 
orly at a vertical canal that opens (Figs. 1 la, d) farther dorsally on 
the anterior process. One of these sulci may carry the lesser petrosal 
nerve. In lateral view (Fig. 10c), the axis of the anterior process is 
reflected down, so that the anterior bullar facet (that part of the 
anterior process normally in contact with the processus tubarius of 



Figure 10. Periotics of Waipatia maerewhenua. squalodontids, and other presumed platanistoids. All coated with sublimed ammonium chloride. All life 
size; scale = 20 mm. A-K. Waipatia maerewhenua, holotype, OU 22095. A-F, right periotic. A. ventral; B. dorsal; C, lateral and slightly ventral; D, medial; 
E, lateral and slightly dorsal; F, posteromedial; G-K, left periotic; G, ventral; H, dorsal; I, medial; J, lateral; K, posteromedial. L-Q, Notocetus marplesi. 
holotype. left periotic, C.75.27. L, ventral; M. dorsal; N. lateral and slightly ventral; O. medial; P. lateral; Q, posteromedial. R-W, un-named squalodontid, 
right periotic, OU 22072. R, ventral; S, dorsal; T, lateral and slightly ventral; U, medial; V, lateral; W, posteromedial, showing reniform fenestra rotunda. X- 
C, unnamed squalodontid. right periotic, OU 21798. X. ventral; Y. dorsal; Z. lateral and slightly ventral; A', medial; B'. lateral; C posteromedial. 



Waipatia maerewhenua. New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



159 




160 



R. Ewan Fordyce 




anterodorsal angle ? 

dorsal crest 

depression internal to dorsal crest, 
at base ot anterior process 

foramen tor greater petrosal nerve? 

facial canal 

foramen singulare 

tractus spiralis foraminosus 

aperture for endolymphatic duct 

aperture for cochlear aqueduct 




anteroventral angle 

anterior bullar facet 

anteroexternal sulcus 

anterointernal sulci 

fovea epitubaria 

lateral tuberosity 

mallear fossa 

fossa incudis 

ventral foramen of facial canal 

fenestra ovalis 

facial sulcus 

stapedial muscle fossa 



D 



anteroventral angle 
anterior keel 
anterodorsal angle ? 
dorsal crest 

lateral tuberosity 
shallow lateral groove 
incipient dorsal tuberosity 
incipient articular rim 

posteroexternal foramen 









f -v s 




/ \ 


V S. jf**^ 





1,1. 



anterointernal sulci 
vertical canal (opens dorsally) 
lateral tuberosity 
hiatus epitympanicus 
facial sulcus 



fenestra rotunda 



posterior bullar facet 



posterior process 




incipient anterior spine 
anterolateral convexity 
broken edge of processus tubarius 



involucrum 



mallear ridge 

sigmoid process 

facet for posterior meatal crest 

facet for posterior process of periotic 
facet for post-tympanic process 



posterior process 



Figure 1 1 . Camera lucida sketches interpreting the key features of the periotic and tympanic bulla of Waipatia maerewhenua. Scale = 20 mm. A-D, right 
periotic, with lateral tuberosity and posterior process reconstructed from left periotic. A, dorsomedial; B, dorsal to dorsolateral: C, ventral; D, ventrolateral; 
E, left tympanic bulla, dorsal. 



Wuiputiu muerewhenua. New Genus and New Species, an Archaic Late Oligocene Dolphin (torn New Zealand 



161 



the bulla; new term; Fig. lie) is steeply inclined; the apex of the 
periotic appears blunt rather than attenuated. There are traces of an 
anterior keel on the anterior process. A prominent nodule on the 
inner face may represent a vestigial anterodorsal angle (sensu 
Fordyce 1983); the presumed anteroventral angle is blunt, not 
acute. The slightly damaged anterior bullar facet is a long shallow 
groove bounded laterally by smooth bone rather than by a thickened 
parabullary ridge (new term, possibly equals "distinct ventral rim" 
or "ventral swelling" of Muizon 1987: 7). More posteriorly, the 
fovea epitubaria (sensu Pompeckj 1922: 58. 66-67. pi. 2: = 
epitubarian fossa) is wide, shallow, and depressed medially. 
(Muizon 1987 used the term epitubarian fossa for what I term the 
anterior bullar facet.) A well-developed anteroexternal sulcus on 
the lateral face of the anterior process is visible in ventral view 
(Figs 10a, lie); its recurved, dorsally concave profile is marked in 
external view (Fig. 10c). The sulcus may mark the path of a loop of 
middle meningeal artery ventral to the periotic. 

The origin for the tensor tympani muscle is an indistinct cleft 
between the base of the anterior process and the perpendicular 
anterior face of the pars cochlearis. The pars cochlearis (Figs. 10a, 
b, d. f) is moderately inflated with abruptly rounded anterointernal 
and posterointernal angles; the posterointernal angle lacks a nodule. 
There is no obvious promontory sulcus. The small suboval fenestra 
rotunda is elongated vertically but is not reniform or fissured 
dorsomedially (Fig. lOf). Dorsally on the pars cochlearis. there is a 
faint raised rim on the long narrow internal auditory meatus, and an 
indistinct groove, perhaps a path for the inferior petrosal sinus, runs 
medial to the rim. Within the meatus, a rather narrow subcircular 
posterior traetus for the acoustic nerve is separated by a cleft from 
the deep narrow fissure into which open the small foramen singulare 
and slightly larger internal or dorsal aperture for the facial canal. A 
supplementary foramen opening 2-3 mm anterior to the internal 
auditory meatus (Figs. 10b. 11a) marks a canal (perhaps for the 
greater petrosal nerve) originating from the facial canal. 

Externally, the dorsal surface of the body of the periotic has a 
long dorsal crest, indistinct posteriorly, but better developed anteri- 
orly where it runs forward from the level of the facial canal on to the 
anterior process (Figs. 10b. 1 lb). Because of its topographic rela- 
tions. I identify the crest as a homolog of the apex of the superior 
process of Archaeoceti (Fig. 9a,b). There is no obvious superior 
petrosal sinus or subarcuate (subfloccular) fossa internal to the 
crest. Between the lateral tuberosity (sensu Barnes 1978; = ventral 
tuberosity of Muizon 1987) and the posterior process, a broad 
shallow lateral groove (Figs. 10b. e. 1 lb) ascends the external face, 
rising toward the level of the dorsal opening of the facial canal. This 
lateral groove complements a depression in the squamosal, so that a 
cavity lies between the periotic and squamosal. 

Ventrally on the body, the dorsoventrally compressed lateral 
tuberosity has a prominent subhorizontal crest (Figs. lOg, j) that 
closely follows the edge of the squamosal (Fig. 3d). Farther posteri- 
orly, the hiatus epitympanicus is indistinctly biconcave (Figs. 10a. 
c). The anterior of this depression receives the spiny process (sensu 
Muizon 1987) at the internal limit of the external auditory meatus, 
while the shallow articular groove at the anterior of the base of the 
posterior process (Fig. 10a) receives the posterior border of the 
spiny process. Of other ventral features on the body, the mallear 
fossa has an indistinct posteroexternal boundary and carries a 
prominent foramen at its inner margin. The fossa incudis is promi- 
nent at the anterior end of a meandering shallow groove. The 
subcircular fenestra ovalis lies far dorsal to the surface of the pars 
cochlearis. The ventral (epitympanic) opening for the facial canal 
opens anteroexternal to the level of the fenestra ovale, while the 
shallow facial sulcus for the facial nerve disappears before the end 
of the fossa for the stapedial muscle. Ridges separate this large, 
concave, rugose, and rather narrow fossa from both the fenestra 
ovalis and the groove for the facial nerve. 



On the posterior process (Figs. 10a, g), the facet for contact with 
the tympanic bulla is long, narrow, smooth, and attenuated, and 
does not extend dorsally onto the posteromedial face of the poste- 
rior process. Dorsally, two raised regions on the base of the poste- 
rior process may be homologous with more prominent structures on 
other platanistoids (Fig. lib). An indistinct bulge on the dorsal 
surface of the posterior process is probably homologous with the 
dorsal tuberosity (sensu Muizon 1987). Further ventrally is a small 
posteroexternal foramen (new term; Figs. lOe. lib), a persistent 
feature amongst archaic Cetacea, although of uncertain function. 
Below the posteroexternal foramen is a bulge on the base of the 
posterior process above the articular groove that is probably ho- 
mologous with the articular rim (sensu Muizon 1 987) or the peglike 
articular process of other Platanistoidea. 

Tympanic bulla. — The bulla (Figs. 4e-j, lie) is crushed. In 
dorsal or ventral view (Figs. 4e, f). it is roughly heart-shaped, 
with a convex outer margin, bilobed posterior face, and straight 
(posteriorly) to gently curved (anteriorly) involucrum. There is 
an incipient anterior spine and an anterolateral convexity (sensu 
Muizon 1987) on the outer lip, but there is no obvious 
anterolateral notch. Posteriorly, an interprominential notch sepa- 
rates the blunt inner prominence (medial lobe) and the narrower, 
deeper, slightly longer and more sharply rounded outer promi- 
nence (lateral lobe) (Figs. 4i, j). There is no obvious horizontal 
ridge between the prominences or across the inner prominence. 
In posterodorsal view, the interprominential notch is deep; below, 
it passes into a deep wide ventral groove that runs forward to 
about level with the sigmoid process (Fig. 4f). Farther forward, 
the groove is shallow and marked by fine to coarse fissures and 
small foramina; it extends to the apex of the bulla. The rough 
surface of the groove perhaps marks the attachment of the fibrous 
sheet known to cover the skull's base in some extant Cetacea 
(Fraser and Purves 1960). Although anteriorly the involucrum is 
depressed abruptly into the tympanic cavity (Fig. 4d), it is broad 
and not obviously invaded by an internally expanded tympanic 
cavity. Coarse striae of uncertain function cross the dorsal sur- 
face of the involucrum. radiating from about the position of the 
sigmoid process. The striae finish at a series of subhorizontal 
creases that traverse the inner face of the involucrum (Fig. 4h) 
and could be associated with tissues of the peribullary sinus 
known to occupy the space between the involucrum and the 
basioccipital crest in some extant Odontoceti (Fraser and Purves 
1960). 

As viewed laterally (Fig. 4j), the sigmoid process has an 
abruptly curved posteroventral profile: in anterior view the profile 
is rounded. The crushed lateral furrow is shallow. There is a robust 
oblique mallear ridge (new term) to which the malleus fuses inter- 
nally at the base of the sigmoid process. The conical process, 
obscured by the sigmoid process, has a flat posterior face and may 
be anteroposterior^ compressed. A wide gap. now distorted, sepa- 
rates the conical process from the posterior process. The distorted 
long posterior process articulates with the squamosal in two ways 
(Fig. 1 le); anterolateral^, the process carries a groove that overlaps 
the posterior meatal crest of the squamosal, while the thinner distal 
12+ mm of the process has a ridged subhorizontal suture (Figs. 4i,j) 
that articulates with the post-tympanic process of the squamosal 
(Figs. 8a, b). In lateral view of the skull (Fig. 6a), the posterior 
process of the bulla is just visible ventral to the post-tympanic 
process. The elliptical foramen is open, deep, and narrow. When the 
bulla is articulated, there is a large cavity, presumably for the 
peribullary sinus, between the bulla and the basioccipital crest. 

Supraoccipital. — The supraoccipital, which slopes forward at 
about 40° from horizontal, is roughly symmetrical, broad, and 
rather flat (Figs. 2a, b). Its blunt rounded anterior margin forms a 
nuchal crest elevated 3-4 mm above the parietal. A broad, low. 
and slightly asymmetrical anterior median ridge (Fig. 2a) bounds 



162 



R Ewan Fordyce 



faint anterolateral depressions. Convex lambdoid crests are 
present laterally. Posteriorly, each crest descends abruptly toward 
the squamosal. 

Basioccipital. — Behind the vomer, the basioccipital forms a 
shallow arcade that deepens posteriorly as the basioccipital crests 
diverge. Each crest is short (Fig. 3a) relative to the basicranial 
length. The crest is transversely thick and robust, with a thin ventral 
margin. Anterolaterally, just behind the carotid foramen, the dorsal 
base of the crest carries part of a large shallow hemispherical fossa 
for part of the pterygoid sinus (Figs. 3e, 8a). A small carotid 
foramen (Fig. 8a) indicates the anterior extent of the basioccipital, 
but there is no clear suture here with the alisphenoid or basi- 
sphenoid. 

Exoccipital. — The hind surface of the exoccipital is gently con- 
vex, other than near the pedicle for the condyle where the surface is 
deeply excavated. The condyloid fossa is excavated deeply into the 
braincase; the condyle has a rather small articular surface and a 
prominent pedicle. The exoccipital is closely applied to the squa- 
mosal along its dorsal and lateral edges, with rounded borders and 
rather curved lateral and ventral profiles. Dorsally. the suture with 
the supraoccipital is fused (Figs. 2b. d). 

Ventrally, the exoccipital forms the posterior portion of the so- 
called basioccipital crest, immediately internal to the shallow jugu- 
lar notch and the internally placed hypoglossal foramen (Figs. 3a-e, 
8a). The paroccipital process is robust, with a prominent but uni- 
dentified groove (Figs. 3a-c. right side) trending dorsomedially 
across the anterior face. Farther dorsally, the region between the 
exoccipital and squamosal-periotic is quite spacious, though there 
is no distinct fossa for a posterior sinus. Laterally, the exoccipital 
contacts the post-tympanic process of the squamosal (Fig. 3a). 

Alisphenoid, basisphenoid, orbitosphenoid. — The alisphenoid 
forms part of the subtemporal crest, but is otherwise not exposed 
within the temporal fossa. Anteriorly, the alisphenoid forms most of 
what remains of the pterygoid sinus fossa. Posteriorly, the 
alisphenoid is notched at a large foramen ovale. The complex 
posterolateral suture with the squamosal is shown in Fig. 8b. The 
alisphenoid carries a broad, shallow groove for the mandibular 
nerve (V,), which runs obliquely from the foramen ovale outward 
beyond the falciform process. Immediately anterior to this groove, 
the alisphenoid carries a large shallow hemispherical depression, 
probably for a lobe of the pterygoid sinus. The basisphenoid is 
probably fused with the alisphenoid: no sutures are apparent. Poste- 
riorly, the carotid foramen marks the likely limit of the basioccipital. 
The orbitosphenoid is not distinct. 

Teeth. — Waipatia maerewhenua is heterodont (Figs. 2e. 6a) and 
polydont. The right maxilla carries 16 alveoli ( 12 teeth are in place), 
suggesting 19 teeth in each upper tooth row. Alveoli in the right 
mandible indicate at least 16 and probably 19 teeth in the lower 
tooth row. Smooth procumbent single-rooted anterior teeth carry a 
crown formed by a single sharp and delicate denticle (Figs. 13 a, b). 
These subhorizontal apical teeth grade back into anterior cheek 
teeth with high crowns, small posterior accessory denticles, and 
fused double roots, in turn succeeded posteriorly by vertically 
positioned posterior cheek teeth with low, rather blunt and robust 
crowns that carry prominent posterior accessory denticles and 
strong ornament (Figs. 1 2a— f. 13c. d). The posterior diastemata are 
rather narrow, so that the upper and lower teeth probably did not 
interdigitate much. Apices of the posterior cheek teeth are worn 
from tooth-to-tooth contact. 

No anterior teeth are in place in the subhorizontal alveoli of the 
premaxilla and mandible. Features of the presumed incisors (Figs. 
13a, b, bottom) include a high smooth crown, subcircular in cross 
section with barely developed keels, and a somewhat inflated root 
that forms most of the height of the tooth. The largest tooth (maxi- 
mum height, apex of crown to apex of root, 76+ mm), presumably 
I 1 , has a gently sigmoid profile; its crown is subcircular in cross 



section. This large tooth was probably quite procumbent. Smaller 
and more recurved single-rooted teeth, presumably I 2 , l\ and C, 
have lower crowns that are recurved buccally and compressed 
laterally with indistinct keels. In lateral view, the axes of these teeth 
are recurved back, so that they were less procumbent than the apical 
teeth. 

Features of the cheek teeth are shown in Figs. 1 2a-f and 1 3a-d. 
The axes of the upper cheek teeth are strongly recurved lingually 
(Fig. 4b), while the lower cheek teeth are roughly straight. The 
posterior two or three lower cheek teeth are inclined slightly out- 
ward, while the other cheek teeth are inclined lingually. Those 
cheek teeth in place are emergent, with the crown well clear of the 
alveolus. Crowns of the middle to posterior cheek teeth (Figs. 12a, 
b) are conspicuously compressed, with a high triangular main (api- 
cal) denticle, two or three posterior denticles, but no anterior den- 
ticles. The apical denticle becomes smaller posteriorly in the tooth 
row as the accessory denticles become larger, and the third denticle 
is better developed on the lower teeth. Buccal ornament is indis- 
tinct, but lingual ornament is strong and. basally, associated with a 
cingulum on most cheek teeth (Figs. 13c, d). In the double-rooted 
teeth, the roots are fused for at least one third of their length; 
anteriorly, roots are divergent, while posteriorly they are roughly 
parallel. The last upper cheek tooth is small and single-rooted with 
a coarsely ornamented subconical crown (Figs. 13c. d, upper left). 

Mandible. — The reconstruction of the mandibles (Fig. 7b) is a 
visual "best fit." determined through aligning the mandibles with 
each other, with the glenoid cavities, and with the rostrum. The 
reconstructed profile in dorsal view is a Y shape, with a symphysis 
1 10- 120 mm long. 

Conspicuous features of each mandible (Figs. 6a, 7b, 12c-f) 
include the relatively long tooth row, 16+ alveoli, the gently curved 
dorsal profile in which the long, narrow, and deep body passes back 
into the low coronoid process, the ventrally and laterally inflated 
"pan bone" (= outer wall of large mandibular foramen, or "man- 
dibular fossa"), and the relatively short unfused mandibular sym- 
physis. Both mandibles are incomplete. With the left jaw articulated 
on the skull, the tooth in the third preserved alveolus occludes 
behind the position of the upper left canine; I identify it provision- 
ally as cheek tooth 1 . Left lower cheek teeth 5-14 are in place, and 
there may have been a. small cheek tooth 15; right lower cheek teeth 
5-10 are in place. 

The dorsal and ventral profiles of the body (Figs. 12c-f) are 
roughly parallel; the apical 80-90 mm of the ventral surface bends 
dorsally forward of the level of the fifth alveolus. The body deepens 
markedly behind cheek teeth 11-12. after which the pan bone is 
progressively inflated. 

The long shallow apical groove on the internal face of each 
mandible probably indicates an unfused symphysis in which the 
bones were not closely apposed in life. The left mandibular foramen 
opens 140-150 mm anterior to the condyle (Figs. 1 2d, e). A robust 
ridge marks the posterodorsal edge of the foramen just below the 
coronoid process, where the foramen is about 90 mm deep. There is 
an equally robust ridge ventrally above the angular process. Inter- 
nally, the condyle is slightly excavated, while its worn outer surface 
protrudes a little beyond the external profile. No distinct fossae are 
apparent for jaw muscle insertions; presumably insertions were as 
in extant Odontoceti (e.g., Howell 1927). Positions of the nine 
mental foramina are shown in Fig. 6a. 

