THOSE QUEER AMPHIBIOUS ANIMALS: DIGGING INTO THE FOSSIL RECORD 
OF OTARIOID PINNIPEDS TO INVESTIGATE THEIR EARLY DIVERSIFICATION, 
LOCOMOTOR EVOLUTION, AND CONSERVATION BIOGEOGRAPHY 
by 
M. KELLUM TATE-JONES
A DISSERTATION 
Presented to the Department of Earth Sciences 
and the Division of Graduate Studies of the 
University of Oregon in partial fulfillment of the 
requirements 
for the degree of 
Doctor of Philosophy 
June 2023 
DISSERTATION APPROVAL PAGE 
Student: M. Kellum Tate-Jones 
Title: Those Queer Amphibious Animals: Digging into the Fossil Record of Otarioid 
Pinnipeds to Investigate Their Early Diversification, Locomotor Evolution, and 
Conservation Biogeography 
This dissertation has been accepted and approved in partial fulfillment of the 
requirements for the Doctor of Philosophy degree in the Department of Earth Sciences 
by: 
Edward B. Davis Chairperson/Advisor 
Samantha S. B. Hopkins Core Member 
Valerie J. Sahakian Core Member 
Stephen R. Frost Institutional Representative 
and 
Krista Chronister Vice Provost for Graduate Studies 
Original approval signatures are on file with the University of Oregon Division of 
Graduate Studies. 
Degree awarded June 2023 
2 
© 2023 M. Kellum Tate-Jones 
 This work is licensed under a Creative Commons 
Attribution-NonCommercial-ShareAlike (United States) License. 
3 
  
DISSERTATION ABSTRACT 
 
M. Kellum Tate-Jones 
 
Doctor of Philosophy 
 
Department of Earth Sciences 
 
June 2023 
 
Title:  Those Queer Amphibious Animals: Digging into the Fossil Record of Otarioid 
Pinnipeds to Investigate Their Early Diversification, Locomotor Evolution, and 
Conservation Biogeography 
 
 
Seals, sea lions, walruses, and their extinct relatives and ancestors have played a 
critical role in marine ecosystems for over 25 million years. Known collectively as 
pinnipeds, these secondarily aquatic animals are unique among modern marine mammals 
for their obligatorily amphibious lifestyle and display remarkable ecological diversity, 
employing a range of strategies for aquatic locomotion and occupying coastal and pelagic 
habitats from the tropics to the poles. In addition to their extant representatives, their 
fossil record contains over a hundred named pinniped species, although the majority of 
these species are relatively derived members of the Phocidae (true seals), Otariidae (eared 
seals, including sea lions and fur seals), and Odobenidae (walruses). In my first chapter, I 
contributed to our understanding of early pinniped evolution by describing the new 
species Eodesmus condoni and performing a phylogenetic analysis to determine its 
placement within Pinnipedia. My analysis resolved E. condoni as the most basal species 
yet described of the extinct pinniped family Desmatophocidae. I then explored the 
evolution of aquatic locomotion in otarioid pinnipeds (otariids, odobenids, and 
desmatophocids) by using three-dimensional geometric morphometric analyses to first 
investigate the relationship between humerus morphology and swimming modes and then 
4 
  
to predict the swimming mode used by fossil otarioid taxa. My results supported the 
plesiomorphy of hindlimb-dominated propulsion in all pinnipeds as well as the 
Otarioidea, with forelimb propulsion evolving three separate times: in the lineages of 
desmatophocid Allodesmus kernensis, odobenid Pontolis barroni, and an ancestor of 
modern Otariidae, likely through exaptation of humeral structures that first evolved for 
pinniped terrestrial locomotion. Finally, I use ecological niche modeling to compare the 
range of environmental conditions inhabited by the sole surviving walrus Odobenus 
rosmarus during the Pleistocene to the conditions in which it currently resides. This study 
reveals that during the Pleistocene, the regions where O. rosmarus lived encompassed 
warmer conditions than its current range. Contrary to predominant narrative about walrus 
response to climate change, these findings suggest that O. rosmarus may indeed be able 
to survive in the seasonally ice-free conditions that will characterize the Arctic in the 
coming decades.  
This dissertation includes previously published and unpublished coauthored 
material. 
 
5 
  
CURRICULUM VITAE 
 
NAME OF AUTHOR:  M. Kellum Tate-Jones 
 
 
GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: 
 
 University of Oregon, Eugene 
 University of Arkansas at Little Rock, Little Rock, AR 
 Harding University, Searcy, AR 
 
 
DEGREES AWARDED: 
 
 Doctor of Philosophy, Earth Sciences, 2023, University of Oregon 
 Bachelor of Science, Earth Science, 2017, University of Arkansas at Little Rock 
 Bachelor of Science, Nursing, 2012, Harding University 
 
 
AREAS OF SPECIAL INTEREST: 
 
 Vertebrate Paleontology 
 
 Marine Tetrapod Evolution 
 
 Conservation Ecology and Paleobiology 
 
 Diversity, Equity, Inclusion, and Justice in the Geosciences 
 
 
PROFESSIONAL EXPERIENCE: 
 
Community Engagement Consultant, Oregon Department of Geology and Mineral 
Industries, 2023 
 
Instructor of Record for ERTH 103 Exploring Earth History, University of 
Oregon, Spring 2021, Spring 2022 
 
Instructor of Record for 407/507 geoSTORIES (STOrytelling for Resilience and 
Inclusion in Earth Sciences) Seminar 
  
 Graduate Employee, University of Oregon, 2017-2023 
 
Paleontology Camp Instructor, Sternberg Museum, 2018-2022 
 
 ICU Registered Nurse, St. Vincent Infirmary, Little Rock, AR 2013-2017 
6 
GRANTS, AWARDS, AND HONORS: 
Women in STEM Conference Award, University of Oregon, 2022 
Department of Earth Sciences Baldwin Scholarship, University of Oregon, 2020, 
2021, 2022 
Community for Minorities in STEM Conference Award, University of Oregon, 
2022 
Gary E. Smith Professional Development Scholarship, University of Oregon, 
2022 
Department of Earth Sciences Teaching Excellence Award, University of Oregon, 
2018, 2021 
Women in STEM Leadership Award, University of Oregon, 2021 
General University Scholarship, University of Oregon, 2018, 2020, 2021 
David S. Easley Memorial Fellowship, University of Oregon, 2020 
Department of Earth Sciences Good Citizen Award, University of Oregon, 2020 
Thomas Condon Outstanding Work in Paleontology, University of Oregon, 2018, 
2019 
Student Research Grant, Geological Society of America, 2018 
L. Lynn Hamilton General Scholarship, University of Oregon, 2018
Science Literacy Program Fellowship, University of Oregon, 2018 
Travel Scholarship, University of Oregon, 2017 
Outstanding Senior Geology Student, University of Arkansas at Little Rock, 2017 
Field Camp Scholarship, Central Arkansas Gem, Mineral and Geology Society, 
2017 
Honors Scholar, Harding University, 2008-2012 
International Baccalaureate Diploma, Jefferson County International 
Baccalaureate School, 2008 
7 
  
PUBLICATIONS: 
 
Tate-Jones, M.K., C.M. Peredo, C.D. Marshall, and S.S.B. Hopkins. 2020. The dawn of 
Desmatophocidae: a new species of basal desmatophocid seal (Mammalia, 
Carnivora) from the Miocene of Oregon, USA. Journal of Vertebrate 
Paleontology 40:1-4.  
 
8 
  
ACKNOWLEDGMENTS 
 
 
 I would like to offer my heartfelt gratitude to all those who have supported me 
through this journey. Just as no one hikes the Pacific Crest Trail or runs a decathlon 
without a dedicated support team, these six years of professional and personal growth 
would never have been possible without all of you. 
 To my family: Aunt Becky for showing me The Land Before Time and thus 
igniting my lifelong love of fossils; Aunt Patti for inspiring the first inklings of my 
geology obsession by bringing me the most fascinating tumbled rocks; my brother Carter 
for his company on my research trips and timely help wrangling my code and truly 
abominable spreadsheets; Dad for all the texts about ducks, pink cars, and bald eagles to 
keep me grounded; Mom for bringing me to see my first pinnipeds at the Birmingham 
Zoo, teaching me to always be curious about the world around me, and always reminding 
me that I can do hard things; my husband Jacob for the countless delicious meals, loads 
of laundry, encouraging pep talks, grocery runs, editing assistance, walks and hikes, cups 
of coffee, and unflagging belief in my ability to live out this dream, even and especially 
when I did not believe in myself; and of course my cats Shupa and Puddles, but Puddles 
especially for keeping me company through hundreds of hours of grading, coding, 
writing, and figure-making.  
 To my friends: Amber, Fletcher, David, and Craig, the tried-and-true Arkansas 
gang, for keeping the windows open to life outside academia; all of the members of 
Rehearsals for Life who have come and gone over the past six years and especially 
Abigail, Manas, Jaclyn, Rory, Kaushal, and Steve, for showing me the messy, imperfect 
beauty of working in community for a common goal; the Rainbow Connection team with 
9 
  
Oregon Country Fair for connecting me to the broader Oregon community and embracing 
me for exactly who I am; Stacie and Beckett for dispensing wisdom, books, hilarity, and 
absurdly delicious meals and cocktails; Ellen and Shawna for the reality TV breaks; 
Michelle for letting me bully her into going birding and truly unforgettable adventures in 
New Orleans and D.C.; Allison for ensuring I never forgot about the importance of 
fantasy fiction and Ted Lasso, even during the crunch times; Monse for keeping me 
connected to the arts and to uncontrollable belly laughs; and Rachel, my soul sister, for 
truly too many things to count but especially teaching me to cross country ski and 
mutually keeping each other afloat as we plummeted through the raging waters of 
everything from health problems to global pandemics. 
 To my colleagues in paleontology: the current and past students of the Oregon 
Vertebrate Paleontology Lab, including Holley, Leonard, Paul, Win, Dana, Helena, 
Amanda, Megan W, Megan P, Rachel, Sarah, Max, Jordan, Andy, Kai, Lyzzy, Deanna, 
Thomas, and Emma, for the laughs, ideas, discussions, encouragement, and gingerbread 
replicas of Jurassic Park; Elena for inviting me across the pond to experience vertebrate 
paleontology, European style, as well as the incredible sights (and tastes!) of Venice; the 
Sternberg Science Camps crew, especially Marjie, Amber, Maggie, and David, for 
keeping silliness and whimsy at the heart of the natural sciences; the other professional 
pinniped nerds for providing knowledge and insight along the way, including Bobby 
Boessenecker, Justin Keller, James Rule, Austin Hocking, and Carlos Peredo; and Lisa 
White for her unparalleled mentorship, perspective, and support in pursuing the next 
stages of my professional path. 
10 
  
 To my undergraduate professors René Shroat-Lewis and Michael “Dr. D” 
DeAngelis not only for inspiring my bone-deep love of Earth science and geologic 
history but also for encouraging me to pursue my dream of earning a doctorate studying 
in this field. None of this would have happened without you. 
 Last but not least, to my committee: Steve Frost for helping me wrap my head 
around geometric morphometrics and taxonomic diversification, especially during 
preparation for my comprehensive exams; Valerie Sahakian for her mentorship with 
regards to my work in DEI spaces and editorial assistance for grant proposals; Sam 
Hopkins for introducing me to “Toby” (who would eventually become my first 
dissertation chapter) and expert perspective on phylogenetic comparative methods and 
evolutionary theory during my second chapter; and Edward Davis for his unwavering 
confidence in my abilities, guidance on ecological niche modeling, and mentoring on 
career pathways both inside and outside academia. 
 
11 
  
 
 
 
 
 
 
 
Dedicated to Jacob F. Jones, who has been the steadfast lift beneath my flippers for this 
entire journey. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12 
TABLE OF CONTENTS 
Chapter Page 
I. INTRODUCTION .................................................................................................... 20 
II. THE DAWN OF DESMATOPHOCIDAE: A NEW SPECIES OF BASAL
DESMATOPHOCID SEAL (MAMMALIA, CARNIVORA) FROM THE
MIOCENE OF OREGON, U.S.A. ........................................................................ 26 
1    Introduction ...................................................................................................... 26 
2    Materials and Methods ..................................................................................... 28 
2.1    Material Examined .................................................................................. 28 
2.2    Anatomical Description  ......................................................................... 28 
2.3    Phylogenetic Methods ............................................................................. 29 
3    Systematic Paleontology and Anatomical Description .................................... 29 
3.1    Systematic Paleontology ......................................................................... 29 
   3.1.1    Emended Diagnosis of Desmatophocidae .................................. 30 
   3.1.2    Type and Included Species of Desmatophocidae ....................... 30 
   3.1.3    Type and Only Included Species of Eodesmus ........................... 30 
   3.1.4    Diagnosis of Eodesmus ............................................................... 30 
   3.1.5    Etymology of Eodesmus  ............................................................ 30 
   3.1.6    Holotype Specimen ..................................................................... 30 
   3.1.7    Etymology of condoni................................................................. 30 
   3.1.8    Diagnosis of Eodesmus condoni ................................................. 31 
3.2    Geographical Location and Geology  ..................................................... 31 
   3.2.1    Locality ....................................................................................... 31 
   3.2.2    Horizon ....................................................................................... 31 
13 
  
 3.3    Anatomical Description .......................................................................... 32 
     3.3.1    Ontogenetic Stage and Sex ......................................................... 32 
     3.3.2    Rostrum ....................................................................................... 33 
     3.3.3    Orbital Region and Zygomatic Arch........................................... 38 
     3.3.4    Braincase and Occiput ................................................................ 39 
     3.3.5    Basicranium ................................................................................ 39 
     3.3.6    Dentition ..................................................................................... 40 
     3.3.7    Comparison to Other Pinnipedimorpha ...................................... 41 
 4    Phylogenetic Results and Discussion............................................................... 42 
 5    Paleobiogeographic Context ............................................................................ 46 
 6    Conclusion ....................................................................................................... 47 
 7    Bridge ............................................................................................................... 47 
II.  FORELIMB-DOMINATED AQUATIC PROPULSION EVOLVED THREE   
TIMES IN OTARIOID PINNIPEDS AS REVEALED BY GEOMETRIC 
MORPHOMETRIC ANALYSES OF MODERN AND FOSSIL HUMERI ........ 49 
 
 1    Introduction ...................................................................................................... 49 
 2    Materials and Methods ..................................................................................... 60 
  2.1    Material Examined .................................................................................. 60 
 2.2    Phylogenetic Information........................................................................ 60 
 2.3    Geometric Morphometrics ...................................................................... 64 
 2.4    Flexible Discriminant Analyses .............................................................. 66 
 3    Results .............................................................................................................. 67 
 3.1    Humeral Morphology and Aquatic Locomotion..................................... 67 
 3.2    Inferring Locomotor Mode from Pinniped Humeral Morphology ......... 70 
14 
4    Discussion ........................................................................................................ 75 
4.1    Implications of Pinniped Humerus Morphology for Locomotor     
Function .................................................................................................. 75 
   4.1.1    Deltoid Crest and Greater Tubercle ............................................ 77 
   4.1.2    Lesser Tubercle ........................................................................... 80 
   4.1.3    Medial Epicondyle ...................................................................... 82 
   4.1.4    Lateral Epicondyle ...................................................................... 83 
4.2    The Evolution of Swimming Modalities in Otarioid Pinnipeds ............. 85 
5    Conclusion ....................................................................................................... 90 
6    Bridge ............................................................................................................... 91 
III. FOSSIL DATA INDICATE ODOBENUS ROSMARUS, THE LAST
ODOBENID, CAN SURVIVE IN SEASONALLY ICE-FREE CONDITIONS 92 
1    Introduction ...................................................................................................... 92 
2    Materials and Methods ..................................................................................... 97 
2.1    Input Data................................................................................................ 97 
   2.1.1    Modern Occurrence Data ............................................................ 97 
   2.1.2    Fossil Occurrence Data ............................................................... 99 
   2.1.3    Environmental Predictors............................................................ 100 
2.2    Analyses .................................................................................................. 102 
   2.2.1    Comparing Niche Space ............................................................. 102 
   2.2.2    Ecological Niche Models ............................................................ 103 
   2.2.3    Binary Maps and Centroid and Area Calculations ..................... 105 
3    Results .............................................................................................................. 105 
3.1    Niche Space Comparison ........................................................................ 105 
15 
  
 3.2    Ecological Niche Modles ........................................................................ 107 
 3.3    Projections to Current Conditions ........................................................... 109 
 3.4    Future Projections ................................................................................... 110 
 3.5    Binary Maps and Area Calculations ....................................................... 110 
 4    Discussion ........................................................................................................ 111 
 4.1    Niche Space Comparison ........................................................................ 113 
 4.2    Current Climate Projections .................................................................... 114 
 4.3    Future Climate Projections ..................................................................... 117 
 4.4    Future Research Directions and Walrus Conservation ........................... 119 
 4    Conclusion ....................................................................................................... 120 
IV. DISSERTATION SUMMARY ............................................................................. 121 
APPENDICES ............................................................................................................. 124 
 A.  DESCRIPTION OF CHARACTERS USED IN THE PHYLOGENETIC 
ANALYSIS OF CHAPTER II ......................................................................... 124 
 
 B.  NEXUS FORMATTED MORPHOLOGICAL CHARACTER MATRIX  
  FOR CHAPTER II ........................................................................................... 137 
 
 C.  FIRST AND LAST APPEARANCE DATA FOR TAXA IN CHAPTER II .. 146 
 D.  TIME-SCALED PHYLOGENY IN NEWICK FORMAT USED IN  
  CHAPTER III .................................................................................................. 147 
 
 E. PRINCIPAL COMPONENTS USED TO TRAIN THE FLEXIBLE 
DISCRIMINANT ANALYSES IN CHAPTER III ......................................... 148 
 
 F.  PLOT OF PRINCIPAL COMPONENTS 1 AND 2 FROM PRINCIPAL 
COMPONENT ANALYSIS OF ENVIRONMENTAL PARAMETERS  
  FROM CHAPTER IV ...................................................................................... 151 
 
 G.  ECOLOGICAL NICHE MODEL SUMMARY STATISTICS ....................... 152 
 
REFERENCES CITED ................................................................................................ 155 
16 
LIST OF FIGURES 
Figure Page 
Chapter II 
1. Map of central Oregon coastline showing exposures of Astoria Formation with
type locality for Eodesmus condoni noted ............................................................. 32 
2. Eodesmus condoni holotype skull in dorsal and ventral views.. ............................ 35 
3. Eodesmus condoni holotype skull in lateral, rostral, and occipital views ............. 36 
4. Phylogenetic tree showing position of Eodesmus condoni .................................... 43 
Chapter III 
1. The three modern pinniped families ...................................................................... 49 
2. Dorsal, medial, ventral, and lateral views of representative humeri of the
Otariidae, Phocidae, and Odobenidae  ................................................................... 57 
3. Time-scaled phylogeny showing the relationships between Ursidae,
Musteloidea, and Pinnipedia, as well as the relationships between and within
the stem and crown pinniped taxa .......................................................................... 62 
4. Geometric morphometric landmarks used in this study shown on the humerus
of Phoca hispida .................................................................................................... 65 
5. Principal components analysis showing the variation between hindlimb- and
forelimb-dominated swimmers .............................................................................. 68 
6. Probability density functions for the uncorrected and phylogenetically
corrected flexible discriminant analyses  ............................................................... 74 
7. Discriminant scores from the phylogenetically corrected discriminant analysis
 plotted across that phylogeny ................................................................................. 86 
Chapter IV 
1. Walrus general appearance and typical habitats and map of known distribution of
Odobenus rosmarus ............................................................................................... 93 
2. Occurrences of O. rosmarus from the dataset used in this study, including
modern and fossil occurrences ............................................................................... 99 
17 
3. Projections of the ecospace for modern and fossil occurrences of O. rosmarus
onto the significant axes of a broken stick principal component analysis ............. 106 
4. Maximum entropy models of environmental suitability derived from modern and
fossil occurrences of O. rosmarus as projected to current and future climate
scenarios ................................................................................................................. 108 
5. Binary maps showing optimal walrus habitat for current and future climate
scenarios for the modern and fossil ecological niche models ................................ 113 
18 
LIST OF TABLES 
Table Page 
Chapter II 
1. Measurements (in millimetres) of cranial dimensions of Eodesmus condoni (gen.
et. sp. nov.) (UOMNCH F-68583). Alveolar measurements are recorded from
 the right toothrow. .................................................................................................. 32 
Chapter II 
1. List of landmarked specimens representing fossil and extant taxa ........................ 61 
2. Landmarks used to quantify the three-dimensional morphology of pinniped
humeri .................................................................................................................... 64 
3. Results of uncorrected and phylogenetically corrected flexible discriminant
analyses  ................................................................................................................. 71 
4. The contribution of each of the first 13 principal components to the two flexible
discriminant analyses ............................................................................................. 73 
Chapter III 
1. Percent contributions of the environmental variables to each of the three
principal components from the broken stick principal components analysis ........ 107 
2. Geographical changes in ecological niche space predicted by the modern ENM
and fossil ENM ...................................................................................................... 111 
19 
CHAPTER I 
INTRODUCTION 
From the study of seals—those queer amphibious animals living at the interface of land and 
sea—we gain new respect for the far-limit possibilities of mammalian adaptation. 
– Victor Scheffer, from Adaptations for a Marine Existence (Riedman, 1990)
1 Background 
For the past 25 million years, pinnipeds have served as critical components of global 
marine ecosystems (Bowen, 1997; Estes et al., 2013; Cammen et al., 2019). Modern pinnipeds 
include 18 species in the Phocidae (true or earless seals), 15 species in the Otariidae (fur seals 
and sea lions or eared seals), and a single extant species of Odobenidae: Odobenus rosmarus, the 
modern walrus (Berta et al., 2018). Additionally, the fossil record contains evidence of over 100 
extinct pinniped species going back to the late Oligocene (Berta et al., 2018; Valenzuela-Toro 
and Pyenson, 2019). Although phocids arose in the North Atlantic before expanding into the 
Southern Hemisphere and eventually the North Pacific (Rule et al., 2020), stem pinnipeds and 
early otarioid pinnipeds (otariids, odobenids, and the extinct desmatophocids) largely diversified 
in the North Pacific and have a particularly rich fossil record from the west coast of the United 
States (Tonomori and Taru, 2022). Pinnipeds perform essential ecosystem services, from 
regulating lower trophic levels (Heithaus et al., 2008) and redistributing nutrients in the water 
column (Theobald et al., 2006; McCauley et al., 2015) to serving as a critical food source for 
other tertiary marine and terrestrial predators (Fay, 1982; Bowen, 1997; Hammerschlag et al., 
20 
 
2006) and restructuring benthic communities through the bioturbation of seafloor sediment (Ray 
et al., 2006).  
Today, however, pinnipeds are experiencing the same disproportionate impacts from 
anthropogenic environmental degradation as other large-bodied marine predators (Springer et al., 
2003; Estes et al., 2011; McCauley et al., 2015). Two species, the Japanese sea lion (Zalophus 
japonicus) and the Caribbean monk seal (Neomonachus tropicalis) have already been hunted 
extinction, and the IUCN considers 30% of the remaining species to be threatened with 
extinction by the consequences of human activity, including, pollution, fishery collapses, 
warming oceans, and habitat disruption and destruction (Kovacs et al., 2012; IUCN, 2022). 
Addressing these urgent threats will require timely and fiscally efficient conservation policies 
based on scientific evidence. Studying the fossil record of modern lineages provides insight into 
the evolutionary roots of their ecological niches as well as their responses to past environmental 
perturbations, thus providing us insight into how they are likely to respond to current 
environmental change according to their respective ecologies (Jablonski and Sepkoski, 1996; 
Dietl and Flessa, 2011; Dietl et al., 2015). Compared to other marine mammals, the pinniped 
fossil record remains relatively understudied (Valenzuela-Toro and Pyenson, 2019), limiting our 
ability to develop the long-term perspective needed for effective and dynamic conservation 
efforts. In this dissertation, I aim to advance our understanding of pinniped evolutionary history 
by describing newly discovered desmatophocid pinniped Eodesmus condoni (Chapter II), 
reconstructing the locomotor evolution of the Otarioidea (Chapter III), and using the fossil record 
of modern walrus O. rosmarus to model its response to a rapidly warming Arctic (Chapter IV). 
 
 
21 
 
 
2 Chapter II 
The extinct pinniped clade Desmatophocidae inhabited the North Pacific from the 
Aquitanian through the early Tortonian stages of the Miocene. The first large-bodied pinnipeds, 
the desmatophocids are primarily represented by around ten species of the geologically younger 
allodesmine desmatophocids (Boessenecker and Churchill, 2013; Tonomori et al., 2018). 
Conversely, only two species are known of the geologically older desmatophocid subclade, 
genus Desmatophoca or the desmatophocines (Condon, 1906; Barnes, 1987; Deméré and Berta, 
2002). Both the desmatophocines and allodesmines are relatively derived compared to the 
morphology of stem pinnipeds like the so-called “enaliarctines,” Pinnarctidion, and 
Pteronarctos. In this chapter, I describe a new fossil pinniped cranium from the Oregon coast and 
through morphological examination and phylogenetic analysis I identify it as the most basal 
desmatophocid species yet discovered, herein named Eodesmus condoni. E. condoni is the sixth 
pinniped species recovered from the Iron Mountain Bed (17.3–16.6 Ma) of the Astoria 
Formation, a remarkable taxonomic diversity that today is reached only in the highly productive 
waters of the polar regions (King, 1983). This diverse fauna lends support to the hypothesis that 
the elevated ocean productivity caused by the Mid-Miocene Climatic Optimum (18.0–16.5 Ma; 
Flower and Kennett, 1993; Böhme, 2003) supported the diversification of marine species (Lipps 
and Mitchell, 1976; Tsuchi, 1992; Pyenson et al., 2009).  
 