Vertebrae. — The atlas (Figs. 13e-g) is slightly distorted 
through crushing and shearing, and surface bone is eroded in 
places. It is moderately thick, not compressed anteroposteriorly, 
and not fused to the axis. The eroded base of the neural spine is 
not massive or inflated. Anterior and posterior facets for contact 
with the skull and axis diverge gently in lateral view. The anterior 
facets are shallow and indistinctly separated ventrally; the poste- 
rior facets are barely raised above the adjacent bone. The 



Waipatia maerewhenua. New Genus and New Species, an Archaic Laie Oligocene Dolphin from New Zealand 



163 



rwvfffflYTi 



k£ - " 







•\<«e ** 







\ 



r *J^5?St* 



Figure 1 2. Mandible and teeth of Waipatia maerewhenua, holotype, OU 22095. All coated with sublimed ammonium chloride. A-B, detail of left cheek 
teeth, both at same scale; scale = 50 mm. A, buccal; B, lingual. C-F, mandibles; all at same scale; scale = 200 mm. C, lateral, left mandible; D, medial, left 
mandible; E, medial, nght mandible; F, lateral, right mandible. 



hypapophysis is small with an elongate base; this process extends 
6-7 mm below and less than 5 mm behind the body. Upper and 
lower transverse processes are separate, not basally confluent. 
The eroded large upper process juts out abruptly, while the lower 
process has an anteroposteriorly short base. Rather small (diam- 
eter 5.5-8 mm) laterally facing transverse foramina perforate the 
robust neural arch. Only a poor natural cast of the anterior face of 
the axis is preserved to reveal a large neural canal, delicate neural 
arches, and a blunt odontoid process. 



COMPARISONS: MORPHOLOGY, 
HOMOLOGY, AND FUNCTION 

This section briefly reviews broader aspects of the skull of W. 
maerewhenua, emphasizing homologies with other taxa and pos- 
sible functional complexes. 

Face. — The soft facial tissues in the Odontoceti include the 
maxillo-naso-labialis muscles, the soft nasal passages, and the nasal 
diverticula (Mead 1975; Heyning 1989). Because these structures 



164 



R Ewan Fordyce 





H { " f: 



*■*< 




Figure 13. A-D, teeth of Waipatia maerewhenua, holotype, OU 22095. All coated with sublimed ammonium chloride. All life size; ruler divisions are 
1 mm. A-B, upper and/or lower anterior teeth. A, lingual; B, buccal. C-D, lower cheek teeth and presumed last upper left cheek tooth. C, lingual; D. buccal. 
E-M, atlas vertebrae, all at same scale. Scale = 100 mm. E-G. atlas of Waipatia maerewhenua, holotype, OU 22095. E, anterior; F, posterior; G, dorsal. H- 
J, atlas of undescnbed squalodontid, OU 22072. H, anterior; I, posterior; J, dorsal. K-M. atlas of Notocetus marplesi, holotype, C.75.27. K, anterior; L. 
posterior; M, dorsal. 



Waipatia maerewhenua. New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



165 



dictate the topography of the facial bones, the structure of facial soft 
tissues can be inferred for fossils. Furthermore, the soft tissues of 
the face probably produce and transmit the high-frequency sounds 
used in echolocation (Mead 1975; Heyning 1989: 40^14). In terms 
of facial structure, W. maerewhenua is notably more derived than 
Archaeodelphis patrius Allen, 1921. in which the supraorbital pro- 
cess extends posteriorly only a little, the orbit is elevated with a 
prominent infraorbital process of the maxilla, the fossa for a facial 
muscle on the cranium is minimal, and the maxillary foramina lie 
roughly level with the antorbital notch. In W. maerewhenua, the 
fossa for the facial muscles is large, the roof of the orbit is depressed 
to lie about level with the posterior portion of the rostrum, so that 
rostral muscle origins and facial muscle origins are roughly on the 
same plane, and the maxilla does not contribute to the orbit. Well- 
developed premaxillary foramina and sulci are associated with a 
"spiracular plate" for the premaxillary sac fossa. Overall, W. 
maerewhenua has fundamentally the same facial structure as do 
many extant Odontoceti; it was probably capable of echolocating. 
In many other odontocetes (e.g., Delphinida, Ziphiidae), the face is 
broader, deeper, and displaced farther posteriorly relative to the 
orbits, so that the postorbital border of the temporal fossa is shorter, 
steeper, and more curved, the frontals and/or parietals are often lost 
from the vertex, and the supraoccipital is less obvious dorsally. 
Such changes probably reflect the continued expansion of the pos- 
terior parts of the maxillo-naso-labialis muscles associated with the 
soft diverticula of the external nares. 

Like most extant Odontoceti. W. maerewhenua shows facial 
asymmetry (involving maxilla, nasals, and frontals) that presumably 
reflects asymmetry of the overlying facial muscles. The asymmetri- 
cal flared margin of the right maxilla (Fig. 2a) is of uncertain 
function; it may be homologous with the maxillary flanges of 
Mesoplodon (Ziphiidae; True 1910b), though the left maxilla lacks 
such a flange. Presumably, such asymmetry indicates muscle asym- 
metry, though it is not clear that this part of the rostrum is a 
significant muscle origin in extant taxa (Mead 1975;Heyning 1989). 
The function of the shallow depression immediately posterior to the 
maxillary flange is also uncertain. A similar asymmetrical profile is 
visible in a cast (USNM 243978) of the skull of Microcetus sharkovi 
Dubrovo, 1971 (in Dubrovo and Sharkov 1971: fig. 2), and in the 
skull of Squaloziphius emlongi Muizon (1991: fig. 1). 

The function of the bifurcated posterior of the premaxilla, a 
feature seen in many odontocetes. is uncertain. The bifurcation 
perhaps marks a boundary for the facial muscles. Similarly uncer- 
tain is the function of the medial cleft in the premaxilla. Possibly 
homologous clefts occur near the boundary between the posterolat- 
eral plate and posteromedial splint in Zarhachisflagellator (figured 
by Kellogg 1926: pi. 2) and sporadically in the Ziphiidae. fudged 
from vascular patterns shown by Schenkkan (1973: fig. 5) for 
Mesoplodon, the cleft carries a vessel from the maxillary artery. 

Feeding apparatus. — The long attenuated rostrum and man- 
dibles, the relatively posterior position of the coronoid process on 
the mandibles, and the moderately large temporal fossae and origins 
for the temporal muscles suggest that W. maerewhenua fed by rapid 
snapping. The origin for the temporal muscles on the supraorbital 
process faces ventrally; in contrast, the temporal muscles' origin on 
the frontals of Archaeoceti and some Odontoceti (e.g., Archaeo- 
delphis and the Physeteridae) faces roughly posteriorly. The more 
ventral position for this origin, widespread amongst the Odontoceti, 
is probably related to lever action of the mandible, but it may also be 
a consequence of the posterior expansion of the maxilla, in turn 
dictated by changes in orientation of the facial muscles. 

Waipatia maerewhenua lacks a bony lateral lamina of the ptery- 
goid sinus fossa, and there is no evidence that an ossified pterygoid 
contacted the falciform process. Among the Odontoceti. the pres- 
ence or absence of such a bony lateral lamina (Cozzuol 1 989; Fraser 
and Purves 1960) perhaps relates to feeding musculature. In the 



extant Phocoena phocoena (see Boenninghaus 1904: figs 3, 4; 
Fraser and Purves I960: 12), which lacks a bony lamina, the medial 
limit of the large internal pterygoid muscle stretches from the 
pterygoid (and palatine?) back to the squamosal and bulla, with an 
extensive origin in the pterygoid ligament. Perhaps the lateral 
lamina is the ossified homolog of some of the pterygoid ligament, 
which would provide a stronger origin for the internal pterygoid 
muscle than would ligament alone. In terms of function, an ossified 
ligament could ( 1 ) compensate for enlargement of the pterygoid 
sinus fossa (functioning in acoustic isolation of the skull base), 
which could otherwise weaken the origin for the internal pterygoid, 
and/or (2) be dictated by an enlarged internal pterygoid muscle 
) functioning in feeding). More information about archaic Cetacea is 
needed to confirm that contact of an ossified lamina of the ptery- 
goid with the squamosal is primitive. Furthermore, the bony lateral 
lamina may be constructed in different ways in platanistids, 
squalodelphids, pontoporiids, and eurhinodelphids (Cozzuol 1989), 
all of which have long, forcepslike jaws; this hints at convergence 
for functional reasons. It is not clear why the apparently 
plesiomorphic pterygoid-falciform contact might be lost. 

Waipatia maerewhenua is polydont, as are most extant and 
fossil Odontoceti and embryonic Mysticeti. For heterodont 
polydont Cetacea, it is not possible to homologize the cheek teeth 
with those of the Archaeoceti or other eutherians(Rothausen 1968). 
Tooth structure in heterodont odontocetes has not been correlated 
with particular food preferences. The gracile procumbent incisiform 
teeth appear delicate, suggesting a reduced role, if any. in feeding. 
Perhaps they were used in display. Notocetus marplesi, some 
undescribed early platanistoids from New Zealand, and the small 
Miocene Kentriodon pernix (Kenlriodontidae) have similarly proc- 
umbent teeth. 

Acoustics: Ear. — Odontocete periotics are conservative elements 
that differ dramatically in overall topography, fossae, sulci, and 
foramina from periotics of other eutherians. Odontocete periotics are 
often diagnostic at the species level (Kasuya 1973), suggesting that 
interspecific differences in morphology reflect interspecific differ- 
ences in acoustic abilities, but their function is understood only 
crudely (e.g., an inflated pars cochlearis presumably correlates with 
changes in cochlear structure associated with high-frequency sound 
reception; Fleischer 1976). The specific functions of most features 
seen in W. maerewhenua. for example, the recurved dorsally con- 
cave anteroextemal sulcus, the anterointernal sulcus, the reduced 
anterodorsal and anteroventral angles on and subcircular cross sec- 
tion of the anterior process, the profile of the pars cochlearis, the 
shape of the lateral tuberosity, the lateral groove on the body, the 
posteroexternal foramen, and the bulge on the posterior process 
(homologous with the articular rim), are uncertain. 

The squamosal and parietal in W. maerewhenua are enrolled 
over the periotic (Fig. 8b, 9b), with the periotic detached from the 
braincase wall and displaced ventrolateral^ relative to the cranial 
cavity. The formerly confluent foramen ovale and posterior lacerate 
foramen are separated by contact of the parietal with opposing 
elements along the border of the basioccipital crest. This pattern of 
the squamosal and parietal in the basicranium is so widespread that 
it is perhaps synapomorphic for the Odontoceti. 

The relationship of the squamosal and periotic may be inter- 
preted with reference to basilosaurid archaeocetes (Figs. 9a, b). In 
the Archaeoceti (e.g., Kellogg 1936: figs. 5. 6), the periotic has a 
roughly flat external wall that rises dorsally to form an elevated 
platelike superior process with a narrow crest. The external wall 
contacts the squamosal just ventral to the parietal on the subvertical 
wall of the braincase (Fig. 9a; Pompeckj 1922: pi. 2). Internally, the 
superior process descends to a depression on the dorsal surface of 
the periotic lateral to the internal auditory meatus. Presumably the 
depression is for the superior petrosal sinus dorsolaterally and the 
subarcuate (subfloccular) fossa ventromedially (e.g., Kellogg 1936: 



166 



R. Ewan Fordyce 



figs. 5, 6). Some homologs of archaeocete structures occur in the 
Odontoceti. The periotic of W. maerewhenua lacks an elevated 
platelike superior process; the homologous structure is the convex 
external to dorsoexternal surface of the body (Figs, lib, 9b) that 
rises from the hiatus epitympanicus to the dorsal crest. The dorsal 
crest is a persistent feature among archaic odontocetes (e.g., the 
platanistoids of Figs. 10, p, v, b'; Prosqualodon australis Lydekker. 
1 894. of True 1 909 ) and is the presumed homolog of the crest on the 
archaeocetes' superior process. Farther internally, a variably present 
small groove may mark the superior petrosal sinus. Although I have 
seen no odontocete with a discrete bony subarcuate fossa, Burlet 
(1913: fig. 10) identified this fossa in an embryo of Phocoena 
phocoena. The position of the superior petrosal sulcus of Pompeckj 
(1922) is uncertain. The convex surface of the periotic body (Fig. 
9b) parallels the overlying periotic fossa in the squamosal, while 
the dorsal crest of the periotic lies ventral to the parietal-squamosal 
suture. The cavity between the periotic and periotic fossa is prob- 
ably vascular in part. In other taxa, Muizon ( 1987: 5) described the 
subcircular fossa (here identified as an enlarged foramen spinosum) 
in Notocetus vanbenedeni, while Kellogg (1926: pi. 5) figured an 
unidentified "foramen 1" dorsal to the periotic in Zarhachis 
flagellator. 

Platanistoids vary in the posterior contact of the periotic with 
the squamosal and bulla. Muizon (1987) discussed and figured 
articulations in the Squalodelphidae and Platanistidae, and men- 
tioned (Muizon 1991) that an articular rim or articular process 
occurs in the Squalodontidae. In some New Zealand Squalodonti- 
dae (OU 22072, Figs. lOr, s; OU 21798, Fig. lOy), a prominent 
articular process is present dorsolateral to the long rough articular 
groove for the spiny process. It is not clear whether this articular 
process is homologous or convergent with that of the Squalodelphi- 
dae. Waipatia maerewhenua (Waipatiidae; Figs. 10b, c) and 
Notocetus marplesi (Squalodelphidae; Figs. 10m. n) have only a 
bulge, rather than a process, at the site where the articular rim 
develops. Periotics referred to the Eurhinodelphidae (Fordyce 1983) 
have a process similar to that of the Squalodontidae. Muizon ( 1987) 
considered the feature in eurhinodelphids as not homologous with 
the articular rim or process, but the case is not clear. In Waipatia 
maerewhenua the posterior process is not fused apically or dorsally 
with the squamosal, but in some Squalodontidae (e.g., "Prosqualo- 
don" hamiltoni Benham, 1937, OM C.02.8; Squalodon calvertensis 
Kellogg. USNM 23537) spongy bone along the dorsal edge of the 
posterior process of the periotic appears to fuse with the adjacent 
squamosal. 

Possible functional explanations for the relationship of the 
periotic to adjacent elements include (1) a need for acoustic isola- 
tion from cranial circulation associated with the brain. (2) a need for 
acoustic separation from the braincase. thus enhancing left-right 
acoustic isolation to provide better directional hearing, or (3) a 
consequence of changes in the braincase dictated by changes in the 
brain itself. 

Acoustics: Pterygoid sinus complex. — Waipatia maerewhenua 
lacks orbital extensions of the pterygoid sinuses. Such extensions in 
the Squalodelphidae, Platanistidae, and some Squalodontidae 
(Muizon 1991) perhaps help acoustically isolate the basicranium 
from the face. There is no fossa for a posterior sinus in W. 
maerewhenua, but a sinus may have been present, for Fraser and 
Purves (1960) showed that the sinus is ubiquitous among the 
Odontoceti while a bony fossa is only variably developed. Fraser and 
Purves ( 1 960) further showed that the middle sinus of the middle ear 
is ubiquitous in extant Odontoceti but absent in the Mysticeti; I 
interpret the middle sinus as a synapomorphy for the Odontoceti. In 
many extant Odontoceti. the middle sinus occupies a distinct 
tympanosquamosal recess (Fraser and Purves 1 960), but the recess is 
only variably developed among archaic Odontoceti; for example, it 
is absent in W. maerewhenua and sporadically present in the Squal- 



odontidae. However, structures immediately in front of the anterior 
meatal crest suggest that, despite the lack of a recess, a middle sinus 
was present in W. maerewhenua and indeed in all archaic Odontoceti. 
In all Odontoceti, the skull lacks a postglenoid foramen and lacks the 
anterior transverse ridge that in the Archaeoceti (e.g., Zygorhiza 
kochii, USNM 11962) and archaic Mysticeti bounds the vestigial 
postglenoid foramen. The site of the postglenoid foramen corre- 
sponds to the posterior poriton of the tympanosquamosal recess in 
those taxa where the recess is distinct. I suggest that the site of the 
postglenoid foramen was probably obliterated with, first, the evolu- 
tion of the middle sinus and, second, the evolution of a tympano- 
squamosal recess to accommodate the sinus. 

CLADISTIC RELATIONSHIPS 

Generalized features and traditional placement. — Waipatia 
maerewhenua shows many generalized features of the Odontoceti, 
while structures diagnostic of extant families (for example, con- 
spicuous derived conditions of the premaxilla, premaxillary sac 
fossa, bony nares, and pterygoid sinus fosse) are not obvious. 
Generalized features include the supraorbital processes of the max- 
illae being relatively narrow rather than inflated, the face not being 
particularly voluminous, the large temporal fossae not being roofed 
fully by the supraoccipital. parietal, frontal, and maxilla, the rem- 
nant intertemporal constriction with the parietals exposed dorsally. 
the prominence of the lambdoid and nuchal crests, the palatines' 
being broadly exposed transversely across the palate and not in- 
vaded by pterygoids or pterygoid sinus fossae, the fossae for the 
pterygoid sinuses being restricted to the basicranium and not exca- 
vated dorsally to extend into the orbit, the teeth being heterodont 
and polydont, theperiotic's having a rather elongate narrow internal 
auditory meatus on a slightly inflated pars cochlearis, and the 
periotic's retaining a dorsal crest and an attenuated apex on the 
posterior process. To some cetologists, such features might warrant 
placing W. maerewhenua in the Squalodontidae, along with some 
other small-toothed heterodont dolphins reviewed below, but cla- 
distic analysis indicates otherwise. 

Cladistic analyses of the Odontoceti. — The traditional families 
and higher subdivisions of the Odontoceti (e.g., Fraser and Purves 
1960; Simpson 1945) have been reappraised in recent cladistic 
studies (Fig. 14), such as those of Barnes (1985, 1990), Heyning 
(1989), Heyning and Mead (1990), and Muizon (1987, 1988a, 
1988b. 1991 ). Barnes. Heyning, and Muizon gave valuable lists of 
characters, many of which I have used (Appendix), but only 
Heyning explicitly discussed character polarities or used computer 
analyses to explore multiple cladograms. The published analyses 
show that relationships among odontocete taxa are still volatile 
(Fig. 14). For example, the Ziphiidae are placed with either the 
Physeteroidea (Muizon 1991) or extant Odontoceti other than the 
Physeteroidea (Heyning 1989). Barnes (1985, 1990) placed the 
Pontoporiidae, Iniidae, and Lipotidae (as the Lipotinae) in the 
Platanistoidea, while Muizon (1988b) used an infraorder 
Delphinida for the Delphinoidea (Kentriodontidae, Delphinidae, 
Monodontidae, Phocoenidae, and Albireonidae), Pontoporiidae, 
Iniidae. and Lipotidae (the last three taxa are "river dolphins"). 
Heyning ( 1989: 56) identified the Platanistoidea in the traditional 
sense as paraphyletic and, like Muizon, recognized a monotypic 
Platanistidae, with the Iniidae (including Inia, Pontoporia, and 
Lipotes) as a sister group to the Delphinoidea. 

The following odontocete families are known well enough to 
be used in a cladistic analysis: the Agorophiidae. Albireonidae, 
Delphinidae. Dalpiazinidae. Eoplatanistidae, Eurhinodelphidae 
(= Rhabdosteidae of recent use), Iniidae, Kentriodontidae. 
Kogiidae. Lipotidae, Monodontidae. Patriocetidae, Phocoenidae, 
Physeteridae. Platanistidae, Pontoporiidae, Squalodontidae. 
Squalodelphidae, and Ziphiidae. Some of these taxa may be 



Waipatia maerewhenua, New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



167 




Barnes 1990 



Mysticeti 

Physeteroidea 

Ziphiidae 

Platanistidae 

Agorophiidae 

Eurhinodelphidae 

Squalodontidae 

Squalodelphidae 

Monodontidae 

Albireonidae 

Kentriodontidae 

Phocoenidae 

Delphinidae 




de Muizon 1988, 1991 



Ziphiidae 

Kogiidae 

Physeteridae 

Squalodontidae 

Dalpiazinidae 

Squalodelphidae 

Platanistidae 

Eurhinodelphidae 

Eoplatanistidae 

Lipotidae 

Iniidae 

Pontoporiidae 

Kentriodontidae 

Phocoenidae 

Albireonidae 

Delphinidae 

Monodontidae 




Heyning 
Heyning 



Mysticeti 
Physeteridae + 
Kogiidae 

Ziphiidae 

Platanistidae 

Iniidae + 
Pontoporiidae 

Monodontidae 

Phocoenidae 

Delphinidae 



1989 

& Mead 1990 



Figure 14. Alternative cladograms of broader relationships of the Odontoceti. Left, Barnes ( 1990); middle, Muizon (1988b, 1991 ); right, Heyning 
(1989, Heyning and Mead 1990). 



given subfamily rank (e.g., the Kogiinae. Lipotinae, and 
Patriocetinae), and, depending on the taxonomist. others are 
paraphyletic (e.g., the Agorophiidae and Kentriodontidae). Other 
nominal families (e.g., the Acrodelphidae, Microzeuglodontidae, 
and Zignodelphidae) are junior synonyms or are too dubiously 
based to be analysed cladistically. 

Cladistics: Approaches. — A cladistic analysis of the relation- 
ships of Waipatia maerewhenua was carried out by means of the 
computer program PAUP, version 3.1.1 (Swofford 1993, Swofford 
and Begle 1993). The final data matrix includes 20 taxa and 67 
characters (Table 2, Appendix). Characters were polarized by 
outgroup comparison (outgroup: Zygorhiza kochii). Uninformative 
characters are omitted. Some potentially useful features discussed 
here were not included in the data matrix because they are not 
preserved or illustrated in enough taxa, or because homologies in 
some taxa are uncertain. Characters included were chosen with the 
aim of elucidating the relationships of W. maerewhenua rather than 
reappraising the relationships of all major odontocete groups. The 
approach is conservative; all characters are treated as unweighted 
and ordered. 

From a spectrum of Odontoceti taxa were chosen to form a 
framework into which W. maerewhenua might be placed. Character 
states were determined from (1 (direct study of specimens (optimal) 
or casts, (2) personal notes or photographs (less satisfactory), and 
(3) published literature, which is often inadequate for the details of 
the basicranium and earbones, so that many characters are coded as 
missing. Taxa and specimens (or principal references) included are 
Zygorhiza kochii (Archaeoceti: Basilosauridae). cast of USNM 
11962. Kellogg (1936); Archaeodelphis patrius (Odontoceti 
incertae sedis), Allen (1921); Physeter catodon (Odontoceti; 
Physeteridae), Kasuya ( 1973) and many published illustrations of 
skulls; Kogia hreviceps and K. simus (Odontoceti: Kogiidae), OM 
A. 84. 14, Kasuya ( 1973) and many published illustrations of skulls; 
Mesoplodon grayi (Odontoceti: Ziphiidae), OM A.64.1; 
Tasmacetus shepherdi (Ziphiidae), OM A.88.177; Eurhinodelphi- 
dae, taxa and/or characters reviewed by Muizon (1988a, 1988b, 
1991); Kentriodon pernix (Odontoceti: Kentriodontidae), cast of 
USNM 10670. Kellogg (1927); Pontoporia blainvillei (Odontoceti: 



Pontoporiidae), Kasuya (1973). Barnes ( 1985), and many published 
illustrations of skulls; Tursiops truncatus (Odontoceti: Delphini- 
dae), OU 21820, Barnes ( 1990), and many published illustrations of 
skulls; Cephalorhynchus hectori (Odontoceti: Delphinidae), OU 
21819; Prosqualodon australis and P. davidis Flynn, 1923 
(Odontoceti: Squalodontidae sensu lata), cast of a skull figured by 
Flynn (1948). Lydekker (1894). and True (1909); Squalodon spp. 
(sensu lato) (Squalodontidae), OU 21798 (Fordyce 1989: 23), 
Kellogg ( 1923a), and Rothausen ( 1968); "Prosqualodon" hamiltoni 
(Squalodontidae), OM C.02.8, Benham (1937); Zarhachis 
flagellator (Odontoceti: Platanistidae), Kellogg (1924, 1926) and 
Muizon (1987); Platanista gangetica (Platanistidae). Kellogg 
(1924) and many published illustrations of skulls: Squalodelphis 
fabianii Dal Piaz, 1917 (Odontoceti: Squalodelphidae), Dal Piaz 
( 1977) and Muizon (1987); Notocetus vanbenedeni (Squalodelphi- 
dae), Lydekker (1894) and Muizon (1987); Notocetus marplesi 
(Squalodelphidae), OM C.75.27, Dickson ( 1964). 

Some odontocete families were not included in the analysis 
because ( 1 ) their relationships seem inadequately established for 
the purposes of this exercise, (2) not enough is published about 
structures needed for a cladistic analysis, or (3) specimens were not 
available for study in New Zealand. The Agorophiidae sensu stricto 
and Patriocetidae were excluded. Barnes (1985, 1990), Heyning 
(1989). and Muizon (1991) demonstrated that the Albireonidae. 
Monodontidae, and Phocoenidae belong with other Delphinoidea. 
Furthermore, these authors suggested that the Iniidae and perhaps 
the Lipotidae are related closely to the Pontoporiidae and in turn to 
traditional Delphinoidea. Muizon ( 1991 ) suggested that the poorly- 
known Eoplatanistidae are a sister group of the Eurhinodelphidae 
and placed the Dalpiazinidae [currently monotypic, including only 
Dalpiazina ombonii (Longhi)] uncertainly as a sister group of the 
Squalodontidae. I excluded the Dalpiazinidae from this analysis 
because I could not identify enough characters from the literature 
and because a new supposed dalpiazinid from New Zealand 
(Fordyce and Samson 1992) is not yet described. 

Computer searches were pursued as follows: ( 1 ) an initial mini- 
mal cladogram of 121 steps was obtained by a general heuristic 
search. (2) 81 nonminimal cladograms of 123 or fewer steps were 



IhS 



R. Ewan Fordyce 



obtained. (3) these 81 nonminimal cladograms were input and 
analysed by varied methods reviewed by Swofford and Begle 
(1990: 32-40, 100-104). 

Cladistic relationships of Waipatia maerewhenua: Results. — A 
single cladogram of 121 steps (consistency index 0.628) was ob- 
tained (Fig. 15). This cladogram shows Waipatia maerewhenua as a 
sister taxon to a clade consisting of the Platanistidae and Squalodel- 
phidae and reinforces Muizon's ( 1991 ) concept of the Platanistoidea 
as an odontocete superfamily including the Squalodontidae, Squa- 
lodelphidae, and Platanistidae. Other clades recognized are ( 1 ) a 
Kentriodon + Pontoporia + Cephalorhynchus + Tursiops group, 
which partly represents the Delphinida of Muizon ( 1988b, 1991 ), 
(2) the Eurhinodelphidae as a sister taxon to the cluster of the 
Delphinida, a relationship also proposed by Muizon (1991: fig. 15), 
and (3) a Physeteroidea (Physeteridae + Kogiidae) + Ziphiidae 
group, also recognised by Muizon (1991: fig. 5) as his Physeterida. 
Of note, the Physeterida appear as a sister group to a clade consist- 
ing of the Delphinida and Eurhinodelphidae. in contrast to the 
position shown by Barnes (1990). Muizon (1991). and Heyning 
(1989) for the Physeteroidea and/or Ziphiidae (Fig. 14). Further- 
more, Prosqualodon australis appears as the sister taxon to the 
Squalodontidae, in contrast to the suggestions of Cozzuol and 
Humbert-Lan (1989) and Muizon (1991). 

The 8 cladograms at 122 steps and 81 cladograms at 123 steps 
show a clade consisting of Waipatia, the Squalodelphidae, and the 
Platanistidae, but the positions of other taxa vary. More study of the 
relationships of the Physeteroidea, Ziphiidae. and Eurhinodelphoi- 
dea is needed. For example, if extant Mysticeti are added to the 
current data set as outgroups, and irreversible soft-tissue (e.g., nasal 
diverticula; see Heyning 1989) and osteological (e.g., premaxillary 



sac fossa, foramen, and sulci) characters are used, the Ziphiidae are 
positioned as a sister group to the Platanistoidea and Delphinida, as 
proposed by Heyning (1989). Below I review the relationships of 
Waipatia in more detail. 

Comparisons with the Agorophiidae. — The Agorophiidae may 
be used narrowly (e.g., Fordyce 1981 ), to include only Agorophius 
pygmaeus, or broadly (e.g., Barnes et al. 1985) as a paraphyletic 
group of archaic and presumably late Oligocene Odontoceti (e.g., 
Archaeodelphis. Xenorophus, Atropatenocetus, and Mirocetus). 
Waipatia maerewhenua is not an agorophiid, differing in possessing 
the following derived features: shorter and wider (almost square) 
face, shorter and wider intertemporal region, and shorter parietals. 
Cladistic analysis (Fig. 15) places Archaeodelphis patrius as a basal 
odontocete, but one having some derived features relative to 
W. maerewhenua (i.e., larger lacrimal and massive pterygoids that 
meet medially above the choanae). Cladistic relationships of 
Xenorophus sloani Kellogg, 1923, the fragmentary Atropatenocetus 
posteocenicus Aslanova, 1977, and the enigmatic Mirocetus 
riabinini Mchedlidze, 1970. are uncertain. 

Comparisons with the Physeteroidea and Ziphiidae. — Waipatia 
maerewhenua lacks the key synapomorphies of sperm and beaked 
whales [see Fig. 15 and osteological characters discussed by 
Muizon ( 1991 )]. W. maerewhenua resembles the ziphiid Mesoplo- 
don grayi in the asymmetrical posterior apices of its premaxillae, 
but this is probably convergent; Heyning ( 1989) reported variable 
bone contacts on the vertex among extant Ziphiidae. 

Comparisons with the Eurhinodelphoidea. — Waipatia maere- 
whenua lacks the key synapomorphies of the Eurhinodelphoidea 
(i.e., Eurhinodelphidae and Eoplatanistidae of Muizon (1988a, 
1991); Rhabdosteidae of Barnes (1990: 20)]. Cladograms of 122 



1,2,3,4,5' 



8', 13', 14, 15, 16', 17, 18, 19,20* 



11, 12 



6', 9, 10 



3', 6, 7, 8 



6", 21,22, 23, 24 
10*, 25,33,34,35, 36, 37 i 



4- 

5*, 25, 26, 27 



28 



29,30,31,32' 



4', 26, 38, 39, 40, 41 



22,34,37,50,51,52 



42 ' 



3V, 36 i 



1* 



Zygorhiza 

Archaeodelphis 

Physeter 

Kogia 

| Mesoplodon 

| Tasmacetus 

Eurhinodelphids 

Kentriodon 

Pontoporia 



41" 

5*, 43' 
9,21,39,41,46' | 



25, 33, 45, 46, 47, 48, 49 



2*, 25", 33', 53, 54 
5', 6', 21  



\ Cephalorhynchus 

iH Tursiops 

Prosqualodon 

— ^ Squalodon 

"P." hamiltoni 



27, 35, 42, 44, 55, 56, 57, 58, 59 



21.60*1 



2*, 6', 30, 32, 33', 45', 60, 61,62 



27*, 30', 33", 36, 42*. 57*, 66, 67 • 



33" i 



46 



d" 



63, 64, 65 



25*', 28 ' 



Waipatia 
Zarhachis 
Platanista 
N. marplesi 
N. vanbenedeni 
Squalodelphis 



Figure 15. Cladogram of relationships of Waipatia maerewhenua. Numbers at each node refer to characters discussed in the text and listed in the 
Appendix. Symbols: ', change from state 1 to state 2; ". change from state 2 to state 3; *, reversal from state I to state 0; **, reversal from state 2 to state 1 . 



Waipatia maerewhenua. New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



i6y 



steps (one over the minimum) suggest various relationships for the 
Eurhinodelphoidea. Affinities with the Delphinida need more study, 
though comparisons are hampered by the lack of published infor- 
mation on eurhinodelphoid basicrania. Similarities, presumably 
convergent, between W. maerewhenua and some eurhinodelphids 
include nodular nasals, as in Argyrocetus patagonicus (Lydekker 
1894: pi. V). and periotic axis (viewed dorsally) being sigmoidal. 
with the anterior process skewed medially and the posterior process 
skewed laterally. 

Comparisons with the Delphinida. — Waipatia maerewhenua 
lacks the key synapomorphies of the Delphinida (sensu Muizon 
1 988b; see also Barnes 1 990: 20, taxa under nodes 23-44 and some 
under node 45). The cladogram (Fig. 15) is consistent with concepts 
of the Delphinidae and associated taxa advanced by Muizon 
(1988b) and Heyning (1989: 56). 

Relationships with the Platanistoidea. — Muizon (1987, 1991) 
abandoned the Platanistoidea sensu Simpson (1945) to propose a 
Platanistoidea encompassing the Platanistidae, Squalodelphidae. 
Squalodontidae. probably Dalpiazinidae, and the enigmatic 
Prosqualodon. A simplified outline of Muizon's proposed relation- 
ships appears in Fig. 14 (which lacks Prosqualodon). Heyning 
(1989) also separated the Platanistoidea from other extant river 
dolphins {Lipotes, Pontoporia, lnia), which Heyning placed in the 
Delphinoidea. Muizon's hypothesis of relationships is broadly sup- 
ported by Fig. 15, which identifies Waipatia as a platanistoid related 
more closely to the squalodelphid-platanistid clade than to the 
Squalodontidae. 

Apparent synapomorphies of the Platanistoidea (Fig. 15; 
Muizon 1987, 1991) are as follows (numbers refer to characters 
listed in Appendix): the profile of the anterior process of the periotic 
is smoothly to abruptly deflected ventrally in lateral view (25); the 
anterior process of the periotic is roughly cylindrical in cross sec- 
tion (47); the periotic has a ridge- or peglike articular process (33); 
the bulla has an anterior spine and an inflated anterolateral convex- 
ity (45, 46); the scapula lacks a coracoid process (49); and the 
acromion process lies on the anterior edge of the scapula, which 
lacks a supraspinous fossa (48). Of note, the last two scapular 
features are not seen consistently in supposed Platanistoidea. 
Cozzuol and Humbert-Lan (1989) stated that the squalodontid 
Phobe radon arctirostris Cabrera. 1926, has a scapula with an ap- 
parent coracoid process, a conspicuous supraspinous fossa, and an 
acromion not located on the anterior edge. Muizon (1987) noted 
that the scapula in Sulakocetus dagestanicus has a narrow coracoid 
process; S. dagestanicus is identified below as probably related to 
W. maerewhenua and thus to other Platanistoidea. 

Future work on new or re-prepared late Oligocene and early 
Miocene platanistoids should further elucidate patterns of homol- 
ogy, including whether character transitions were reversible or 
irreversible). Thus the detailed pattern of platanistoid relationships 
shown in Fig. 15 is likely to change. 

Relationships with the Squalodontidae. — Waipatia maerewhe- 
nua lacks the key synapomorphies for Squalodontidae as defined 
below, but shares synapomorphies of the periotic and basicranium 
with the Platanistidae and Squalodelphidae. The cladogram (Fig. 
15) is broadly consistent with the concept of squalodontid relation- 
ships proposed by Muizon (1991). Some discussion of the Squal- 
odontidae is needed, however, for many heterodont Cetacea, in- 
cluding Waipatia-Wke taxa. have been referred to this family. The 
following brief review incorporates some revisions made by 
Muizon (1987. 1991). 

The Squalodontidae derive their identity in nomenclature from 
Squalodon gratelupi Pedroni (= Squalodon typicus Kellogg. 1923). 
the type species of Squalodon (see Rothausen 1968). The holotype 
of Squalodon gratelupi (early Miocene) is a partial rostrum 
(Grateloup 1840: fig. I: Kellogg 1923a; Rothausen 1968). 
Squalodon is well known from skulls, such as those referred to 5. 



bariensis (Jourdan) and S. calvertensis, and other specimens (e.g., 
S. melitensis, S. kelloggi) represented by teeth and partial jaws are 
reasonably assigned to Squalodon. Overall, Squalodon provides a 
sound typological base for diagnosing the Squalodontidae. Of note, 
some nominal species of Squalodon based on teeth probably do not 
represent Squalodon, the Squalodontidae. or even the Odontoceti; 
e.g.. Squalodon serralus Davis (archaic Mysticeti?) and Squalodon 
(MicrozeuglodonT) wingei Ravn (archaic Odontoceti?). Beyond 
Squalodon. concepts of the Squalodontidae vary markedly. Kellogg 
( 1 923a) stressed that many heterodont odontocetes had been placed 
arbitrarily in the Agorophiidae. Microzeuglodontidae, 
Patriocetidae. or Squalodontidae. Later. Kellogg (1928) listed 
Squalodon, Prosqualodon, Microcetus and 1 1 other genera as 
squalodontids. recognized the Agorophiidae (Agorophius, Xenoro- 
phus), and placed Patriocetus and Agriocetus (with Archaeo- 
delphis) as Cetacea incertae sedis. Simpson (1945) proposed a 
superfamily Squalodontoidea, but otherwise largely followed 
Kellogg. Rothausen ( 1961 ) confirmed the squalodontid affinities of 
Microcetus. discussed below. Rothausen (1968) placed the 
Patriocetinae (Agriocetus, Patriocetus) in a grade Squalodontidae. 
Muizon ( 1987) alluded to the possible polyphyly of the Squalodon- 
tidae but later ( Muizon 1 99 1 ) listed synapomorphies of the Squal- 
odontidae within the Platanistoidea. Cozzuol and Humbert-Lan 
( 1 989) excluded Prosqualodon australis (including P. davidis) from 
the Squalodontidae, suggesting relationships with the Delphinida. 
More broadly, Cozzuol and Humbert-Lan (1989) questioned the 
synapomorphies used by Muizon to include the Squalodontidae in 
the Platanistoidea. 

I use the name Squalodontidae conservatively here, to include 
Squalodon, Eosqualodon. Kelloggia, and Phoherodon (Cabrera 
1926; Muizon 1991; Rothausen 1968), and "Prosqualodon" 
hamiltoni. Whitmore and Sanders (1977) and Fordyce (1989: 23) 
mentioned skulls of new squalodontids, not yet formally described; 
elements of the latter (OU 21798) are figured here (Figs. 10x-c'). 
"Prosqualodon" marplesi is a squalodelphid: see below. 
Prosqualodon australis is discussed below. None of the other 
"shark-toothed dolphins" is demonstrably close to Waipatia. For 
example, Neosqualodon and Patriocetus are of uncertain relation- 
ships (cf. Rothausen 1968). Agriocetus, Austrosqualodon, 
Metasqualodon, Microzeuglodon, and Tangaroasaurus are based 
on fragmentary specimens 1 regard as Odontoceti incertae sedis. 
Microcetus, Sachalinocetus, and Sulakocetus are not clearly 
squalodontids but are perhaps related to the Squalodelphidae and 
Waipatia. as discussed below. 

Waipatia maerewhenua lacks the following key synapo- 
morphies seen in the skulls of Squalodontidae (as defined above): 
skull long (estimated condylobasal length >700 mm in adults); 
rostrum robust and long with expanded apex (Muizon 1991); ros- 
trum proximally deep, probably a consequence of a narrow deep 
mesorostral groove; cheek teeth triangular, large (>20 mm long), 
high-crowned, denticulate, and elongate but somewhat inflated lat- 
erally; and crowns of anterior to middle cheek teeth with rather 
small denticles widely spaced on anterior and posterior cheek-tooth 
keels. Furthermore. Waipatia maerewhenua lacks the following key 
synapomorphies (numbers refer to characters listed in Appendix) 
seen in the tympano-periotics of squalodontids (see squalodontids 
OU 22072, Figs. lOr-w, and OU 21798, Figs. 10x-c'): the 
subcylindrical anterior process has a prominent tubercule on the 
apex (53); the dorsal surface of the anterior process is smoothly 
curved (in lateral view) so that the apex of the process lies ventrally; 
there is no anteroexternal sulcus for the middle meningeal artery: 
there is no subhorizontal anterointernal sulcus, though multiple fine 
vertical vascular grooves run across the internal surface of the 
anterior process (54); the fenestra rotunda is reniform. prolonged 
dorsomedially, and associated with a fissure and posterior ridge that 
run dorsally to the aperture for the cochlear aqueduct (22); the 



170 



R. Ewan Fordyce 



Table 1. Measurements of Waipatia maerewhenua, OU 22095, 
holotype (mm). 