3 Chapter III 
While the publication of fossil otarioids like E. condoni is continuing to expand our 
understanding of their early diversity, we have yet to solve the mystery of their locomotor 
evolution. Significant evidence exists to support the plesiomorphy of hindlimb-dominated 
22 
 
 
swimming, with multiple recent studies identifying stem pinnipeds Puijila darwini, 
Potamotherium valletoni, and Enaliarctos mealsi as such (Bebej, 2009; Furbish, 2015; Paterson, 
2017; Paterson et al., 2020). All modern phocids are highly adapted to hindlimb-propelled 
swimming, likely representing an extreme specialization on the ancestral condition (Hocking et 
al., 2021). Among otarioids, otariids rely on forelimb propulsion to swim while the walrus O. 
rosmarus uses a swimming somewhat intermediate to phocids and otariids, relying on forelimb 
propulsion at low speeds and hindlimb propulsion at higher speeds (Berta et al., 2018). This 
character distribution raises the question of whether the ancestor of the otarioids was forelimb-
propelled with O. rosmarus representing a partial reversal to the ancestral condition, or was the 
otarioid progenitor hindlimb-propelled and otariids evolved forelimb propulsion separate from 
the minimal use of the forelimbs in O. rosmarus?  
The fossil record represents an opportunity for investigating this conundrum. Indeed, 
recent analyses have resolved derived allodesmine desmatophocid Allodesmus kernensis as a 
forelimb-dominated swimmer (Bebej, 2009; Furbish, 2015). However, the linear morphometric 
analysis (Bebej, 2009) and ancestral trait reconstruction (Furbish, 2015) used in these and other 
similar studies rely on the availability of nearly complete specimens. While the Otarioidea are 
fairly well-represented in the fossil record in terms of number of specimens, most consist of 
isolated elements (Valenzuela-Toro and Pyenson, 2019), precluding the use of these previous 
methodologies. Therefore to investigate otarioid locomotor evolution, I used three-dimensional 
geometric morphometric methods to 1) quantify humerus morphology of forelimb- and 
hindlimb-propelled pinnipeds using a novel landmarking schema and Procrustes 
superimposition, 2) explore the variation of humerus morphology between locomotor categories 
via principal components analysis, 3) assess the utility of humerus morphology for diagnosing 
23 
 
swimming mode with flexible discriminant analysis, and finally 4) predict the swimming modes 
of fossil otarioids with the discriminant analysis. I found that the discriminant analysis correctly 
classified the modern taxa 90% of the time. Fossil classifications supported an ancestrally 
hindlimb-propelled state for Pinnipedia as well as for Otarioidea, with desmatophocid 
Allodesmus kernensis, odobenid Pontolis barroni, and the ancestor of modern Otariidae each 
separately acquiring forelimb propulsion, likely through the coopting of humeral structures that 
first evolved to facilitate the lunging terrestrial locomotion style used by modern otarioids. 
4 Chapter IV 
The range of environmental conditions within which a species can thrive forms another 
important dimension of ecology and is particularly relevant considering the rapidly accelerating 
environmental change mediated by human activity. Nowhere is this change more apparent than 
in the Arctic, which is warming two to four times as quickly as the rest of the planet (Yletyinen, 
2019; Huntington et al., 2020; Turner et al., 2020; Jacobs et al., 2021; Rantanen et al., 2021). 
Consequently, sea ice is expected to disappear seasonally from the Arctic within the next two to 
three decades (Overland and Wang, 2013; Jahn et al., 2016; Notz and Stroeve, 2018; Diebold and 
Rudebusch, 2021). Multiple species of Arctic pinnipeds rely on ice for activities such as hauling 
out between feedings, pupping, maternal care, and mating (King, 1983), including the recent 
poster child of climate change, the walrus Odobenus rosmarus. However, the Pleistocene and 
Holocene fossil record for O. rosmarus includes specimens found far south of the southernmost 
extent of sea ice during even the coldest glacial periods. The paleodistribution of walruses 
suggests that despite the modern correlation between walrus range and sea ice, the full breadth of 
their ecological niche may include ice-free conditions.  
24 
 
To test this hypothesis, I first evaluated whether modern and fossil walrus ecospace 
occupation is significantly different using a permutational multivariate analysis of variance 
(PERMANOVA) and a principal components analysis (PCA). I then constructed two sets of 
ecological niche models (ENMs) from modern and fossil occurrences of O. rosmarus, 
respectively, using the maximum entropy algorithm, and projected these models to 2050 and 
2100. The PERMANOVA supported a significant difference between the ecospace inhabited by 
modern and fossil walrus populations (p = 0.0001; R2 = 0.125). Similarly, the PCA revealed that 
modern walrus populations indeed inhabit only a portion of their past ecospace, including the 
ice-free conditions likely to characterize the Arctic in the upcoming decades. The ENM 
projections also supported these findings. Furthermore, while the future projections derived from 
the modern ENM predict an 18% decrease in suitable walrus habitat by 2100, the fossil model 
projections predict no appreciable change in range over the next century. Although I consider an 
intermediate scenario more realistic than either endmember, our results nonetheless present 
encouraging evidence that with adequate conservation, walruses may adapt to the changing 
Arctic climate if other anthropogenic elements such as overfishing, oceanic pollution, and habitat 
disruption through shipping, drilling, and seafloor mining are mediated. 
 
5 Co-authorships and Previous Publications 
Chapter II is co-authored with C. M. Peredo, C. D. Marshall, and S. S. B. Hopkins and is 
published in the Journal of Vertebrate Paleontology. Chapter III is also co-authored with S. S. B. 
Hopkins and is in preparation for submission to the Journal of Vertebrate Paleontology. Chapter 
IV is co-authored with E. B. Davis and is in review with the journal Frontiers of Biogeography. 
 
25 
 
CHAPTER II 
THE DAWN OF DESMATOPHOCIDAE: A NEW SPECIES OF BASAL 
DESMATOPHOCID SEAL (MAMMALIA, CARNIVORA) FROM THE 
MIOCENE OF OREGON, U.S.A. 
From Tate-Jones, M. K., C. M. Peredo, C. D. Marshall, and S. S. B. Hopkins. 2020. The 
dawn of Desmatophocidae: a new species of basal desmatophocid seal (Mammalia, Carnivora) 
from the Miocene of Oregon, U.S.A. Journal of Vertebrate Paleontology. DOI: 
10.1080/02724634.2020.1789867. 
1 Introduction 
The pinniped family Desmatophocidae originated in the early Miocene and dispersed 
throughout the North Pacific before their final extinction approximately 10 Ma (Barnes, 1987; 
Boessenecker and Churchill, 2018). Not only were Desmatophocidae the first large-bodied 
pinnipeds, but they also represent the only extinct family within crown Pinnipedia (Churchill et 
al., 2015; Berta et al., 2018). Desmatophocids were relatively diverse and, unlike their sister 
clade Phocidae (Dewaele et al., 2018), left behind a particularly rich fossil record of numerous 
well-preserved specimens, with a significant amount of articulated and cranial material. Two 
monophyletic subfamilies have been identified within Desmatophocidae: the Desmatophocinae 
and Allodesminae (Mitchell, 1966, 1968). Much attention has been given in particular to the 
geologically younger allodesmine pinnipeds, centering on debates regarding the number of 
species, how they relate to one another, and what led to their eventual extinction (Barnes, 1972; 
26 
Barnes and Hirota, 1994; Boessenecker and Churchill, 2018; Tonomori et al., 2018). Knowledge 
of desmatophocid diversity prior to the appearance of the allodesmines, however, has been 
limited to only the two species of desmatophocines, the type species Desmatophoca oregonensis 
and Desmatophoca brachycephala (Condon, 1906; Barnes, 1972; Deméré and Berta, 2002). The 
lack of knowledge of earlier, less derived desmatophocid species hinders our ability to 
confidently resolve the placement and structure of Desmatophocidae within the pinniped 
phylogeny.  
This study sheds new light on family Desmatophocidae by describing a new genus and 
species of desmatophocid represented by a nearly complete cranium. Originating from the Lower 
Miocene Iron Mountain Bed of the Astoria Formation in Oregon, the specimen we describe here 
was collected nearly twenty years ago, erroneously identified as Zalophus, and stored without 
further examination until the present study. As noted by Valenzuela-Toro and Pyenson (2019), a 
considerable number of pinnipedimorph fossils remain unstudied or incorrectly catalogued in 
museum collections, some of which undoubtedly represent species that could illuminate 
previously uncertain regions of the pinniped tree. This study presents a morphological and 
cladistic analysis of just such a specimen. 
Institutional Abbreviations—UOMNCH — University of Oregon Museum of Natural and 
Cultural History, Eugene, Oregon, U.S.A.; UO—University of Oregon, Eugene, Oregon, U.S.A.; 
USNM—United States National Museum of Natural History, Washington, D.C., U.S.A. 
 
 
 
 
27 
 
2 Material and Methods 
2.1 Material Examined 
For the purpose of this study, we examined Allodesmus kernensis (USNM 25227; 
Kellogg, 1922), Desmatophoca brachycephala (USNM 251201, cast of holotype LACM 
120199; Barnes 1987), Desmatophoca oregonensis (UO MNCH F-735, holotype; USNM 
335478; Condon, 1906; Deméré and Berta, 2002), Enaliarctos barnesi (USNM 314295; Berta, 
1991), Enaliarctos emlongi (USNM 250345; Berta, 1991), Enaliarctos mitchelli (USNM 
175637; Barnes, 1979), Enaliarctos tedfordi (USNM 206273; Berta, 1991), Erignathus barbatus 
(USNM 16116, 269126), Imagotaria downsi (USNM 335594; Mitchell, 1968),  Monachus 
monachus (USNM 219059), Pinnarctidion rayi (USNM 250321, 314325; Berta, 1994), 
Proneotherium repenningi (USNM 335526, 215068, 205334; Kohno et al., 1994), Pteronarctos 
goedertae (USNM 250320, 335432; Barnes, 1989), Zalophus californianus (USNM 396062, 
21735), and USNM 335445, an unpublished pinniped skull. USNM 335445 represents an 
undescribed desmatophocid-like pinniped from the Astoria Formation of Central Oregon. 
Preliminary work indicates that it may be sister to UOMNCH F-68583 (described herein) and we 
have included it in our phylogenetic analysis to determine its placement, a full description and 
analysis of USNM 335445 and its relationship to other pinnipeds is beyond the scope of this 
study and will be undertaken separately.  
2.2 Anatomical Description 
Anatomical terminology follows Deméré and Berta (2002) and Boessenecker and 
Churchill (2018) with ontogenetic stage assignment following Siversten (1954). The description 
addresses whichever side of the skull is more complete unless otherwise stated. Measurements 
were taken with Mitutoyo 500-171-30 Digimatic 150 mm digital calipers to the nearest tenth of a 
28 
millimeter. Angles were calculated using the angle tool in tpsDig2 2.31 (Rohlf, 2017). 
Photographs were taken with a Google Pixel 2. 
2.3 Phylogenetic Methods 
We added the new species to the 95-character, 24-taxon data matrix of Boessenecker and 
Churchill (2018) with the addition of three characters from the Tonomori et al. (2018) matrix. 
Our character list is provided in Appendix A. We also included the recently published 
Allodesmus uraiporensis (Tonomori et al., 2018) and the undescribed specimen USNM 335445. 
We could not include the enigmatic ‘Desmatophocine B’ (Barnes, 1972) because it lacks 
identified cranial material. Allodesmus sadoensis was also excluded from the analysis as we 
agree with the assessment of Tonomori et al. (2018), considering this taxon to be a junior 
synonym of Allodesmus sinanoensis. The character matrix was prepared in Mesquite 3.51 
(Maddison and Maddison, 2018) and phylogenetic analysis was performed in TNT 1.5 (Goloboff 
and Catalano, 2016). We coded those characters with polymorphic states as ordered and the 
remainder as unordered. We conducted a new technology search using 10,000 replicates using 
sectorial and tree-fusing search options and an implied weight of K = 3. We assessed support for 
each node using a standard bootstrap analysis with 1,000 replicates. 
 
3 Systematic Paleontology and Anatomical Description 
3.1 Systematic Paleontology 
MAMMALIA Linnaeus 1758 
CARNIVORA Bowdich 1821 
PINNIPEDIA Illiger 1881 
DESMATOPHOCIDAE Hay 1930 
29 
 
3.1.1 Emended Diagnosis of Desmatophocidae—Pinnipedia differing from Phocidae, 
Odobenidae, and Otariidae, in possessing bulbous cheek teeth; a mortised, interlocking contact 
between the jugal and squamosal; a corner-like extension of the pterygoid process of the maxilla 
into a suborbital shelf; and a stylomastoid foramen distinct from the tympanohyal pit. 
3.1.2 Type and Included Species of Desmatophocidae—Eodesmus condoni gen. et. 
sp. nov., Desmatophoca oregonensis, type species (Condon, 1906),  Desmatophoca 
brachycephala (Barnes, 1987), Atopotarus courseni (Downs, 1956), Allodesmus kernensis 
(Kellogg, 1922), Allodesmus naorai (Kohno, 1996), Allodesmus packardi (Barnes, 1972), 
Allodesmus sinanoensis (Nagao, 1941), Allodesmus demerei (Boessenecker and Churchill, 2018), 
Allodesmus uraiporensis (Tonomori et al., 2018). 
EODESMUS gen. nov. 
3.1.3 Type and Only Included Species—Eodesmus condoni gen. et. sp. nov. 
3.1.4 Diagnosis—as for type and only species. 
3.1.5 Etymology—Greek Eo- meaning “dawn”; -desmus (Greek, masculine) from the 
naming conventions used for other taxa of Desmatophocidae and originating from the Greek “to 
bind.” 
EODESMUS CONDONI, sp. nov. 
3.1.6 Holotype—UOMNCH F-68583, nearly complete cranium of a male individual 
lacking the incisive edge of the rostrum; both canines; the posterior edge of the left mastoid; 
right P1, P4, M1, and M2; and left P3, M1, and M2. 
3.1.7 Etymology—In honor of Thomas Condon, who described the first desmatophocid 
(Condon, 1906). 
30 
 
3.1.8 Diagnosis—Eodesmus condoni can be diagnosed from other desmatophocids by 
wide anterior nares laterally bounded by the maxilla rather than premaxilla, the presence of a 
nasiolabialis fossa, a circular rather than dorsoventrally compressed foramen magnum, and 
posteroventrally protruding paraoccipital processes. 
3.2 Geographical Location and Geology 
3.2.1 Locality—The type specimen was found in float at UO Locality 4663 Moolack 
Beach (alternatively spelled as Moloch Beach in some publications, e.g., Deméré and Berta 
(2002) just south of Beverly Beach State Park, Lincoln County, Oregon (Fig. 1). This specimen 
was donated to the UOMNCH in December 2015 after collection by Andrew Berkholtz in the 
early 2000s. This locality is Paleobiology Database Locality Number 190995.  
3.2.2 Horizon—The type specimen comes from the Iron Mountain Bed, a prominent 
horizon representing a near-shore facies in the Newport Embayment of the Astoria Formation 
(Armentrout, 1981). The Iron Mountain Bed consists of calcareously cemented, well-sorted 
volcaniclastic sandstone characterized by numerous concretions of varying sizes (Armentrout, 
1981). The type specimen was preserved in a fine-grained, extremely hard concretionary boulder 
discovered in the boulder fields at Moolack Beach (spelled “Moloch” on the U.S. Geological 
Survey, Yaquina, Oregon, 15-minute topographic map, 1959 edition). Such concretions are 
characteristic of the Iron Mountain bed, which outcrops for 1.5 km at Moolack Beach (Munthe 
and Coombs, 1979; Armentrout, 1981; Barnes, 1990). Other higher and lower sections of the 
Astoria Formation, the singular Tertiary rock unit that outcrops at Moolack Beach (Barnes, 
1990), largely consist of friable, poorly sorted sandy mudstones that differ significantly in 
lithology from the Iron Mountain bed (Colbath, 1985). The Iron Mountain bed has produced 
31 
Figure 1. Map of central Oregon coastline with exposures of Astoria Formation with type 
locality for Eodesmus condoni gen. et. sp. nov. (UOMNCH F-68583) noted.  
numerous vertebrate fossils preserved in similar concretions, including the type specimens for 
Proneotherium repenningi, Pacifocotaria hadromma, and Desmatophoca oregonensis (Condon, 
1904; Barnes, 1992; Kohno et al., 1994; Deméré and Berta, 2002). Enaliarctos specimens have 
also been recovered from this horizon (Poust and Boessenecker, 2018). Notable non-
pinnipedimorph taxa from this horizon include the baleen whale Cophocetus oregonensis 
(Packard and Kellogg, 1934) and the desmostylian Desmostylus hesperus (Domning, 2008), as 
well as terrestrial taxa such as the chalicothere Tylocephalonyx and the rhinoceros Aphelops 
(Coombs, 1979). Paleomagnetic data by Prothero et al. (2001) have dated the Iron Mountain Bed 
to 17.3–16.6 Ma, and these data along with the terrestrial fauna preserved here correlate this bed 
with the Hemingfordian. 
3.3 Anatomical Description 
Measurements taken from the skull are recorded in Table 1. 
3.3.1 Ontogenetic Stage and Sex—The type specimen displays well-fused cranial 
sutures (suture age 35; Siversten, 1954). Its pronounced sagittal and lambdoidal crests indicate 
that the Eodesmus condoni holotype likely represents an adult male. 
32 
 
Table 1. Measurements (in millimetres) of cranial dimensions of Eodesmus condoni (gen. et. 
sp. nov.) (UOMNCH F-68583). Alveolar measurements are recorded from the right toothrow. 
 
  
Total length, premaxilla to intercondylar notch 281.1  
Facial length 171.6  
Orbit length 62.1  
Temporal fossa length 40.8  
C1-M2 toothrow length 97.7  
Braincase length 108.4  
Braincase depth 72.7  
Greatest transverse braincase width 103.6 
Greatest width of anterior nares 43.3 
Greatest length nasals 66.4 
Greatest width nasals 30.7 
Transverse width of rostrum at C1 64.4 
Narrowest width of rostrum 60.9 
Transverse width of palate at M2 74.3 
Bizygomatic width  161.0 
Transverse width at mastoid 163.0 
Transverse width across tympanic bullae 103.0 
Transverse width at paraoccipital process 122.7 
Transverse width of condyles 68.4 
Foramen magnum width 21.3 
Foramen magnum depth 23.0 
Anteroposterior length, mastoid and paraoccipital process 57.5 
Width of internal choanae between pterygoid hamuli 23.2 
Infraorbital foramen, depth/transverse width 9.3/12.7 
Least interorbital width 39.1 
Anteroposterior length of sagittal crest 124.8  
Left P1 crown height/anteroposterior length 9.0/9.1 
Right P2 crown height/anteroposterior length 12.4/11.2 
Left P2 crown height/anteroposterior length 11.4/10.4 
Right P3 crown height/anteroposterior length 11.5/11.3 
Left P4 crown height/anteroposterior length 9.2/11.4 
C1 alveolus anteroposterior length/transverse width 13.1/9.1 
M1 alveolus anteroposterior length/transverse width 8.0/5.2 
M2 alveolus anteroposterior length/transverse width 6.1/6.0 
  
 
3.3.2 Rostrum—The rostrum is anteroposteriorly elongate (74.6 mm as measured from 
the anterior margin of the narial opening to the anterior extent of the orbit; 27% of condylobasal 
length; Figs. 2, 3). The premaxilla forms the anterior rostral margin and is slightly anteriorly 
expanded to form a small prenarial shelf. The rostral margin is smoothly rounded, lacking the 
33 
 
distinctive transverse expansion for accommodation of the canines seen in dorsal view of other 
desmatophocids. The premaxillary-maxillary suture is well-fused and extends from the 
anterolateral margin of the rostral border to the lateral portion of the floor of the external nares.  
Unlike in all other desmatophocids, the premaxilla does not contact the nasals laterally and does 
not contribute to the lateral walls of the external nares at all. Instead, it defines only the floor of 
the premaxilla, forming an anteroposteriorly elongate process on either side of the nares, 
extending posterolaterally from the rostral border into the nares. These processes bound a medial, 
posteriorly flaring fossa between them (Fig. 3). The anterior narial opening is wide compared to 
the rest of Desmatophocidae at 66% of the rostral width at the canine alveoli, compared to 38-
45% for specimens of Desmatophoca oregonensis and less for other desmatophocids as 
published in Deméré and Berta (2002). This comparative width appears to derive from the lack 
of contribution of the premaxilla to the lateral borders of the external nares. 
Directly anterior to the orbit, the maxilla preserves a shallow but distinct nasolabialis 
fossa, a plesiomorphic characteristic seen in stem pinnipeds and the fossil phocid Devinophoca 
but not in other desmatophocids. Just lateral to the nasolabialis fossa, the confluence of the 
maxillofrontal suture with the anterior rim of the orbit forms a poorly developed antorbital 
process, a characteristic trait of desmatophocids. Along the lateral border of the maxilla, a low 
ridge extends posteriorly from the anterior aspect of the canine alveolus to the anterior root of the 
zygomatic arch, forming a shallow fossa below it and above the toothrow. Just anterolateral to 
the anterior extent of the maxilla-nasal suture, the maxilla projects slightly over the external 
nares in a low, blunted, anteromedially directed process. Laterally, the maxilla is slightly flared at 
the level of the canines. Anteriorly, the nasals protrude over the nares in an M-shaped projection 
and extend posteriorly to form a narrow V-shaped suture with the frontal at approximately the 
34 
 
 
Figure 2. Eodesmus condoni gen. et. sp. nov. (UOMNCH F-68583). Holotype skull in dorsal (a, 
b) and ventral (c, d) views. 
35 
 
 
Figure 3. Eodesmus condoni gen. et. sp. nov. (UOMNCH F-68583). Holotype skull in lateral (a, 
b), rostral (c, d), and occipital (e, f) views. 
36 
 
level of the origination of the sagittal crest. The lateral edges of the nasals are approximately 
parallel to one another anterior before narrowing at the frontal-nasal-maxillary juncture, where 
they then converge into a narrow V-shaped nasal-frontal suture.  
On the ventral aspect of the rostrum, the palate displays slight dorsal arching in the 
transverse plane and is widest at the level of M2. Two distinct, ovoid incisive foramina separated 
by a thin septum are preserved between the canine alveoli. At the level of P4, palatal foramina 
are also preserved. The anterior set are proceeded by shallow grooves and the second set are just 
posterior to the first. The toothrows are roughly parallel to one another up to the level of P4, 
where they flare slightly, extending onto the suborbital shelf formed by the palatine process of 
the maxilla. The lateral border of this shelf contacts the ventral root of the zygomatic process at a 
45-degree angle just anterior to the M2 alveolus. The palatine-maxillary suture occurs at the level
of the M2 alveolus and is medially transverse before abruptly flexing posterolaterally and 
extending to the posterior extent of the suborbital shelf. All other palatal sutures have been 
obliterated. From the lateral extent of the palatine-maxillary suture, the lateral borders of the 
palatine contract medially, forming the posterior corner of the suborbital shelf, before extending 
posteriorly to contact the pterygoid. The posterior border of the palatine is smooth and bluntly 
rounded, forming the anterior walls of the internal nares together with the pterygoids forming the 
posterior walls.  
In lateral view, the pterygoid hamuli are triangular, gracile, laterally recurved, and 
laterally compressed. The pterygoid struts are widest at the level of the hamuli. They display a 
rugose texture and are characterized by two ridges extending posterolaterally from laterally 
adjacent to the hamuli and subsequently merging at the level of the common pit for the foramen 
37 
ovale and alisphenoid canal; the resultant singular ridge extends to the medial corner of the 
auditory bulla.  
3.3.3 Orbital Region and Zygomatic Arch—In lateral aspect, the interorbital region is 
high and of approximately equal height with the braincase. The interorbital region as comprised 
by the frontals is relatively narrow. The sagittal crest initiates at approximately the level of the 
termination of the nasals. Along its length, the sagittal crest rises between 9.8 and 11.7 mm from 
the skull roof. The inner walls of the orbits are damaged, obscuring sutures, foramina, and the 
presence or absence of the orbital fissure. The lacrimal bone is not preserved. 
Anteriorly, the jugal-maxillary suture is fully fused. Within the maxillary root of the 
zygomatic process, the infraorbital foramen is ovoid and exhibits slight anteroventral 
compression. Dorsally, its posterior outlet is visible within the orbit. The floor of the infraorbital 
foramen merges posteriorly with the suborbital shelf/pterygoid process of the maxilla. The 
zygomatic arch is laterally compressed and quite narrow from the dorsal perspective. In dorsal 
view, it extends laterally from the rostrum at the level of P3 at approximately 140° before 
abruptly recurving medially at the level of the glenoid fossa, where the zygoma is at its widest. 
At this juncture, it becomes dorsoventrally flattered. This flattened facet of the zygomatic root of 
the squamosal forms the squamosal fossa along with the dorsal aspect of the mastoid process. A 
slight ridge divides these two anterior and posterior portions of the squamosal fossa, which are 
approximately equal in size. This ridge overlies the external auditory meatus. The jugal-
squamosal articulation is mortised and dorsoventrally expanded (Fig. 3). The posteroventral root 
of the jugal extends posteriorly to the level of the lateral extent of the glenoid fossa, forming a 
sigmoidal suture with the squamosal. The dorsal root of the jugal is triangular and forms the 
postorbital process; this process extends dorsally and interlocks posteriorly with the squamosal.  
38 
 