Skull (± 1 mm; following Pemn 1975) 
Condylobasal length 
Rostrum length 
Rostrum width at base 
Rostrum width at preserved mid length 
Premaxillary width dorsally at level of preserved 

midlength of rostrum 
Premaxillary width dorsally at level of antorbital notches 
maximum premaxillary width dorsally. about level 

with mid-orbit 
Distance from level of antorbital notches to most 

anterior border of nasals 
Distance from level of antorbital notches to 

border of internal nares (pterygoids missing) 
Cranial length (averaged to compensate for distortion) 
Preorbital width at level of lacrimal-frontal suture 
Postorbital width, maximum across postorbital processes 
Palatine length, in midline 
Maximum width of external nares (between margins 

of premaxillae immediately anterior to nasals) 
Width of left frontal at level of apex of premaxilla 
Width of right frontal at level of apex of premaxilla 
Minimum width, intertemporal constriction 
Distance from anterior of inter-nasal suture to apex 

of supraoccipital 
Maximum width across zygomatic processes 
Point-to-point distance, apex of supraoccipital to 
dorsal intercondylar notch 
Periotic (±0.5 mm; right periotic unless specified) 
Anteroposterior length 
Width, internal margin of pars cochleans to external 

margin at hiatus epitympanicus, level with fenestra ovalis 
Length of pars cochleans, from groove for tensor 
tympani to mid-point of stapedial muscle fossa 
Length of internal auditory meatus 
Length of posterior bullar facet (left) 
Tympanic bulla (±0.5 mm; after Kasuya 1973) 
Standard length, anterior apex to apex of outer 

posterior prominence 
Length, anterior apex to apex of inner posterior prominence 
Distance, outer posterior prominence to apex 

of sigmoid process 
Width at level of sigmoid process 
Dorsoventral depth of involucrum immediately in front 

of posterior pedicle 
Elliptical foramen >5 (high) 

Maximum point-to-point length of posterior process 
Mandible (± 1 mm; following Perrin 1975) 
Length of left tooth-row, from posterior margin 
of most posterior alveolus to tip of mandible 
Maximum length of right mandible 
Maximum length of left mandible 
Maximum height of right mandible, perpendicular 
to maximum length 
Atlas (± 1 mm) 

Maximum vertical diameter, parallel to anterior face 
Maximum vertical diameter of neural canal 
Maximum vertical diameter of neural canal 
Minimum anteroposterior diameter of centrum ventrally 
(just lateral to hypapophysis) 



>556 
>320 

147 
59.5 

27 
95 

113.5 

71 

69+ 
235 
199 
237 

89 

42.5 
20 
28 
115 

63 

244 

102 

40.0+ 

20.4 

19.3 
10.0 
19.0 



48.8+ 

45+ 

33.5+ 
33.2 

18.4 
x >1 (wide) 
25.2 



278+ 
446+ 
458+ 

133 

72+ 
c. 30 
c.40 

27.5 



lateral face of the periotic between the internal auditory meatus and 
hiatus epitympanicus is wide and flat to gently convex (50) (Muizon 
1991: 305); the apex of the posterior process of the periotic is 
attenuated (Muizon 1991: 305). narrow, and dorsoventrally deep, 
with a porous to spiny dorsal surface (37); and the bullar facet on 
the posterior process extends dorsally onto the posteromedial face 



of the process (51). The atlas is less compressed than in the 
squalodontid OU 22072 (Figs. 13h-j). 

Other structures cited as characteristic of the Squalodontidae 
(see Kellogg 1923a; Rothausen 1968; Mchedlidze 1984; Muizon 
1987, 1991; Cozzuol and Humbert-Lan 1989) are not all reliable 
synapomorphies. For example, large robust apical teeth also occur 
in the Archaeoceti. Heterodont teeth, a symmetrical cranium, and 
large temporal fossae are primitive features seen in archaeocetes 
and the odontocete Agorophius pygmaeus. Many odontocetes have 
a well-telescoped skull in which the maxillae closely approach or 
contact the supraoccipital, so that the parietals are not exposed in a 
continuous band across the vertex [but I disagree with Muizon 
( 1 99 1 : fig. 15, character 1 ) in his use of a contact of the maxilla with 
the supraoccipital as a key synapomorphy of Odontoceti]. The 
frontals contact the apex of the supraoccipital, excluding the pari- 
etals from the vertex (e.g., Eurhinodelphidae, Ziphiidae, and 
Kentriodontidae). The apex of the pterygoid hamulus is also elon- 
gated, subcorneal, and not excavated by the pterygoid sinus in 
Eurhinodelphidae and in an undescribed "agorophiid" (USNM 
256517). The lateral lamina of the pterygoid contacts the falciform 
process of the squamosal in Zygorhiza, the Balaenopteridae, and the 
Platanistidae, for example, and the dorsal region of the pterygoid 
sinus fossa primitively lies below the level of the orbit in 
archaeocetes and some undescribed "agorophiids." 

Cozzuol and Humbert-Lan ( 1989) and Muizon ( 1991 ) excluded 
the enigmatic Prosqualodon australis from the Squalodontidae and 
placed it incertae sedis. Initially I followed this assignment, but the 
taxon emerges in the squalodontid clade of Fig. 15. The study of 
specimens, rather than casts, of P. australis may alter characters in 
Table 2, thus modifying the proposed relationships. 

Comparisons with the Dalpiazinidae. — Muizon (1988a) pro- 
posed the new monotypic family Dalpiazinidae and new genus 
Dalpiazina for Champsodelphis ombonii. Dalpiazina ombonii is 
known from a partial rostrum, partial skull, and periotic, which 
Muizon (1988a) described and figured. Later, Muizon (1991: fig. 
15) identified Dalpiazinidae as a possible sister taxon to the Squal- 
odontidae. Waipatia maerewhenua lacks the presumed synapomor- 
phies of D. ombonii; for example, it lacks homodont teeth, deep 
premaxillary sulci, an enlarged exposure of vomer on the rostrum, 
and a short wide vertex. Waipatia maerewhenua is more derived in 
that its mandibles have a shorter unfused symphysis and the ante- 
rior process of the periotic is relatively larger and more inflated 
transversely, with a blunter apex reflected more abruptly ventrally. I 
doubt that Waipatia maerewhenua is descended from or ancestral to 
D. ombonii. 

Comparisons with the Squalodelphidae and Platanistidae — 
My results suggest that Waipatia maerewhenua is the sister taxon to 
the Platanistidae and Squalodelphidae (Fig. 15). Though W. 
maerewhenua is more primitive than Platanistidae and Squalodel- 
phidae in many characters. I doubt that it is merely a generalized 
"ancestral" squalodelphid. Compared with these taxa, it is derived 
in some features, and it is not demonstrably descended from any 
known platanistid or squalodelphid. 

The family Squalodelphidae Dal Piaz. 1917 (sensu Muizon 
1987, 1988a), includes Squalodelphis fabianii, Notocetus vanbene- 
deni, Notocetus marplesi, Phocageneus venustus Leidy. 1869 (see 
Kellogg 1957). and Medocinia tetragorinus (Delfortrie. 1875); all 
are early Miocene. (Notocetus spp. and 5. fabianii are included in 
Fig. 15. though in S. fabianii many sutures are uncertain). Thus 
delimited, squalodelphids possess several cranial features more 
derived than those of Waipatia maerewhenua, some of which are 
included in Fig. 15. Published comments (Barnes 1990; Dal Piaz 
1917; Lydekker 1894; Moreno 1892; Muizon 1987, 1991; True 
1910a) and interpretations of published figures suggest that these 
features include the following: the median cranial elements are 
more asymmetrical and more skewed to the left; the maxillae are 



Waipatia maerewhenua, New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



171 



Table 2. Data matrix used with PA UP 3. 1. 1 in the cladistic analysis of Waipatia maerewhenua. Number of taxa, 20; number of characters, 
67. Symbols used are 0. 1.2, 3. Missing characters are eoded *; irrelevant characters are coded -. An "equate" macro was used thus: 
-= *; a = (01); b = (12); c = (123); d = (23). See text for details. 

















Character 














Taxon 


5 


10 


15 


20 


25 


30 


35 


40 


45 


50 


55 


60 


65 




Zygorhiza 


00000 


00000 


oo*** 


**oo* 


aOOOO 


00000 


*00*0 


OOOOO 


00*00 


ooooo 


ooooo 


ooooo 


000-0 


00 


Archaeodelphis 


1 * 1 ** 


000*0 


*o*** 


**oo* 


*000* 


00*00 


*0*o* 


***O0 


0**0* 


*o*** 


*000* 


***o* 


00*00 


0* 


Physeler 


*12*1 


2101* 


11011 


onto 


0000* 


0-00(1 


ooooo 


0000* 


*0 


ooooo 


00(100 


00-00 


ooooo 


00 


Kogia 


1 12** 


2101* 


11011 


oiiio 


00000 


00000 


ooooo 


0000* 


— *00 


ooooo 


ooooo 


00-00 


ooooo 


00 


Mesoplodon 


11210 


inn 


11100 


10001 


inn 


11000 


ooooo 


ooooo 


ooooo 


ooooo 


ooooo 


00-00 


ooooo 


00 


Tasmacetus 


11211 


him 


11100 


10001 


imo 


00 1 00 


0*000 


ooooo 


ooooo 


ooooo 


ooooo 


00-00 


ooooo 


00 


Squalodon 


10211 


21100 


00*** 


**00* 


01 001 


ooooo 


*0210 


01000 


00*01 


urn 


Hill) 


-00** 


ooooo 


00 


"P." hamilloni 


**21* 


all*0 


oo*** 


**oo* 


01001 


00000 


**2*o 


010*0 


0**0* 


Mill 


11110 


ooooo 


ooooo 


0* 


Prosqualodon 


*121* 


11110 


oo*** 


**oo* 


1*002 


000*0 


*0110 


0*010 


10*** 


01111 


*1000 


0*0*0 


00*00 


0* 


Eurhinodelphid 


11211 


21110 


00*** 


**00* 


0*001 


0001* 


* 1 1 1 1 


11000 


0-**0 


0*000 


0000* 


0*-00 


ooooo 


00 


Waipatia 


11210 


011*0 


00*** 


**00* 


10002 


01000 


*01*1 


ooooo 


01*11 


11**0 


00001 


imo 


000(10 


00 


Notocetus murplesi 


1*21* 


*1 1*0 


oo*** 


»*oo* 


00002 


01001 


** 1 * J 


*00*0 


01*1* 


*1110 


00001 


111*1 


inn 


00 


N. vanbenedeni 


*0211 


21100 


00*** 


**oo* 


00001 


01101 


*!">*] 


ooooo 


0***2 


11110 


0000* 


urn 


urn 


00 


Squalodelphis 


**21 1 


* | **Q 


oo*** 


**o*+ 


000*b 


0*1** 


**2*i 


00**0 


****2 


11**0 


00001 


**11* 


1*1** 


** 


Zarhachis 


10211 


21100 


oo*** 


**oo* 


1000b 


000*2 


* 1301 


10000 


0***") 


1 ***Q 


*0001 


10-10 


11000 


11 


Platanista 


10211 


21100 


00100 


10001 


0*00b 


00002 


01301 


10000 


000*2 


0*110 


0000* 


[*-l] 


1IOOO 


11 


Kentriodon 


11201 


21111 


oo*** 


**00* 


00000 


1001 1 


*10*0 


001*1 


11*00 


ooooo 


ooooo 


00-00 


OOOOO 


00 


Cepkalorhynchus 


01200 


21111 


00100 


10001 


00000 


10011 


11000 


00111 


20100 


ooooo 


ooooo 


00-00 


ooooo 


00 


Tur stops 


11200 


21111 


00100 


10001 


00000 


10011 


11000 


00111 


20110 


ooooo 


OOOOO 


00-00 


ooooo 


00 


Pontoporia 


*1201 


2IIII 


00100 


10001 


00000 


10011 


21000 


10111 


200*0 


ooooo 


OOOOO 


0*-00 


ooooo 


00 



markedly thickened to form crests above the orbits and have a 
"squared off" posterior profile (dorsal view) at the contact with the 
nuchal crest of the supraoccipital. where the parietals are eliminated 
from the vertex; the premaxillae overhang the mesorostral groove 
more: the face between the level of the antorbital notch and the 
nasals is more foreshortened, with a more curved premaxillary- 
maxillary suture concentric around the nares; the internarial suture 
and nasofrontal suture are deep and narrow; the palatine is not 
exposed broadly from side to side on the rostrum, since the apex of 
the pterygoid here contacts the maxilla, but is exposed laterally; the 
narrow pterygoid sinus fossae are excavated dorsally (as seen from 
below), with a continuous lateral lamina of pterygoid extending 
back to contact the falciform process of the squamosal; there are 
marked orbital fossae in thickened frontals for orbital extensions of 
the pterygoid sinuses; and the supraoccipital is asymmetrical, with 
a skewed median ridge and rather abrupt anterolateral corners. The 
periotics of the Squalodelphidae are more derived than those of 
W. maerewhenua in having a prominent to peglike articular process 
in most, a more circular pars cochlearis, a more prominent lateral 
groove on the periotic. so that the profile in dorsal view is more 
sigmoidal, a larger aperture for the cochlear aqueduct in most, a 
rounder internal auditory meatus, and a posterior process with a 
long smooth parallel-sided facet for contact with the bulla. De- 
scribed squalodelphid teeth are nearly homodont. Synapomorphies 
listed by Muizon ( 1987, 1991 ), some of which are included in Fig. 
15, adequately separate the Platanistidae from Waipatia. 

A new family Waipatiidae. — These comparisons suggest that 
Waipatia maerewhenua warrants a new family. The species is a 
platanistoid more closely related to the Squalodelphidae and Pla- 
tanistidae than to the Squalodontidae, but it differs from the former 
taxa in possessing the following derived features: the mandibles 
have a shorter unfused symphysis (5); the nasals are short and 
broad; the pterygoid sinus fossa posteromedial to the foramen ovale 
is larger (6) (this fossa is absent in the squalodelphid Notocetus 
marplesi); the falciform process is bifid, without clear evidence of 
contact with an ossified lateral lamina of the pterygoid; the anterior 
process of the periotic is relatively larger and more inflated trans- 



versely (21 ), with a blunter tip and an axis more abruptly reflected 
ventrally (see Notocetus marplesi. Fig. lOn); and the atlas lacks a 
long hypapophysis. 

Other odontocetes superficially similar to W. maerewhenua are 
known. Of these, Agriocetus incertus (Brandt. 1874) (see Abel 
1914: pis. 4—5) is similar in size and age to Waipatia maerewhenua. 
The holotype is part of a cranium (late Oligocene, Austria) too 
heavily encrusted with matrix to reveal the suture detail necessary 
for useful comparisons. Agriocetus incertus cannot be assigned to a 
family, and the holotype is so uninformative that the name may be a 
nomen dubium. 

Rothausen (1961) reviewed the heterodont Phoca ambigua 
Meyer. 1840 (late Oligocene. Germany), which is the type species 
of the supposed squalodontid genus Microcetus Kellogg. 1923. The 
species is known only from cheek teeth, which are smaller than 
those of Waipatia maerewhenua. It shares with W. maerewhenua 
the (derived?) loss of anterior denticles on the cheek teeth. In M. 
amhiguus, the cheek-tooth crowns are lower and thicker, with finer 
apical ornament, and the roots are thicker. Characters of heterodont 
teeth of Cetacea are understood too poorly to enable cladistic 
comparison of Microcetus with Waipatia, but the stated differences 
probably separate these genera in terms of traditional taxonomy. 
Microcetus ambiguus is probably not a squalodontid but may be a 
waipatiid. 

Other nominal species of Microcetus are not clearly congeneric 
with M. amhiguus. Microcetus sharkovi is based on a crushed 
partial skull and incomplete mandible (late Oligocene, Kazakhstan). 
The skull is similar in size and profiles to W. maerewhenua, but its 
sutures are indistinct. As in W. maerewhenua, the base of the 
rostrum, antorbital notches, and preorbital processes are asym- 
metrical, though the right process is more pronounced. These fea- 
tures and the small cheek teeth eliminate M. sharkovi as a 
squalodontid. The species differs from W. maerewhenua in that the 
premaxilla overhangs the mesorostral groove more, the posterolat- 
eral sulcus is deeper, and the worn mandibular cheek teeth are 
smaller and less emergent. This species does not clearly belong in 
Microcetus. It may be a waipatiid. 



172 



R. Ewan Fordyce 



Microcetus hectori Benham, 1935, is known only from the 
holotype (NMNZ Ma 653), collected in the Waitaki Valley, near the 
type locality of W. maerewhenua. The holotype includes a distorted 
partial cranium, the described incomplete right mandible with 5 
small heterodont cheek teeth in place, and loose teeth. The holotype 
is from about the middle of the Maerewhenua Member of the 
Otekaike Limestone, about lower Waitakian Stage (= earliest Mio- 
cene, about 23 Ma). Benham (1935) assigned the species to 
Microcetus because its cheek teeth lack anterior denticles. 
Microcetus hectori differs from M. ambiguus in that the former has 
cheek teeth on which the crowns are relatively higher, more intlated 
laterally, and smoother. Rothausen (1961) suggested that these 
species are probably not congeneric. Rothausen ( 1970: fig. 1 ) pro- 
posed the generic name Uncamentodon for M. hectori without 
further diagnosis. Microcetus hectori is similar in size to W. 
maerewhenua, and also has deeply rooted and presumably procum- 
bent incisors, but M. hectori differs in the following features: 
middle to posterior mandibular cheek teeth subcortical, smaller, and 
more inflated laterally, with reduced ornamentation; tympano- 
squamosa) recess more pronounced and more pitted posteriorly; 
and foramen spinosum (an incipient subcircular fossa) markedly 
larger. These species are not conspecific, and are probably not 
congeneric. The large foramen spinosum indicates that Microcetus 
hectori is probably a squalodelphid; it is not a squalodontid (cf. 
Fordyce 1982: Rothausen 1961). 

The New Zealand species "Prosqualodon" marplesi is known 
only from the holotype (OM C.75.27), which includes an incom- 
plete skull (Figs. 16e, f), an undescribed periotic, and assorted 
elements listed or described by Dickson ( 1964). The type locality is 
"Trig Z," near Otiake. Waitaki Valley. The holotype is probably 
from the "lower shell bed" at the top of the Maerewhenua Member 
of the Otekaike Limestone, about middle Waitakian Stage (= earli- 
est Miocene, about 22-23 Ma: Fordyce et al. 1985: Hornibrook et 
al. 1989). This is younger than Microcetus hectori and Waipatia 
maerewhenua. The skull of "Prosqualodon" marplesi differs mark- 
edly from that of squalodontids and Prosqualodon in its asymmetry 
and other features noted in the cladistic analysis. Despite its small 
size, procumbent anterior teeth, and probable heterodont dentition. 
"Prosqualodon" marplesi is not conspecific or congeneric with W. 
maerewhenua; rather. "Prosqualodon" marplesi resembles 
Notocetus vanbenedeni (early Miocene, Patagonia) in its deeply 
sutured nodular asymmetrical frontals. "squared off posterior mar- 
gin of the maxillae along the contact with the supraoccipital, asym- 
metrical supraoecipital. larger hypapophysis (Figs. 13k-m) on the 
atlas, and a range of features on the previously undescribed periotic 
(e.g., acute anterointernal apex, sigmoidal profile in dorsal view, 
prominent dorsal crest on the dorsal surface of the periotic at the 
junction between the body and anterior process, and long smooth 
parallel-sided facet on the posterior process). "Prosqualodon" 
marplesi is here formally transferred from the Squalodontidae to 
Notocetus (Squalodelphidae) (see Fig. 15). 

Sulakocetus dagestanicus is a late Oligocene supposed 
squalodontid. based on a holotype from the Caucasus. The incom- 
plete skull (Figs. 16e>d: Mchedlidze 1984: pis. 13, 14: Pilleri 1986: 
pis. 5-8) is small and heterodont. with a rostrum moderately wide at 
the base and attenuated distally. Mchedlidze (1984) outlined gen- 
eral features of the skull; most details of the sutures are uncertain, 
and the periotic is unknown. Sulakocetus dagestanicus is not clearly 
a squalodontid. In lateral view, the skull is similar in profile to that 
of W. maerewhenua. The mandibular teeth (Mchedlidze 1984: pis. 
14, 15) are smaller and more gracile than those of the Squalodonti- 
dae, resembling those of W. maerewhenua. Sulakocetus perhaps 
belongs in the Waipatiidae but is known too poorly for cladistic 
analysis. Waipatia maerewhenua apparently differs from 5. 
dagestanicus as follows: preorbital process not as thick dorsoven- 
trally; premaxillary-maxillary suture on rostrum less pronounced; 



premaxillary sulci shallower; premaxilla overhangs mesorostral 
groove less; premaxilla has transversely flatter profile in front of 
nares; nasals appear more nodular; posterolateral plate has more 
convex profile (lateral view); vertex is not as elevated or rounded in 
lateral view; mandibular cheek teeth more emergent with less trian- 
gular crowns; body of mandible more robust; and pan bone of 
mandible less inflated ventrally. It is not clear whether the maxilla 
contacts the supraoccipital in S. dagestanicus (cf. Muizon 1987). 
Sulakocetus dagestanicus is not clearly conspecific with other de- 
scribed heterodont taxa. 

Sachalinocetus cholmicus Dubrovo. 197 1 , is an early or middle 
Miocene supposed squalodontid from Sakhalin, northwest Pacific. 
The holotype skull is about 600 mm long. Dubrovo's ( 1971 ) recon- 
structions (Figs. 16a. b) suggest that the skull is similar in profile to 
XV. maerewhenua in dorsal and ventral views, but a lateral view of 
the skull reveals a deeper fossa for facial muscles. On the vertex, 
the frontals appear to be longer and narrower than in W. 
maerewhenua. Not enough is shown of skull sutures to allow de- 
tailed comparisons. The teeth are heterodont, and the slender long 
incisors were probably procumbent. Posterior cheek teeth lack 
much ornamentation on the crowns and have reduced posterior 
denticles. In a traditional approach to classification, similarities 
between Sachalinocetus and Waipatia would probably see these 
genera in the same family. Contrary to Dubrovo's (1971) conclu- 
sions, Sachalinocetus is not clearly a squalodontid. I suspect that 
Sachalinocetus belongs in the Waipatiidae, and that the ciade thus 
ranges into the Miocene. 

CONCLUSIONS 

Waipatia maerewhenua is sufficiently generalized that it might 
be placed in one of several odontocete clades. Dorsal structures on 
the cranium in W. maerewhenua. traditionally used in odontocete 
classification, indicate that cranial asymmetry arose by the late 
Oligocene, but otherwise suggest only that the species perhaps is 
not a squalodontid. What remains of the pterygoid sinus complex is 
also generalized, apart from the posteromedial expansion of the 
sinus. Features of the tympano-periotic and basicranium allow W. 
maerewhenua to be placed in the Platanistoidea and in a new 
family, Waipatiidae, as a sister group to the Squalodelphidae and 
Platanistidae. Some described Oligocene and earlier Miocene 
"squalodontids" may also be waipatiids. but most are too incom- 
plete or too poorly described to be sure. The range of described 
Waipatia-\\ke species hints at a significant diversity of the 
Waipatiidae later in the Oligocene. Waipatiids were perhaps the 
ecological equivalents of medium-sized extant delphinids with ro- 
bust rostra, such as Tursiops truncatus. Judged from New Zealand 
late Oligocene specimens such as the Squalodon-Yike OU 21798 
(Fordyce 1989: 23), contemporaneous squalodontids were larger 
predators with no clear modern analogs. Squalodelphids and a 
Da/piazina-like small odontocete lived in New Zealand waters 
during the latest Oligocene or earliest Miocene (Fordyce and 
Samson 1992), and early Miocene representatives (Muizon 1991) 
are well known elsewhere. Such fossils suggest that platanistoids 
were globally diverse and ecologically important earlier than sus- 
pected. Later Neogene long-beaked Zarhachis-like taxa, which 
reveal little of this older history of platanistoids, foretell the origins 
of the fluviatile Platanista spp. — the near-extinct relicts of a once- 
diverse marine taxon. 

ACKNOWLEDGMENTS 

I wish to dedicate this article to Frank C. Whitmore. Jr.. with 
grateful thanks for over a decade of wide-ranging counsel on fossil 
cetaceans. Through his thoughtful, supportive, and temperate com- 
ments, through ideas shared freely with many colleagues, and 






Waipatia maerewhenua. New Genus and New Species, an Archaic Late Oligocene Dolphin from New Zealand 



173 




Notocetus marplesi 



Figure 16. Reconstructions of skulls of some archaic platanistoids, not to same scale. A-B. Sachalinocetus cholmicus, based on Dubrovo (1971). A, 
lateral; B. dorsal. C-D, Sulakocetus dagestanicus. based on Mchedlidze (1976. 1984). C. lateral; D, dorsal. E-F, Notocetus marplesi, based on Dickson 
( 1964) and on holotype. E. lateral; F. dorsal. 



through an appreciation of the human element in science. Dr. 
Whitmore has done much to further the study of fossil Cetacea 
world-wide. 

I also thank the following for their help. The Harvey family 
gave permission to work on their property. Andrew Grebneff as- 
sisted in the field, skillfully prepared the holotype of Waipatia 
maerewhenua, and helped with illustrations. Greg Ferguson helped 
with preparation, literature work, and photography. Bob Connell 
assisted with field work. Christian de Muizon and Catherine R. 
Samson provided very useful comment on the manuscript and/or 



specimens. The editors, M. Gottfried, and an anonymous referee 
also gave constructive comments. Lawrence G. Barnes, Mario A. 
Cozzuol, James G. Mead, and Frank C. Whitmore, Jr.. discussed 
cetacean systematics. N. de B. Hornibrook and Michael A. Ayress 
helped with biostratigraphy. Jeffrey D. Stilwell provided transla- 
tions. John T. Darby (Otago Museum), Alan N. Baker (National 
Museum of New Zealand), Clayton E. Ray (Smithsonian Institu- 
tion), and Richard H. Tedford (American Museum of Natural His- 
tory) provided access to specimens. Field work was aided by the 
Francis, McKenzie, Parker, Simpson, and Williamson families and 



174 



R. Ewan Fordyce 



others of the Duntroon district. The holotype and other specimens 
were collected and prepared with support from the National Geo- 
graphic Society (grants 4024-88 and 434 1-90; field work and prepa- 
ration), the New Zealand Lottery Research Board (equipment), and 
the Research Committee of the University of Otago (preparation). 

LITERATURE CITED 

Abel. O. 1914. Die Vorfahren der Bartenwale. Denkschriften der 

Akademie der Wissenschaften, Wien. Mathematisch-naturwissen- 

schaftliche Klasse 90:155-224. 
Allen, G. M. 1921. A new fossil cetacean (Archaeodelphis patrius gen. 

etsp. nov.). Bulletin of the Museum of Comparative Zoology 65:1- 

14. 
Barnes, L. G. 1978. A review of Lophocetus and Liolithax and their 

relationships to the delphinoid family Kemriodonhdae (Cetacea: 

Odontoceti). Natural History Museum of Los Angeles County 

Science Bulletin 28: 1-35. 
. 1985. Fossil pontoponid dolphins (Mammalia: Cetacea) from 

the Pacific coast of North America. Natural History Museum of 

Los Angeles County Contributions in Science 363. 
. 1990. The fossil record and evolutionary relationships of the 



genus Tursitips. Pp. 3-26 in S. Leatherwood and R R. Reeves 
(eds.). The Bottlenose Dolphin. Academic Press, San Diego, Cali- 
fornia. 

, D. P. Domning, and C. E. Ray. 1985. Status of studies on fossil 



marine mammals. Marine Mammal Science 1:15-53. 

Benham, W. B. 1935. The teeth of an extinct whale, Microcetus hectori 
n. sp. Transactions of the Royal Society of New Zealand 65:239- 
243. 

. 1937. Fossil Cetacea of New Zealand III. — The skull and other 

parts of the skeleton of Prosqualodon hamiltoni n. sp. Transactions 
of the Royal Society of New Zealand 67:8-14. 

Boenninghaus, G. 1904. Das Ohr des Zahnwales. Zoologische 
Jahrbucher. Abteilung fiir Anatomie und Ontogenie der Tiere 
19:189-360. 

Burlet, H. M. de. 1913. Zur entwicklungsgeschiehte des walschadels. 1. 
Uber das primordialcranium eines embryo von Phocaena 
communis. Gegenbaurs Morphologisches Jahrbuch 45:523-556. 

Cabrera. A. 1 926. Cetaceos fosiles del Museo de la Plata. Revista Museo 
de La Plata 29:363-41 1. 

Cozzuol, M. A. 1484. An alternative interpretation of the evolutionary 
significance of the bony lateral lamina and its implications for 
odontocete systematics. Abstracts of Papers and Posters. Fifth In- 
ternational Theriological Congress, Rome. 22-29 August 1989, 
1:481^482. 

Cozzuol, M. A., and G. Humbert-Lan. 1989. On the systematic position 
of the genus Prosqualodon Lydekker, 1 893, and some comments 
on the odontocete family Squalodontidae. Abstracts of Papers and 
Posters, Fifth International Theriological Congress, Rome, 22-29 
August 1989, 1:483^484. 

Dal Piaz, G. 1917. Gli Odontoceti del Miocene Bellunese III 
Squalodelphis fabianii. Memorie dell'Istituto Geologico della R. 
Universitadi Padova 5:1-34. 

Dal Piaz, G. 1977. Gli Odontoceti del Miocene Bellunese. V-X. 
Cyrtodelphis, Acrodelphis, Protodelphinus, Ziphiodelphis, Scaldi- 
cetus. Memorie dell'Istituto Geologico della R. Universita di 
Padova, Publ. dal Prof. G. Dal Piaz, Allegato al vol. 4 (1916): 
1-128. 

Dickson, M. R. 1964. The skull and other remains of Prosqualodon 
marplesi, a new species of fossil whale. New Zealand Journal of 
Geology and Geophysics 7:626-635. 

Dubrovo, 1. A. 1971. Novyi rod kitoobraznykh (Sachalinocetus 
cholmicus gen. et sp. nov.) izmiotsenao-va Sakhalin [Anew genus 
of cetaceans {Sachalinocetus cholmicus gen. et sp. nov.) from the 
Miocene of the island of Sakhalin], Akademia Nauk SSSR. Trudy 
Paleontologicheskh Institut 130:87-103. 

. and A. A. Sharkov. 1971. Kit iz verkhnego Oligotsena 

Mangyshlaka [A whale from the Upper Oligocene of Mangyshlak], 
Doklady Akademia Nauk SSSR 1 94: 1 403- 1 406. 

, and — . 1972. A whale from the Upper Oligocene of 

Mangyshlak [Translation by American Geological Institute]. 



Doklady Akademia Nauk SSSR 198: 140-143. 

Fleischer. G. 1976. Hearing in extinct cetaceans as determined by co- 
chlear structure. Journal of Paleontology 50: 133-152. 

Flynn, T. T. 1948. Description of Prosqualodon davidi Flynn, a fossil 
cetacean from Tasmania. Transactions of the Zoological Society of 
London 25:153-197. 

Fordyce, R. E. 1981. Systematics of the odontocete Agorophius 
pygmaeus and the family Agorophiidae (Mammalia: Cetacea). 
Journal of Paleontology 55:1028-1045. 

— . 1 982. The fossil vertebrate record of New Zealand. Pp.629-698 
in P. V. Rich and E. M. Thompson (eds.). The Fossil Vertebrate 
Record of Australasia. Monash University Offset Printing Unit, 
Clayton, Victoria, Australia. 

— . 1983. Rhabdosteid dolphins (Mammalia: Cetacea) from the 
middle Miocene, Lake Frome area. South Australia. Alcheringa 7: 
27^40. 

— . 1989. Fossil whales, dolphins and porpoises. Earth Science 42 
(2):20-23. 

— , N. de B. Hornibrook, and P. A. Maxwell. 1985. Field tnp guide 
to Cenozoic geology of North Otago and South Canterbury. Geo- 
logical Society of New Zealand Miscellaneous Publication 33B.I- 
50. 

— . and C. R. Samson. 1992. Late Oligocene platamstoid and 
delphinoid dolphins from the Kokoamu Greensand-Otekaike 
Limestone, Waitaki Valley region, New Zealand: An expanding 
record [abstract]. Geological Society of New Zealand Miscella- 
neous Publication 63A:66. 

Fraser, F. C. and P. E. Purves I960. Hearing in cetaceans: Evolution of 
the accessory air sacs and the structure of the outer and middle ear 
in recent cetaceans. Bulletin of the British Museum (Natural His- 
tory), Zoology 7:1-140. 

Gage, M. 1957. The geology of Waitaki subdivision. New Zealand 
Geological Survey Bulletin 55:1-135. 

Grateloup, J. P. S. 1840. Description d'un fragment de machoire fossile, 
d'un genre nouveau de reptile (saunen), de taille gigantesque, 
voisin de VIguanodon, trouve dans le Gres Marin, a Leognan, pres 
Bordeaux. Actes de l'Academie National des Sciences, Belles- 
lettres et Artes de Bordeaux 2:201-210. 

Heyning, J. E. 1989. Comparative facial anatomy of beaked whales 
(Ziphiidae) and a systematic revision among the families of extant 
Odontoceti. Natural History Museum of Los Angeles County Con- 
tributions in Science 405:1-64. 

— , and J. G. Mead. 1990. Evolution of the nasal anatomy of 
cetaceans. Pp. 67-79 in J. Thomas and R. Kastelein (eds.). Sensory 
Abilities of Cetaceans. Plenum. New York. New York. 

Homibrook. N. de B. 1966. The stratigraphy of Landon (or Boundary) 
Creek, Oamaru. New Zealand Journal of Geology and Geophysics 
9:458-470. 

— . R. C. Brazier, and C. P. Strong. 1989. Manual of New Zealand 
Permian to Pleistocene foraminiferal biostraligraphy. New Zealand 
Geological Survey Paleontological Bulletin 56:1-175. 

Howell. A. B. 1927. Contribution to the anatomy of the Chinese finless 
porpoise Neomeris phocaenoides. Proceedings of the United States 
National Museum 70 (13):1— 43. 

Kasuya, T. 1973. Systematic consideration of recent toothed whales 
based on the morphology of tympanopenotic bone. Scientific Re- 
ports of the Whales Research Institute, Tokyo 25:1-103. 

Kellogg, A. R. 1923a. Description of two squalodonts recently discov- 
ered in the Calvert Cliffs. Maryland; and notes on the shark-toothed 
dolphins. Proceedings of the United States National Museum 62 
(6): 1-69. 

. 1923b. Description of an apparently new toothed cetacean from 

South Carolina. Smithsonian Miscellaneous Collections 76 (7): 1-7. 
— . 1924. A fossil porpoise from the Calvert Formation of Mary- 
land. Proceedings of the United States National Museum 63 (14): 
1-39. 

1926. Supplementary observations on the skull of the fossil 



porpoise Zarhachis flagellator. Proceedings of the United States 

National Museum 67 (28):1-18. 
. 1927. Kentriodon pernix, a Miocene porpoise from Maryland. 

Proceedings of the United States National Museum 69 ( 19): 1-55. 
— . 1928. History of whales — their adaptation to life in the water. 

Quarterly Review of Biology 3:29-76, 174-208. 



Waipatia maerewhenua. New Genus and New Speeies, an Archaic Late Oligocene Dolphin from New Zealand 



175 



. 1936. A review of the Archaeoceti. Carnegie Institute of Wash- 
ington Publication 482. 

— . 1957. Two additional Miocene porpoises from the Calvert 
Cliffs, Maryland. Proceedings of the United States National Mu- 
seum 107 (3387):279-337. 

Lydekker, R. 1 894. Contributions to a knowledge of the fossil verte- 
brates of Argentina. Part 11. Cetacean skulls from Patagonia. 
Annates del Museo de La Plata 1893. 2:1-14. 

Mchedlidze, G. A. 1970. Nekotorye Obshie Cherty Istorii Kitoobraz- 
nykh (Some general characteristics of the evolution of cetaceans. 
Part 1]. Metsniereba. Tbilisi. Georgia. 

— . 1984. General Features of the Paleobiologies Evolution of 
Cetaeea [translation of 1976 title]. Amennd. New Delhi, India. 

Mead, J. G. 1975. Anatomy of the external nasal passages and facial 
complex in the Delphinidae (Mammalia: Cetaeea). Smithsonian 
Contributions to Zoology 207:1-72. 

Moreno, F. P. 1892. Lijeros apuntes sobre dos generos de cetaceos 
fosiles de la Repiiblica Argentina. Revista Museo de La Plata 
3:381^*00. 

Muizon. C. de. 1987. The affinities of Notocetus vanbenedeni. an Early 
Miocene platanistoid (Cetaeea, Mammalia) from Patagonia, south- 
ern Argentina. American Museum Novitates 2904. 

. 1988a. Le polyphyletisme des Acrodelphidae, Odontocetes 

longirostres du Miocene europeen. Bulletin du Museum Nationale 
d'Histoire Naturelle, Section C, 4eme serie, 10 (l):31-88. 

. 1988b. Les relations phylogenetiques des Delphinida (Cetaeea. 

Mammalia). Annales de Paleontologie 74:159-227. 

. 1991. A new Ziphiidae (Cetaeea) from the early Miocene of 

Washington State (USA) and phylogenetic analysis of the major 
groups of odontocetes. Bulletin du Museum National d'Histoire 
Naturelle, Section C, 4eme serie. 12 (3^0:279-326. 

Pernn, W. F. 1975. Variation of spotted and spinner porpoise (genus 
Stenella) in the eastern tropical Pacific and Hawaii. Bulletin of the 
Scripps Institution of Oceanography 21:1-206. 

Pilleri, G. 1986. Beobachtungen an den Fossilen Cetaceen des 
Kaukasus. Hirnanatomisches Institut. Ostermundigen, Switzerland. 

Pompeckj, J. F. 1922. Das Ohrskelett von Zeuglodon. Senckenbergiana 
4 (3, 4):43-100. 

Rothausen, K. 1961. Uber Microcetus, einen kleinen Squalodontiden 
aus dem Oberoligozan. Neues Jahrbuch fiir Geologie und 
Palaontologie Abhandlungen 1 12:106-1 18. 

. 1 968. Die systematische Stellung der europaischen Squalodon- 

tidae (Odontoceti: Mamm.). Palaontologische Zeitschrift 42:83- 
104. 

. 1970. Marine Reptilia and Mammalia and the problem of the 

Oligocene-Miocene boundary. Giornale di Geologia (series 2) 
35:181-189. 

Schenkkan. E. J. 1973. On the comparative anatomy and function of the 
nasal tract in odontocetes (Mammalia. Cetaeea). Bijdragen Tot de 
Dierkunde43:127-159. 