3.3.4 Braincase and Occiput—The braincase meets the interorbital region at an 
approximately right angle, forming a distinct corner. The braincase appears to have been 
dorsoventrally compressed post-burial, exhibiting fracturing along its lateral edges. These 
fractures obscure the region of the pseudosylvian sulcus. From its origin at the nasofrontal 
suture, the sagittal crest extends medially along the length of the frontals and parietals until 
merging with the lambdoidal (nuchal) crests. The lambdoidal crests diverge posterolaterally from 
the interparietal suture at 235° degrees to one another and posteriorly overhang the occipital 
shield. They then curve ventromedially before merging laterally with the mastoid process. The 
mastoid processes are large and dorsally triangular. The paraoccipital processes are not visible in 
dorsal view and are discussed below (see Basicranium). As a whole, the dorsal surface of the 
paraoccipital processes, the posterior surface of the mastoid processes, the posterolateral 
braincase, and the overhanging, anteriorly directed lambdoidal crests form deep lateral fossae on 
the braincase. The occipital shield exhibits some diagenetic fracturing but is otherwise relatively 
smooth. From its crest, the shield is slightly concave. From the lateral extent of the lambdoidal 
crests, the occipital shield narrows sharply before intersecting the dorsomedial surface of the 
occipital condyles. The occipital condyles are positioned unusually close to one another for a 
desmatophocid, bounding a small, rounded foramen magnum rather than the typical wide, 
dorsoventrally compressed morphology seen in the rest of Desmatophocidae (Fig. 3). 
3.3.5 Basicranium—On the whole, the basicranium is broadly similar to that of 
desmatophocines. Anteriorly, the basisphenoid-basioccipital suture is visible but crushed, 
obscuring its exact morphology. Overall, the basioccipital is generally pentagonal. It is also 
characterized by the presence of a low, posteriorly narrow sagittal ridge that widens anteriorly 
before flattening at the level of the anterior extent of the auditory bullae. Anterior to this ridge, 
39 
 
the basioccipital is gently concave. Just anterior to the occipital condyles and on either side of 
this low ridge are the hypoglossal foramina, which are small and blind. The paraoccipital 
processes are large and triangular with excavated ventral surfaces. Unusually, they project from 
the posterior aspect of the basicranium at a sharp ventrolateral angle as opposed to the 
posteriorly directed paraoccipital processes of other desmatophocids. The mastoid processes 
project laterally and are approximately equal in size to the posteriorly located paraoccipital 
processes. Together, the anterior border of the mastoid process and the postglenoid process form 
an arched, bony roof that overhangs the anterolaterally directed external auditory meatus.  
The auditory bullae are relatively flattened. Anteromedially, a small hooked projection extends 
medially over the common fossa for the auditory tube and median lacerate foramen. The ventral 
surfaces of the bullae are characterized by a recurved ridge originating at the posteromedial 
corner of the bulla before forking. One branch runs posterolaterally, just lateral to the carotid 
canal and posterior lacerate foramen to the anterolateral corner of the bulla. The other curves 
anterolaterally before flattening just medial to the external auditory meatus. The bullae protrude 
slightly posteroventrally, extending partially over the stylomastoid foramen and tympanohyal pit. 
The stylomastoid foramen and tympanohyal pit are small and of approximately equal size, 
obliquely ovoid, and directed dorsomedially into the bulla. Posteromedially to the bullae, the 
posterior lacerate foramina are large and elliptical with their main axes located transversely 
rather than anteroposteriorly as in Desmatophoca oregonensis. 
3.3.6 Dentition—The preserved dentition and alveoli indicate an upper dental formula 
of I1-3, C1, P1-4, M1-2. Although the incisive border and incisor alveoli are not preserved, all 
pinnipeds except the most derived odobenids have three incisors and we have no reason to 
expect this species to differ (Drehmer et al., 2015; Winer et al., 2016). Preserved teeth include 
40 
 
the left P1, right and left P2, right P3, and left P4. Overall, the teeth are bulbous, peg-like, and 
homodont with vertically oriented roots. The left P1 is single-rooted and the alveolus of the right 
P1 both display single roots. P1 and the alveoli of M1 and M2 are single-rooted; P2-P4 are 
double-rooted. Each tooth features a single, bulbously triangular principal cusp. Buccally, the 
enamel is smooth and slightly more bulbous at the base. Lingually, a faintly crenulated cingulum 
is apparent on all of the teeth. Wear facets are present on the posteroventral surface of each of the 
teeth as well. The alveoli of M1 and M2 are distinctly smaller than those of the premolars. 
3.3.7 Comparison to Other Pinnipedimorpha— Eodesmus condoni represents an 
important specimen for understanding the origination of Desmatophocidae as it possesses a 
unique combination of basal and derived traits. Unlike the basal pinnipedimorphs, E. condoni 
possesses characteristics of crown Pinnipedia such as homodont teeth, absence of the lacrimal 
bone and lacrimal foramen, and an expanded infraorbital foramen (Berta et al., 2018). However, 
it also possesses more basal traits such as the presence of a nasiolabialis fossa—a basal trait it 
shares with basal pinnipedimorphs and Devinophoca, the earliest fossil seal known from cranial 
material (Koretsky and Holec, 2002). Additionally, while the interorbital morphology is not 
preserved in Eodesmus condoni, the possible sister taxon of E. condoni, USNM 335445, lacks an 
orbital vacuity, as do the basal pinnipedimorphs and Desmatophoca. 
Several characteristics preclude inclusion of E. condoni with any of the other three crown 
pinniped families. Presence of an M2, prenarial shelf, alisphenoid canal, pronounced 
paraoccipital processes, flattened tympanic bullae, and the shelf-like morphology of the 
pterygoid process of the maxilla exclude this specimen from Phocidae. E. condoni differs from 
Otarioidea (Otariidae + Odobenidae) in its V-shaped nasal-frontal suture, thin and laterally 
directed pterygoid strut, divided squamosal fossa, reduced M1 that is smaller than the premolars, 
41 
 
and its lack of antorbital and supraorbital processes. Additionally, its slightly divergent tooth 
rows, the smoothly convex anterolateral margin of the braincase, and the pentagonal shape of the 
basioccipital all rule out membership in Otariidae. Finally, it shares the diagnostic characters of 
its mortised jugal-squamosal articulation (Fig. 3) and the presence of a tympanohyal pit that is 
distinct from the stylomastoid foramen (Fig. 2) with the rest of Desmatophocidae. 
 
4 Phylogenetic Results and Discussion 
Our phylogenetic analysis with implied weighting recovered two most parsimonious trees 
with a best score of 32.661, consistency index = 0.443, and retention index = 0.683. We show the 
50% consensus tree and report support values for this tree in Figure 4. As this paper focuses on 
desmatophocids, we limit our discussion of specific level relationships and synapomorphies to 
this clade. Overall, the relationships among major pinniped clades remain similar to the original 
results published with the Boessenecker and Churchill matrix (2018). Enaliarctos, Pteronarctos, 
and Pinnarctidion form a successive grade into crown Pinnipedia. Within Pinnipedia, our 
analysis resolved two lineages, Phocoidea (Desmatophocidae + Phocidae) and Otarioidea 
(Otariidae + Odobenidae). Eodesmus condoni sp. nov. and a yet undescribed specimen, USNM 
335445, resolve as sister taxa in the most basal clade within Desmatophocidae. Together, these 
two specimens form a sister group to Desmatophocinae + Allodesminae in a poorly supported 
topology. This new taxon can be distinguished from the rest of Desmatophocidae by one 
ambiguous synapomorphy, the lack of contact between the premaxilla and the lateral border of 
the nasals (shared with Odobenus; character 1: state 2), and a symplesiomorphy, the presence of 
a nasiolabialis fossa (shared with basal pinnipedimorphs and Devinophoca; 11:0).  
 
42 
 
Figure 4. Phylogenetic tree showing position of Eodesmus condoni gen. et. sp. nov (UOMNCH 
F-68583). Time-calibrated most parsimonious tree from the analysis using implied weighting (K
= 3). Terminal horizontal bars represent the age range of the taxa as determined by Boessnecker
and Churchill (2018). We assigned the age range of Eodesmus condoni and USNM 335445 as
17.3–16.6 Ma because each of these specimens comes from the same horizon as Proneotherium
repenningi, which was assigned to this time range by Boessenecker and Churchill (2018). Time-
scaling was applied to the tree using the strap package (Bell and Lloyd, 2015) in R (R Core Team,
2019).
43 
Additionally, inclusion of Eodesmus condoni gen. et sp. nov. provided better resolution to 
the topology of Desmatophocidae as a whole. Desmatophocidae is united by two unambiguous 
synapomorphies: the pterygoid process of the maxilla expanded laterally as into a corner-like 
suborbital shelf (8: 1) and a stylomastoid foramen distinct from the tympanohyal pit (33: 1). 
Desmatophocidae also possesses one ambiguous synapomorphy that it shares with Imagotaria, 
Gomphotaria, and Odobenus: round and bulbous (as opposed to transversely narrow) lower 
postcanines. Similarly, Desmatophoca is here recovered with moderate support (BS = 76) rather 
than poor (Boessenecker and Churchill, 2018). It is diagnosed by a single unambiguous 
synapomorphy: a mortised jugal-squamosal articulation without the dorsoventral expansion 
characteristic of other desmatophocids (22:1). Desmatophoca also shares an ambiguous 
synapomorphy with basal pinnipedimorphs, the lack of an orbital vacuity (17:0).  
Allodesminae (Atopotarus + Allodesmus) is recovered with weak support (BS = 53) 
showing two unambiguous synapomorphies: divergent lower cheek tooth orientation (59:1) and 
the presence of a scapular undulation in place of the secondary spine on the scapulae (72:1). 
Ambiguous synapomorphies of allodesmines include an absent or indistinct genial tuberosity of 
the mandible (40:0; reversal to basal character trait; 40:0), the absence of a cingulum on the 
lower postcanines (61:2; shared with Gomphotaria and Odobenus; 61:2), and a large, 
dorsoventrally elliptical anterior narial opening with thick borders (shared with Imagotaria, 
Gomphotaria and Odobenus; 96:1). Within Allodesminae, Atopotarus resolves as the sister taxon 
to Allodesmus. The remaining allodesmine species are united with moderate support (BS=79). 
Allodesmus packardi and Allodesmus naori form a successively more derived grade leading to 
the final clade, a weakly supported polytomy (BS=52) containing Allodesmus uraiporensis, 
Allodesmus kernensis, Allodesmus demerei, and Allodesmus sinanoensis. Notably, we recovered 
44 
 
a very different placement for Al. uraiporensis and Al. kernensis than seen in the Tonomori et al. 
(2018) publication of this specimen. Tonomori et al. (2018) recovered Al. uraiporensis and Al. 
kernensis in a weakly supported (BS=53) clade intermediate between Atopotarus and the more 
derived allodesmines. However, as neither topology is well-supported, we do not consider this 
difference to be notable beyond the indication that the topology of Allodesminae is as yet 
unsettled. 
One of the most notable features of this and similar phylogenetic analyses is the failure to 
conclusively resolve high level relationships within Pinnipedia. In our analysis, Pinnipedia itself 
is relatively poorly resolved (BS = 59), and both Phocoidea and Otarioidea are poorly supported. 
Even the monophyly of Odobenidae, one of the four traditionally recognized families within 
Pinnipedia, is poorly supported. These weaknesses parallel those of other recent phylogenetic 
analyses of pinnipeds (Deméré and Berta, 2002; Dewaele et al., 2017; Boessenecker and 
Churchill, 2018; Tonomori et al., 2018). Although most recent morphological phylogenetic 
analyses have recovered Phocidae as the immediate sister group to Desmatophocidae, this 
relationship has been poorly supported in each study. As noted by Boessenecker and Churchill, 
non-cladistic analyses have generally grouped Desmatophocidae with otariids and odobenids 
(Downs, 1956; Barnes, 1987, 1989; Barnes and Hirota, 1994). Though our results support the 
cladistically derived hypothesis, the poor support for both Phocoidea and Otarioidea in our 
analysis and those of others call for more extensive future investigation of these family level 
relationships with emphasis on the placement of Desmatophocidae. The lack of robust support 
for the placement of this extinct family demonstrates that the interfamilial relationships in crown 
Pinnipedia have yet to be fully settled. 
 
45 
 
5 Paleobiogeographic Context 
Geospatially, Eodesmus condoni co-occurs with several basal pinnipedimorphs and early 
crown pinnipeds also found at the Iron Mountain bed. Basal taxa include Enaliarctos sp. (Poust 
and Boessenecker, 2018), Pacificotaria hadromma (Barnes, 1992), and Pteronarctos goedertae 
(Berta, 1994). Besides E. condoni, crown pinnipeds from this locality include the desmatophocid 
Desmatophoca oregonensis (Condon, 1906; Deméré and Berta, 2002) and the earliest known 
odobenid, the small-bodied walrus Proneotherium repenningi (Kohno et al., 1994). The presence 
of at least six pinnipedimorph species at a single locality represents a remarkable taxonomic 
diversity, particularly when compared to other similar localities from the early Miocene and late 
Oligocene, which typically only preserve one or two species of pinnipedimorph. Notably, this 
assemblage coincides with the Mid-Miocene Climatic Optimum (MMCO), a period of relative 
warming and subsequently increased ocean productivity from 18.0–16.5 Ma (Flower and 
Kennett, 1993; Bohme, 2003). Previous studies have indicated that this elevated productivity 
drove an increase in marine species richness (Lipps and Mitchell, 1976; Tsuchi, 1992; Pyenson et 
al., 2009). The diversity of the Iron Mountain bed represents an intriguing case study for further 
investigation of this hypothesized process. However, greater understanding of the genesis of the 
Iron Mountain bed with particular attention to the effects of time-averaging and other 
taphonomic biases is needed to satisfactorily investigate the relationship between 
pinnipedimorph diversity and the MMCO (Pyenson et al., 2009).  
Although Eodesmus condoni represents the most basal desmatophocid yet described, the 
more derived Desmatophoca brachycephala precedes it chronologically. The holotype and single 
known specimen of D. brachycephala derives from the lower Astoria Formation and is dated to 
23.0–20.4 Ma (Barnes, 1987). This occurrence of a more derived desmatophocid at least 3.1 
46 
 
million years before the appearance of Eodesmus condoni indicates the presence of a ghost 
lineage of eodesmine desmatophocids during the Aquitanian at the very least. Further fieldwork 
at older localities is needed to illuminate the history of this clade. 
 
6 Conclusions 
Here we describe a new genus and species of desmatophocid pinniped, Eodesmus 
condoni gen. et. sp. nov. from the Iron Mountain bed of the Astoria Formation of Oregon based 
on a nearly complete cranium. Our phylogenetic analysis resolves this new taxon as the most 
basal desmatophocid species identified to date. Our analyses also provide better resolution to the 
topology of the Desmatophocidae. Eodesmus condoni expands our knowledge of early 
pinnipedimorph diversity and further elucidates the relationships among fossil desmatophocids. 
However, our failure to recover strong support for high-level relationships among the major 
pinniped clades highlights the need for further phylogenetic work with an expanded subset of 
taxa and characters. 
 
7 Bridge 
 As our understanding of the early diversity of otarioids expands with the description of 
new species like E. condoni, we gain new insight into when and how pinnipeds integrated into 
marine ecosystems. The way pinnipeds and other organisms interact with their respective 
ecosystems is largely a product of their locomotor mode. Whereas true seals and walruses 
produce thrust while swimming through lateral oscillation of their hindflippers, eared seals swim 
using forelimb propulsion in a stroke cycle sometimes known as “underwater flying.” How and 
why these idiosyncratic swimming modes arose remains an outstanding mystery in pinniped 
47 
 
paleoecology. In the next chapter, I turn my attention specifically to the evolution of otarioid 
locomotor modes to investigate the development of modern pinniped ecology. 
48 
 
 
CHAPTER III 
 
FORELIMB-DOMINATED AQUATIC PROPULSION EVOLVED THREE   
TIMES IN OTARIOID PINNIPEDS AS REVEALED BY GEOMETRIC 
MORPHOMETRIC ANALYSES OF MODERN AND FOSSIL HUMERI 
 
 
1 Introduction 
 The Pinnipedia (Illiger, 1811) are one of the most species-rich clades of secondarily 
aquatic vertebrates in the modern ocean. Second among marine tetrapods only to the cetaceans 
for number of species (Berta, 2020), modern pinnipeds include three major taxa: the Phocidae 
(Gray, 1821), also known as the true or earless seals (Fig. 1a); the Otariidae (Gray, 1825), the fur 
seals and sea lions (Fig. 1b); and the Odobenidae (Fig. 1c; Allen, 1880), sister taxon to the  
FIGURE 1. The three modern subclades of Pinnipedia. A, Phocidae, or true seals, showing 
clockwise from top lefthand corner: Mirounga leonina, the southern elephant seal, in the 
Kerguelen Islands (photo by Antoine Lamielle); a juvenile Pagophilus groenlandicus, the harp 
seal, at False Cape Beach in Virginia, U.S.A. (photo by Albert Herring); and Lobodon 
carcinophaga, the crabeater seal, hauled out on the ice near Almirante Brown Station in Antarctica 
(photo by W. Bulach). B, Otariidae, or eared seals, showing from top to bottom: Eumetopias 
jubatus, the Steller sea lion, in a rookery in Alaska, U.S.A. (public domain photo by U.S. Fish and 
Wildlife Service); and Zalophus californianus, the California sea lion, resting near La Jolla, 
California, U.S.A. (photo by Mike’s Birds). C, Odobenus rosmarus, the walrus and sole extant 
member of Odobenidae, on an ice floe in the Arctic (public domain photo by U.S. Fish and Wildlife 
Service). All photos are from Wikimedia Commons.  
49 
 
 
Otariidae and today represented by the singular extant walrus species Odobenus rosmarus 
(Linnaeus, 1758). Unlike other extant marine mammals like cetaceans and sirenians, pinnipeds 
have a uniquely amphibious lifestyle, relying on both the terrestrial and aquatic realms for 
survival (King, 1991; Berta et al., 2018). Furthermore, the swimming modes employed by 
pinnipeds differ significantly not only from those seen in other marine mammals but also 
between each of the three families. Phocids chiefly propel themselves via lateral oscillation of 
their hind flippers whereas otariids employ a bilateral stroke of the forelimbs for propulsion; O. 
rosmarus employs a type of combination swimming, sometimes using an otariid-like stroke 
cycle of the forelimbs at low speeds but relying on pelvic oscillation for high speed swimming 
(Howell, 1929; Backhouse, 1962; English, 1977; Gordon, 1981; Berta et al., 2018). 
Such diversity of locomotor modalities is distinct from all other modern marine mammals 
which swim via dorsoventral undulation of either their tails, in the case of whales and sirenians, 
or their tails and hindlimbs, in the case of sea otters (Fish, 1996, 2016; Gingerich, 2003). Indeed, 
the disparity between pinniped swimming modes formed one of the primary arguments for 
pinniped diphyly, with otariids and odobenids hypothetically arising from one terrestrial ancestor 
and the phocids from another (Koretsky and Holec, 2002; Kuhn and Frey, 2012; Koretsky et al., 
2016). However, molecular phylogenetic analyses have unequivocally supported a monophyletic 
Pinnipedia (Berta and Wyss, 1994; Arnason et al., 2006; Fulton et al., 2010; Nyakatura and 
Bininda-Emonds, 2012; Luan et al., 2013), raising the question of how these differing locomotor 
modes evolved from a single ancestral pinniped and what selective forces may have influenced 
the development of such different swimming modes. 
Members of carnivoran clade Arctoidea, pinnipeds share a common ancestor with ursids 
(bears) and musteloids, which include the mustelids (weasels), procyonids (raccoons), mephitids 
50 
 
 
(skunks), and ailurids (red pandas) (Finarelli, 2008; Nyakatura and Bininda-Emonds, 2012; Luan 
et al., 2013). Both the ursids and musteloids also include secondarily aquatic species. Among the 
Ursidae, the polar bear Ursus maritimus is considered a marine mammal and although primarily 
adapted for ambulation on ice and land in terms of locomotor behavior, U. maritimus can swim 
long distances via alternating paddle-strokes of its forelimbs while dragging its hindlimbs (Fish, 
1996; Lone et al., 2018). Several fossil and modern musteloids display varying degrees of 
aquaticism, from the semi-aquatic American mink Neovison vison and European river otter Lutra 
lutra to the fully aquatic sea otter Enhydra lutris (Mori et al., 2015), which is more behaviorally 
aquatic than any pinniped (Berta, 2020). Musteloids employ a range of swimming modes. Less 
aquatically specialized species swim using strategies from the quadrupedal paddling used by the 
mink Mustela vison to the somewhat unusual pectoral paddling observed in the domestic ferret 
Mustela putorius furo to the range of forelimb and/or hindlimb paddling and dorsoventrally 
undulatory swimming modes seen in American river otter Lutra canadensis (Fish, 1994). E. 
lutris uses its dorsoventrally flattened tail and flipper-like hindlimbs to provide thrust while 
swimming via dorsoventral undulation (Tarasoff et al., 2008), not altogether unlike the 
dorsoventral caudal undulation employed by whales and sirenians (Fish, 1996). 
Perhaps surprisingly, the locomotor modes used by pinnipeds differ significantly from the 
swimming methods used by other arctoids, obscuring the trajectory by which the pinniped 
swimming modes may have arisen (Fish, 2016). When phocids or O. rosmarus are swimming 
rectilinearly, the animals propel themselves via alternating strokes of each hind flipper using 
lateral undulation of their hindlimbs, held with solar surfaces facing one another, via alternating 
adduction and abduction of the pes in the horizontal plane to produce thrust (Fish, 1996). They 
hold their forelimbs adducted at the side, pressed into bilaterally present axillary depressions in 
51 
 
the blubber of the torso to produce a hydrodynamic body shape for the reduction of drag 
(Backhouse, 1962; Fish, 1996). Although phocids and odobenids primarily rely on the hindlimbs 
for generating thrust while swimming, the forelimbs also play important roles in aquatic 
locomotion (Howell, 1929; Backhouse, 1962; Gordon, 1981, 1983). To execute turns, phocids 
and O. rosmarus will employ the forelimbs to produce lift by abducting the forelimb, extending 
the joints distal to the elbow, and maintaining the manus in a mid-prone-supine position so that 
the radius and radial digit form the leading edge of a hydroplane (Howell, 1929; Backhouse, 
1962; Gordon, 1981). Occasionally, phocids will also use the foreflippers to paddle using 
alternating strokes when swimming slowly or will paddle on one side of the body to provide 
extra thrust for a tight turn. O. rosmarus uses this modality even more frequently than phocids, 
particularly while foraging along the seafloor (English, 1977; Gordon, 1981). Paddling is 
accomplished by first moving the forelimb into position via the recovery stroke, performed by 
extending the forearm and manus via elbow flexion, followed by the power stroke, in which the 
foreflipper is adducted, retracted, and pronated with the digits spread to provide maximal water 
resistance and therefore thrust (Backhouse, 1962). O. rosmarus also uses its forelimbs as 
stabilizers when it is feeding on the sea floor (Fay, 1982). 
In contrast with both phocids and odobenids, otariids primarily employ pectoral 
oscillation of the forelimbs for thrust production with the hindlimbs passively trailed behind 
(English, 1977; Feldkamp, 1987a; Fish, 1996; Kuhn and Frey, 2012; Berta et al., 2018). At the 
beginning of the power stroke, the shoulder joint is in an extended position with the humerus in 
an abducted-protracted position, the elbow flexed, and the manus slightly extended (Howell, 
1929; English, 1977). Subsequently, the shoulder joint undergoes flexion and the forelimb is 
simultaneously medially rotated, adducted, and retracted, with extension of the elbow joint and 
52 
 
flexion of the manus, so that at the end of the stroke the palmar surface of the flippers is 
appressed against the flank (Howell, 1929; English, 1977). To reposition the forelimb for the 
next power stroke via the recovery stroke, the foreflipper is then laterally rotated before the 
shoulder joint undergoes extension, protraction, and abduction, the elbow joint is flexed, and the 
manus is slightly extended (Howell, 1929; English, 1977). The hindlimbs are apparently less 
involved in otariid locomotion than the forelimbs are in that of phocids and odobenids, but they 
do function as rudders and occasionally brakes (English, 1977). 
Given the disparity between pinniped swimming modes and those of their fissiped 
(terrestrial) arctoid relatives, understanding how and when these disparate swimming methods 
evolved requires investigation into the aquatic locomotion of fossil pinnipeds. The pinniped 
fossil record consists of over 100 extinct species (Valenzuela-Toro et al., 2018), including a 
number of stem “pinnipedimorphs” (herein referred to as stem pinnipeds), the enigmatic 
Desmatophocidae (Hay, 1930) which lived in the North Pacific from the mid-Arikareean to the 
early Hemphillian (Boessenecker and Churchill, 2018), and numerous extinct members of the 
three extant clades. Most previous studies have focused on reconstructing the locomotor modes 
of stem pinnipeds Puijila darwini (Paterson, 2017; Paterson et al., 2020) and Enaliarctos mealsi 
(Berta et al., 1989; Berta and Ray, 1990; Berta and Adam, 2001; Bebej, 2009; Furbish, 2015) and 
derived desmatophocid Allodesmus kernensis (synonymized with Allodesmus kelloggi by 
Deméré and Berta, 2002, and Boessenecker and Churchill, 2018; Berta and Adam, 2001; Bebej, 
2009; Furbish, 2015), as they are each represented by nearly complete specimens. Notably, each 
of these three taxa display combinations of traits that, in modern species, are typically associated 
with either hindlimb- or forelimb-dominated propulsion, leading previous workers to infer that 
these species used a sort of combination-style swimming that involved both the forelimbs and 
53 
 
 
hindlimbs in propulsive strokes, not dissimilarly to O. rosmarus (Berta and Ray, 1990; Bebej, 
2009; Furbish, 2015; Paterson, 2017). Nevertheless, each seems to show a greater morphological 
affinity for one style over the other. Basal taxa P. darwini and E. mealsi have generally been 
interpreted as combination swimmers with a preference for hindlimb propulsion (Bebej 2009, 
Paterson 2017). These results lend support to the hypothesis of an ancestral hindlimb-dominated 
swimmer, which Hocking et al. (2018) attribute to the need to liberate the forelimbs for prey 
manipulation while traveling in the aquatic environment. By this logic, the phocid swimming 
modality is a further specialization of the ancestral hindlimb-dominated swimming mode 
(Hocking et al., 2018, 2021). 
Although significant evidence exists supporting the plesiomorphy of hindlimb propulsion 
in pinnipeds, the evolution of forelimb propulsion remains perplexing. A. kernensis, the only 
otarioid fossil known from a generally complex skeleton, has typically been resolved as a 
forelimb-dominated combination swimmer (Bebej, 2009; Furbish, 2015). The phylogenetic 
placement of the desmatophocids is controversial (Dewaele et al., 2017b; Boessenecker and 
Churchill, 2018; Paterson et al., 2020; Tate-Jones et al., 2020), but if Desmatophocidae is an 
otarioid as the most recent total evidence phylogenies indicate (Paterson et al., 2020; Rule et al., 
2020), a forelimb-propelled A. kernensis in conjunction with the forelimb propulsion seen in 
modern otariids suggests that pectoral oscillation may indeed be ancestral to the Otarioidea, with 
reversal to the plesiomorphic condition at some point in the odobenid lineage (Gordon, 1979; 
Bebej, 2009). Such a scenario is supported by the reconstruction of partially known odobenid 
Dusignathus seftoni as a forelimb-dominated swimmer (Furbish, 2015) and morphological 
evidence that the highly derived Pleistocene odobenid Valenictus chulavistensis relied 
predominantly on forelimb propulsion (Deméré, 1994). However, these characterizations of D. 
54 
 
 
seftoni and V. chulavistensis are far from certain; Furbish (2015) notes that many of the 
characters used for trait reconstruction were absent for D. seftoni, decreasing the certainty of this 
classification, and Giffin (1992) interpreted the swimming of V. chulavistensis as comparable to 
that of O. rosmarus through examination of neural canal anatomy. So while the locomotor 
evolution of otarioid pinnipeds remains unresolved, this previous work provides a testable 
hypothesis: forelimb propulsion arose once at the base of Otarioidea with a partial reversal to the 
ancestral condition of hindlimb-dominated combination swimming at some point in the odobenid 
lineage. 
As discussed, most previous locomotor studies examine species that are represented by 
nearly complete skeletons or which, in the case of D. seftoni, at least have multiple limb 
elements available, severely limiting exploration of locomotor modes in basal members of the 
otarioid families, none of which are represented by complete specimens and very few of which 
are represented by elements other than crania and humeri (Valenzuela-Toro et al., 2018). 
However, this latter bone offers a largely untapped resource for exploring the evolution of 
otarioid locomotion. The mammalian humerus displays a remarkable range of morphological 
diversity that has been shown to strongly correlate with function (Andersson, 2004; Fabre et al., 
2014; MacLaren and Nauwelaerts, 2016; Jones et al., 2022), including locomotion. Furthermore, 
Muñoz (2021) found that the morphology of the carnivoran humerus is not only strongly 
correlated with locomotor mode but is also more influenced by locomotion than that of the 
femur, which tends to be more conserved. 
Well over a century of work focusing specifically on the functional morphology of 
pinniped humeri provides a foundation for predicting how humeral shape may vary with specific 
aquatic locomotor functions, particularly with regards to phocids and otariids. With their reliance 
55 
 
 
on forelimb propulsion, the morphology of the otariid humerus (Fig. 2a) has been considered 
especially associated with swimming modality. Otariid humeri display an elongated deltoid crest 
and high greater tubercle which provides a long moment arm for forelimb adduction during 
thrust production via the pectoralis muscle, increasing the efficiency of the power stroke (Howell 
1929). The same condition allows for an extended attachment of the deltoid muscle, which is 
responsible for generating the majority of the force of the recovery stroke (Howell, 1929; 
English, 1977), and a similarly extended attachment of the humerotrapezius muscle, allowing for 
strong, rapid extension of the humerus during this recovery stroke. Howell (1929) attributes the 
elevation of the greater tubercle above the level of the humeral head and the lesser tubercle in the 
California sea lion Zalophus californianus to the provision of increased leverage for the 
extension of the humerus via the supraspinatus muscle during the recovery stroke. Otariids also 
possess enlarged medial epicondyles compared to phocids, ostensibly for providing an anchor for 
the powerfully developed digital flexors as needed for the power stroke (Howell, 1929). 
As previously discussed, although the phocid foreflipper is not generally used for primary 
thrust production, it nonetheless serves numerous important locomotor functions and thus phocid 
humeral morphology (Fig. 2b) has also been interpreted in relation to aquatic locomotion 
(Howell, 1929; Backhouse, 1962). Similar to otariids, the deltoid crest in phocids is significantly 
developed relative to the usual condition in fissiped carnivorans, but the phocid deltoid crest is 
generally proximodistally shorter and lateromedially wider than that of otariids. Howell (1929) 
________________________________ 
 