Schulte, H. von, and M. de F. Smith. 1918. The external characters, 
skeletal muscles, and peripheral nerves of Kogia breviceps 
(Blainville). Bulletin of the American Museum of Natural History 
38:7-72. 

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. 

Swofford, D. L. 1993. PAUP: Phylogenetic analysis using parsimony, 
version 3.1.1. Computer program distributed by Illinois Natural 
History Survey, Champaign, Illinois. 

, and Begle, D. P. 1993. PAUP: Phylogenetic analysis using 

parsimony, version 3.1.1. User's manual. Laboratory of Molecular 
Systematics, Smithsonian Institution/Illinois Natural History Sur- 
vey. Champaign, Illinois. 

True, F. W. 1909. A new genus of fossil cetaceans from Santa Cruz 
territory, Patagonia; and description of a mandible and vertebrae of 
Prosqualodon. Smithsonian Miscellaneous Collections 52:441- 
456. 

. 1910a. Description of a skull and some vertebrae of the fossil 

cetacean Diochoticus vanbenedeni from Santa Cruz, Patagonia. 
Bulletin of the American Museum of Natural History 28:19-32. 

. 1910b. An account of the beaked whales of the family Ziphiidae 

in the collection of the United States National Museum, with 



remarks on some specimens in other American museums. United 
States National Museum Bulletin 73:1-89. 
Whitmore, F. C, and A. E. Sanders. 1977. Review of the Oligocene 
Cetaeea. Systematic Zoology 25 (for 1976):304-320. 

APPENDIX: CHARACTERS USED IN CLADISTIC 

ANALYSIS OF THE RELATIONSHIPS 

OF WAIPATIA MAEREWHENUA 

These characters are discussed in the text and/or by Barnes 
(1990), Heyning (1989). and Muizon (1987, 1988a, 1988b, 1991). 

0, primitive; 1-3. derived. 

1. Posterior lacerate foramen confluent with foramen ovale to 
form "cranial hiatus": 0, yes; 1, no, parietal and/or squamosal 
contact basioccipital to separate posterior lacerate foramen from 
foramen ovale. 

2. Foramen "pseudo-ovale": 0, present; 1, absent. The foramen 
"pseudo-ovale" marks the exit of the mandibular branch of the 
trigeminal nerve from the region of the pterygoid sinus fossa. The 
foramen is bounded by the pterygoid and falciform processes of the 
squamosal and normally by the ossified outer lamina of the ptery- 
goid. Present in Archaeoceti, Mysticeti. and those Odontoceti (e.g., 
Platanista) in which an extensive ossified outer lamina of the 
pterygoid contacts the falciform process. The palatine may contact 
the falciform process in some Delphinoidea. 

3. Overlap of maxilla onto frontal in supraorbital region: 0, no 
overlap; 1. partial overlap; 2. supraorbital process of maxilla ex- 
tends posterior to mid-orbit. 

4. Form of anterior bullar facet of periotic: 0, facet flat or 
absent; 1, facet depressed with shallow groove; 2, facet depressed 
with deep groove. 

5. Mandibles fused at symphysis: 0, no; 1, yes. 

6. Depth of pterygoid sinus fossa in basicranium: 0, shallow or 
little excavated; 1, deep, excavated dorsal to level of foramen ovale; 
2, deep and extended dorsally toward or into orbit. Functional 
reasons for apparent reversals from state 1 to or 2 to 1 are 
uncertain; irreversibility seems likely. 

7. Maxilla present in anterior wall or in floor of orbit: 0, yes; 1, 
no. 

8. Position and orientation of origin for temporal muscle on 
supraorbital process of frontal: 0. origin lies on the posterior face of 
the supraorbital process and is directed roughly posteriorly; 1, 
origin lies on posteroventral face of supraorbital process and is 
directed roughly ventrally. 

9. Ossified lateral lamina of pterygoid present and in contact 
with falciform process: 0. yes; 1, ossified lamina reduced or absent. 

10. External auditory meatus: 0, wide; 1, narrow. 

11. Contact of enlarged posterior process of bulla with 
paroccipital: 0, no contact; I , sutural contact. 

12. Accessory ossicle of periotic: 0, small to medium, not well 
fused; 1. enlarged, subspherical, and fused tightly to periotic. 

13. Blowhole ligament present: 0, no; 1. yes. Not known for 
fossils. In extant Mysticeti. not included in this analysis, the blow- 
hole ligament is absent; its absence in the Physeteroidea is probably 
primitive, rather than a result of reversal. Heyning ( 1989) discussed 
the soft anatomy of the face (e.g., characters 13-17). 

14. Nasal passage — distal sac developed: 0, no; 1, yes. Derived 
for the Physeteroidea. 

15. Nasal passage — proximal sac evolves into frontal sac: 0, no; 

1, yes. 



176 



R. Ewan Fordyce 



16. Nasal passage — proximal sac evolves into sac complex: 0, 
no; 1 , yes. Regarded as derived for extant Odontoceti other than the 
Physeteroidea; the absence in Physeteroidea is probably primitive, 
rather than a result of reversal as indicated in Figure 15. 

17. Spermaceti organ present: 0, no; 1, yes. 

18. Supracranial basin in skull: 0, absent; 1, present. 

19. Number of nasals: 0. two; 1, one or both lost. 

20. Nasal passages confluent distal to bony nares: 0, no; 1, yes. 
In extant Mysticeti. the nasal passages are separate distal to the 
bony nares; separation in the Physeteroidea is probably primitive, 
rather than a result of reversal. 

21. Anterior process of periotic: 0. not thickened tranversely; 1. 
thickened tranversely by expanded internal and external faces at 
some point beyond the base of the process. 

22. Fenestra rotundum of periotic reniform, with a dorsal fissure 
directed toward the aperture for the cochlear aqueduct: 0, no; 1, yes. 

23. Premaxilla with a transversely flattened vertical face and 
prominent lateral crest at the level of the nares: 0, no; 1, yes. 

24. Enlarged dorsal lamina of pterygoid tightly fused with 
alisphenoid anterior to foramen ovale: 0, no; 1, yes. 

25. Profile of anterior process of periotic ventrally deflected in 
lateral view: 0, no. has crudely rectangular profile; 1, smoothly 
deflected; 2. abruptly deflected. 

26. Periotic - parabullary ridge developed laterally along ventral 
border of anterior process: 0, ridge absent; 1, ridge present. 

27. Long posterior apex of premaxilla lies posterior to nasals 
wedged between elevated edge of maxilla and frontal on vertex; 
apices show left-right asymmetry: 0, no; 1, yes. 

28. Cochlear aqueduct on periotic large with a thin edge: 0. no; 
1, yes. 

29. Articulation of posterior process of tympanic bulla with 
squamosal: 0, process contacts post-tympanic process of squamosal 
and posterior process of periotic: 1 . bulla contacts periotic only. 

30. Frontal excavated for orbital extensions of pterygoid 
sinus(es): 0. not excavated; 1, slightly excavated with shallow- 
edged depression; 2, deeply excavated. 

31. Nasal passage — vestibular sac: 0. absent; 1. present; 2. 
hypertrophied. 

32. Palatine invaded by or modified by pterygoid sinus fossa: 0, 
no: 1 . yes. The palatine is progressively narrowed to ventral view 
between maxilla and pterygoid as the pterygoid sinus fossa invades 
the palatine. 

33. Articular process on periotic: 0. process absent; 1, incipient 
ridge present; 2, strong ridge present; 3, peg present. 

34. Apex of pterygoid hamulus solid, robust, long and 
subconical in ventral view: 0, no; 1, yes. 

35. Lateral groove or lateral depression affects profile of periotic 
as viewed dorsally: 0, no obvious vertical groove dorsal to hiatus 
epitympanicus; 1, groove present so that overall profile of periotic 
is slightly to markedly sigmoid in dorsal view. 

36. Rostral suture between premaxilla and maxilla deeply 
grooved: 0, no; 1, yes. 

37. Dorsal edge of posterior process of periotic spongy and 
fused or tightly articulated with adjacent squamosal: 0, no; 1, yes. 

38. Dorsal surface of involucrum of bulla markedly depressed 
or excavated anterior to the base of the posterior process, so that the 
involucrum has parallel dorsal and ventral profiles in medial view: 
0, no; 1, yes. 

39. Palatine with ossified lateral lamina directed posterolater- 
ally from about the level of the choanae: 0, no; 1, yes. 



40. Anterior bullar facet lost from periotic: 0, no; 1, yes. 

41. Relationship of ascending process of premaxilla with nasal: 
0, left and right processes extend posteriorly beyond anterior of 
nasals; 1, processes contact only front of nasal; 2. one or no process 
contacts nasal. 

42. Incisors relatively delicate and procumbent: 0, no; I , yes. 

43. Nasal passage — posterior sac lost: 0, no; 1, yes. 

44. Pterygoid sinus fossa present in alisphenoid and/or basioc- 
cipital, dorsolateral to basioccipital crest and posteromedial to fora- 
men ovale: 0. no; 1, yes. 

45. Anterior spine present on bulla: 0, no; 1. spine small to 
moderate; 2. spine long. 

46. Bulla with inflated anterolateral convexity that may be 
associated with an anterolateral notch: 0, no; 1, yes. 

47. Anterior process of periotic roughly cylindrical in cross 
section: 0, no; 1, yes. 

48. Scapula — acromion process lies on anterior edge, with loss 
of supraspinous fossa: 0, no; 1, yes. 

49. Scapula — coracoid process: 0. present; 1, absent. 

50. Periotic with low, wide, and regularly convex transverse 
profile across dorsal surface (= across dorsal process, sensu 
Muizon): 0. no; 1, yes. 

51. Bullar facet on posterior process of periotic extends dorsally 
onto the posteromedial face of the posterior process: 0, no; 1, yes. 

52. Posterior portion of rostrum robust and deep, with open and 
deep mesorostral groove: 0, no; 1, yes. 

53. Apex of anterior process of periotic tuberculate: 0, no; 1, 
prominent small tubercule present. 

54. Anterior process of periotic with multiple subvertical fine 
fissures on the internal face: 0, no; 1, yes. 

55. Anteroposterior ridge on dorsal side of anterior process and 
body of periotic, associated with the development of a depression 
adjacent to groove for tensor tympani: 0. absent; 1, present. 

56. Anteroexternal sulcus profile on periotic recurved so that it 
is concave dorsally (seen in external view): 0, no; 1, yes. 

57. Foramen spinosum enlarged to form a subcircular fossa 
dorsal to periotic: 0, no; 1, yes. 

58. Crown of heterodont teeth: 0, long (>10 mm); 1, short (<10 
mm). 

59. Bulla — ventral groove: 0, groove not marked anteriorly; 1, 
groove present anteriorly (shallow or deep, may include anterior 
spine). 

60. Atlas vertebra — relative size of dorsal transverse process: 0, 
moderate; 1, large. 

6 1 . Pars cochlearis of periotic inflated with subrectangular pro- 
file: 0, no; 1, yes. 

62. Posterior maxillary (infraorbital) foramen placed 
posteromedially. near the bifurcation in the posterior of the premax- 
illa: 0, no; 1, yes. 

63. Facet for bulla on posterior process of periotic relatively 
narrow, long, and parallel-sided: 0, no; 1, yes. 

64. Posterior margin of maxilla elevated, with "squared off 
profile as viewed dorsally: 0, no; 1, yes. 

65. Nodular frontals prominent on vertex, separated by a promi- 
nent medial groove: 0, no; 1. yes. 

66. Ridge or crest of maxilla/frontal, pneumatized ventrally. 
present along lateral margin of face above orbit: 0. no; 1. yes. 

67. Bulla with thin outer lip that is smoothly overarching and 
high relative to transverse width of bulla: 0, no; 1 . yes. 



A Phylogenetic Analysis of the Sirenia 

Daryl P. Domning 

Laboratory of Paleobiology, Department of Anatomy, Howard University. Washington, D.C. 20059 

ABSTRACT. — Analysis of 62 crania] and denial characters of 36 species and subspecies of sirenians, by means of the Hennig86 computer 
program without character weighting, produced 60 maximally parsimonious trees (length 152, consistency index 0.55, retention index 0.83). With 
successive character weighting, these were reduced to six maximally parsimonious trees, of which the Nelson consensus tree is presented here 
(length 162, consistency index 0.76, retention index 0.91). Sample size and intrapopulational variation are insufficiently studied problems in 
cladistic analysis, and a statistically based method for scoring variable characters is introduced. The tree's topology is least certain in three groups of 
taxa: Eocene dugongids, dugongines (here including rytiodontines), and species of Metaxytherium. The most novel results of this study: ( 1 ) The 
Miosireninae are the sister group of the Tnchechidae as previously defined, and are here placed in that family; a subfamily Trichechinae is formally 
erected for the remaining trichechids. (2) The Tnchechidae in this broader sense appear to have arisen somewhat later than previously supposed (late 
Eocene or early Oligocene rather than middle Eocene) and are rooted well within the Dugongidae instead of being derived separately from the 
Protosirenidae. (3) Dugong lies within the clade heretofore called the Rytiodontinae, on the basis of the first strong evidence of where among the 
Dugongidae the living dugong's phyletic affinities lie. The name Dugonginae is extended to this entire clade in place of the junior name 
Rytiodontinae. Except within the Dugonginae, age rank and clade rank are highly correlated, suggesting (hat the fossil record provides a good picture 
of the history of the Sirenia. A revised provisional classification is proposed for the sirenian taxa analyzed here. 



INTRODUCTION 

The first formal cladogram of the order Sirenia to be published 
was that of Savage ( 1977). Since then, cladistic analyses have been 
presented for several subsets of the order: the Tnchechidae 
(Domning and Hayek 1986), the Rytiodontinae (Domning 1989a.b, 
1990), and the European species of Metaxytherium (Domning and 
Thomas 1987). In this paper I revise and extend this previous work 
to encompass all of the better-known Sirenia. 

This study has been done in the context of much recent work 
that strongly supports the strict monophyly ( = holophyly) of the 
order Sirenia and its membership in a supraordinal group 
(Tethytheria) with the Proboscidea and Desmostylia (e.g.. Domning 
et al. 1986; Shoshani 1986; Tassy and Shoshani 1988; Novacek 
1990; Thewissen and Domning 1992; and references cited therein). 
Although a few characters of the order gleaned from these studies 
are noted here. I do not review this body of work in detail or attempt 
to identify the sister group of the Sirenia but instead refer the reader 
to these sources for evidence on the relationships of sirenians to 
other mammals. 

This paper is a preliminary report, based on a systematic revi- 
sion still in progress. 

MATERIALS AND METHODS 

Thirty-six species and subspecies of sirenians were analyzed. 
Several other nominal species were excluded because they are 
known only from very incomplete material, because I have not 
examined the original specimens, and/or because I have serious 
doubts about their validity. For example, Thalattosiren petersi 
(Abel, 1904) was excluded because I suspect that the known skulls 
may represent merely immature Metaxytherium. 

Moeritheriwn (Proboscidea) and Paleoparadoxia (Desmo- 
stylia) were used as outgroups for polarization of characters be- 
cause of the evidence (cited above) that these two orders are the 
closest relatives of the Sirenia and because these genera are the 
most primitive adequately known members of their respective or- 
ders. However, both of these are apparently derived, relative to 
other mammals, in their imperforate lacrimals and single-rooted 
canines, whereas early sirenians display the primitive states (pos- 
session of a lacrimal foramen and double-rooted canines, respec- 
tively). More primitive proboscideans and desmostylians are known 
(anthracobunids and Behemotops, respectively; see Ray, Domning, 
and McKenna 1994, this volume) but are represented at present by 



little or no cranial material and cannot be scored for most of the 
characters used here. 

This analysis is based on some 108 morphological characters of 
the skull, mandible, and dentition (excluding cheek-tooth cusp 
patterns) that I have examined in detail in almost all of the known 
taxa of fossil and living sirenians. Of these 108 characters. I elimi- 
nated 46 that I was unable to score consistently or that were cladis- 
tically uninformative for the taxa included here (e.g., because they 
vary only in taxa that were excluded). The 62 informative charac- 
ters (Table 1) were analyzed with the Hennig86 computer program 
(Farris 1988). Three multistate characters were treated as unordered 
because in these cases I had significant doubts that the states formed 
a single transformation series. Some other significant cranial and 
postcranial characters not used in the analysis corroborate and 
supplement certain parts of it. 

Two aspects of cladistic data sets that are normally ignored are 
explicitly addressed here: sample size and intraspeciftc variation. 
Table 1 lists for each taxon the largest number of specimens exam- 
ined for which any character could be scored. For any given charac- 
ter, the actual number of specimens scored was often much less than 
this maximum; however, separate citation of a sample size for each 
character of each taxon [as Domning and Thomas ( 1987) did for a 
much smaller data set] would have made the table prohibitively 
large and cumbersome. The present compromise at least provides 
an approximation of the sample sizes available for this study. As for 
variation, since Hennig86 does not accept multiple states of a 
character for a given taxon, polymorphisms had to be scored unam- 
biguously as one of two states. The following procedure was 
adopted. 

For the available samples, confidence limits for proportions and 
critical values of sample fractions (Xln = frequency of a state in a 
sample of size n) were determined (these are given in graphic or 
tabular form in standard statistics tables), using a confidence coeffi- 
cient of 0.95. For example, if four specimens in a sample of five 
display a derived state. Xln = 4/5 = 0.8. The probability that the 
frequency of occurrence of the derived state in the sampled popula- 
tion was between 0.995 and 0.284 (the 95% confidence limits) is 
0.95. (I here designate the lower confidence limit. 0.284 in this 
example, as the LCLyv) 

If a state (either primitive or derived) was present in the major- 
ity of the sample and its LCL,,, > 0.5, the taxon was scored as 
having that state. If the majority state had an LCL Q5 0.5. the 
scoring depended on the taxon's position relative to the character's 
distribution in the trees obtained from preliminary analyses: 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. 29:177-189, 1994 



178 



Daryl P. Domning 



taxon was scored whichever way was more congruent with other 
characters (i.e., whichever way did not imply a reversal). If the 
sample was evenly divided, it was likewise scored as having the 
congruent state. If the taxon lay at the borderline of the character 
transformation (so that neither scoring choice would imply a rever- 
sal), it was scored as having the majority state. If it was both located 
at the borderline and evenly divided, it was scored as having the 
more primitive state (this is arbitrary; the opposite rule was tried 
also, but in this analysis the choice did not affect the geometry of 
the final tree). The rationale and implications of this procedure are 
discussed below (see under Comments on Methods). 

CHARACTERS USED 

The following 62 characters are those that have proven most 
informative and were used in the computer analysis, out of 108 
characters (assigned numbers between 1 and 158) that I have stud- 
ied in some detail. For simplicity of record-keeping, the numbers 
originally assigned to these characters are retained here. Numbers 
are not assigned to some other cranial characters whose only effect 
would be to define terminal taxa or strengthen nodes already ad- 
equately supported, as in the case of the hydrodamalines; these 
characters would therefore have no effect on the geometry of the 
final tree, though they would alter the tree's statistics. Likewise 
unnumbered are postcranial characters, because data on these are 
missing for many taxa. None of these unnumbered characters was 
included in the computer analysis, though they are listed below at 
the appropriate nodes. The data matrix for the 62 included charac- 
ters is shown in Table 1 . As usual, designates the most primitive 
state observed among the taxa studied. 

3. Rostrum: (0) small relative to cranium; ( 1) enlarged (length of 
premaxillary symphysis > 0.27 x condylobasal skull length) (see 
Fig. 1). (The ratio 0.27. like other ratios used below, was chosen 
because it separates what appear visually to be significantly differ- 
ent character states.) 

6. Nasal process ofpremaxilla: (0) thin and tapering at posterior 
end, having lengthy overlap with frontal and/or nasal; (1| broad- 
ened and bulbous at posterior end, having more or less vertical joint 
surface in contact with frontal (Domning 1989a,b). 

7 '. Nasal process ofpremaxilla: (0) long; (I) very short (see 
Fig. 1 ). 

8. External nares: (0) not retracted; ( I ) retracted and enlarged, 
reaching to or beyond the level of the anterior margin of the orbit. 

9. Premaxilla: (0) does not contact frontal; ( 1 ) contacts frontal. 
1 1. Zygomaticoorbital bridge of maxilla: (0) nearly level with 

palate; ( 1 ) elevated above palate, with its ventral surface lying > 1 
cm above the alveolar margin (cf. Domning 1978: fig. 8). 

13. Infraorbital foramen: (0) small (about 15x10 mm or less); 
( 1 ) large (greater than 15x10 mm). 

14. Zygomatic— orbital bridge of maxilla: (0) long antero- 
posterior^ (vertical thickness < 0.40 x minimum length); ( 1 ) short- 
ened (thickness 0.40 x length; cf. Domning 1978: fig. 24); (2) 
shortened and transformed into transverse vertical wall (Domning 
1989b). 

16. Palate: (0) thin or incomplete at level of penultimate cheek 
tooth; (1) > 1 cm thick at level of penultimate tooth. 

31. Nasals: (0) meet in midline; (1) separated in midline by 
frontals or an incisure, or absent. 

32. Nasals: (0) large (length of internasal suture 0.5 x length 
of interfrontal suture exposed dorsally ); ( 1 ) smaller, or separated in 
midline, or absent. 

36. Supraorbital process of frontal: (0) well developed, with 
prominent, dorsoventrally flattened posterolateral corner; ( 1 ) dor- 
soventrally thickened, with posterolateral corner variably devel- 
oped; (2) reduced, rounded, lacking posterolateral corner (see 




Figure 1. Skulls of sirenians in right lateral view, illustrating eight of the 
characters of the anterior part of the skull used in this analysis. Not drawn to 
same scale. See text for explanations of characters and states. Dashed lines 
indicate parts restored; dotted lines outline tusks within alveoli. Abbrevia- 
tions: f. frontal; j, jugal; p. premaxilla; s. zygomatic process of squamosal. 
A, Trichechus senegalensis: 3(0). 7(0), 36(0), 43(0). 85(0), 89(0), 139(1), 
140(0). B. Halitherium schinzii: 3(1). 7(0), 36(0). 43(0). 85(1), 89(0). 
1 39(0), 140( 1 ). C, Dioplotherium manigaulti: 3( 1 ). 7(0). 36(— or 1 ), 43( 1 ). 
85(2). 89(1), 139(0), 140(2). D, Rytiodus sp.: 3(1), 7(1), 36( — or I), 43(1), 
85(2), 89(0), 139(0). 140(2). E, Metaxytherium floridanum: 3(1), 7(0), 
36(1). 43(0). 85(2), 89(0), 139(0), 140(0). 



Fig. 1 ; state 2 not illustrated). (This character was treated as inappli- 
cable to "rytiodontines" because these follow a somewhat different 
transformation series, here expressed by character 43; however, it 



A Phylogenetic Analysis of Ihe Sirenia 



179 



would probably be equally correct, and would not alter the tree's 
topology, if all of these taxa were scored I for this character.) 

37. Nasal incisure at posterior end of mesorostral fossa: (0) 
absent or small (does not extend posterior to the supraorbital pro- 
cess); (I) deep and narrow (extends posterior to the supraorbital 
process); (2) comparably deep but broad, with the anterior frontal 
margin displaying a median convexity. (Unordered character.) 

38. Lamina orbitalis of frontal: (0) thin or absent; (1) 1 cm 
thick. 

42. Frontal roof. (0) convex, or more or less flat between 
temporal crests (if latter present); (1) deeply concave, sloping 
steadily ventrad to anterior margin (cf. Domning 1 990: fig. 4E). 

43. Supraorbital process of frontal: (0) flattened in more or less 
horizontal plane, with dorsal surface inclined relatively gently 
ventrolaterad; ( 1 ) turned markedly downward, with dorsal surface 
inclined strongly ventrolaterad and posterolateral corner projecting 
posteriorly (see Fig. 1; Domning 1989a,b, 1990). 

5 1 . Sagittal crest: (0) present; ( 1 ) absent. 

66. Exoccipitals: (0) meet in a suture dorsal to foramen mag- 
num; ( 1 ) do not meet in a suture (this is a reversal to the condition 
found in primitive mammals; Shoshani 1986). 

67. Supracondylar fossa of exoccipital: (0) absent; (1) distinct 
but shallow, directly dorsal to condyle; (2) deep and extending 
across entire width of occipital condyle; (3) reduced and located 
dorsomedial to condyle, or lost. 

70. Dorsolateral border of exoccipital: (0) rounded and more or 
less smooth, not flangelike; ( 1 ) thick and overhanging posteriorly 
as a flange; (2) greatly thickened, forming rugose overhanging 
flange (Domning 1978; Domning and Hayek 1986). 

73. Posttympanic process of squamosal: (0) absent (i.e., no facet 
projecting for sternomastoid muscle); ( 1 ) present; (2) enlarged and 
clublike. 

74. Sigmoid ridge of squamosal: (0) present and prominent; ( 1 ) 
reduced or absent (cf. Domning 1978: fig. 7). 

75. External auditory meatus of squamosal: (0) long 
mediolaterally (> 1 cm); (1) short ( 1 cm). 

76. Squamosal: (0) does not extend to temporal crest; ( 1 ) ex- 
tends to temporal crest. 

77. Processus retroversus of squamosal: (0) absent; ( 1 ) present, 
moderately inflected; (2) present, not inflected (cf. Domning 1978: 
fig. 7). In Dugong dugon, it is strongly inflected (an autapomorphy ). 
(Unordered character.) 

82. External auditory meatus of squamosal: (0) narrow and 
slitlike (anteroposterior breadth less than dorsoventral); ( 1 ) about as 
wide anteroposteriorly as high; (2) very broad and shallow, wider 
anteroposteriorly than high. 

84. Zygomatic process of squamosal: (0) medial side not swol- 
len, appears relatively flat or concave and inclined inward dorsally; 
(1) medial side markedly swollen, inclined inward ventrally or 
forming a vertical wall (Domning and Hayek 1986). 

85. Ventral extremity ofjugal: (0) lies posterior to orbit; ( 1 ) lies 
approximately under posterior edge of orbit, but forward of jugal's 
postorbital process (if present); (2) lies ventral to orbit (see Fig. 1). 

87. Preorbital process ofjugal: (0) does not contact premaxilla; 
( 1 ) contacts premaxilla. 

88. Preorbital process of jugal: (0) relatively flat and thin 
(posteromedial-anterolateral breadth of portion lateral to 
maxillojugal suture > anteromedial-posterolateral thickness); (1) 
thick and robust (breadth thickness). 

89. Posterior (zygomatic) process of jugal: (0) as long as or 
longer than diameter of orbit; ( 1 ) shorter than diameter of orbit (see 
Fig. 1 ). 

91. Lacrimal: (0) with foramen (nasolacrimal canal); ( I ) with- 
out foramen, but still large; (2) vestigial or absent. 

97. Posterior border of palatine: (0) not incised, merely shal- 



lowly concave; (1) incised or deeply indented; (2) very deeply 
incised, to as far forward as level of M 1 . 

99. Palatines: (0) extend anteriorly beyond posterior edge of 
zygomaticoorbital bridge; ( I ) do not extend so far forward. 

101. Alisphenoid canal: (0) present: (I) absent. (Though this 
polarity is debatable in mammals generally, it is well supported for 
the Paenungulata, including the taxa considered here; Thewissen 
and Domning 1992.) 

102. Pterygoid fossa: (0) absent; (1) present. (The polarity of 
this character is problematical, in view of the fossa's evident pres- 
ence in Prorastomus but absence in Paleoparadoxia and Protosiren. 
This character is also scored in Moeritherium, but this is appar- 
ently variable, as the fossa is present in one specimen but absent in 
another; J. Shoshani and J. G. M. Thewissen, pers. comm.) 

103. Foramen ovale: (0) enclosed by bone: ( I ) opened to form a 
notch or incisure (this is a reversal to the condition found in primi- 
tive mammals; Novacek 1990). 

1 15. Periotic: (0) fused to alisphenoid; (1 ) not fused with any 
other skull bone, set in closely fitting socket in squamosal. 

121. Mandibular symphysis: (0) laterally compressed, with nar- 
row masticating surface scarcely wider than the two rows of tooth 
alveoli it bears; ( 1 ) broad. 

122. Ventral border of horizontal mandibular ramus: (0) straight 
or only slightly concave; (1) moderately concave, sharply 
downtumed anteriorly: (2) moderately and evenly concave; (3) 
strongly concave (see Fig. 2). 

123. Accessory mental foramina: (0) present, in addition to and 
usually posterior to the large principal foramen: (1) absent (see 
Fig. 2). 

125. Posterior border of mandible: (0) descends ventrally or 
posteroventrally from condyle without marked interruption or 
abrupt change of direction; ( 1 ) bears a steplike process (processus 
angularis superior) below condyle; (2) has no distinct processus 
angularis superior but does have broadly convex outline beginning 
well below condyle (see Fig. 2). 

126. Anterior border of coronoid process: (0) approximately 
vertical; ( I ) extends slightly anterior to base of process; (2) extends 
very far anterior to base (see Fig. 2). 

127. Mandibular dental capsule: (0) completely enclosed by 
bone of mandible; ( 1 ) exposed posteroventrally; (2) absent. 

128. Horizontal ramus of mandible: (0) slender (minimum dor- 
soventral height < 0.25 x length of mandible); ( 1 ) broad dorsoven- 
trally (height 0.25 x length of mandible) (see Fig. 2). 

129. Ventral border of horizontal ramus of mandible: (0) tan- 
gent to angle; ( 1 ) not tangent to angle (see Fig. 2). 

136. First upper incisor: (0) with enamel on all sides, forming 
complete enamel crown; ( 1 ) with enamel mainly on lateral side. 

137. First upper incisor: (0) enamel crown distinct from root; 
( 1 ) enamel extends entire length of tusk. 

138. First upper incisor: (0) not strongly curved; (1) strongly 
curved in parasagittal plane. (Polarity uncertain.) 

1 39. First upper incisor: (0) present; ( 1 ) vestigial or absent (see 
Fig. 1). 

140. Depth of I' alveolus: (0) much less than half the length of 
the premaxillary symphysis; ( 1 ) about half the length of the sym- 
physis; (2) much greater than half the length of the symphysis (see 
Fig. 1 ). 

141. Cross section of l' crown: (0) suboval or subelliptical; (1) 
lens-shaped, with sharp anterior and posterior edges; (2) lozenge- 
shaped (Domning 1978: fig. 3B; 1989a: fig. 4A); (3) broad and 
extremely flattened mediolaterally (Domning 1990: fig. 4). (Unor- 
dered character.) 

142. First upper incisor: (0) with enamel on all sides, forming 
complete enamel crown; ( I ) with enamel mainly on medial side. 

143. Second and third upper incisors, first through third lower 



180 



Daryl P. Domning 




Figure 2. Right mandibles of sirenians in lateral view, illustrating six of the characters used in this analysis. Not drawn to same scale. See text for 
explanations of characters and states. A. Prorastomus sirenoides: 122(0). 123(0), 125(0). 126(0), 128(0), 129(0). B, Prototherium veronense: 122(2), 
123(0), 125(1), 126(0), 128(0), 129(0). C, Trichechus senegalensis: 122(2), 123(0), 125(2), 126(2), 128(0). 129(1). D, Haliiherium schinzii: 122(1), 123(0), 
125(2), 126(1), 128(0), 129(1). E, Metaxytheriumfloridanum: 122(3), 123(1), 125(2), 126(1), 128(1), 129(1). 



incisors: (0) present, at least in part; ( 1 ) all absent. 

144. Canines: (0) double-rooted; ( 1 ) single-rooted; (2) absent. 

146. Fifth permanent premolars: (0) present; ( 1 ) absent; i.e., no 
replacement occurs at P 5 and P 5 loci. 

150. Supernumerary molars: (0) absent; ( 1 ) present and replen- 
ished indefinitely by horizontal replacement (Domning 1982). 

151. Functional cheek teeth: (0) present in adult; ( I ) present in 
juvenile only; (2) absent (Domning 1978; Domning and Demere 
1984). 

155. Postcanine dental formula: (0) Pl-4. Ml-3; (1) Pl-5. 
Ml-3, or secondarily reduced from this condition by loss of ante- 
rior premolars. [It is still unresolved whether the five premolars of 
early sirenians are a synapomorphy of the order, as assumed here, 
or a retention of a primitive placental trait. However, I still lean 
toward the latter opinion, as expressed in Domning et al. (1982, 
1986). In any case the decision would not affect the analysis within 
the Sirenia since five premolars are clearly primitive for the order. 
See Thewissen and Domning (1992) for further discussion.] 

156. Cheek-tooth enamel: (0) smooth; ( I ) wrinkled. 

157. Permanent premolars: (0) some double- or triple-rooted; 
(1) all single-rooted; (2) all absent. 

158. Molars: (0) unreduced; (1) conspicuously reduced in size 
relative to skull and mandible, without loss of total occlusal area [as 
a result of increased number of molars (Domning 1982); however, 
character state 1 50( 1 ) also occurs in the absence of this one). 

RESULTS OF CLADISTIC ANALYSIS 

The analysis of the 36 sirenian taxa using the 62 unweighted 
characters above and the mh*;bb*; routine (which constructs trees 
with branch-swapping and retains all trees for each initial one 
found) in Hennig86 produced 60 maximally parsimonious trees, all 
of them 152 steps long with a consistency index of 0.55 and a 
retention index of 0.83. A Nelson consensus tree of these 60 re- 
vealed that the variation among them was due entirely to different 



combinations of variants in the topology of some Eocene dugongids 
(node 6 in Fig. 3) and in that of the rytiodontine-dugongine clade 
(nodes 20-2.3). The remainder of the tree was stable. 

Use of Hennig86's successive-weighting option reduced the 
number of trees from 60 to 6 and eliminated most of the variation in 
the rytiodontine-dugongine clade. leaving this part of the consen- 
sus tree much better resolved (Fig. 3) and increasing the consis- 
tency and retention indices to 0.76 and 0.91. respectively, with a 
tree length of 162. However, as discussed below, the resolution of 
the rytiodontine-dugongine clade in Fig. 3 may well be incorrect. 
Character fits and weights for this consensus tree are given in 
Table 2. 

Because missing data have been shown to cause problems in 
cladistic analysis (Platnick et al. 1991; Huelsenbeck 1991), I reran 
the analysis omitting the nine taxa lacking data for 20 or more 
characters (Eosiren aheli, E. stromeri, Ribodon limbatus, 
Potamosiren magdalenensis, Anomotherium langewieschei. 
Haliiherium christolii, Rytiodus capgrandi, Corystosiren varguezi, 
Xenosiren yucateca). The mh*;bb*; routine produced two trees 140 
steps long with a consistency index of 0.58 and a retention index of 
0.83. Successive weighting reduced these two to a single tree that 
departed from the topology shown in Fig. 3 in only one respect: 
Haliiherium schinzii was shifted downward two nodes, becoming 
the sister group of the other taxa included within node 8 of Fig. 3 
(namely, of the Trichechidae, Dugonginae. Caribosiren, Meta- 
xytherium, and Hydrodamalinae). In all other respects the tree 
remained stable. 

The character transformations at the nodes of the tree in Fig. 3 
(or in terminal taxa within these nodes) are listed below. Also listed 
are characters (e.g.. postcranial characters) not used in the analysis 
but supporting various parts of this tree. The letters r and c after 
character-state changes denote reversals and convergences, respec- 
tively; the numbers after the letter c indicate the other nodes at 
which the convergence occurred (or, in the case of convergences in 
terminal taxa, the nodes under which the convergence is discussed 



181 



— in oo 



in m 

m — 

m O 

*+ so 



** o 

m o> 

m oo 

m r- 

m sc 

cm O"- 

CM OO 

oi r~~ 
CM ^D 
cm m 



= 
u 



o\ 


c^ 


o* 


r— 


CT\ 


— 


oo 


o* 


oo 


oo 


oo 


r— 


oo 


U~) 


oo 


-3- 


oo 


cm 


r— 


r^ 


r- 


vO 



r- o 

w, — i 



o. 

3 

o 



s< E :t 



o 

X 

■c 






OOOOOOOOOO — — — — OO 
O O O — — — — — C- — CM CM CM CM <"M CM 

O O — —  ______ 

oooooooooooooooo 
oooooooooo — — — — — o 

ooooo — — — — o  — — — — — — 



ooooooooooooooooo OO I I 
— i — ' — c" n n n n n n rj n N <N fN m N cm cm cm cm 

OOOOOOOOOOOOOOOOO O O O o- 



ooooooooooooooooo 
ooooooooooooooooo 



O O O — CM 

ooooo 



o o 
o o 



o o o o o o o 
O e-- O o c-- c- e- 



O O O O i 



I 



o o 

O o 

o o 

o o 

o o 

o o 

o o 

o o 

— o 

o o 

o o 

o o 

I I 

o o 

o o 



OC--0000 — ooooo- 
ooooooo— —  

O t>« O O O O O I I 

O t>- O O O O O I I 
O C- o o c- 

_ , . o o 

o o o o o ° o 



 e- o c-- O O o 

 o o o o o e- 



— O O — '■ 



o  e- cm 



i O O P- c— C-- O 

' O O O C-- — t>- 

i o c-- O c~- e— — 

 t>- O C— O O O 

 o O o- c- o o 

> O O O P" o o 

 O O C-- c- o O 

' o o o o o o 

o o o <>• 1 

. o o — — 

. o o — — — — 

i o o  

o o o o o o 

, O O — — — — 

i o o o e- o o 

> o p- O CM 

) o o o o o o 



OOf- O' — O IC--0000 O I I I I 

o o e— en r*- ( t  cm cm o o i o- o o o o o I I I I 

n - ?■ n n n n fl n « o o o O o - r-i o O O O 

oop-- oooooooo- ooooo o — —  

OOP- OOOOOOO IOOOOO O I I I I 

— o o- — — — — — — OIOOOOO Ol I I I 

O O o- | I I IOIOIOOOOO Ol I I I 



— tN CM CM CM tr " P" 
i-l CM CM CM CM c " C— 

O- o o o o ° o 

— cm CM n n N n 



CM CM ( 



CM P— C" 



— o 

oi r-i r~i r-i 



, rn r*~. r*i  



P-- CM CM CM CM  

o o o o o ^ 

o o o o o c 

o o o o o c 

o o o o o e 

o o — o r 

e— p— r~\ cm cm cn 

 o o o o T 



O O O O " 



OOO — 

o o o — T 
o o o o c 



ooo p-- oooooooo 
oooooooooooo 

r-- OOP- OOOOOOOO 
OOO P-- OOOOOOOO 

ooo p-- oooooooo 

ooop-oooo — — — — 
oooooooo———— 
oooooooooooo 

O O O ^- — — P-- — 

— ooo — ooo — — 



o o o — C- o P- c— o o — 

O- C- C" »• O O O- 

P-- o O e- 

CM CM CM CM 



— c-- p- 

O O o 

O O p- 

C-. 0> P- 

cm — e* 

oooooooo- oooooooo 



 O- O O — O O I 



— n r J CM CM 

ooooo 

(-■ OC" o o 
CM CM CM CM CM 

o o o o o 



— CM ol CM O 



 e-oooooooo o o 



o o O- O- o 
csl cm e- c- — 

O O O o- O c-  



OOOOC-- ooooo 
CM ol CM C^l C-- fN cm ri CM en 

00—OC--0  — 



o — ' 



o> o o — C-- — — — o — oooooo 

ooo — — — — — — ooooooo 

— OOC-C-- o^- ooooooooo 
OOO' — CM — — — ri — OOOOOO 

— O O I I I I i — i — — 



OOOOO' 
OOOOO 

ooooo 

■ooooo 



O O O C-- o o 

O O O c— o o 
O — c-- c-. o — 



O O O O O O e- 
c-- — O O O O e- 



OOO- — T--  _ _ _- o- — . , ___ 

— OO^-O- — — o — ooooooo ooooo 

O O O- O o- CM oo — ooooo — o ooo — — 

ooo f-C-- oooooo — — — — — — — — — — 



OOO- — — ' 
O O c-- — c-- 



oooooooooo 
— — oooooooo 



o o o o o 
o o o o o 



sG T — r*- t — _-__[-— — o, IT) CM — — — <~l r-l 



— -ri" — CM OS W"> C*"l ' 



m oo — m oi 



■^ ^ 



"^ s h 5 











B. 




c 






t 


^ 


_>*. 




