FIGURE 2 (next page). Dorsal, medial, ventral, and lateral views of the right humerus of the 
three modern pinniped clades: A, Otariidae, represented by the New Zealand fur seal 
Arctocephalus forsteri; B, Phocidae, represented by the harp seal Pagophilus groenlandicus; and 
C, Odobenidae, represented by the walrus Odobenus rosmarus. Each scale bar represents 5 cm. 
56 
 
57 
 
hypothesized that the elevation of the crest in phocids resulted from the effort of the 
humerotrapezius and atlantoscapularis, each of which have broader if shorter attachment sites in 
phocids than otariids, to limit forelimb range of motion and to immobilize the forelimb against 
the body during sustained periods of swimming. Another major distinction between phocid and 
otariid humeri is the lesser tubercle (Berta et al., 2018). Although enlarged in all pinnipeds 
relative to other carnivorans (Wyss, 1988), this condition is even more pronounced in phocids, 
which possess massively developed lesser tubercles that extend further proximally than the 
greater tubercle or the humeral head. Howell (1929) attributes this condition partially to the 
greater development of the subscapularis in phocids, which holds the humerus in a more 
abducted resting position relative to otariids, although he considers this to be insufficient 
explanation for the enormity of the lesser tubercle. Laakkonen and Nihtilä (2021) accredit the 
elevation of the lesser tubercle above the humeral head to the provision of a long moment arm 
for the protraction and rotation of the humerus. These movements are of particular importance 
for braking, turning, and use of the forelimb as a hydroplane, considered to be the primary uses 
of the forelimb in phocids (Backhouse, 1962). While the phocid medial epicondyle is notably 
less developed than the otariid condition, the phocid lateral epicondyle is laterally protuberant 
with a medially convex edge compared to the comparatively faint otariid state (Berta et al., 
2018), assumedly to anchor the respectively larger forearm extensors which allow for dorsal 
thrusts of the manus for initiating dives or ventral bending of the anterior portion of the body 
(Howell, 1929). 
The humerus of O. rosmarus largely resembles the otariid condition (Murie, 1870; 
Howell, 1929) with several notable differences (Fig. 2c). Like otariids, it also possesses an 
elongated deltoid crest with an elevated greater tubercle, a large lesser tubercle whose proximal 
58 
 
 
extent does not quite extend to the level of the humeral head or the greater tubercle, and a 
medially protuberant medial epicondyle. Although this condition is similar to that which allows 
for forelimb-dominated swimming in otariids, Gordon (1983) attributes the elongation of the 
deltoid crest in O. rosmarus to the provision of a long moment arm for the forelimb adduction 
needed “push-up” phase of the forward lunge performed during terrestrial locomotion, raising 
the question of whether a proximodistally long deltoid crest first arose to assist with efficient 
terrestrial locomotion or, if otarioids are ancestrally forelimb-propelled, it evolved for forelimb-
dominated aquatic locomotion, and was subsequently exapted for the terrestrial gait of O. 
rosmarus. Unlike otariid humeri, however, the proximal ventrolateral surface of the deltoid crest 
of O. rosmarus is wide like the condition seen in many phocids. Additionally, the lateral 
epicondyle protrudes in a laterally convex curve from the condyle before sloping back towards 
the axis of the shaft at its proximal aspect in a condition that differs from either otariids or 
phocids (Murie, 1870). However, phocid lateral epicondyles do display a greater degree of lateral 
protrusion than those of otariids (although the muscle attachment for the forearm extensors rises 
further proximally than in O. rosmarus), so the possibility remains that these conditions are 
functionally linked with hindlimb-dominated propulsion. This work on the functional 
morphology of pinniped humeri provides a foundation for exploring the relationship between 
swimming mode and humeral shape and thus for predicting the locomotor affinities of fossil 
otarioids. 
To quantitatively analyze humeral morphology and its functional implications, I used 
three-dimensional geometric morphometrics (GMM) methods to first investigate the relationship 
between humeral morphology and swimming mode in modern pinnipeds and then to predict the 
locomotor affinities of fossil otarioids (Zelditch et al., 2004; Marramà and Kriwet, 2017). 
59 
 
 
Through these analyses, I can test the formerly stated hypothesis that forelimb propulsion is the 
plesiomorphic state for Otarioidea with a reversal to hindlimb-dominated combination swimming 
in the odobenid lineage. 
2 Materials and Methods 
2.1 Material Examined 
 We examined, photographed, and landmarked the humeri of 23 extant pinniped species 
represented by 41 individual adult specimens and 17 species of fossil pinniped. A list of all taxa 
and specimens included in this study is available in Table 1. This sample includes the smallest 
and second largest species of phocids (Pusa sibirica and Mirounga angustirostris, respectively) 
and the smallest and second largest species of otariids (Arctocephalus galapagoensis and Z. 
californianus, respectively). When possible, I landmarked right humeri; the landmarks of left 
humeri were reflected during the Procrustes analysis. When available, I landmarked a male and 
female humerus to include the widest range of intraspecific morphological variation and because 
the fossil record likely consists of both male and female specimens. All modern specimens are 
housed in the Smithsonian National Museum of Natural History (USNM) and fossil specimens 
are housed at both the USNM and the Natural History Museum of Los Angeles County (LACM). 
2.2 Phylogenetic Information 
Using all the taxa included in this study, I constructed a phylogeny in Mesquite using 
previously published trees as I subsequently explain (Fig. 3). Because I landmarked at least two 
specimens for most species, I included multiple representatives of a single species in our tree as a 
polytomy. For the placement of stem pinnipeds Potamotherium valletoni and Pinnarctidion rayi, 
I used the topology of Paterson et al. (2020). Family-level topology derives  
60 
 
 
Table 1. List of landmarked specimens representing fossil and extant taxa. USNM = Smithsonian National 
Museum of Natural History; LACM = Natural History Museum of Los Angeles County. 
 Taxon Specimen(s)  Taxon Specimen(s) 
    
Fossil Taxa Extant Phocids 
1. Potamotherium valletoni USNM 214984 28. Cystophora cristata USNM 188962 
2. Pinnarctidion rayi USNM 250321 29. USNM 593970 
3. Desmatophoca oregonensis USNM 33524 30. Erignathus barbatus USNM 16116 
4. Allodesmus kernensis LACM 4320 31. USNM 269126 
5. Pithanotaria starri LACM 52773 32. Halichoerus grypus USNM 446405 
6. Thalassoleon mexicanus LACM 128005 33.  USNM 446406 
7. Neotherium mirum LACM 4319 34. Histriophoca fasciata USNM 504959 
8. Imagotaria downsi USNM 23865 35.  USNM 504960 
9. Pontolis barroni LACM 214848 36. Hydrurga leptonyx USNM 396931 
10. Dusignathus seftoni USNM 314263 37. Leptonychotes weddellii USNM 504871 
11. Pliopedia pacifica USNM 187328 38. USNM 550118 
12. Valenictus imperialensis LACM 3926 39. Mirounga angustirostris USNM 12441 
   40. Monachus monachus USNM 219059 
Extant Otarioids 
 
13. Odobenus rosmarus USNM 21331 41. Neomonachus schauinslandi USNM 334577 
14.  USNM 500252 42.  USNM 395997 
15. USNM 550040 43. USNM 395999 
16. Arctocephalus australis USNM 23331 44. Ommatophoca rossii USNM 275206 
17.  USNM 501120 45.  USNM 302975 
18. Arctocephalus forsteri USNM 396921 46. Pagophilus groenlandicus USNM 593976 
19.  USNM 504891 47.  USNM 594013 
48. Phoca vitulina USNM 594207 
20. Arctocephalus galapagoensis USNM 392266 
21. Callorhinus ursinus USNM 219847  49. Pusa capsica USNM 175689 
22. USNM 219848 50. Pusa hispida USNM 16106 
23. Neophoca cinerea USNM 484382 51.  USNM 7102 
24. Otaria flavescens USNM 484912 52. Pusa sibirica 550038 
25. USNM 95063    
26. Zalophus californianus USNM 14410    
27. USNM 504991 
 
from the phylogenies of Furbish (2015) and Paterson et al. (2020): phocids are the outgroup to 
the Otarioidea with Desmatophocidae as the earliest diverging member of the otarioids, with 
placement of the desmatophocids resolved with moderate support (0.72 and 0.75, respectively). I 
acknowledge that significant debate exists regarding the placement of the extinct pinniped family 
Desmatophocidae. Initially, Kellogg (1922) classified desmatophocids as basal otariids, a  
61 
 
 
 
FIGURE 3. Phylogenetic topology used for this study. Relationships between arctoid subclades, 
stem pinnipeds, and the crown pinniped subclades Phocidae, Desmatophocidae, Otariidae, and 
Odobenidae after Paterson et al. (2020). Interclade relationships of Phocidae, Desmatophocidae, 
Odobenidae, and Otariidae after Rule et al. (2020), Tate-Jones et al. (2020), Magallanes et al. 
(2018), and Boessenecker and Churchill (2014), respectively. Time-scaling applied to tree using 
the strap package (Bell and Lloyd, 2015) in R (R Core Team 2021).  
 
 
placement supported by later workers as well (Simpson, 1945; Mitchell, 1966, 1968; Barnes, 
1989; Barnes and Hirota, 1994). Others also allied Desmatophocidae with Otarioidea but as the 
sister taxon to Odobenidae (Berta et al., 1989; Rule et al., 2020). A number of relatively recent 
phylogenetic analyses have alternately resolved desmatophocids as sister taxon to Phocidae 
(Berta and Wyss, 1994; Boessenecker and Churchill, 2015, 2018; Tanaka and Kohno, 2015; 
Dewaele et al., 2017a; Tonomori et al., 2018; Velez-Juarbe, 2018; Tate-Jones et al., 2020). First, 
Furbish (2015) is the only phylogeny that specifically examines the placement of 
62 
 
 
desmatophocids. Secondly, only Furbish (2015), Paterson et al. (2020), and Rule et al. (2020) use 
both molecular and morphological data for the phylogenetic reconstruction of family-level 
relationships in Pinnipedia and support for the Rule et al. 2020 resolution was lower (0.59) than 
that of the other two studies. Thirdly, all currently known desmatophocids hail from the North 
Pacific, which is also the region where odobenids and otariids first diversified, whereas all early-
diverging phocids have been found in the North Atlantic and Tethys basins (Berta et al., 2018). 
Finally, current literature strongly supports the use of probabilistic phylogenetic reconstruction 
methods and total evidence datasets with both molecular and morphological data for the most 
accurate topological resolution and highest nodal support (O’Reilly et al., 2018; Puttick et al., 
2019; Smith, 2019; Mongiardino Koch et al., 2021). For the topology for intrafamilial 
relationships, I also relied on total evidence phylogenies where available, referring to 
Boessenecker and Churchill et al. (2014) for Otariidae, to Rule et al. (2020) for Phocidae, and to 
Boessenecker and Churchill (2018), Tonomori et al. (2018), and Tate-Jones et al. (2020) for 
Desmatophocidae. As O. rosmarus is the only modern species of Odobenidae, total evidence 
phylogenies are not available, so I used the morphological phylogenies from Magallanes et al. 
(2018) and Biewer et al. (2020) for the odobenid topology.  
To timescale the phylogeny, I followed the protocol of (Morales-García et al., 2021) 
using the “equal” method (Brusatte et al., 2008) in package ‘paleotree’ (Bapst, 2012). I retrieved 
first and last appearance data for fossil taxa from the Paleobiology Database (pbdb.org) and 
divergence dates for nodal constraint from the literature (Higdon et al., 2007; Velez-Juarbe, 
2017; Magallanes et al., 2018; Biewer et al., 2020; Paterson et al., 2020; Rule et al., 2020; Tate-
Jones et al., 2020). The time-scaled phylogeny is available in Appendix D. 
 
63 
 
 
2.3 Geometric Morphometrics 
Geometric morphometric methods (GMM) are a set of statistical tools used to analyze the 
morphological variation of biological structures and can be used to infer functional adaptations 
including locomotor modes (Zelditch et al., 2012). To quantify the morphology of pinniped 
humeri, I developed a landmarking schema consisting of 26 fixed landmarks including a 
combination of Type I (homologous) and Type II (based on maxima and minima) landmarks 
(Fig. 4; Table 2). This schema derives from both our own observations and previous studies 
examining humeral morphology (Milne et al., 2009; Richards et al., 2022). I landmarked all  
Table 2. Landmarks used to quantify the three-dimensional morphology of pinniped humeri. 
1. Proximal apex of the lesser tubercle 
2. Distal extent of the lesser tubercle 
3. Lateral extent of the lesser tubercle 
4. Medial extent of the lesser tubercle 
5. Proximal apex of the greater tubercle 
6. Medial extent of head at confluence with the lesser tubercle 
7. Head at distodorsal edge 
8. Lateral extent of head at confluence with greater tubercle/deltoid crest 
9. Proximoventral extent of head at top of bicipital groove  
10. Head at center of free surface 
11. Bicipital groove  
12. Laterodistal extent of deltoid crest 
13. Distal extent of deltoid crest 
14. Proximolateral extent of ventral surface of humeral condyle 
15. Distolateral extent of humeral condyle 
16. Proximolateral extent of dorsal surface of humeral condyle 
17. Proximal extent of dorsal surface of humeral condyle 
18. Point of maximal depth of olecranon fossa 
19. Proximomedial extent of dorsal surface of humeral condyle 
20. Distomedial extent of humeral condyle 
21. Proximomedial extent of ventral surface of humeral condyle 
22. Inflection point of proximal edge of ventral surface of humeral condyle 
23. Inflection point of distal extent of humeral condyle 
24. Distomedial extent of medial epicondyle 
25. Proximomedial extent of medial epicondyle 
26. Proximal extent of lateral epicondyle 
  
64 
 
 
 
FIGURE 4. In order of right to left, dorsal, lateral, ventral, and medial views of the landmarks used 
in this study as shown on the left humerus of the spotted seal Phoca hispida. Numbers correspond 
to the landmarks listed in Table 2. 
 
 
humeri using a Revware Microscribe G2LX with a reported position accuracy of 0.30 mm; the 
3D coordinates (x, y, z) were imported into Excel using the Revware MicroScribe Utility 
Software package. 
To analyze shape variation between humeri associated with specific swimming 
modalities, I designated otariids as forelimb-propelled swimmers and phocids and odobenids as 
hindlimb-propelled swimmers. I acknowledge that these designations collapse the true diversity 
of pinniped locomotor modes into an artificial binary but chose this framework for the sake of 
analytical simplicity. I then ingested the landmark data into R (R Core Team, 2021) and used the 
‘gpagen’ function in R package geomorph version 4.0.4 (Baken et al., 2021; Adams et al., 2022) 
to perform a generalized Procrustes analysis (GPA) on the landmark data to remove non-shape 
variation from between the specimens, such as size, orientation, and positioning (Rohlf and 
65 
 
 
Slice, 1990). I then performed a principal components analysis (PCA) to reduce the 
dimensionality of the dataset and to visualize its variation with R package ggplot2 (Wickham, 
2016; Marramà and Kriwet, 2017). 
2.4 Flexible Discriminant Analyses 
To test the hypothesis that the three dimensional morphology of pinniped humeri 
distinguishes between hindlimb- and forelimb-propelled swimmers, I used flexible discriminant 
analyses (FDA). Similar to linear discriminant analysis, FDA finds a combination of predictors 
that maximize the separation between groups, but FDA allows for non-linear decision boundaries 
by fitting flexible functions to the data rather than being limited to a line (Hastie et al., 1994). I 
performed two flexible discriminant analyses (FDAs), one without phylogenetic correction 
(npFDA) and one with (pFDA), so that I could assess the degree of impact phylogeny had on 
shape variation and classification rates. I conducted the npFDA using the ‘fda’ function in R 
package mda version 0.5.3 (2022) and the pFDA following the protocol of Morales-García et al. 
(2021) which uses source code ‘phylo.fda.v0.2.R’ (Motani and Schmitz 2011). I trained the 
FDAs on extant taxa with known locomotor modes with the principal components (PCs) that 
accounted for up to 90% of the variation in the dataset. As part of the pFDA, I also calculated the 
optimal lambda, a parameter that quantifies the degree to which the trait data of a group of 
species are influenced by their shared evolutionary history (Pagel, 1999; Motani and Schmitz, 
2011). I calculated confusion matrices for the FDAs, the posterior classification rates for the 
training (modern) dataset, and the relative contribution of each PC to the discriminant function. 
Finally, I used the FDAs to classify the fossil taxa and calculated posterior probabilities and 
discriminant scores for each. I then plotted the mean discriminant score for each species on a 
66 
 
 
phylogeny using the R package phytools (Revell, 2012). The coefficients of the FDAs, by which 
each PC is multiplied to form the discriminant decision rule, are also calculated. 
 
3 Results 
3.1 Humeral Morphology and Aquatic Locomotion 
Using 3D GMM, I found that humeral morphology correlates with pinniped swimming 
mode.  In Fig. 5, I show a visualization of PC1 and PC3, as these were the two PCs that 
contributed the most to the FDAs (see following section). PC1 accounts for 32.7% of the 
variation and PC3 for 10.4% of the variation. PC1 largely discriminates between otarioids on the 
negative end of the axis and phocids on the positive end of the axis. Taxa with negative PC1 
scores have a relatively long deltoid crest that runs parallel to the shaft and extends into an 
exaggerated greater tubercle that reaches further proximally than either the humeral head or the 
lesser tubercle, which is located just ventrally to the comparatively slightly more proximally 
elevated head. The two tubercles are separated by a narrow intertubercular fossa. The medial 
epicondyle is proximodistally short and its distal extent arises from the shaft slightly proximally 
to the medial keel of the condyle and its proximal extent merges again with the shaft at 
approximately the same level as does the lateral epicondyle. The olecranon fossa is small and 
limited to a shallow indentation just proximal to the ventroproximal edge of the condyle. 
Comparatively, taxa with positive PC1 values have a shorter deltoid crest that tilts away from the 
shaft in the ventrodistal direction. The lesser tubercle is bulbous, medially offset from the 
humeral head, and extends further proximally than the humeral head or the comparatively 
reduced greater tubercle. A wide intertubercular fossa separates the two tubercles. The medial  
67 
 
 
 
FIGURE 5. Scatter plot of the results of the principal components analysis showing principal 
components (PCs) 1 and 3, the two main contributors to the flexible discriminant analyses. Extant 
taxa are indicated by solid points and extinct taxa by unfilled points. Convex hulls are shown for 
extant hindlimb-dominated swimmers (teal) and forelimb-dominated swimmers (gold). The 
humeri of selected taxa are shown to demonstrate the range of morphologies associated with the 
maxima and minima of the PCs. See Table 1 for taxon names. 
 
68 
 
 
and lateral epicondyles are proximodistally longer than the condition seen in taxa with negative 
PC1 values, with the lateral epicondyle extending further proximally than the medial epicondyle. 
The olecranon fossa is broad and shallowly concave. 
The PC3 axis forms a continuum between two distinct morphologies that are less 
taxonomically discriminatory than PC1, although the otariids do fall exclusively towards the 
positive end of the spectrum. The minimum for PC3 represents humeri with a broad deltoid crest 
that extends proximally past the humeral head to approximately the same level as the robust, 
medially protuberant lesser tubercle. The tubercles are separated by a deep, wide intertubercular 
fossa. The medial and lateral epicondyles are each fairly symmetrical around the humeral 
condyle and are approximately equal in proximodistal length to one another. The olecranon fossa 
is extremely shallow. The positive end of the PC3 axis represents humeri with a narrow deltoid 
crest that extends proximally into a greater tubercle that rises past the humeral head and the 
comparatively smaller lesser tubercle. The medial epicondyle is proximodistally shorter than that 
structure in specimens represented by negative PC3 values but is also bulbously protuberant 
rather than flat. The lateral epicondyle extends further proximally than the medial epicondyle to 
the approximate halfway between the distal extent of the humerus and the distal extent of 
humeral head. The olecranon fossa is slightly deeper and wider than the condition at the PC3 
minimum. Monk seals cluster towards the minimum of PC3. O. rosmarus, a male Mirounga 
angustirostris (the northern elephant seal), a male Halichoerus grypus (the grey seal), and a 
female Neophoca cinerea (the Australian sea lion) cluster around the median. The remaining 
taxa cluster towards the PC3 maximum. 
We also projected the humeral shape of fossil taxa into the morphospace of extant taxa. 
Stem pinnipeds P. valletoni and P. rayi fall near the upper limit of the hindlimb morphospace, in 
69 
 
 
line with previous predictions that hindlimb-dominated swimming is the ancestral state in 
pinnipeds (Furbish, 2015; Paterson et al., 2020; Hocking et al., 2021). Early diverging 
desmatophocid Desmatophoca oregonensis falls within the hindlimb-dominated morphospace 
intermediate to phocids and odobenids, but derived desmatophocid A. kernensis falls outside of 
modern morphospace. The two fossil otariids, the basal Pithanotaria starri and later branching 
Thalassoleon mexicanus, similarly fall in the space intermediate to the modern hindlimb and 
forelimb morphospaces, but P. starri has a higher PC1 value than any of the modern otariids and 
falls very close to the two stem pinnipeds. Except for Pontolis barroni, all of the fossil walruses 
fall within the hindlimb-dominated morphospace. The early walruses Neotherium mirum and 
Imagotaria downsi cluster relatively close to early otariid P. starri and stem pinniped P. rayi, 
while P. barroni falls just outside the modern otariid morphospace. The neodobenids D. seftoni, 
Pliopedia pacifica, and Valenictus imperialiensis all cluster near modern O. rosmarus. Notably, 
most of the fossil species fall outside regions occupied by modern pinnipeds. 
3.2 Inferring Locomotor Mode from Pinniped Humeral Morphology 
To train the FDAs, I used the first 13 PCs (Appendix E), which accounted for 88.8% of 
the total variation in the dataset. The posterior classification rates and discriminant scores for 
each species are recorded in Table 3 and probability density functions displaying the distribution 
of discriminant scores for each FDA are shown in Figure 6. The coefficients of the FDAs, which 
quantify the degree to which each PC contributes to the discriminant function, are recorded in 
Table 4. Both FDAs performed remarkably well: the npFDA correctly classified both hindlimb- 
and forelimb-dominated swimmers 100% of the time and the pFDA had an overall correct 
classification rate of 90% with correct identification of all of the hindlimb swimmers and four 
out of 12 forelimb swimmers incorrectly classified as hindlimb swimmers (Table 3). The  
70 
 
 
 
 
71 
 
Table 3. Results of phylogenetically uncorrected and corrected flexible discriminant analyses (FDAs). Numbers correlate with specimens 
listed in Table 1; DS = discriminant score. Continued onto the subsequent page. 
 
 Uncorrected FDA Results Phylogenetically Corrected FDA Results 
 
Taxon Name Predicted Forelimb Hindlimb Class Probability Probability DS 
Predicted Forelimb Hindlimb 
Class Probability Probability DS 
 Fossil Taxa         
1. Potamotherium vallentoni Hindlimb 0.000 1.000 0.06 Hindlimb 0.324 0.676 -0.30 
2. Pinnarctidion rayi Hindlimb 0.000 1.000 0.60 Hindlimb 0.231 0.769 0.30 
3. Desmatophoca oregonensis Hindlimb 0.000 1.000 0.84 Hindlimb 0.131 0.869 1.17 
4. Allodesmus kernensis Forelimb 0.972 0.028 -1.99 Forelimb 0.651 0.349 -2.03 
5. Pithanotaria starri Hindlimb 0.000 1.000 -0.28 Hindlimb 0.158 0.842 0.89 
6. Thalassoleon mexicanus Forelimb 1.000 0.000 -2.98 Hindlimb 0.340 0.660 -0.39 
7. Neotherium mirum Hindlimb 0.327 0.673 -1.36 Hindlimb 0.436 0.564 -0.91 
8. Imagotaria downsi Hindlimb 0.001 0.999 -0.46 Hindlimb 0.261 0.739 0.09 
9. Pontolis barroni Forelimb 1.000 0.000 -3.45 Forelimb 0.724 0.276 -2.46 
10. Dusignathus seftoni Hindlimb 0.123 0.877 -1.17 Hindlimb 0.340 0.660 -0.39 
11. Pliopedia pacifica Hindlimb 0.032 0.968 -0.96 Hindlimb 0.426 0.574 -0.85 
12. Valenictus imperialensis Hindlimb 0.000 1.000 -0.10 Hindlimb 0.220 0.780 0.37 
Ext ant Otarioids         
13. Odobenus rosmarus Hindlimb 0.000 1.000 0.16 Hindlimb 0.294 0.706 -0.12 
14.  Hindlimb 0.000 1.000 1.51 Hindlimb 0.176 0.824 0.73 
15.  Hindlimb 0.000 1.000 0.29 Hindlimb 0.231 0.769 0.29 
16. Arctocephalus australis Forelimb 1.000 0.000 -3.26 Hindlimb 0.321 0.679 -0.28 
17.  Forelimb 1.000 0.000 -5.33 Forelimb 0.591 0.409 -1.70 
18. Arctocephalus forsteri Forelimb 1.000 0.000 -5.89 Forelimb 0.651 0.349 -2.03 
19.  Forelimb 1.000 0.000 -5.50 Forelimb 0.627 0.373 -1.90 
20. Arctocephalus galapagoensis Forelimb 0.999 0.001 -2.60 Hindlimb 0.254 0.746 0.13 
21. Callorhinus ursinus Forelimb 1.000 0.000 -5.27 Forelimb 0.605 0.395 -1.78 
22.  Forelimb 1.000 0.000 -4.60 Forelimb 0.571 0.429 -1.60 
23. Neophoca cinerea Forelimb 1.000 0.000 -5.58 Forelimb 0.706 0.294 -2.35 
24. Otaria flavescens Forelimb 1.000 0.000 -4.13 Hindlimb 0.442 0.558 -0.94 
25.  Forelimb 1.000 0.000 -4.86 Forelimb 0.553 0.447 -1.51 
 
 
 
72 
 
Table 3. Continued from previous page. 
  Uncorrected FDA Results Phylogenetically Corrected FDA Results 
 
Taxon Name Predicted Forelimb Hindlimb Predicted Forelimb Hindlimb Class Probability Probability DS Class Probability Probability DS 
26. Zalophus californianus Forelimb 1.000 0.000 -4.04 Hindlimb 0.432 0.568 -0.89 
27.  Forelimb 1.000 0.000 -5.22 Forelimb 0.604 0.396 -1.77 
 Extant Phocids         
28. Cystophora cristata Hindlimb 0.000 1.000 1.30 Hindlimb 0.280 0.720 -0.03 
29.  Hindlimb 0.000 1.000 2.21 Hindlimb 0.212 0.788 0.44 
30. Erignathus barbatus Hindlimb 0.000 1.000 3.07 Hindlimb 0.138 0.862 1.10 
31.  Hindlimb 0.000 1.000 1.67 Hindlimb 0.254 0.746 0.14 
32. Halichoerus grypus Hindlimb 0.000 1.000 3.83 Hindlimb 0.111 0.889 1.41 
33.  Hindlimb 0.000 1.000 1.85 Hindlimb 0.259 0.741 0.10 
34. Histriophoca fasciata Hindlimb 0.000 1.000 2.21 Hindlimb 0.198 0.802 0.55 
35.  Hindlimb 0.000 1.000 1.93 Hindlimb 0.242 0.758 0.22 
36. Hydrurga leptonyx Hindlimb 0.000 1.000 2.08 Hindlimb 0.222 0.778 0.36 
37. Leptonychotes weddellii Hindlimb 0.000 1.000 1.35 Hindlimb 0.287 0.713 -0.08 
38.  Hindlimb 0.000 1.000 1.88 Hindlimb 0.237 0.763 0.25 
39. Mirounga angustirostris Hindlimb 0.000 1.000 2.54 Hindlimb 0.180 0.820 0.69 
40. Monachus monachus Hindlimb 0.000 1.000 2.91 Hindlimb 0.139 0.861 1.09 
41. Neomonachus schauinslandi Hindlimb 0.000 1.000 3.05 Hindlimb 0.112 0.888 1.40 
42.  Hindlimb 0.000 1.000 1.79 Hindlimb 0.237 0.763 0.25 
43.  Hindlimb 0.000 1.000 2.69 Hindlimb 0.149 0.851 0.98 
44. Ommatophoca rossii Hindlimb 0.000 1.000 1.23 Hindlimb 0.250 0.750 0.16 
45.  Hindlimb 0.000 1.000 4.27 Hindlimb 0.080 0.920 1.87 
46. Pagophilus groenlandicus Hindlimb 0.000 1.000 0.67 Hindlimb 0.350 0.650 -0.45 
47.  Hindlimb 0.000 1.000 0.49 Hindlimb 0.422 0.578 -0.83 
48. Phoca vitulina Hindlimb 0.000 1.000 1.58 Hindlimb 0.285 0.715 -0.06 
49. Pusa caspica Hindlimb 0.000 1.000 2.79 Hindlimb 0.157 0.843 0.90 
50. Pusa hispida Hindlimb 0.000 1.000 1.64 Hindlimb 0.230 0.770 0.30 
51.  Hindlimb 0.000 1.000 2.18 Hindlimb 0.199 0.801 0.54 
52. Pusa sibirica Hindlimb 0.000 1.000 3.07 Hindlimb 0.132 0.868 1.16 
 
 
Table 4. The contribution of each of the first 13 principal components (PCs) to 
the phylogenetically uncorrected FDA (npFDA) and phylogenetically corrected 
FDA (pFDA), which quantify the contribution of each principal component (PC) 
to the discriminant function. 
PC npFDA Coefficient pFDA Coefficient 
1 36.35 8.73 
2 -1.62 -0.87 
3 -35.33 -8.22 
4 -4.40 -1.14 
5 -20.46 -2.79 
6 -16.04 -2.76 
7 3.31 0.96 
8 8.36 2.38 
9 12.23 2.18 
10 -4.37 -1.33 
11 7.15 2.14 
12 -1.95 -0.24 
13 -13.87 -2.88 
 
incorrectly classified specimens from the pFDA include a male Z. californianus with a 
phylogenetically corrected discriminant score (pDS) of -0.89 and uncorrected discriminant score 
(npDS) of -1.20, a male Otaria flavescens (pDS = -0.94, npDS = -1.22), a male Arctocephalus 
australis (pDS = -0.28, npDS = -0.96), and an Arctocephalus galapagoensis of unknown sex 
(pDS = 0.13, npDS = -0.77). The locomotor groups are well-separated in both FDAs with some 
overlap between the groups in the pFDA (Fig. 6).  
The swimming mode predictions for fossil taxa are recorded in Table 3 along with the 
posterior classification rates and discriminant scores. Predictions were consistent between the 
npFDA and the pFDA with the exception of otariid T. mexicanus. Both stem pinnipeds were 
classified as hindlimb-dominated swimmers (P. valletoni: pDS = -0.30, npDS = 0.06; P. rayi: 
pDS = 0.30, npFD = 0.60). This finding is consistent with the morphospace occupied by these 
species in the PCA and with previous predictions that hindlimb-dominated swimming is  
73 
 
 
 
FIGURE 6. Probability density functions for the uncorrected flexible discriminant analysis (FDA, 
A) and the phylogenetically corrected FDA (B). Unfilled points show the discriminant scores for 
each of the fossil taxa. See Table 1 for taxon names. 
 
74 
 
 
ancestral for pinnipeds (Bebej, 2009; Paterson, 2017; Hocking et al., 2018). Similarly, the most 
basal members of the otarioid clades were classified as hindlimb-dominated swimmers: otariid P. 
starri (pDS = 0.89, npDS = -0.28), desmatophocid D. oregonensis (pFDA pDS = 1.17, npDS = 
0.84), and odobenids N. mirum (pDS = -0.91, npDS = -1.36) and I. downsi (pDS = 0.09, npDS = 
-0.46). Later-diverging odobenids D. seftoni (pDS = -0.39, npDS = -1.17), P. pacifica (pDS =     
-0.85, npDS = -0.96), and V. imperialensis (pDS = 0.37, npDS = -0.96) were all resolved as 
hindlimb-dominated swimmers. The only difference in predicted swimming modes for fossil taxa 
between the two FDAs was the classification of the otariid Thalassoleon mexicanus, which 
diverged after P. starri but before crown Otariidae, as a forelimb-dominated swimmer in the 
npFDA (npDS = -2.98) and as hindlimb-dominated swimmer in the pFDA (pDS = -0.39). 
 
4 Discussion 
4.1 Implications of Pinniped Humerus Morphology for Locomotor Function 
Overall, I found that humeral shape is significantly influenced not only by phylogeny 
(optimal lambda = 0.52) but also by aquatic locomotor mode. The results of our PCA indicate 
that the humeral morphospace occupied by forelimb-dominated swimmers is limited towards the 
negative end of PC1 and the positive end of PC2 and is characterized by a long, narrow deltoid 
crest that extends into a tall greater tubercle that reaches proximally past the humeral head and 
the lesser tubercle; a shallow, narrow intertubercular fossa; a proximodistally short but medially 
bulbous medial epicondyle; a broad, shallowly concave olecranon fossa; and a proximodistally 
long lateral epicondyle. The humeri of hindlimb-dominated swimmers, alternately, occupy a 
wider morphospace than the forelimb-dominated swimmers, with representatives from both 
75 
 
 
extremes of PC3 and occupying nearly the entire range of PC1 as well, although phocids are 
largely limited to the positive end of the PC1 axis. Both the markedly smaller morphospace for 
forelimb-dominated swimmers compared to hindlimb-dominated swimmers and the positioning 
of A. kernensis, T. mexicanus, and P. barroni close to the convex hull surrounding the modern 
forelimb swimmers indicates that forelimb propulsion constrains the overall morphology of the 
humerus. Alternately, the greater area of the morphospace occupied by hindlimb-dominated 
swimmers likely reflects both the wider phylogenetic and ecological diversity of this group. It 
also raises the possibility that primary dependence on the hindlimbs for aquatic locomotion 
releases the forelimbs from locomotor constraint, allowing for a wider range of forelimb 
morphologies as well as functions. Indeed, hindlimb-dominated swimmers utilize their forelimbs 
for a wide range of activities other than aquatic locomotion, including prey manipulation 
(Backhouse, 1962; Hocking et al., 2017, 2018), underwater “clapping” against the torso by grey 
seals (H. grypus) (Hocking et al., 2020), extensive scratching (Backhouse, 1962), digging 
through ice to create breathing holes (Stirling, 1977; Laakkonen and Nihtilä, 2021), and 
hollowing out snowbanks into lairs for hauling out, birthing, and protecting young (Helle et al., 
1984; Lydersen and Gjertz, 1986). Grey seals (H. grypus) even use their forelimbs during 
terrestrial locomotion to propel the body forwards while the pelvis is still lifted from the ground, 
primarily through hyperextension and flexion of the phalanges and metacarpals (Backhouse 
1962). 
As expected, the specific morphology of the major structures of the humerus correlates 
with swimming modes in generally predictable ways. Here I discuss the implications of the 
particular morphology of each of the major structures for aquatic locomotion as identified by our 
analyses. 
76 
 
 
4.1.1 Deltoid Crest and Greater Tubercle—At the PC1 minimum, the deltoid crest is 
proximodistally long, extending proximally past both the humeral head and the lesser tubercle, 
with a ventral edge that is relatively narrow and runs approximately parallel to the shaft. The 
positive end of PC1 represents a proximodistally shorter crest that rises proximally to a level 
intermediate between the proximal extent of the lesser tubercle and the lower humeral head; the 
ventral surface of the deltoid crest is wider than the condition at the opposite end of PC1 and it 
also tilts away from the shaft towards is distal extent. Along PC3, the minimum is represented by 
a proximodistally long deltoid crest that runs parallel to the shaft with its lateral-most extent 
located just proximal to halfway along its length, with a greater tubercle than extends proximally 
past the head to approximately the same level as the lesser tubercle. At the PC3 maximum, the 
deltoid crest is slightly shorter than the condition seen at the minimum and tilts distally away 
from the shaft with its lateral-most extent located just proximolaterally to its furthest distal 
extent; the greater tubercle extends proximally to both the humeral head and the lesser tubercle.  
With at least ten different muscles attaching to the deltoid crest and greater tubercle, 
these structures have significant relevance to pinniped forelimb movements. The two largest 
muscles attaching to the deltoid crest are the deltoid muscle, inserting to the lateral aspect of the 
deltoid crest, and the pectoralis muscles, inserting to the medial aspect of the deltoid crest, which 
have opposing functions. The deltoid muscle works with other muscles that attach to the deltoid, 
including the humerotrapezius, atlantoscapularis, supraspinatus, brachioradialis, 
cephalohumeralis, and infraspinatus, to accomplish both the recovery stroke and the abduction-
protraction needed to execute braking or turning. The pectoralis works with the latissimus dorsi, 
which attaches to the insertional edge of the pectoralis, to produce the power stroke in Z. 
californianus (Howell 1929); in Phoca vitulina, the pectoralis and latissimus dorsi, each of 
77 
 
 
which insert to the medial lip of the deltoid crest, produce the rhythmic lateral undulations of the 
caudal portion of the body used during hindlimb propulsion (Howell 1929). In phocids, the 
atlantoscapularis and humerotrapezius also work to hold the forelimb against the body to 
maintain a hydrodynamic shape during hindlimb propulsion. Howell (1929) attributes the 
shortness of the greater tubercle in P. vitulina and other phocids to a comparatively 
underdeveloped supraspinatus, responsible for humeral extension and therefore less important to 
phocids than otariids. Notably, although the deltoid crest in P. valletoni is relatively short, 
similar to the phocid condition, its greater tubercle extends proximally past the height of the head 
and lesser tubercle. This trait could indicate that P. valletoni relied more on forelimb propulsion, 
perhaps via paddling as seen in semi-aquatic musteloids (Samuels et al., 2013; Fabre et al., 
2015), than later stem pinnipeds like P. rayi and E. mealsi, each of which have greater tubercles 
that do not reach the level of the head or lesser tubercle. Furthermore, unlike other otarioids, 
early otariid P. starri has a relatively short deltoid crest that, other than extending past the level 
of the humeral head and lesser tubercle in a tall greater tubercle, is more similar to the condition 
seen in phocids and stem pinnipeds than other otarioids. This condition is also seen in early 
desmatophocid otarioid D. oregonensis. This lends further support to their classification as 
hindlimb-dominated swimmers.  
Our findings that a proximodistally long deltoid crest and tall, proximally extended 
greater tubercle are associated with a forelimb-dominated swimming style are expected as this 
condition provides a long moment arm for the adduction, abduction, and rotation of the forelimb 
needed for the stroke cycle (Howell, 1929; Gordon, 1983). However, as previously stated, this 
condition is not solely indicative of forelimb propulsion as the elongated deltoid crest also allows 
for the forelimb adduction needed during the forward lunge phase of terrestrial locomotion in O. 
78 
 
 
rosmarus (English, 1977; Gordon, 1981). Furthermore, this condition also characterizes all the 
fossil otarioids in our analysis with the exception of P. starri and D. oregonensis, regardless of 
whether the FDAs classified them as hindlimb- or forelimb-dominated swimmers. Although 
previous workers have discussed an apparent tradeoff between the functional constraints of 
aquatic versus terrestrial locomotion (Gingerich, 2003; Botton-Divet et al., 2016; Berta et al., 
2018; Hocking et al., 2021), our findings raise the possibility that this condition originally 
evolved not for aquatic forelimb propulsion but rather to allow for efficient terrestrial locomotion 
with an aquatically adapted bauplan. This condition was then exapted for forelimb propulsion in 
otariids, A. kernensis, and P. barroni. Whereas the majority of locomotor propulsion in fissiped 
carnivorans originates in the hindlimbs, otariids generate the majority of the force for terrestrial 
locomotion with their forelimbs (English, 1976, 1977; Beentjes, 1990). Further investigation of 
functional morphology of the forelimb with regards to terrestrial locomotion in otarioids is 
needed. 
In opposition to the proximodistally long deltoid crest seen in otarioids, most phocids 
possess a deltoid crest that is lateromedially wider than the condition in most otariids, seen in the 
PCA at the maximum of PC1 and at the minimum of PC3. O. rosmarus also possesses a laterally 
expanded deltoid crest. This trait appears to accommodate a lateral placement of the 
humerotrapezius in phocids compared to its attachment in Z. californianus along the ventral edge 
of the deltoid crest. Although the longer attachment of the humerotrapezius in Z. californianus 
allows for stronger, faster extension of the humerus during the recovery stroke needed for rapid 
forelimb propulsion (English, 1977), its shorter, more laterally placed attachment in phocids 
seems to represent an adaptation for immobilizing the forelimb against the body during sustained 
periods of rectilinear swimming (Howell, 1929). The broader attachment for the humerotrapezius 
79 
 
 
both limits the range of motion and allows for more energy efficient muscle contraction while 
the forelimbs are being held against the body. Notably, this lateromedially wide surface of the 
deltoid crest does not appear in any of the fossil stem pinnipeds or otarioids, indicating a strong 
phylogenetic component to this condition. 
The phocid deltoid crest also forms an acute angle to the humeral shaft from its vertex at 
the greater tubercle as opposed to running parallel to the shaft as seen in most otarioids. The 
phocid condition also appears in stem pinnipeds P. valletoni, P. rayi, and D. oregonensis. This 
condition potentially results from a broader attachment site for the brachialis muscle, which 
attaches to the shaft just dorsal to the lateral edge of the deltoid crest and assists with flexion of 
the elbow joint, a critical movement in braking and turning. Although previous workers have 
considered biceps muscle to be the primary actor in elbow flexion/braking (Howell, 1929; 
English, 1977), the large insertion of the brachialis in phocids indicates that it may also be 
significantly involved. The presence of this condition in the stem pinnipeds as well as D. 
oregonensis suggests that it is the plesiomorphic condition for pinnipeds. 
4.1.2 Lesser Tubercle— Another point of differentiation between the humeral 
morphology associated with each swimming mode is the lesser tubercle.  At the negative end of 
the PC 1 axis, where the otariids and O. rosmarus fall, the lesser tubercle is located just 
mediodorsally to the head, is relatively appressed to the medial edge of the head and extends 
proximally neither as far as the top of the head nor the top of the greater tubercle. At the positive 
end of PC1, where the phocids fall, the lesser tubercle is broader than the condition at the 
negative end, strongly medially offset from the head, and extends proximally past the top of the 
head and the greater tubercle. The negative end of PC3, where the monk seals fall, is 
characterized by a wide, medially offset lesser tubercle that is approximately equal in height to 
80 
 
 
the greater tubercle; at the positive end of PC3, where the otariids and phocine seals fall, the 
lesser tubercle is still relatively wide but is slightly less offset from the humeral head and does 
not extend as far proximally as either the head or the greater tubercle. 
Compared to fissiped carnivorans, the lesser tubercle is highly developed in all pinnipeds 
(Wyss, 1988; Berta et al., 2018), even in basal taxa P. valletoni (Savage, 1957), E. mealsi (Berta 
and Ray, 1990), and P. rayi (Berta, 1994), allowing for a broad attachment site of the 
subscapularis muscle, which is responsible for medial rotation and adduction of the humerus. 
However, the lesser tubercle is particularly enormous in phocids, extending further proximally 
than the head or the greater tubercle. In otarioids, while the lesser tubercle is still substantial, it is 
less prominent than in phocids and does not extend as far proximally as the greater tubercle or 
the top of the humerus head. As previously discussed, the phocid condition likely results both 
from the more abducted resting position of the humerus in phocids relative to otarioids (Howell, 
1929) and the provision of a longer moment arm for protraction and rotation of the humerus for 
braking, turning, and use of the forelimb as a hydroplane (Backhouse, 1962; Laakkonen and 
Nihtilä, 2021). Dewaele et al. (2017) further discuss this condition and posit that the large lesser 
tubercle allows for energy efficiency in maintaining the medial rotation of the humerus while the 
forelimb is held against the torso during hindlimb-propelled rectilinear swimming. 
In stem pinnipeds P. valletoni and P. rayi, the proximal extent of the lesser tuberosity is 
already inflated compared to the fissiped condition but its proximal extent stops well short of the 
level reached by the proximal extent of the humeral head or greater tubercle (Savage, 1957; 
Berta and Ray, 1990; Berta, 1994). I also observed this condition in otariid T. mexicanus and 
odobenids N. mirum, I. downsi, and P. barroni. In stem otariid P. starri, the height of the lesser 
tubercle is intermediate between the tall greater tubercle and the humeral head. Each of these 
81 
 
 
similar conditions implies a less developed subscapularis in these taxa than seen in modern 
phocids, possibly because they utilized their forelimbs more regularly when swimming than 
phocids. In derived odobenids P. pacifica and especially V. imperialensis, the lesser tubercle is 
rather enormous despite its proximal extent falling short of the humeral head and greater 
tubercle; it bows distomedially outwards from the shaft of the humerus and its proximal 
confluence with the humeral head. This latter condition likely also contributes to efficient medial 
rotation and adduction of the humerus during hindlimb-propelled swimming, indicating that P. 
pacifica and V. imperialensis might have relied more heavily on hindlimb propulsion than their 
earlier relatives. 
4.1.3 Medial Epicondyle—On PC1, negative values represent a medial epicondyle that 
is medially inflated with a base that is proximomedially offset from landmark 20 (the mediodistal 
extent of the humeral condyle) and reaches proximally just short of the proximal extent of the 
lateral epicondyle. Positive PC1 values are associated with a medial epicondyle with a base that 
is just mediodorsal to landmark 20 and is proximodistally shorter than the lateral epicondyle. At 
the PC3 minimum, the distal base of the medial epicondyle is closely associated with landmark 
20 and it is approximately equal in proximodistal length to the lateral epicondyle. At the PC3 
maximum, the medial epicondyle is proximomedially offset from the condyle and 
proximodistally shorter than the lateral epicondyle.  
The medially expanded medial epicondyle associated with the forelimb-dominated 
morphospace provides a more secure elbow joint by providing an anchor for the enlarged medial 
collateral ligament (English 1977) and allows for the attachment of well-developed forearm and 
hand flexors (Milne et al., 2009), increasing the efficiency of the power stroke in modern otariids 
(Howell, 1929; English, 1976). The plesiomorphic condition for this trait appears to be a medial 
82 
 
 
epicondyle that forms a large, broad flank at approximately the same level as the base of the 
trochlea, as observed in P. darwini (Paterson, 2017), P. valletoni (Savage, 1957), and P. rayi 
(Berta, 1994). This is similar to the condition in mustelids (Schutz and Guralnick, 2007). This 
condition is also retained in D. oregonensis (Deméré and Berta, 2002) and A. kernensis. Stem 
otariid P. starri possesses a somewhat poorly developed medial epicondyle compared to living 
otariids; while the medial epicondyle of T. mexicanus is more prominent than that of P. starri, it 
is also less medially protuberant than that of living otariids. The medial epicondyle of the fossil 
odobenids is somewhat varied in morphology, from the comparatively weakly developed 
epicondyle in N. mirum and I. downsi to the subtriangular, intermediately protuberant condition 
seen in P. pacifica to the massively medially enlarged epicondyle seen in D. seftoni and 
particularly V. imperialiensis where the medial epicondyle nearly doubles the width of the distal 
end of the humerus. 
Although Howell (1929) attributed the greatly developed medial epicondyle seen in Z. 
californianus to the flexors of the lower arm as needed for a more efficient power stroke, the 
presence of this condition in hindlimb-dominated modern O. rosmarus as well as D. seftoni and 
V. imperialiensis, both categorized by our analysis as hindlimb-dominated, calls this assumption 
into question. Gordon (1981) reports that during the forward lunge phase of the terrestrial gait of 
O. rosmarus, the manus remains flexed as the forelimb is retracted, but it seems unlikely that 
such a behavior would account for the enormous development of the medial epicondyle seen in 
derived odobenids. The functional role of the morphology of this structure in walruses as well as 
how and if it relates to aquatic locomotion remains an open question. 
4.1.4 Lateral Epicondyle—In our PCA, the negative end of both PC1 and PC3 
represents a lateral epicondyle that extends approximately as far proximally as the medial 
83 
 
 
epicondyle and flares proximolaterally from its base. At the positive end of both PC1 and PC3, 
the lateral epicondyle extends further proximally than the medial epicondyle to slightly less than 
halfway up the length of the shaft; it also flares at a slight angle proximally from its base. This 
landmark configuration limits the utility of understanding the functionality of these character 
states as it does not fully capture the variation in this structure because the only landmark 
describing the lateral epicondyle is landmark 26 (apex of the lateral epicondyle). This landmark 
largely shows only the relative height of this structure along the bone and not its differentiation 
from the shaft. Further investigation into the variation in this structure is therefore needed.  
From qualitative observations, I noted that the apex of the lateral epicondyle in otariids 
was higher than the condition in phocids, but the condyle itself was largely confluent with the 
humeral shaft, whereas the lateral epicondyle formed a structure distinct from the humeral shaft 
in phocids and, notably, O. rosmarus. In the latter, the lateral epicondyle is proximodistally 
shorter than the medial epicondyle and bows laterally out from its confluence with the lateral 
base of the condyle in a convex curve that is even more laterally protuberant than the condition 
of many phocid species. I also noted that the relative height of the lateral epicondyle varied 
significantly between species within the Phocidae. I consider it likely that this condition varies 
according to influences other than locomotion, but it remains unclear whether this results from 
other ecological adaptations or random phylogenetic variation. As the lateral epicondyle 
primarily serves as the attachment site for forearm extensors (Howell, 1929), this could be a 
point of variation related to the various uses of the forelimb and manus available to hindlimb-
dominated swimmers. 
In the fossil taxa, stem pinnipeds P. valletoni and P. rayi possess a tall, flank-like lateral 
epicondyle that rises proximally to approximately halfway along the shaft. D. oregonensis has a 
84 
 
 
low lateral epicondyle that consists only of a slight crook in the lateral edge of the shaft just 
proximal to the proximodorsal edge of the condyle; it is approximately equal in proximodistal 
height to the medial epicondyle. Interestingly, A. kernensis, which was categorized as a forelimb-
dominated swimmer by our analysis as well as other studies (Bebej, 2009; Furbish, 2015), 
possesses a lateral epicondyle that more resembles the condition in phocids as opposed to 
modern otariids. Its lateral edge forms a convex curve from the laterodistal edge of the condyle 
before an inflection point at which it sharply curves back towards the shaft, forming a distinctive 
structure from the main shaft. The lateral epicondyle of early otariid P. starri is relatively similar 
to the tall, flank-like condition seen in the stem pinnipeds, although it does not extend as far 
proximally as the stem species. T. mexicanus has a lateral epicondyle that forms a broad lateral 
convex bend in the lateral edge of the distal shaft. Although the O. rosmarus condition is 
approached in Valenictus, in the other odobenids, the lateral epicondyle approximates the 
morphology of D. oregonensis. The reason for the greater development of the lateral epicondyle 
in Odobenus and Valenictus compared to early odobenids is unclear, although it is possibly a 
product of allometry. Extension of the manus seems unimportant in the terrestrial locomotion of 
modern walruses (Gordon 1981), and it is not readily apparent how and if the greater 
development of the lateral epicondyle in these two derived odobenids would impact aquatic 
locomotion. 
4.2 The Evolution of Swimming Modalities in Otarioid Pinnipeds 
In Figure 7, I show the average discriminant scores associated with each taxon plotted 
across the pinniped phylogeny. According to the results of our FDAs, both the stem pinnipeds 
and the earliest diverging members of otarioid clades Desmatophocidae, Otariidae, and 
Odobenidae were likely all hindlimb-propelled swimmers. The classification of P. valletoni and  
85 
 
 
 
86 
 
FIGURE 7. Discriminant scores from the phylogenetically corrected flexible discriminant analysis plotted across that 
phylogeny. Specimen-specific discriminant scores can be viewed in Table 3. Cool colors represent a forelimb-dominated 
swimming style and warm colors represent a hindlimb-dominated swimming style.  
 