4 

^ 


3 


a 


1 


~5 


"C 


| 


c 




3 


§ 




& 


eg 






c 

f 




5 


s 


"i£ 


^ 


%i 


%1 
























s 


c£ 


i 


a. 


UJ 


c§ 


c 


tS 


UJ 




S 


-C 



a -5 a ~ = 5 



■£ -e 3 t -= 'S 

? tij >-~ — 'Z Jy 



E5 v 3 a 3 



-c -s: -c  



■- Si - 



S = ? S3 S 



^ ^ ^ ■= 



^-^^^!U^^3 



H -2,-a s 



-3 ^3 55 ^ -3 -3 
^r 3 z* c r> r 



5 .O -° IC X S IE ■-; 5 3 ~~ ,3 3 3 ^ 3 <u ^3 -3 3 C 



h. i- a; 0_ rc 



S«4'' 



c § 



Iff p 

^t ■»> i* ^u 

§ § § 5 






Q Q a:' a: 



§ 3 



E T 

.|-S 
S -3 
s < 



182 



Daryl P. Domning 




Moeritherium 
Paleoparadoxia 
Prorastomus sirenoides 
Protosiren fraasi 
Eotheroides aegyptiacum 
Eosiren abeli 
Protolherium veronense 
Eosiren libyca 
Eosiren stromeri 
Prototherium intermedium 
Trichechus m. manatus 
Trichechus m. latirostris 
Trichechus senegalensis 
Trichechus inunguis 
Ribodon limbatus 
Potamosiren magdalenensis 
Anomotherium langewieschei 
Miosiren kocki 
Halitherium schinzii 
Halilherium chrislolii 
Rytiodus capgrandi 
Corystosiren varguezi 
Xenosiren yucateca 
Diopiotherium allisoni 
Dioplotherium manigaulti 
Dugong dugon 
Crenatosiren oiseni 
Caribosiren turner! 
Metaxytherium krahuletzi 
Metaxytherium medium 
Metaxytherium calvertense 
Metaxytherium tloridanum 
Metaxytherium serresii 
Metaxytherium subapenninum 
Dusisiren jordani 
Dusisiren dewana 
Hydrodamalis cuestae 
Hydrodamalis gig as 



Figure 3. Nelson consensus tree of sirenian taxa and outgroups, generated by Hennig86 using 62 characters and the successive weighting option. Tree 
length, 162 steps; consistency index, 0.76; retention index, 0.91. Character fits and weights are given in Table 2. Note that node 27 is probably spurious (see 
text). 



below). Autapomorphies of terminal taxa are listed if any are 
known. When the node at which a transformation occurred is 
uncertain because of missing data, the transformation is listed under 
the first node or terminal taxon by which it had certainly occurred, 
with an indication of the earlier node at which it may questionably 
have first occurred. Significant polymorphisms are also noted 
where they occur. 

Places where the nodes of the tree correspond to traditionally 
recognized taxa are indicated. Only one new name is introduced 
here: inclusion of the Miosireninae within the Trichechidae necessi- 
tates the recognition of the new nominotypical subfamily 
Trichechinae. This and other suggested modifications to the present 
classification of the taxa here considered are shown in the Appendix. 



Basal Radiation of the Sirenia; Prorastomidae 

Node 1 (order Sirenia; one branch forms the possibly para- 
phyletic family Prorastomidae): 8(1), 9(1), 51(1). 155(1). Also, 
mastoid inflated and exposed through occipital fenestra (Novacek 
and Wyss 1987); ectotympanic inflated and droplike (Tassy and 
Shoshani 1988); pachyostosis and osteosclerosis present in skel- 
eton (Domning and de Buffrenil 1991). The possession of five 
premolars. 155( 1 ). is here provisionally treated as a synapomorphy 
of the Sirenia rather than a primitive retention, in view of the strong 
evidence placing the Sirenia well within the Ungulata. which are 
characterized by only four (Thewissen and Domning 1992). Al- 
though possession of double-rooted canines, 144(0), is here treated 






A Phylogenetic Analysis of the Sirenia 



183 



Table 2. Character fits and weights for the tree in Figure 3. 



Character 


Steps" 


Consistency index 


Retention index 


Weight'' 


3 


3 


33 


77 


2 


6 


1 


100 


1(10 


10 


7 


1 


100 


100 


1(1 


8 


1 


100 


100 


10 


9 


1 


100 


1(H) 


10 


II 


5 


20 


75 


1 


13 


1 


100 


100 


10 


14 


5 


40 


25 


1 


16 


4 


25 


62 


1 


31 


2 


50 


91 


4 


32 


4 


25 


40 


1 


36 


3 


66 


91 


6 


37' 


3 


66 


80 


5 


38 


1 


100 


100 


10 


42 


1 


100 


100 


10 


43 


2 


50 


75 


3 


51 


1 


100 


100 


10 


66 


2 


50 


88 


4 


67 


7 


42 


82 


3 


70 


3 


66 


75 


5 


73 


2 


100 


100 


10 


74 


1 


100 


100 


10 


75 


1 


100 


100 


10 


76 


3 


33 


60 


"> 


77' 


4 


50 


83 


4 


82 


2 


100 


100 


10 


84 


1 


100 


100 


10 


85 


6 


33 


84 


2 


87 


2 


50 


75 


3 


88 


1 


100 


100 


10 


89 


1 


100 


100 


10 


91 


5 


40 


70 


T 


97 


3 


66 


83 


5 


99 


4 


25 


72 


1 


101 


1 


100 


100 


10 


102 


2 


50 


66 


3 


103 


2 


50 


75 


3 


115 


1 


100 


100 


10 


121 


2 


50 


66 


3 


122 


6 


50 


85 


4 


123 


4 


25 


78 


2 


125 


2 


100 


100 


10 


126 


5 


40 


62 


2 


127 


2 


100 


100 


10 


128 


2 


50 


88 


4 


129 


2 


50 


90 


4 


136 


1 


100 


100 


10 


137 


3 


33 


71 


2 


138 


1 


100 


100 


10 


139 


2 


50 


85 


4 


140 


8 


25 


68 


1 


141' 


3 


100 


100 


10 


142 


2 


50 


75 


3 


143 


1 


100 


100 


10 


144 


3 


66 


90 


5 


146 


1 


100 


100 


10 


150 


1 


100 


100 


10 


151 


2 


100 


100 


10 


155 


1 


100 


100 


10 


156 


1 


100 


1(H) 


10 


157 


3 


66 


93 


6 


158 


1 


100 


100 


10 



"Number of transformations undergone by the character on this tree. 
^Calculated by the successive weighting option of Hennig86. 
' Unordered character. 



as a primitive retention in Prorastomus, it may be that the same 
reasoning should apply to this character. Autapomorphies of P. 
sirenoides: ll(l)c5, 11,25, 136(1), 1 37( 1 )c 10,20, 138(1), 
140( l)c7.28; also, extension of premaxilla-maxilla suture forward 
of rear end of premaxillary symphysis; enlargement of P, . Scoring 
of this species was based on a redescription of the holotype and 
examination of fragmentary new material (including a tusk) from 
Jamaica by Savage et al. (in press). 

Protosirenidae and Early Dugongidae 

Node 2 (one branch forms the possibly paraphyletic family 
Protosirenidae): 32(1), 67(1), 103(1), 115(1), 122(1), 144(1), 
157(1). Also, increase in rostral deflection; reduction of wing of 
atlas; loss of costal groove on ribs. Autapomorphies of Pwtosiren 
fraasi: 3( 1 )c6?, 1 02(0)r; however, 3( 1 ) here may be spurious, due to 
distortion (Andrews 1906: 204). 

Node 3 (paraphyletic family Dugongidae; paraphyletic subfamily 
Haiitheninae): 73(1). 75(1), 76(1), 77(1), 101(1), 102(1) (node 1?), 
121(1), 125(1) (node 2?), 127(1) (node 2?). Autapomorphy of 
Eotheroides aegyptiacum: 123( 1 )cl0,17. Characters 13 and 82 are 
derived in exactly half the sample of E. aegyptiacum (actual frequen- 
cies 1/2 and 2/4, respectively); they were arbitrarily scored here as 
primitive for this species, which appears to be genuinely transitional in 
regard to these two characters. Other polymorphisms and frequencies 
observed in this species: 32(0), 1/3; 67(2). 1/3; and possibly 103(0), 1/ 
3. A fourth specimen, definitely displaying 103(0) according to Abel 
(1913), was made the type of Eosiren abeli by Sickenberg (1934). 

Node 4: 13(1), 82(1), 97(1), 146( 1 ) (node 3?). Also, reduction 
of pubis and probable loss of terrestrial locomotor ability (node 3?). 

Node 5: 141(1). Autapomorphies of Prototherium veronense: 
ll(l)cl, 11,25, 32(0)rcl0, 67(0)rcl4, 121(0)r, 122(2)cll,29; also, 
pronounced narrowing of skull roof. Polymorphism and frequency 
observed: 76(0), 1/2. I scored the processus retroversus as present, 
77( 1 ), in P. veronense, contrary to Sickenberg (1934). The holotype 
of Eosiren abeli was destroyed in World War II; scoring of this 
species is based on the description by Sickenberg (1934) and on 
unpublished new material provisionally referred to this species. 

Node 6: 3( 1 )c2?, 67(2). 91(1 ) (node 4?), 126(1) (node 2?). Also, 
broadening of supraspinous fossa of scapula; loss of symphyseal 
contact between pubic bones. Polymorphisms and frequencies ob- 
served in Eosiren libyca: 32(0). 1/11; 43(1). 1/8; 122(2). 1/4. 
Autapomorphies of E. stromeri: frontals much longer than parietals 
in midline; M' smaller than M\ 

Node 7: 125(2), 140(l)cl,28. 

Node 8: 143(1). 144(2). 

Trichechidae 

Node 9 (family Trichechidae): 3(0)r, 77(0)rc31. 82(2); also, 
reduction of neural spines; possible tendency to enlargement and (at 
least in Trichechus) anteroposterior elongation of thoracic centra. 

Node 10 (subfamily Miosireninae): 32(0)rc5, 38(1). Possible 
autapomorphy of Anomotherium langewieschei: 123( l)c3,17 (node 
10?). Autapomorphies of Miosiren kocki: 16(l)cl2,20 (node 10?), 
36(l)cl8, 73(2), 85(2)c20,26 (node 10?), 97(0)r (node 10?), 
137(1 )cl,20 (node 9?). 140(2)c20,28 (node 10?); also, reduction 
and simplification of M\ 

Node 11 (subfamily Trichechinae): 1 1( 1 )c 1.5,25, 99(1 )c27, 
I22(2)c5,29, 157(2)cl8; also, thickening of molar enamel (node 
9?) (Domning, in press). Potamosiren magdalenensis is here taken 
to include Metaxytherium ortegense (Domning, in press). 

Node 12: 16(l)cl0,20. 1 50( 1 ); also, thinning of molar enamel 
(reversal; Domning. in press). Ribodon limbatus is here taken to 
include the maxilla (U.S. National Museum 167655) referred to 
Ribodon sp. by Domning ( 1982). 



184 



Daryl P. Domning 



Node 13 (genus Trichechus): 3 1( 1 )c 1 8 (node 11?). 67(l)rc22 
(node 11?), 91(2)c29 (node 11?), 126(2) (node 9?). 139(l)c29 
(node 11?), 140(0)rc24(node 11?), 158(1). Also, reduction of cervi- 
cal vertebrae to six (node 11?); elongation of acromion process of 
scapula (node 9?); reduction of bicipital groove of humerus (node 
1 1?): reduction of ilium (node 1 1?). Polymorphisms include 1 1(0) 
in all species. 67(0), 84( 1 ), 99(0), 1 29( 1 ), and 1 56( 1 ) in T. inunguis, 
156(0) in T. manatus, 156(1) in T. senegalensis, and 67(1). 84(0), 
122(1), and 129(0) in both T. manatus and T. senegalensis; the 
frequencies of these states have not been determined in all cases. 
Autapomorphies of T. inunguis: 70( 1 )c29; also, division of foramen 
incisivum; lateral projection of temporal crests with postorbital 
apophyses on frontal frequent; inflation of supraoccipital; elonga- 
tion of mandibular symphysis; increase in number of accessory 
mental foramina; reduction of DP' and DP,; increase in complexity 
and further decrease in size of molars; reduction of thoracic verte- 
brae to 14—16; elongation of forelimb; loss of nails. See Domning 
and Hayek ( 1986) for details regarding Trichechus. 

Node 14: 67(0)rc5, 84(1), 129(l)cl6. Also, loss of bicipital 
groove of humerus. Autapomorphies of Trichechus senegalensis: 
shortening of rostrum; decrease in rostral deflection; more trans- 
verse orientation of posterolateral sides and constriction of bases of 
supraorbital processes; presence of longitudinal crests on floor of 
mesorostral fossa; broadening of zygomatic arch and coronoid 
process. 

Node 15 (Trichechus manatus): 156(1). Also, elongation of 
vomer; more transverse orientation of median portion of 
frontoparietal suture; broadening of ribs. Autapomorphies of T. in. 
latirostris: widening of foramen magnum and straightening of its 
dorsal border; increase in rostral deflection; increase in height of 
mandibular symphysis. 

Later Dugongidae 

Node 16 (paraphyletic genus Halitherium): 85(1) (node 8?). 
I29( 1 )cl4. Also, development of cetaceanlike triangular flukes in 
place of a rounded caudal fin. Polymorphism and frequency ob- 
served in H. schinzii: 13(0). 4/6; though in the majority, this state 
has an LCL, 5 of only 0.223, and is also incongruent. 

Node 17: 122(3), 123(l)c3,10, 128(1). 

Node 18:31(l)cl3(node 17?). 36(l)cl0. 157(2)cll (node 17?). 

Dugonginae, Including Rytiodontinae 

Node 19 (subfamily Dugonginae, formerly Rytiodontinae): 
37( 1 ), 43( 1 ), 88( 1 ). Autapomorphies of Crenatosiren olseni: fusion 
of nasals with frontals; elongation of bases of supraorbital pro- 
cesses; deepening of nasal incisure. 

Node 20: 16(l)cl0.12. 42(1). 85(2)cl0.26, 137(l)cl,10, 
140(2)cl0,28. 142(1). Autapomorphies of Dugong dugon: 
14(l)c26,31, 37(2)c22, 43(0)r, 66(1 )c26; also, strong inflection of 
processus retroversus of squamosal: constant presence in juveniles 
of deciduous I 1 , and frequent presence in adults of vestigial lower 
incisors (these are atavisms, seemingly due to neoteny): sexual 
dimorphism in size and eruption of permanent I 1 tusks; functional 
loss of enamel crowns of cheek teeth; persistently open roots of 
M 2 " 3 and M 2 _ 3 . Although the zygomatic process of the jugal of the 
adult Dugong is long, 89(0). the process is much shorter in fetuses 
and neonates, suggesting that the ancestors of Dugong may have 
had the derived state 89( 1 ), like Dioplotherium and Xenosiren 
(below). Trichechus, in contrast, has a long process in both fetuses 
and adults, so a short process is not simply a condition of early 
ontogeny. 

Node 21: 6(1). 141(2). 

Node 22: 7(1), 141(3). Autapomorphies of Corystosiren 
varguezi: 37(2)c20, 67( 1 )rc 13 (node 22°). 76(0)rc23. Separation of 



the squamosal from the temporal crest, 76(0), may reflect the great 
and uniquely derived thickening of the parietals characteristic of 
Corystosiren. 

Node 23: 89( 1 ). Also, incipient blockage of infraorbital canal by 
a transverse wall; apparent fusion of nasals with frontals. 
Autapomorphies of Dioplotherium manigaulti: 16(1 )r?. 97(2), 
142(0)r?; these "reversals" more likely indicate that this entire 
clade should be rooted farther down in the tree. Possible 
autapomorphies of D. allisoni: 76(0)rc22 (condition unknown in its 
possible descendant Xenosiren); L23(0)r (node 21?). D. allisoni 
here includes referred specimens from Brazil (Toledo and Domning 
1991). Autapomorphies of X. yucateca: 14(2), 85(1 )r; also, accen- 
tuation of concavity of frontal roof; thinning and medial concavity 
of preorbital process of jugal. 

Carihosiren and Metaxytherium 

Node 24: 140(0)rcl3. It is uncertain whether the tusks of 
Caribosiren were really absent (an autapomorphy ) or merely small. 

Node 25 (paraphyletic genus Metaxytherium): ll(l)c 1,5,1 1. See 
Domning and Thomas (1987) and Domning (1988) for details. 
Polymorphism and frequency observed in M. krahuletzi: 66( 1 ), 1/2; 
evidently a genuinely transitional condition, scored arbitrarily as 
primitive. 

Node 26: 66( 1 )c20, 85(2)cl0,20. Autapomorphy of 
Metaxytherium floridamtm: 14(l)c20,31. Polymorphisms and fre- 
quencies observed in M. floridanum: 11(0), 8/26. 14(0), 1/3; 67(1), 
1 5/26; 85( 1 ). 1 2/20. The latter two majority states have LCL^s of 
only 0.369 and 0.361, respectively, and are both incongruent. 

Node 27: 67(3), 99(1 )cll. I believe that this node is spurious 
and that these changes were actually evolved in parallel by Euro- 
pean Pliocene Metaxytherium and North Pacific hydrodamalines 
(i.e., at nodes 28 and 29 of this tree, respectively). 

Node 28: 140(l)rcl,7; this increase in tusk length was inter- 
preted by the program as a re-reversal of the reduction at node 24. 
The body of M. serresii is smaller than that of the European Mio- 
cene Metaxytherium; I interpret this as ecophenotypic dwarfism 
that was reversed in M. subapenninum (Domning and Thomas 
1987). Polymorphisms and frequencies observed: in M. serresii, 
31(0), 2/3: in M. subapenninum. 66(0), 2/3. In each case, the 
majority state has an LCL,, 5 of only 0.094 and is incongruent. 
Autapomorphy of M. subapenninum: 140(2)cl0.20. This name is 
accepted by Pilleri ( 1988) as a valid senior synonym of M. forestii. 

Hydrodamalinae 

Node 29 (subfamily Hydrodamalinae; paraphyletic genus 
Dusisiren): 70(l)cl 3, 77(2), 87(l),91(2)c 1 3, 122(2)rc5,H, I28(0)r, 
139( I )cl3. Also, decreased rostral deflection; increased body size 
(to about 4.5 m in D. jordani). See Domning (1978) for details. 
Polymorphisms and frequencies observed in D. jordani: 66(0), 2/6; 
67( 1 ). 1/5. Apeculiarity of the available specimens of D. jordani is 
separation of the palatines in the midline, a condition seen in no 
other sirenian. Although the palatal incisure is consequently very 
deep, because of the different anatomical basis of this condition 
(compared to Dioplotherium manigaulti, where the incisure is deep 
despite the median juncture of the palatines), character 97 was here 
scored 1 rather than 2. Whe