 
 
P. rayi as hindlimb-propelled swimmers is consistent with previous studies that recovered the 
plesiomorphy of hindlimb propulsion in pinnipeds (Bebej, 2009; Paterson, 2017; Hocking et al., 
2018, 2021; Paterson et al., 2020). Additionally, the resolution of P. valletoni, N. mirum, I. 
downsi, D. seftoni, and P. pacifica as close to the mean discriminant score also supports their 
reliance on a more intermediate swimming mode than modern hindlimb swimmers. However, 
our recovery of early otarioids as hindlimb-propelled runs contrary to the hypothesis that 
forelimb swimming evolved at the base of Otarioidea with a partial reversal to hindlimb-
dominated combination swimming at some point prior to the divergence of O. rosmarus. Instead, 
our results suggest that forelimb-dominated swimming actually evolved three different times in 
the otarioid lineage: once in an ancestor of desmatophocid A. kernensis, as previously recovered 
in Bebej (2009) and Furbish (2015); once in an ancestor of large-bodied odobenid P. barroni; 
and once either in an ancestor of crown Otariidae, as supported by the pFDA, or an ancestor of T. 
mexicanus and crown Otariidae, as supported by the npFDA. Notably I consider it possible that 
the pFDA misclassified T. mexicanus for several reasons. First, T. mexicanus possessed a 
humerus with long deltoid crest that runs parallel to the shaft, a prominent greater tubercle, and a 
protuberant medial epicondyle, all structures that correlated with forelimb-dominated propulsion. 
Furthermore, it misclassified the modern forelimb-dominated training taxa 33% of the time, and 
T. mexicanus had a posterior forelimb classification rate of 0.34. Given these circumstances, I 
consider it likely that T. mexicanus was an intermediate swimmer relying more on forelimb 
propulsion than its hindlimb-propelled ancestors but not yet as strongly adapted to forelimb 
propulsion as modern otariids.  
Regardless of the true locomotor ecology of T. mexicanus, the evolution of forelimb-
propelled swimming three separate times in Otarioidea raises questions about both why and how 
87 
 
 
this modality arose repeatedly in this lineage. As noted by Backhouse (1962), forelimb 
propulsion is highly unusual among marine mammals as the majority, including phocids and 
odobenids, produce most of their thrust with their hindlimbs. Cetaceans, sirenians, and the sea 
otter E. lutris all generate thrust via dorsoventral caudal propulsion; cetaceans and sirenians 
undulate their wide, flattened tails in the vertical plane while E. lutris undulates its tail and 
webbed hindlimbs (Fish, 1996). In the fossil record, both the Miocene desmostylian Desmostylus 
hesperus (Domning, 2002; Gingerich, 2005) and aquatic giant sloth Thalassocnus natans 
(Amson et al., 2015) appear to have primarily relied on alternating paddling strokes of the 
forelimb, similar to polar bears (Ursus maritimus), but this modality differs substantially from 
the bilateral pectoral oscillation used by otariids. The otariid mode of forelimb propulsion is 
indeed more similar to the “underwater flying” modality used in penguins and sea turtles 
(Feldkamp, 1987b; Gutarra and Rahman, 2022).  
One possible driver for the evolution of forelimb-propelled swimming could be for 
counteracting torque. Torque increases as the distance increases between the source of the 
propulsive force and the center of mass of the object. Forelimb propulsion places the source of 
propulsive force closer to the center of mass, decreasing torque and therefore reducing energy 
expenditures for counteracting torque. Indeed, otariids have a remarkably high aerobic 
efficiency, the ratio of power expended to mechanical work accomplished, among marine 
mammals (Fish, 1996). Otariids are also the fastest and most aquatically agile of the pinnipeds, 
with California sea lion Z. californianus able to swim up to 30 miles per hour (Campagna et al., 
2021). As suggested by English (1976), the higher swimming speed achievable by otariids 
compared to other pinnipeds grants access to faster prey items that slower predators cannot 
reliably capture. As discussed by Hocking et al. (2021), leopard seals rely more on forelimb 
88 
 
 
propulsion than any other modern phocid and their flippers converge on the otariid morphology; 
they also are able to achieve underwater speeds of up to 23 miles per hour and typically rely on 
speed, not stealth, to catch their fast-moving prey, including penguins (Muir et al., 2006). The 
ability to exploit this fast-swimming guild of prey species could have acted as a strong 
evolutionary attractor throughout pinniped evolution.  
As I have previously noted, nearly all the fossil otarioids I examined in this study possess 
an elongated deltoid crest regardless of whether the FDAs classified them as hindlimb- or 
forelimb-propelled. O. rosmarus, which is primarily hindlimb-propelled, also possesses this trait, 
ostensibly to provide a long moment arm for forelimb adduction during the forward lunge in the 
terrestrial gait cycle (English, 1977; Gordon, 1983). Although otariid terrestrial locomotion 
differs from that of O. rosmarus, primarily in that they fully support their weight on their limbs 
whereas O. rosmarus rests its bulk on its venter, otariids still produce the majority of thrust 
during terrestrial locomotion from forelimb adduction (Beentjes, 1990). Given that a 
proximodistally long deltoid crest is not intrinsically indicative of forelimb-dominated aquatic 
propulsion and seems to have evolved prior to this locomotor ecology, I posit that this trait may 
have first arisen as an adaptation to accommodate terrestrial locomotion in a lineage whose 
forelimb morphology is constrained by the need to effectively locomote both on land and in the 
water. The prior availability of this structure thus provides a mechanism for the repeated 
evolution of forelimb-dominated propulsion. 
Previous workers have posited that phocids are the most aquatically specialized of the 
pinnipeds as they lack external ears, are no longer able to walk on land, and generally spend 
more time at sea than their otarioid cousins (Howell, 1929; Ray, 1963; King, 1983; Berta et al., 
2015). Our results present an alternative viewpoint: after diverging from a common aquatic 
89 
 
 
ancestor, phocids further specialized the ancestral hindlimb-dominated intermediate swimming 
style through a variety of skeletal changes, including the telescoping of the neck and the locking 
of their hindlimbs in the backwards position by a massively enlarged astragalar process and the 
resulting tension in the flexor hallucis longus muscle. Otariids, in turn, are not “less” aquatically 
adapted than phocids but have simply solved the same ecological problem in a different way, 
specializing in rapid, agile swimming in the marine environment while retaining the ability to 
move relatively quickly while on land. Odobenids are somewhat intermediate to these two 
conditions and appear to have largely maintained the ancestral combination swimming style with 
hindlimb domination at higher speeds. Together with observations that many phocid species also 
use forelimb propulsion to varying degrees (Backhouse, 1962; Hocking et al., 2018, 2021), this 
trait distribution demonstrates that forelimb-dominated propulsion and hindlimb-dominated 
propulsion form a continuum rather than two diametrically opposed conditions. 
 
5 Conclusion 
I tested the hypothesis that humeral morphology is diagnostic for swimming modality in 
pinnipeds using three-dimensional geometric morphometrics and then applied this methodology 
to predicting the swimming modes of fossil pinnipeds. Our results supported a hindlimb-
dominated mode of aquatic propulsion as the plesiomorphic condition for all pinnipeds as well as 
otarioids, with three separate acquisitions of forelimb-dominated propulsion in desmatophocid A. 
kernensis, odobenid P. barroni, and in the Otariidae. The primary use of the forelimbs in aquatic 
locomotion seems to constrain humeral morphology and the range of other associated functional 
uses available to these taxa.  As our analyses indicate that an elongated deltoid crest arose before 
90 
 
 
forelimb propulsion in otarioids, this condition seems to have predisposed otarioids to the 
evolution of forelimb propulsion. Overall, I find that when considering pinniped locomotion 
from the perspective of the fossil record as well as their modern ecology, the two main 
locomotor modes represent a continuum of swimming modalities. 
 
6 Bridge 
 Understanding how pinniped locomotion has evolved provides critical insight into how 
their positionality within marine ecosystems has changed through time regarding factors from 
accessible prey items and predator avoidance to cooperative interactions with other predators and 
the particular marine and terrestrial environments they typically inhabit. However, locomotor 
abilities are only one of the dimensions influencing pinniped ecology. Abiotic environmental 
variables such as water temperature, coastal landforms, and presence or absence of ice also 
determine the geographic range and resources available to these marine mammals. As climate 
changes through time, so too does the biogeography of the organisms that live under those 
conditions. This coupling between the range of an organism and shifts in climate within 
evolutionary time provides a sort of paleoecological laboratory that conservation paleobiologists 
can use to better understand the fundamental niche of the organism in question, that is, the range 
of environmental conditions it is capable of tolerating as opposed to those it is currently known 
to inhabit. In the next chapter, I turn to the fossil record of the walrus Odobenus rosmarus, the 
last living odobenid, to investigate its paleobiogeography with respect to the availability of sea 
ice in order to model its response to the next hundred years of anthropogenic climate change. 
91 
 
 
CHAPTER IV 
 
FOSSIL DATA INDICATE ODOBENUS ROSMARUS, THE LAST 
ODOBENID, CAN SURVIVE IN SEASONALLY ICE-FREE CONDITIONS 
 
 This chapter is planned for submission to Frontiers of Biogeography, co-authored with E. 
B. Davis. The project idea, data collection, coding and ecological niche modeling, and writing 
were all performed by M. K. Tate-Jones. Editorial assistance was provided by E. B. Davis. 
 
1 Introduction 
 With the Arctic Circle warming two to four times faster than the rest of the planet, its 
constituent ecosystems are undergoing similarly rapid changes (Yletyinen 2019, Huntington et 
al. 2020, Turner et al. 2020, Jacobs et al. 2021, Rantanen et al. 2021). As anthropogenic climate 
change and polar amplification increase Arctic temperatures, the melting of sea ice is 
accelerating (Meehl et al. 2018, Diebold and Rudebusch 2021). Even conservative models 
predict a high likelihood of a seasonally ice-free Arctic by 2050 or earlier (Overland and Wang 
2013, Jahn et al. 2016, Notz and Stroeve 2018, Guarino et al. 2020, Årthun et al. 2021, Diebold 
and Rudebusch 2021). As sea ice plays a critical role in Arctic marine ecosystems (Wassmann et 
al. 2011), its retreat will drastically alter their structuring and composition (Post et al. 2013). The 
loss of sea ice is particularly threatening for the wide diversity of pagophilic (ice-dependent) 
species from microscopic algae (Clarke and Peck 1991), bacteria (Amiraux et al. 2021), and 
zooplankton (Werner 1997) to seabirds (Divoky et al. 2015), polar bears (Rode et al. 2014), and 
pinnipeds, including six seal species and the walrus (Laidre et al. 2015). Understanding how the 
decline in sea ice availability will affect these organisms as well as the broader ecosystems with 
92 
 
which they interact is crucial for mitigating the impact of climate change on the Arctic 
environment. 
 One such ice-dependent organism is the walrus, Odobenus rosmarus (Fig. 1a, b; Linnaeus 
1758). The only living member of the pinniped family Odobenidae, O. rosmarus is a large-
bodied molluscivorous marine mammal easily identified by its prominent tusks and bristling 
whiskers, or vibrissae (Fay 1985). Today, the walrus has a discontinuous circumpolar distribution 
and consists of two genetically distinct subspecies, the Atlantic (O. rosmarus rosmarus) and 
Pacific (O. rosmarus divergens) walrus (Fay 1985, Lindqvist et al. 2009; Fig. 1c). Because of its  
Figure 1. Walrus general appearance and typical habitat and map of known distribution of the 
modern walrus Odobenus rosmarus. a) Female Pacific walrus and calf on a summer ice floe; 
photograph taken by United States Fish and Wildlife Service, public domain. b) Atlantic walruses 
hauled out at a terrestrial rookery in Svalbard, Norway; photograph taken by Jerzy Strzelecki and 
provided under Creative Commons Attribution. Photographs were downloaded from Wikimedia 
Commons. c) Range of the two subspecies of walrus: the Atlantic walrus (yellow; O. rosmarus 
rosmarus) and the Pacific walrus (purple; O. rosmarus divergens) (Higdon and Stewart 2018).  
 
93 
 
unique feeding modality, O. rosmarus serves as an Arctic ecosystem engineer (Hocking et al. 
2021). While foraging, walruses uncover infaunal bivalves by sweeping away sediment with 
their flippers or through oral hydraulic jetting. This sediment bioturbation restructures benthic 
communities over thousands of square kilometers of ocean seafloor and changes ocean 
productivity by redistributing nutrients within the water column (Ray et al. 2006). Furthermore, 
walruses serve as a critical food source for Alaskan and Siberian indigenous peoples; historical 
collapses of local walrus populations have led to famine and starvation in these communities 
(Metcalf and Robards 2008, Jay et al. 2011, Ray et al. 2016), demonstrating the importance of 
walruses for food security within the Arctic Circle. 
The crucial position walruses hold within Arctic ecosystems and cultures necessitates a 
robust understanding of how sea ice loss will impact them. Currently, walrus dependence on sea 
ice varies seasonally and according to sex. In the winter, walruses gather in mixed herds in ice-
free polynyas deep within the pack ice where they forage for bivalves in the shallow waters of 
the continental shelf (Fay 1982). In late spring, males typically remain in more southerly regions 
where they haul out at terrestrial rookeries. Alternately, females and calves follow the retreating 
sea ice northwards, using ice as a resting platform between foraging trips (Beatty et al. 2016). As 
a result, the overall decline in summer sea ice as well as its increasing distance from the bivalve 
beds where walruses typically forage raises the question of how and whether walruses will adapt 
to its total loss. Jay et al. (2017) found that when walruses lack access to sea ice during the 
summer, they spend significantly less time foraging or resting than when sea ice is available, 
diminishing their ability to develop endogenous fat stores. Models developed by Udevitz et al. 
(2017) predicted that by the second half of the 21st century, this relationship between sea ice loss 
and fat storage would lead female walruses to weigh 7–12% less by the end of the summer 
94 
 
foraging season than they do currently. Although the broader demographic impacts of such 
changes remain unknown, the impact of this predicted trend could be substantial, especially since 
walruses have the lowest reproductive rate of all pinnipeds, calving only every two years (Fay 
1985). Already walrus behavior has been shifting. Increasingly, female walruses and their calves 
are hauling out in large numbers along coastlines and islands by late summer as sea ice becomes 
too distant and unreliable (Jay et al. 2012). This change increases their vulnerability not only to 
polar bear predation but also to trampling mortality as walruses tend to stampede when disturbed 
(Øren et al. 2018). Because trampling deaths disproportionately affect calves and young 
walruses, this trend could be particularly detrimental to walrus populations (Udevitz et al. 2013). 
With the ambiguity surrounding the future of the walrus, turning to the past can offer new 
insights. Historically, extirpated walrus populations inhabited significantly warmer climes than 
current walrus populations (Boessenecker and Churchill 2021). Both the Canadian Maritimes O. 
rosmarus, hunted to extinction by the end of the 18th century (McLeod et al. 2014), and the 
recently identified Icelandic O. rosmarus, eradicated following Norse settlement in medieval 
times (Keighley et al. 2019), seemed to have thrived in seasonally ice-free conditions. The fossil 
record provides further evidence that modern walruses may indeed inhabit only a subset of the 
environmental conditions they are capable of tolerating. Paleontologists have identified 
Pleistocene fossils of O. rosmarus much further south than they range today, including in Tokyo 
(Harington 2008), San Francisco Bay (Jefferson 1991, Harington 2008), and Edisto Island off the 
southern coast of South Carolina (Roth and Laerm 1980). Even during the coldest periods of the 
Pleistocene, these areas were warmer than current walrus habitat and likely ice-free much or all 
of the year during both glacial and interglacial intervals (Batchelor et al. 2019).  
95 
 
While the historical and paleontological distribution of O. rosmarus suggests that 
walruses may indeed be capable of surviving warmer, more seasonally ice-free conditions than 
they do today, this hypothesis has yet to be quantitatively tested. Ecological niche models 
(ENMs) provide a robust methodology for predicting range shifts by correlating geographic 
occurrence data with environmental predictor variables (Elith et al. 2010). These models can then 
be projected to the conditions predicted by future climate simulations. Despite their utility, such 
models are limited by the difference between the realized niche and fundamental niche of an 
organism. Although a species may currently occupy a certain set of environmental parameters, 
known as its realized niche, these conditions do not necessarily reflect the total breadth of 
conditions within which the species can survive, its fundamental niche (Svenning et al. 2011). 
The inclusion of fossil occurrences of extant organisms can illuminate potential gaps between 
realized and fundamental niches (Davis et al. 2014, Dietl et al. 2015, Maguire et al. 2015, Myers 
et al. 2015, Lima-Ribeiro et al. 2017). For instance, Jarvie et al. (2021) used fossil occurrences of 
the tuatara, a New Zealand rhynchocephalian reptile, to identify areas of suitable habitat that fall 
outside the ecological limits of its current distribution. Even more notably, Mallorcan midwife 
toads, believed to be extinct until found in remote gorges on the island in 1980, have been 
successfully reintroduced to suitable areas as identified by EMNs generated using fossil 
occurrence data (Bloxam and Tonge 1995). Here, we are the first to apply these techniques to the 
walrus.  
In this study, we test the hypothesis that the fundamental niche of O. rosmarus is broader 
than the range of conditions they currently occupy and that walrus fundamental niche in fact 
includes the seasonally ice-free conditions predicted to characterize the Arctic within the next 
two to three decades. First, we evaluate whether the environmental conditions inhabited by 
96 
 
walruses during the Pleistocene are significantly different from current conditions using two 
datasets of fossil and modern geographic occurrences, respectively. We then construct two sets of 
ENMs derived from these two datasets to quantify the current ecological and paleoecological 
niche of O. rosmarus. Finally, we extrapolate each of these ENMs to future climate scenarios and 
compare the level of threat to walrus survivorship implied by each set of projections. We 
conclude by exploring the implications of our results for further research into odobenid ecology 
as well as the development of conservation policy. 
 
2 Materials and Methods 
2.1 Input Data 
 2.1.1 Modern Occurrence Data— We compiled modern occurrence data both from 
online databases and literature sources for both walrus subspecies. Occurrence records for O. 
rosmarus were downloaded from the Global Biodiversity Information Facility (GBIF) on July 
30, 2021, and from the Ocean Biodiversity Information System (OBIS) databases on November 
30, 2021. Because occurrences from GBIF and OBIS were concentrated around North America 
and Svalbard, we supplemented these database occurrences with additional records drawn from 
the literature to represent walrus sightings from the northern coast of Norway (Gjertz et al. 1993) 
as well as populations from the Kara, Laptev, and Chukchi Seas (Zyryanov and Vorontsov 2009, 
Lydersen et al. 2012, Glazov et al. 2013). We then cleaned the data of duplicates, fossil 
occurrences, and points with latitude and longitude that fell out of the expected range (e.g., 
occurrences that fell well inland or in the Southern hemisphere). After cleaning, the dataset 
contained 4586 occurrence points for modern O. rosmarus (Fig. 2).  
97 
 
Because of geographic sampling bias, spatial autocorrelation is ubiquitous in databases like 
GBIF and OBIS (García-Roselló et al. 2015) as well as regional occurrences records from 
specific literature sources (Dennis and Thomas 2000, Kadmon et al. 2004). Such autocorrelation 
can overinflate measures of model performance, leading to overfitting and diminished model 
applicability (Veloz 2009). Several studies have demonstrated that thinning occurrences 
according to a spatial filter can reduce the effects of sampling bias (Pearson et al. 2007, Kramer-
Schadt et al. 2013, Boria et al. 2014, Fourcade et al. 2014). Therefore we used the R package 
spThin (Aiello-Lammens et al. 2015) to spatially filter the modern occurrences so that no 
occurrence fell within 50 km of any other occurrence. The 50 km spatial filter reduced the 
density of well-sampled areas like the Bering Sea and Svalbard to approximate the distribution of 
occurrences in areas with fewer recorded occurrences like the coast of Siberia. Following 
thinning, 383 total modern occurrences remained in the modern dataset. 
2.1.2 Fossil Occurrence Data—We downloaded fossil occurrence data from the 
Paleobiology Dabatase (PBDB; http://paleobiodb.org) on July 21, 2021, for Odobenus rosmarus 
from the Pleistocene and Holocene. Again, we cleaned this dataset of duplicate occurrences. 
None of the downloaded fossil occurrences fell outside of the expected latitude and longitude. 
Because most fossils are found on land as a result of collection bias, the majority of fossil 
occurrences in our dataset were located on terrestrial raster cells. Consequently, we manually 
moved fossil occurrence datapoints that fell on terrestrial cells to the closest laterally adjacent 
marine cell using QGIS (QGIS Development Team, 2022). Following cleaning, our fossil dataset 
contained 103 occurrences of O. rosmarus. We did not thin the fossil occurrences both because 
preliminary models performed poorly when using thinned fossil datasets and because the 
98 
 
 
Figure 2. Occurrences of O. rosmarus from the dataset used in this study. Modern walrus 
occurrences are indicated by orange circles and were compiled from the Global Biodiversity 
Information Facility (GBIF), Ocean Biodiversity Information System (OBIS), Gjertz et al. (1993), 
Zyryanov and Vorontsov (2009), Lydersen et al. (2012), and Glazov et al. (2013). Fossil 
occurrences are indicated by green triangles and were downloaded from the Paleobiology Database 
(PBDB). 
99 
 
taphonomic biases inherently present in the fossil record have already naturally thinned the fossil 
sample.  
Our fossil dataset exhibits relatively poor age constraint both because many of the 
specimens come from poorly dated Pleistocene terrace deposits and because the PBDB assigns 
fossil ages using North American Land Mammal Age rather than absolute dates (Peters and 
McClennen 2015). Additionally, we used paleoenvironmental raster data from the Last Glacial 
Maximum (LGM) because marine raster data were not available for other time periods and 
constructing them was outside the scope of this project. However, not all of our fossil 
occurrences have age ranges that include this time period. Despite these problems, we decided to 
include all of the available fossil occurrences for O. rosmarus for several reasons. First, we 
assume climatic conditions during the LGM are approximately equivalent to those of other 
glacial periods during the Pleistocene (Lisiecki and Raymo 2005, Zachos et al. 2008, Alonso-
Garcia et al. 2011, Verbitsky et al. 2018, Batchelor et al. 2019). Second, although some number 
of these occurrences undoubtedly fall during warmer transitional or interglacial periods, 
considering them to all occur during the LGM allows for the most conservative temperature 
estimate of their fundamental ecological niche. 
2.1.3 Environmental Predictors—To train our ENMs and quantify the realized 
ecological niche of walruses, we used three climatic variables: minimum annual sea surface 
temperature (SSTmin), annual sea surface temperature range (SSTrg), and annual sea surface 
salinity range (SSSrg). Although previous studies have found strong relationships between walrus 
presence and the distribution of sea ice (Beatty et al. 2016, Udevitz et al. 2017), layers directly 
quantifying the presence of sea ice are not available from paleo-MARSPEC and future 
predictions of such are considered overly uncertain for use in ENM construction (Bonan et al. 
100 
 
2021). Therefore we chose to use SSTmin, SSTrg, and SSSrg because together they mechanistically 
impact sea ice formation and distribution. When sea ice forms, the salinity of the remaining 
liquid water increases; when ice melts in the summer, seawater salinity decreases again. 
Additionally, the polynyas where walruses congregate in winter are associated with high winter 
salinity which prevents the water there from freezing (Winsor and Björk 2000). Because of these 
factors, areas with a larger relative SSSrg likely correlate with the seasonal presence of sea ice. 
While the variance of sea surface salinity would be preferable to range, variance layers were not 
available for the LGM or for future projections and creating these layers is outside the scope of 
this project. Furthermore, Gebruk et al. (2021) demonstrated that walrus distribution also 
correlates with the location and biomass of bivalves, their main prey item, and both salinity and 
temperature influence the distribution and biomass of bivalves (Green 1973, Van Dolah et al. 
1999, Feng and Papeş 2017, Pourmozaffar et al. 2020). Although previous studies have 
identified a correlation between walrus presence and water depth (Beatty et al. 2016), we did not 
include bathymetry as one of our predictor variables because our fossil dataset hails almost 
exclusively from the terrestrial environment, artificially biasing the fossil dataset towards 
shallow water. Instead, we incorporated bathymetry post hoc by cropping our final binary 
presence/absence projections to depths of 200 meters or less since walruses rarely forage in 
deeper waters (Born et al. 2005, Lowther et al. 2015, Garde et al. 2018) and because the wind-
driven polynyas inhabited by walruses during the winter do not typically exceed 200 meters in 
depth (Winsor and Björk 2000). Finally, the calculated variable inflation factors (VIFs) for these 
three variables were less than 5, reducing the impact of multicollinearity on the models 
(Dormann et al. 2013). We calculated VIFs calculated using R package usdm (Naimi 2015).  
101 
 
We downloaded modern and future environmental raster data from bio-ORACLE 2.2 
(Ocean Rasters for Analysis of CLimate and Environment; Tyberghein et al. 2012, Assis et al. 
2018) and paleoclimatic data for the Last Glacial Maximum (LGM; 21 ka) from paleo-
MARSPEC (MARine SPatial ECology; Sbrocco 2014) at a resolution of 5 arcmin using the R 
package sdmpredictors (Bosch et al. 2017). We chose predicted future raster layers for the years 
2050 and 2100 based on Representative Concentration Pathway (RCP) 8.5 which represents a 
climate scenario for relatively high greenhouse gas emissions (Riahi et al. 2011, van Vuuren et 
al. 2011).  
We took several steps to prepare the rasters for analysis. In the modern and future raster 
layers, SSTmin did not extend lower than -1.94 °C, the approximate freezing temperature of 
water; however, SSTmin for the LGM extended below -1.94 °C. To ensure comparability between 
the three climate scenarios, we changed all values for SSTmin that were below -1.94 °C in the 
LGM raster to this value. We also recalculated SSTrg for the LGM by subtracting the new SSTmin 
raster from an SST maximum layer so the SSTrg rasters would be comparable between the 
modern and LGM datasets. Raster manipulations were accomplished using the R package raster 
(Hijmans et al. 2015).  
2.2 Analyses 
All analyses were performed in R (R Core Team 2021) unless otherwise noted. 
 2.2.1 Comparing Niche Space— To compare the ecospace represented by the modern 
and fossil occurrence datasets for O. rosmarus, we first extracted the values of SSTmin, SSTrg, 
and SSSrg associated with each occurrence and calculated the mean and standard deviation for 
each of the sets of extracted environmental variables. Next, we visualized the variance between 
the environmental space occupied by each respective dataset using broken-stick principal 
102 
 
components analysis (PCA) (Legendre and Legendre, 1998). We then conducted a permutational 
multivariate analysis of variance (PERMANOVA; Anderson 2001) to quantitatively test the 
hypothesis that Pleistocene populations of O. rosmarus inhabited a significantly different niche 
space than modern populations. Because PERMANOVA assumes that the multivariate dispersion 
of the variables in question is roughly equal, we began by evaluating this assumption using the 
PERMDISP2 procedure of Anderson (2006) to assess for multivariate homogeneity of group 
dispersions using the “betadisper()” function in the R package vegan (Oksanen et al. 2018). This 
test demonstrated that the variance of the modern and fossil datasets was significantly different 
(p = 0.001). Although the results of PERMANOVA are robust to such heterogeneous group 
dispersions, this condition applies only when the compared groups are the same size (Anderson 
and Walsh 2013). To meet this criterion, we used a modified bootstrapping process to produce 
1000 datasets of equal size to the fossil dataset by randomly sampling without replacement from 
the modern dataset. Next, we ran the PERMANOVA using vegan function “adonis()” with 9999 
permutations using the Euclidean method for calculating pairwise distance between individuals 
on each of the 1000 modern sample datasets using a for-loop. We report the average p-value 
from this modified bootstrap as well as the effect size (R2). 
2.2.2 Ecological Niche Models—To estimate suitable habitat range for walruses, we 
implemented the maximum entropy (Maxent) algorithm using the R package ENMeval version 
2.0 (Kass et al. 2021). Maxent uses a machine learning algorithm to identify the environmental 
variables that best discriminate between occurrence points and background points (Phillips et al. 
2004, Elith et al. 2006). We chose to use Maxent for our ENMs because it performs well with 
small sample sizes (Hernandez et al. 2006, Pearson et al. 2007), requires relatively little 
computational power relative to other ENM methodologies (Kaky et al. 2020), outperforms 
103 
 
conventional methods when analyzing complex nonlinear relationships (Elith et al. 2006, 
Shoemaker et al. 2018), and balances overfitting and model complexity with overgeneralization 
when used in conjunction with species-specific model parameters (Radosavljevic and Anderson 
2014).  
Following the protocol of Simoes et al. (2020), we first built a buffer around the modern 
occurrence points using a radius of 3000 km based on the dispersal ability of walruses (Reeves et 
al. 1992). We then defined the calibration area by masking the modern environmental layers 
according to this buffer. Using the “randomPoints” function in R package dismo (Hijmans et al. 
2017), we next generated 100,000 random background points. To calibrate the model, we tested 
four Maxent features classes (linear, linear quadratic, product, and linear quadratic product) and 
six regularization multipliers (0.1, 0.25, 0.5, 1, 2, and 4). Following Jarvie et al. (2021), we 
evaluated the appropriateness of the resulting models using the average area under the curve 
(AUC) and the omission rate of the minimum training presence value (ORMTP). AUC measures 
predictive accuracy on a scale from 0 to 1 with a value of 0.5 equivalent to random predictive 
accuracy and a value of 1 denoting absolute accuracy. ORMTP signifies goodness of fit on a scale 
of 0 to 1 with higher values indicating greater degrees of overfitting. 
Out of the models generated under these parameters, we selected the model with the 
lowest Akaike Information Criterion corrected for small sample sizes (AICc) value, which 
optimizes model fit with model complexity (Warren and Seifert 2011, Muscarella et al. 2014). 
We chose this model selection criterion because previous studies have shown AICc provides the 
most conservative predictions and also outperforms other criteria in terms of selecting the true 
model as the best model in simulations (Katz and Zellmer 2018) and because AICc optimizes 
transferability across different time periods because of its penalization of model complexity 
104 
 
(Moreno-Amat et al. 2015). For the final model, we selected complementary log-log (cloglog) as 
the output format per the recommendation of Phillips et al. (2017) and evaluated the relative 
contribution of each of the three environmental variables using a jackknife approach. We then 
projected this final model to present environmental conditions and visually assessed how well it 
fit the current known distribution of walruses. Lastly, we projected this model to 2050 and 2100 
for RCP 8.5 to evaluate how walrus distribution is predicted to shift according to walrus current 
ecological niche. 
We then repeated this process for the fossil dataset. The optimal model generated from 
the fossil dataset was subsequently projected to modern conditions and visually compared to 
known walrus range and finally projected to future conditions for 2050 and 2100 under RCP 8.5. 
2.2.3 Binary Maps and Centroid and Area Calculations—To assess the differences 
in the ranges predicted by the modern and fossil models, we calculated the area and centroid of 
the latitude of each of the projected ranges for current and future conditions from the two ENMs. 
To quantify these differences, projections were first restricted to the latitudes inhabited by 
walruses during historical times according to the range defined by Higdon and Stewart (2018) as 
we consider it unlikely walruses will establish new populations south of this range. We then 
restricted the projections to water depths of 200 meters or less as previously explained. We next 
converted these restricted ranges from continuous suitability indices ranging from 0 to 1 to 
binary predictions of presence or absence by rejecting cells with a suitability index less than 
0.25. We then calculated the centroid latitude of the range and the area of the predicted range 
using R and QGIS (QGIS Development Team, 2022) respectively. 
 
 
105 
 
3 Results 
3.1 Niche Space Comparisons 
Extracting environmental variables revealed that modern occurrences were associated 
with an SSTmin of -1.85 ± 0.55 °C, SSTrg of 10.84 ± 2.43 °C, and an SSSrg range of 4.84 ± 1.88 
PSS (Practical Salinity Scale) (mean ± standard deviation); fossil occurrences were associated 
with an SSTmin of 0.57 ± 4.73 °C, SSTrg of 7.87 ± 5.00 °C, and an SSSrg of 2.90 ± 2.92 PSS 
(mean ± standard deviation). Our broken-stick PCA identified principal components (PCs) 2 and 
3 as significant, accounting for 32.02% and 22.09% of the variation, respectively (Fig. 3). 
Although PC1 accounted for 45.89% of the variation, it was nonsignificant according to this test  
 
Figure 3. Projections of the ecospace for modern (maroon) and fossil (light blue) occurrences of 
O. rosmarus onto the significant axes of a broken stick principal component analysis (PCA). The 
arrows indicate how sea surface temperature minimum, sea surface temperature range, and sea 
surface salinity range are distributed along principal components 2 and 3. See Fig. S1 for plot of 
PC1 and PC2. 
106 
 
  
Table 1. Percent contribution of the environmental variables sea surface temperature 
minimum (SSTmin), sea surface temperature range (SSTrg), and sea surface salinity range 
(SSSrg) to each of the three principal components (PCs) from the broken-stick principal 
c omponents analysis. The significant PCs are demarcated in grey.    
Environmental 
variable PC1 PC2 PC3 
SSTmin 23.85 60.27 18.80 
SSTrg 35.67 35.45 37.27 
SSSrg 40.48 4.28 43.92 
 
(Appendix D). The percent contribution of each environmental variable to the three PCs is 
summarized in Table 1. For the PERMANOVA comparing the fossil and modern ecospace, the 
average p-value for the 1000 bootstrapped sample runs was 0.0001 ± 0.0000 (mean ± standard 
deviation) with significance level α = 0.0005. The average R2 = 0.125 ± 0.010 (mean ± standard 
deviation). 
3.2 Ecological Niche Models 
The predictive ability of both the modern and fossil models was excellent. The modern 
model produced an average AUC = 0.885 ± 0.011 (mean ± SE) and AUC range = 0.872–0.900, 
and the fossil model produced an average AUC = 0.849 ± 0.012 (mean ± SE) and AUC range = 
0.829–0.865 (Appendix E). Neither model was overfit, with the modern model average ORMTP = 
0.010 ± 0.003 (mean ± SE) and range = 0.002–0.010 and the fossil model average ORMTP = 
0.017 ± 0.006 (mean ± SE) and range = 0.009–0.031 (Appendix E). In the modern ENM, the  
variable that contributed the most to the model was minimum SST followed by SST range and 
subsequently SSS range; for the fossil ENM, the minimum SST had the highest contribution 
followed by SSS range and finally SST range. For the modern model, the iteration with the 
lowest AICc (10343.94) used the linear quadratic function and 0.1 regularization multiplier; for  
107 
 
 
 
108 
 
Figure 4. Maximum entropy (Maxent) models of environmental suitability derived from modern (a–c) and fossil (d–f) occurrences of Odobenus 
rosmarus as projected to current (a and d) and future (b, c, e, and f) climate scenarios. We constructed the fossil model using climate data from 
paleo-MARSPEC (Sbrocco 2014); for building the modern model and for the current and future projections of each model, we utilized climate data 
from Bio-ORACLE 2.0 (Assis et al. 2018). The future climate scenarios derive from Representative Concentration Pathway (RCP) 8.5. In general, 
the modern model predicts that suitable conditions for walruses will shift far northward by the end of the 21st century. In contrast, the fossil model 
predicts relatively little change in environmental suitability. Map scale and coordinate grid are the same as Figs. 1 and 2. 
 
the fossil model, the lowest AICc (2770.164) was achieved with the same feature class and 
regularization multiplier as the modern model. 
3.3 Projections to Current Conditions 
When projected to current environmental conditions, the modern model explained the 
known range of walruses remarkably well overall, with some exceptions (Fig. 4a; for known 
range of walruses, see Born et al. 1995, Jay et al. 2011, Lydersen et al. 2012, Glazov et al. 2013, 
Higdon and Stewart 2018). This model did identify high suitability (> 0.8) in areas where 
walruses are not known currently or in recent historical times, such as the western East Siberian 
Sea, the Beaufort Sea, the western Canadian Arctic Archipelago, the eastern-central and 
southeastern coast of Kamchatka, and the Sea of Okhotsk. Interestingly, the waters around the 
Canadian Maritimes received a relatively low suitability index (< 0.25) except for the eastern 
extent of the Bay of Fundy (0.40–0.52). Additionally, while the majority of the area around 
Iceland displays a low suitability index (< 0.10 south of the island and < 0.20 north of the 
island), our model identified the ocean surrounding northwestern Iceland with medium suitability 
(< 0.5), the same area where Keighley et al. (2018) identified the highest density of the medieval 
Icelandic walrus population. 
Conversely, when we projected the fossil model to current conditions, several 
discrepancies with the known range of walruses arise (Figure 4d). The fossil model predicts very 
high suitability (>0.90) in areas where walruses are not known to occur outside of individual 
vagrants, such as the western coast of Canada, the Sea of Japan, the Yellow Sea, the northeast 
coast of the United States, the southern coast of Norway, and the southern North Sea. However, 
the fossil model does predict extremely high suitability around the entirety of the Canadian 
Maritimes (<0.99) and also better predicts the distribution of walruses on the coast of Iceland as 
109 
 
mapped by Keighley et al. (2019) with a suitability index of up to 0.80 in the northwestern 
corner of the island, up to 0.73 in its northeastern extent, and up to 0.66 in the southwestern 
reaches. 
3.4 Future Projections 
According to the modern model, suitable conditions shift only slightly northwards by 
2050 (Fig. 4b). The most noticeable change is the development of higher suitability around 
Svalbard, Franz Josef Land, and Severnaya Zemlya compared to modern conditions and 
decreased suitability off the eastern coast of Chukotka Province. However, by 2100, the 
suitability of much of the current range of walruses diminishes appreciably (Fig. 4c). The 
suitability of the Bering Sea drops precipitously, particularly along the eastern coast of 
Kamchatka and Chukotka in northeastern Russia. The southern Chukchi Sea, where female 
Pacific walruses and their calves migrate in the summer, becomes less suitable as well, albeit not 
as severely as the Bering Sea. Suitability also decreases sharply around Svalbard, Franz Josef 
Land, and in the Barents and Kara Seas, areas formerly characterized by high suitability. Hudson 
Bay suitability indices drop by more than 40% throughout most of its extent, especially in the 
southern reaches of the bay where only a small relict population of walruses remains today 
(Higdon and Stewart 2018). Notably, the Laptev Sea remains highly suitable by 2100. 
Conversely, the fossil model essentially indicates that the Arctic will, overall, become better 
suited for walruses over the next 80 years as the biome warms. No areas where walruses 
currently reside become less suitable from current conditions to 2050 or 2100 and in most cases 
these areas become more suitable (Fig. 4e, f). 
 
 
110 
 
3.5 Binary Maps and Area Calculations 
Consistent with predictions of Arctic warming, our results predict decreasing ecological 
niche area and north-shifting latitudinal centroids for walruses by 2100 under both models (Table 
2), although these trends are extremely small for the fossil model. For the modern model, the 
centroid of the current range has a latitude of 70.54 °N; by 2050, the centroid latitude is 
predicted to shift to 71.04 °N and by 2100 to 71.82 °N. The fossil model predicts little 
appreciable shift in the centroid latitude under the projected scenarios from 71.10 °N in current 
conditions to 71.17 °N by 2050 and 71.19 °N by 2100. The modern model predicts that current  
suitable habitat occupies approximately 6.77 x 106 km2, and the fossil model predicts 7.10 x 106 
km2 of current suitable habitat (Fig. 5a, d). Under RCP 8.5, the modern model predicts that this 
range will grow by 1.26% by 2050 but shrink by 11.31% by 2100 (Fig. 5b, c). However, the 
fossil model predicts almost no change in suitable habitat, with an increase of 0.07% by 2050 
and a decrease 0.001 % by 2100 (Fig. 5e, f).  
4 Discussion 
Our findings show that during the Pleistocene, O. rosmarus occupied a significantly 
wider ecospace than its current representatives inhabit, supporting our hypothesis that walrus 
fundamental niche does indeed include the relatively warm, seasonally ice-free conditions likely 
to characterize the Arctic in the near future. Below, we discuss how the outcomes of the PCA and  
PERMANOVA relate to our understanding of walrus fundamental niche. We then interpret the 
modern and future projections of both the modern and fossil ENMs regarding this understanding 
and predict how walrus populations will respond to climate change and the associated loss of sea 
ice. Finally, we recommend future research directions and consider the implications of our study 
for walrus conservation in the face of Arctic warming. 
111 
 
 
Table 2. Geographical changes in ecological niche space predicted by the modern ENM and 
fossil ENM.  
  
 
Centroid Distance Shifted Percent Change Time Period Latitude from Current Binary Area from Projected Centroid Latitude Current Range 
 
Modern model 
Current 70.54 °N n/a 6.77 x 106 km2 n/a 
2050 71.04 °N 55.5 km north 6.85 x 106 km2 1.26 
2100 71.82 °N 142.08 km north 6.00 x 106 km2 -11.31 
Fossil model 
Current 71.18 °N n/a 6.98 x 106 km2 n/a 
2050 71.17 °N 11.1 km south 6.95 x 106 km2 -0.36 
2100 71.19 °N 22.2 km north 6.95 x 106 km2 -0.37 
 
4.1 Niche Space Comparisons 
Plotting our PCA shows that the ecospace inhabited by the fossil dataset is wider than 
that of the modern dataset with respect to both PC2 and PC3, the significant PCs according to the 
broken-stick criterion (Fig. 3). The results of our PERMANOVA, which support the hypothesis 
of a significantly different niche space between the two populations (p = 0.0001), demonstrate 
that the difference in ecospace shown in the PCA is statistically significant, if relatively small in 
terms of real-world impacts (R2 = 0.125). Together, these analyses indicate that walrus 
fundamental niche is not fully realized by its current distribution and that this fundamental niche 
does in fact include seasonally ice-free condition.  
Furthermore, our results probably still underestimate the true breadth of walrus 
fundamental niche because of biases within our analyses, particularly with regard towards 
warmer climates. Our fossil occurrence dataset is likely biased towards interglacial occurrences 
for a number of taphonomic reasons. Because sea level was often greater than 100 meters lower 
during glacial periods than it is today (Berends et al. 2021), many of the coastal facies likely to 
preserve walrus fossils from glacial periods are currently below sea level and thus largely 
112 
 
 
 
113 
 
Figure 5. Binary maps showing optimal walrus habitat for current and future climate scenarios for the modern model (a–c) and fossil 
model (d–f). We generated these maps by constraining the optimal habitat to cells with a suitability index of 0.25 or greater and cropping 
the areas to water depths of 200 meters or less. Overall, neither model predicted a significant change in optimal habitat by 2050. In 
contrast, the predictions of the two models diverge substantially by 2100, with the modern model predicting a large northwards 
contraction of walrus range and the fossil model predicting little appreciable change. Map scale and coordinate grid are the same as 
Figs. 1 and 2. 
 
inaccessible to fossil collectors. While some few specimens have been recovered through 
dredging (e.g., the San Francisco Bay specimens; Harington 2008), almost all of the O. rosmarus 
fossils included in our study have been recovered on land in areas that would have been coastal 
or underwater during past interglacial periods but far from the shore during glacial periods. For 
instance, the Edisto Island, South Carolina, specimen (Roth and Laerm 1980) was likely 
deposited during an interglacial period, as sea level would have been lower and the coastline 
nearly 100 miles further east during glacial periods (De Groeve et al. 2022). These taphonomic 
considerations strongly suggest that specimens from glacial periods are underrepresented while 
interglacial occurrence records are overrepresented in our fossil dataset. Additionally, we 
extracted environmental variables associated with fossil occurrence localities from rasters based 
on LGM climatic conditions both because interglacial raster data were not available and because 
we chose to take a conservative approach to our analyses. So even though the temperatures 
associated with the fossil dataset (average SSTmin = 0.57 °C) are overall higher than those 
associated with the modern dataset (average SSTmin = -1.85 °C), they are still likely cooler than 
the actual conditions in which many of these individuals lived. Improving age constraints for our 
fossil dataset and constructing raster layers for interglacial periods would offer clarity on these 
limitations. 
4.2 Current Climate Projections 
Current suitable habitat projected by the modern ENM and fossil ENM display several 
important differences. Overall, the geographical projection of the modern model covers a 
smaller, more northerly area than the much wider range projected by the fossil model. It also 
correlates more closely with the actual geographic range inhabited by modern walruses 
compared to the paleontological ENM, which predicts suitable habitat over a much wider area 
114 
 
than walruses actually occupy, congruent with the findings of our niche space comparisons. 
Notably, it also predicts moderately suitable habitat in areas where walruses do not currently 
reside but were historically present despite our decision not to include historical records in our 
modern dataset. It identified both northwestern Iceland and the Bay of Fundy as moderately 
suitable, both of which were home to walrus populations prior to extirpation 1,000 and 250 years 
ago, respectively (McLeod et al. 2014, Keighley et al. 2019).  
However, several noteworthy inconsistencies exist between the modern ENM and actual 
current and historical walrus geographic range. Although the zooarchaeological record 
demonstrates that walruses were widely present throughout the Maritimes (McLeod et al. 2014), 
our model only identified the Bay of Fundy as moderately suitable. Similarly, although the 
greatest density of Icelandic walrus specimens identified by Keighley et al. (2019) fell in areas of 
moderate suitability according to the modern model, they also list specimens from areas our 
model identified as unsuitable. Notably, both the Icelandic and Maritimes walrus were 
genetically distinct from the modern Atlantic walrus (McLeod et al. 2014, Keighley et al. 2019) 
and the Maritimes walrus was morphologically distinct as well on the basis of their larger body 
size and more asymmetrical tusks. It is certainly possible that each of these populations were 
relicts of more warm-adapted Pleistocene walrus populations, as suggested by Boessenecker and 
Churchill (2021), and that these more heat-tolerant walruses did not survive the end-Pleistocene 
megafaunal extinction event, eliminating this variation from surviving populations. More precise 
dating of southerly fossil specimens could reveal the merit of this hypothesis. However, we 
consider it implausible that walruses have, on the whole, become more cold-adapted as average 
temperatures across their range have risen by 4–20 °C since the LGM (Osman et al. 2021). 
Instead, we find it more likely that other factors have constrained their range to its current 
115 
 
boundaries, where ice provides convenient haul-outs but is not a crucial element for their 
survival. Further investigation is needed to determine the identity and nature of these limiting 
factors. 
An additional inconsistency between the modern model and documented walrus range is 
the identification by the modern ENM of the Sea of Okhotsk as highly suitable. Although 
paleontological and zooarchaeological evidence indicates that walruses were present in this 
region during the late Pleistocene and early Holocene (Borisiak 1930; Vasilevskii 1971), an 
exhaustive analysis of historical records and archeological material by Slobodin (2010) 
determined that walruses have been absent from the Sea of Okhotsk for at least the last 2,500 
years. The fact that our modern model identifies this area as highly suitable for walruses raises 
the question of why they went extinct in this region. Investigating this quandary could lend 
further insight into the factors that determine walrus distribution. 
Although on the whole, the southern extent of the fossil model projections reaches much 
further south than walruses usually range, summer and fall sightings of vagrant walruses around 
Britain, Ireland, and the northern coast of continental Europe appear to have been increasing 
according to numerous news articles reporting walrus sightings around Europe1. This pattern has 
not been thoroughly investigated, obfuscating the degree to which it is the result of true 
ecological change or simply alterations in reporting. However, while these nomads are far from 
self-sustaining populations, the relative health of these individuals, some of whom were observed 
for several months, does provide at least anecdotal evidence that walruses can survive in ice-free 
 
1 https://www.cornwalllive.com/news/cornwall-news/wally-walrus-could-latvia-year-7323451; 
https://www.politico.eu/article/wally-walrus-ireland-cork-coast/; https://www.bbc.com/news/uk-england-hampshire-
63940472; https://www.nbcnews.com/news/world/freya-the-walrus-norway-sinking-boats-rcna40174; 
https://www.scotsman.com/arts-and-culture/arctic-walrus-rare-sighting-orkney-beach-328507; all accessed 
December 15, 2022; https://www.smithsonianmag.com/smart-news/the-unusual-european-journey-of-thor-the-
walrus-180981734/; accessed April 18, 2023 
116 
 
and much warmer conditions than the majority currently do. So perhaps the predictions of 
suitable walrus habitat in the North Sea by the fossil model are less erroneous than they would 
appear on first inspection. 
4.3 Future Climate Projections 
The two sets of ENMs display considerably disparate outcomes for the walrus over the 
next seventy years. The modern model predicts a severe decrease in suitability by 2100 for 
several regions where walruses currently live, including the Bering Sea, Svalbard, the Laptev 
Sea, the southern Kara Sea, and the Barents Sea (Fig. 4c). Conversely, according to the fossil 
model projections, no areas currently inhabited by walruses display significant decreases in 
suitability and indeed many become even more suitable (Fig. 4f). Similarly, according to our 
binary maps, the modern ENM predicts that by 2100, suitable habitat for walruses will decrease 
by 18.47% and that the average latitude of suitable habitat will shift northwards by 1.09° or ~120 
km (Table 2; Fig. 5c). However, the fossil model predicts no appreciable change either in the 
total area of suitable habitat or in the centroid latitude of suitable habitat before the end of the 
21st century (Table 2; Fig. 5f).  
Unfortunately, no species-wide demographic surveys of walruses have been undertaken 
in the last several decades, limiting our ability to assess which of these scenarios most closely 
aligns with current trends. Studies of individual walrus populations present a relatively equivocal 
picture of the current status of O. rosmarus and its response to ongoing Arctic warming 
(Fischbach et al. 2009, Jay and Fischbach 2010, Garlich-Miller et al. 2011, Jay et al. 2012, 
MacCracken 2012). Across the Arctic, female walruses and their calves have been increasingly 
utilizing terrestrial haul-outs in late summer and fall as sea ice becomes unavailable, 
demonstrating adaptive behavioral plasticity (Jay et al. 2012, Lydersen et al. 2012, Udevitz et al. 
117 
 
2017, Higdon and Stewart 2018). A study of changes in Kara Sea marine mammal distribution 
and density between 1996 and 2013 found little to no change in Atlantic walrus occurrences 
(Mäkinen and Vanhatalo 2018). Similarly, a study examining the energetic impacts of increasing 
distance between feeding grounds and haul-outs in the Chukchi Sea found that as long as 
walruses were able to rebuild their endogenous fat reserves while overwintering in the Bering 
Sea, the increased energetic costs of longer foraging trips would not likely alter walrus 
population dynamics (Jay et al. 2017). Less encouragingly, Zagrebelny and Kochnev (2017) 
reported that numerous historical terrestrial haul-outs in eastern Chukotka and Kamchatka have 
disappeared with their previous inhabitants apparently shifting to a smaller number of much 
larger haul-outs, increasing the risk of depleting benthic foraging grounds that are also being 
altered by climate change. Several mass-mortality events for walruses, particularly calves and 
juveniles, have also been reported in the last few decades with causes ranging from trampling to 
prolonged exposure to turbulent seas (Ovsyanikov et al. 2008; Fischbach et al. 2009, Goertz et 
al. 2017). Thus we consider a future intermediate to the highly optimistic fossil model and the 
direr situation presented by the modern model to be more likely than either endmember. 
Although the retreat of sea ice is generally predicted to have a negative consequence on 
walrus populations, its seasonal disappearance from the high Arctic may open up new refugia for 
the species. For instance, both the modern and fossil ENMs predict that the northern reaches of 
Greenland and the Arctic Archipelago will be highly suitable for walruses by 2100 (Fig. 4c, f). 
These areas are not currently accessible to walruses as they are ice-locked year-round. However, 
with the northward retreat and eventual loss of the Arctic ice cap, these coastlines will open up as 
potential terrestrial haul-outs. The species has already demonstrated the ability to establish 
rookeries in novel locations within both the Atlantic and Pacific subpopulations (Kovacs et al. 
118 
 
2014; Zagrebelny and Kochnev 2017). Investigating how and why walruses establish new 
terrestrial haul-outs could reveal the likelihood of walrus dispersal to such previously 
inaccessible localities. 
4.4 Future Research Directions and Walrus Conservation 
Our study primarily tests the hypothesis that walruses can survive without the availability 
of year-round sea ice. While our results offer hope for the survival of walruses in the face of 
decreasing sea ice, multiple other environmental biotic and abiotic factors influence walrus 
distribution and health, including proximity to and condition of feeding grounds (Beatty et al. 
2016), the presence and density of predators such as orcas and polar bears (Kryukova et al. 2012, 
Øren et al. 2018), frequency and intensity of storm events (Campagna et al. 2021), and 
increasing toxin exposure from harmful algae blooms (Lefebvre et al. 2022). As climate change 
will impact all of these variables, we recommend incorporating dynamic quantifications of these 
elements into future models. Furthermore, constructing ENMs from mechanistic rather than 
correlative patterns can improve their applicability by using knowledge of the critical maximum 
and minimum tolerable values for salient environmental variables (Shin et al. 2021). Research 
into the range of seawater temperatures, ambient air temperatures, and seawater salinity that 
walruses are physiologically capable of tolerating would allow for the construction of such 
mechanistic ENMs.  
We also note that our models do not incorporate directly anthropogenic elements that are 
likely to impact walrus demographics. As sea ice melts, human activity in this region will also 
change and without concerted international conservation efforts, these changes will likely have 
negative consequences on the long term likelihood of walrus survival. Nation states are already 
vying for the rights to increasingly accessible Arctic shipping lanes that pass through critical 
119 
 
walrus habitat (Theocharis et al. 2018), increasing risk of collision-related mortality and 
contamination from fuel spillage as well as damage from ice-breaking ships to pupping platforms 
(Gross 2018, Wilson et al. 2020).  Additionally, vast fossil fuel reserves have been identified in 
such crucial walrus and walrus prey habitats as the Kara, Pechora, Beaufort, and Barents Seas 
(Spencer et al. 2011). The overlap in undeveloped fossil fuel reserves and walrus range 
highlights the need to transition to a cleaner energy future not only to reduce carbon emissions 
but also to preserve the ecosystems where these resources are located. Given the status of O. 
rosmarus as an Arctic keystone species and ecosystem engineer (Ray et al. 2006, Hocking et al. 
2021), policies guiding Arctic resource development should include robust plans for protecting 
walrus haul-outs, migration corridors, and feeding grounds in order to maintain overall 
ecosystem health. 
5 Conclusion 
In essence, our models support the hypothesis that walrus fundamental niche 
encompasses the seasonally ice-free conditions that will characterize the Arctic in the coming 
decades. While the acceleration of polar warming and sea ice loss will continue to alter Arctic 
ecosystems, the results presented here demonstrate that if humans protect walruses from other 
sources of anthropogenic environmental degradation, they are likely to adapt to the next century 
of climate change. 
 
 
 
120 
 
 
CHAPTER V 
 
DISSERTATION SUMMARY 
 
 The goals of this dissertation were to a) expand our knowledge of the early evolution of 
the extinct pinniped family Desmatophocidae through the description of new basal species 
Eodesmus condoni, b) investigate the locomotor evolution of otarioid pinnipeds by exploring the 
utility of humerus morphology for diagnosing swimming mode in modern and fossil taxa, and c) 
quantify and compare the modern and paleoecological niches of the walrus Odobenus rosmarus 
in order to model its response to the next 70 years of climate change. First, I described new fossil 
cranium UOMNCH F-68583 from the Iron Mountain Bed of the Astoria Formation where it 
outcrops at Moolack Beach on the Oregon coast. After determining that the specimen represented 
a desmatophocid pinniped based on the morphology of its mortised jugal-squamosal suture, I 
identified it as a new species after comparing it to other known species of desmatophocid and 
performing a phylogenetic analysis to resolve its placement within the Desmatophocidae. New 
species Eodesmus condoni nests at the base of the desmatophocid phylogeny and is the earliest 
diverging desmatophocid yet identified. The identification of another desmatophocid species at 
the already pinniped-rich Iron Mountain Bed suggests that the approximately synchronous mid-
Miocene Climatic Optimum played a role in this increase in pinnipedimorph species diversity. 
 Next, I found that humeral morphology is discriminatory for aquatic locomotor mode and 
recovered a hindlimb-propelled as the ancestral condition for Pinnipedia as a whole as well as 
subclade Otarioidea (eared seals, walruses, and desmatophocid seals). Using three-dimensional 
geometric morphometrics, I landmarked the humeri of modern phocids, otariids, and O. rosmarus 
121 
 
as well as fossil stem pinnipeds and otarioids. Following data collection, I performed a 
Procrustes superimposition and principal components analysis (PCA) to explore the between-
group variation; I then used the principal components to construct two flexible discriminant 
analyses (FDAs), one with phylogenetic correction and one without, to classify the fossil taxa 
according to their predicted aquatic locomotor mode. As the literature suggested, the humeral 
structures that most distinguished between the hindlimb- and forelimb-propelled swimming were 
the deltoid crest, greater tubercle, lesser tubercle, and lateral and medial epicondyles, as 
identified by the PCA. The FDAs classified all the fossil pinnipeds as hindlimb-propelled with 
the exception of desmatophocid Allodesmus kernensis and odobenid Pontolis barroni, and otariid 
Thalassoleon mexicanus in the nonphylogenetic FDA, strongly suggesting that the otarioid 
ancestor was hindlimb-propelled like the stem pinnipeds. The presence of a proximodistally 
elongated deltoid crest in hindlimb-propelled species like O. rosmarus as well other earlier 
otarioids raises the possibility that this trait first arose as an adaptation for the peculiar lunging 
gait modern otariids and odobenids use on land before its exaptation for forelimb-propelled 
swimming in these three lineages, implying that terrestrial and aquatic locomotor modes are 
evolutionarily coupled in otarioids.  
 Finally, I found that modern walrus populations indeed inhabit only a portion of their past 
ecospace, including the ice-free conditions likely to characterize the Arctic in the upcoming 
decades. The permutational multivariate analysis of variance (PERMANOVA) supported a 
significant difference between the ecospace inhabited by modern and fossil walrus populations 
and the PCA showed that during the Pleistocene, walruses inhabited a wider range of 
environmental conditions with a higher temperature maximum than they do today. Furthermore, 
while the future projections derived from the modern ecological niche model predict an 18% 
122 
 
decrease in suitable walrus habitat by 2100, the fossil model projections predict no appreciable 
change in range over the next century. Although we consider an intermediate scenario more 
realistic than either endmember, our results nonetheless present encouraging evidence that with 
the enactment of conservation policies addressing other sources of anthropogenic environmental 
degradation, walruses may survive and adapt to a seasonally ice-free Arctic climate. These 
findings contrast the widespread narrative of probable walrus population collapse with the loss of 
summer sea ice, inviting further investigation into how conservation strategies can enhance O. 
rosmarus survivorship in an increasingly ice-free Arctic Ocean. 
 In conclusion, this dissertation advances our knowledge of early desmatophocid 
diversification, the locomotor evolution of otarioid pinnipeds, and the fundamental ecological 
niche of O. rosmarus with respect to predicting their response to climate change in the Arctic. 
This provides a framework for further exploration of how pinniped fauna have evolved with 
respect to taxonomic and locomotor ecology as well as the range of environmental parameters 
within which pinnipeds can survive. As human exploitation continues to deteriorate global and 
marine and coastal ecosystems, understanding the paleoecology of disproportionately threatened 
tertiary consumers like pinnipeds will provide a critical baseline for the establishment of 
evidence-based conservation policies.  
123 
 
APPENDIX A 
 
DESCRIPTION OF CHARACTERS USED IN THE 
PHYLOGENETIC ANALYSIS OF CHAPTER II 
 
 
 
1.  Rostrum, premaxilla-nasal contact. Boessenecker and Churchill (2018: character 
1) Deméré and Berta (2002: character 1), Kohno (1996: character 2). 
0 = extensive along lateral border of nasal; ~50% or greater of total nasal length.  
1 = reduced along lateral border of nasal; ~25%-50% of total nasal length 
2 = reduced; does not contact lateral border of nasal and only contacts anterior 
margin 
 
2.  Rostrum, nasal-frontal suture. Boessenecker and Churchill (2018: character 2), 
Deméré and Berta (2002: character 2), Kohno (1996: character 3, 6). 
0 = transverse 
1 = v-shaped 
2 = w-shaped 
 
3.  Rostrum, prenarial shelf. Boessenecker and Churchill (2018: character 3), Kohno 
(1996: character 1). 
0 = absent 
1 = present 
2 = present and transversely expanded anteriorly 
 
4. Rostrum, alveolar margin of maxilla in lateral view. Boessenecker and Churchill 
(2018: character 4). 
0 = straight or slightly concave 
1 = distinctly concave leading to posteroventrally inclined maxillary root of 
zygomatic arch 
2 = strongly concave, but leading to no change in the orientation of the maxillary 
root of the zygomatic arch 
 
5. Nasals, anterior margin. Boessenecker and Churchill (2018: character 5). 
0 = at level of canine or P 
1 = well posterior 
 
6. Premaxilla, prenarial process. Boessenecker and Churchill (2018: character 6), 
Berta and Wyss (1994: character 3), Kohno et al. (2007: character 1), 
Boessenecker and Churchill (2013: character 1). 
0 = absent or indistinct 
1 = present 
124 
 
7. Premaxilla, posterior termination. Boessenecker and Churchill (2018: character 
7). 
0 = well anterior to level of infraorbital foramen 
1 = extends posteriorly to level of infraorbital foramen 
 
8. Pterygoid process of the maxilla. Boessenecker and Churchill (2018: character 8), 
Berta and Wyss (1994: character 6), Deméré and Berta (2002: character 3), 
Kohno (1996: character 16). 
0 = does not form anteromedial floor to orbit 
1 = posterolaterally expanded as corner-like shelf below orbit 
 
9. Infraorbital foramen. Boessenecker and Churchill (2018: character 9), Berta and 
Wyss (1994: character 11), Deméré and Berta (2002: character 17). 
0 = large and broad, visible in anterior view 
1 = narrow, not visible in anterior view  
 
10. Antorbital process. Boessenecker and Churchill (2018: character 10), Deméré 
(1994: character 6), Kohno (1996: character 7). 
0 = present 
1 = absent 
2 = prominent 
 
11. Maxilla, nasolabialis fossa. Boessenecker and Churchill (2018: character 11), 
Berta and Wyss (1994: character 7), Boessenecker and Churchill (2015: 12). 
0 = present 
1 = absent 
 
12. Palate arching. Boessenecker and Churchill (2018: character 12), Deméré and 
Berta (2002: character 4), Kohno (1996: character 9). 
0 = relatively flat (palatal arch approximately 0.18-0.19) 
1 = transversely arched (palatal arch greater than 0.25) 
 
13. Palate: Incisive foramina. Boessenecker and Churchill (2018: character 13), 
Deméré and Berta (2002: character 7), Kohno (1996: character 4). 
0 = divided by medial septum 
1 = medial septum absent 
2 = foramina reduced  
 
14.  Palate, toothrows. Boessenecker and Churchill (2018: character 14), Berta and 
Wyss (1994: character 13), Kohno (1996: character 8). 
0 = slightly divergent 
1 = strongly divergent 
2 = parallel 
 
125 
 
15. Supraorbital process of frontals. Boessenecker and Churchill (2018: character 15), 
Berta and Wyss (1994: character 17), Deméré and Berta (2002: character 8), 
Kohno (1996: characters 11, 12). 
0 = low but distinct ridge 
1 = polymorphic for states 0 and 2  
2 = absent or indistinct  
3 = small and posteriorly positioned along intertemporal bar 
4 = strongly developed rectangular or triangular supraorbital shelf  
 
16. Orbit size. Boessenecker and Churchill (2018: character 16). 
0 = Large (18-21% BL) 
1 = Very Large (+21% BL) 
2 = Small (< 18%) 
 
17. Orbit, orbital vacuity. Boessenecker and Churchill (2018: character 17), Berta and 
Wyss (1994: character 12), Kohno (1996, character 13). 
0 = absent 
1 = present 
 
18. Orbit, optic foramen. Boessenecker and Churchill (2018: character 18), Kohno 
(1996: character 17). 
0 = located anterodorsally 
1 = positioned posteroventrally anterior to braincase 
 
19. Orbit, optic foramina. Boessenecker and Churchill (2018: character 19), Kohno 
(1996: character 19). 
0 = bilateral and separated 
1 = merged at sagittal plane 
 
20. Lateral wall of alisphenoid canal. Boessenecker and Churchill (2018: character 
20), Berta and Wyss (1994: character 20); Furbish (2015: character 22). 
0 = thick and well-developed  
1 = thin  
2 = absent 
 
21. Zygomatic arch: dorsal margin of maxillary root. Boessenecker and Churchill 
(2018: character 21), Deméré and Berta (2002: character 6), Kohno (1996: 
character 10). 
0 = not retracted dorsal to infraorbital foramen 
1 = retracted dorsal to infraorbital foramen 
 
22. Zygomatic arch: squamosal-jugal articulation. Boessenecker and Churchill (2018: 
character 22), Berta and Wyss (1994: character 39), Deméré and Berta (2002: 
character 11), Kohno (1996: character 15). 
0 = splint like 
126 
 
1 = mortised, zygomatic process of squamosal exhibits minimal dorsoventral 
expansion 
2 = mortised, zygomatic process of squamosal greatly expanded dorsoventrally 
 
23. Zygomatic arch: jugal-maxillary suture. Boessenecker and Churchill (2018: 
character 23), Deméré and Berta (2002: character 12), Kohno (1996: character 
14). 
0 = jugal with anterodorsal and anteroventral splints 
1 = polymorphic for states 0 and 2 
2 = jugal with anterodorsal splint only 
3 = elongate anteroventral splint extends anteriorly to level of M1 
 
24. Zygomatic arch, ventral tuberosity of anterior zygomatic root. Boessenecker and 
Churchill (2018: character 24), Kohno et al. (2007: character 7). 
0 = absent 
1 = present 
 
25. Zygomatic arch, postorbital process. Boessenecker and Churchill (2018: character 
25). 
0 = well developed and triangular 
1 = indistinct, dorsal surface smooth and flush with zygomatic process 
 
26. Zygomatic arch, transverse width. Boessenecker and Churchill (2018: character 
26). 
0 = widest point at approximate level of anterior border of the glenoid fossa 
1 = anterior to glenoid fossa  
 
27. Braincase, pseudosylvian sulcus. Boessenecker and Churchill (2018: character 
27), Kohno (1996, character 25). 
0 = visible externally on braincase 
1 = absent, lateral wall of braincase smoothly convex  
 
28. Braincase, anterolateral margin in dorsal view. Boessenecker and Churchill 
(2018: character 28), Boessenecker and Churchill (2015: character 58). 
0 = forms a corner 
1 = smoothly convex 
 
29. Braincase, squamosal fossa. Boessenecker and Churchill (2018: character 29), 
Deméré and Berta (2002: character 18). 
0 = undivided  
1 = divided, posterior and anterior portions of approximately equal in size 
2 = divided, posterior portion smaller than anterior 
 
30. Basicranium, paroccipital process. Boessenecker and Churchill (2018: character 
30), Berta and Wyss (1994: character 45). Deméré and Berta (2002: character 9), 
Kohno (1996: character 29). 
127 
 
0 = small and blunt 
1 = large and excavated  
2 = large and not excavated  
3 = small and crest-like 
 
31. Basicranium, tympanic bulla. Boessenecker and Churchill (2018: character 31), 
Deméré and Berta (2002: character 10). 
0 = only ectotympanic inflated  
1 = ectotympanic flat with minimal inflation 
2 = entotympanic and ectotympanic greatly inflated and bulbous 
 
32. Styloid process of tympanic bulla. Boessenecker and Churchill (2018: character 
32), Berta and Wyss (1994: character 41), Deméré and Berta (2002: character 13). 
0 = present 
1 = absent, anterior corner of entotympanic retracted posteriorly 
2 =absent, ectotympanic reduced 
 
33.  Stylomastoid foramen. Boessenecker and Churchill (2018: character 33), Deméré 
and Berta (2002: character 14). 
0 = in common fossa with tympanohyal pit 
1 = separated from tympanohyal by a raised strut or ridge  
 
34.  Basicranium, pterygoid strut. Boessenecker and Churchill (2018: character 34), 
Deméré (1994: character 13), Kohno (1996: character 18).  
0= slender 
1= dorsoventrally and laterally projected 
2= dorsoventrally thick and transversely broad 
3= transversely thin and "rolled" 
 
35. Basicranium, shape of basioccipital. Boessenecker and Churchill (2018: character 
35), Berta and Wyss (1994: character 42), Kohno (1996, character 26). 
0 = parallel sided 
1 = pentagonal in shape, gradually but noticeably narrows anteriorly 
 
36. Basicranium, posterior lacerate foramen. Boessenecker and Churchill (2018: 
character 36), Berta and Wyss (1994: character 43), Kohno (1996, character 28). 
0 = anteroposteriorly expanded 
1 = transversely expanded 
2 = fissure 
 
37. Mandible, angular process. Boessenecker and Churchill (2018: character 37), 
Berta and Wyss (1994: character 21). 
0 = well developed  
1 = polymorphic for states 0 and 2 
2 = reduced or absent  
128 
 
38. Mandible, digastric insertion. Boessenecker and Churchill (2018: character 38), 
Berta and Wyss (1994: character 52), Deméré (1994: character 44), Deméré and 
Berta (2002: character 22), Kohno (1996: character 30). 
0 = small and indistinct 
1 = well developed  
 
39. Mandible: depth of horizontal ramus. Boessenecker and Churchill (2018: 
character 39), Boessenecker and Churchill (2013: 43). 
0 = deepest part of horizontal ramus posterior to mandibular symphysis 
1 = deepest part of horizontal ramus at posteroventral terminus of symphysis 
 
40. Mandible: mandibular condyle. Boessenecker and Churchill (2018: character 40), 
Berta and Wyss (1994: character 53). 
0 = at or slightly above level of tooth row 
1 = elevated above tooth row 
 
41. Mandible, genial tuberosity. Boessenecker and Churchill (2018: character 41), 
Boessenecker and Churchill (2013: character 37). 
0 = absent or indistinct 
1 = present; developed as small tubercle or process on anterior portion of ramus 
2 = present and well developed; extends well below ventral margin of ramus 
 
42. Symphyseal angle of mandible (angle between longest axis of symphysis and 
alveolar margin of mandible in medial view). Boessenecker and Churchill (2018: 
character 43). 
0 = >50°  
1 = 40-50°  
2 = <40° 
 
43. Mandible: length of lower postcanine tooth row length. Boessenecker and 
Churchill (2018: character 43), Berta and Wyss (1994: character 69), 
Boessenecker and Churchill (2013: character 65). 
0 = long (length of tooth row greater than 40% of the mandible length).  
1 = short (length of tooth row less than or equal to 40% of the mandible length). 
 
44. Dentition, I3 orientation. Boessenecker and Churchill (2018: character 44), 
Deméré and Berta (2002: character 23), Kohno (1996: character 31). 
0 = vertical or subvertical 
1 = procumbent and laterally directed 
 
45. I3 size and shape. Boessenecker and Churchill (2018: character 45). 
0 = same size or only slightly larger than I1-2  
1 = greatly larger than I1-2 and canine-like 
2 = similar in morphology to premolars 
 
129 
 
46. C1, carinae. Boessenecker and Churchill (2018: character 46), Deméré and Berta 
(2002: character 24), Kohno (1996: character 32). 
0 = posterior and medial carinae present  
1 = posterior and medial carinae absent  
 
47. Upper postcanine tooth roots orientation. Boessenecker and Churchill (2018: 
character 47), Deméré and Berta (2002: character 28) 
0 = vertical  
1 = anteriorly inclined 
 
48. P1-3 crowns, cingulum. Boessenecker and Churchill (2018: character 48), Berta 
and Wyss (1994: character 67), Boessenecker and Churchill (2013: character 71; 
2015: character 82). 
0 = narrow, smooth lingual cingulum 
1 = well-developed cuspate cingulum 
2 = no cingulum 
 
49. P2-P3 rooting. Boessenecker and Churchill (2018: character 49), Berta and Wyss 
(1994: character 62), Boessenecker and Churchill (2013: character 74; 2015: 
character 82) 
0 = double 
1 = polymorphic for states 0 and 2 
2 = single 
 
50. P4 rooting. Boessenecker and Churchill (2018: character 50), Berta and Wyss 
(1994: character 64), Boessenecker and Churchill (2013: character 76; 2015: 
character 82) 
0 = Triple 
1 = polymorphic for states 0 and 2 
2 = Double 
3 = Single 
 
51. P4, metacone. Boessenecker and Churchill (2018: character 51), Deméré and 
Berta (2001: character 16). 
0 = distinct 
1 = reduced or absent 
 
52. P4, protocone shelf. Boessenecker and Churchill (2018: character 52), Berta and 
Wyss (1994: character 63), Boessenecker and Churchill (2013: character 75), 
Deméré (1994: character 36), Kohno et al. (2007: character 47). 
0 = anteromedially placed 
1 = posteromedially placed with small cuspules 
2 = reduced or absent 
 
53. Dentition, size of M1 relative to premolars. Boessenecker and Churchill (2018: 
character 53), Berta and Wyss (1994: character 66), Kohno (1996: character 40). 
130 
 
0 = equal in size 
1 = reduced in size 
 
54. M1 rooting. Boessenecker and Churchill (2018: character 54), Berta and Wyss 
(1994: character 65), Boessenecker and Churchill (2013: character 77; 2015: 
character 82). 
0 = Triple 
1 = polymorphic for state 0 and 2 
2 = Double  
3 = polymorphic for state 2 and 4 
4 = Single 
 
55. M2. Boessenecker and Churchill (2018: character 55), Berta and Wyss (1994: 
character 68), Kohno (1996: character 39). 
0 = present  
1 = polymorphic for states 0 and 2 
2 = absent 
 
56. M1, metacone. Boessenecker and Churchill (2018: character 56). 
0 = distinct, prominent cusp 
1 = absent or indistinct  
 
57. M1, protocone. Boessenecker and Churchill (2018: character 57). 
0 = distinct, prominent and posteromedially positioned, resulting in a triangular-
shaped M1 
1 = indistinct or absent, resulting in a largely round or square M1 
 
58. Dentition, lower cheek teeth, size. Boessenecker and Churchill (2018: character 
58), Berta and Wyss (1994: character 66), Kohno (1996: character 40). 
0 = p4-m1 larger than p2-3  
1 = p2-m1 all approximately equal size  
2 = m1 distinctly smaller than p2-4  
 
59. Dentition, orientation of lower cheek teeth. Boessenecker and Churchill (2018: 
character 59). 
0 = teeth parallel to adjacent cheek teeth, oriented dorsoventrally   
1 = cheektooth orientation is divergent, with p1 directed anterodorsally and 
posterior teeth posterodorsally 
 
60. Lower postcanines, crowns. Boessenecker and Churchill (2018: character 60), 
Berta and Wyss (1994: character 75), Boessenecker and Churchill (2013, 
character 63; 2015: character 74). 
0 = transversely narrow 
1 = round and bulbous 
 
131 
 
61. Lower postcanines, cingulum. Boessenecker and Churchill (2018: character 61), 
Boessenecker and Churchill (2013: character 67). 
0 = present lingually and labially  
1 = present lingually only  
2 = absent 
 
62. p2-4 rooting. Boessenecker and Churchill (2018: character 62), Boessenecker and 
Churchill (2013: 72, 73; 2015: character 92). 
0 = double rooted 
1 = polymorphic for states 0 and 2 
2 = single rooted 
 
63. p2-4 crowns, paraconid and hypoconid cusps. Boessenecker and Churchill (2018: 
character 63), Berta and Wyss (1994: character 70); Boessenecker and Churchill 
(2013: character 66, 68; 2015: 93). 
0 = trenchant  
1 = conical  
2 = absent or reduced 
 
64. m1 rooting. Boessenecker and Churchill (2018: character 64), Boessenecker and 
Churchill (2013: 79; 2015: 92). 
0 = double 
1 = polymorphic for states 0 and 2 
2 = single 
 
65. m1, talonid basin. Boessenecker and Churchill (2018: character 65), Berta and 
Wyss (1994: character 71), Boessenecker and Churchill (2013: 69). 
0 = present as slight concavity or small shelf 
1 = absent 
 
66. m1, metaconid. Boessenecker and Churchill (2018: character 66), Berta and Wyss 
(1994: character 72), Boessenecker and Churchill (2013: 70). 
0 = present 
1 = reduced 
2 = absent 
 
67. m2. Boessenecker and Churchill (2018: character 67), Berta (1995: character 31), 
Berta (1994: character 19), Berta and Wyss (1994: character 85), Churchill et al. 
(2014: character 83), Boessenecker and Churchill (2015: character 98) 
0 = present  
1 = polymorphic for states 0 and 2 
2 = absent 
 
68. Manubrium, proportions. Boessenecker and Churchill (2018: character 68). 
0 = anterior and posterior halves similar in transverse diameter  
1 = anterior portion transversely wider 
132 
 
69. Scapula, supraspinous fossa. Boessenecker and Churchill (2018: character 69), 
Berta and Wyss (1994: character 85). 
0 = slightly larger than infraspinous fossa. 
1 = much larger than infraspinous fossa.  
 
70. Scapula, acromion process. Boessenecker and Churchill (2018: character 70), 
Berta and Wyss (1994: character 83). 
0 = knoblike 
1 = reduced 
 
71. Scapula, hook-like process for teres major. Boessenecker and Churchill (2018: 
character 71), Berta and Wyss (1994: character 82). 
0 = present 
1 = absent 
 
72. Scapula, secondary spine. Boessenecker and Churchill (2018: character 72), Berta 
and Wyss (1994: character 86), Boessenecker and Churchill 2015: character 103). 
0 = absent 
1 = absent, but a scapular undulation is present  
2 = present  
 
73. Humerus, greater tuberosity. Boessenecker and Churchill (2018: character 73). 
0 = at level of humeral head 
1 = extends far proximal to humeral head and lesser tuberosity 
 
74. Humerus, length of deltopectoral crest. Boessenecker and Churchill (2018: 
character 74), Berta and Wyss (1994: character 88). 
0 = less than or equal to 1/2 humerus length 
1 = 1/2 to 3/4 humerus length  
 
75. Humerus, distal termination of deltopectoral crest. Boessenecker and Churchill 
(2018: character 75), Berta and Wyss (1994: character 88). 
0 = gradual 
1 = abrupt 
 
76. Humerus, supinator ridge. Boessenecker and Churchill (2018: character 76), Berta 
and Wyss (1994: character 89). 
0 = well developed 
1 = absent or reduced 
 
77. Humerus, diameter of humeral trochlea. Boessenecker and Churchill (2018: 
character 77), Berta and Wyss (1994: character 93), Deméré (1994: character 47). 
0 = medial and lateral edges equal  
1 = medial edge much larger 
 
133 
 
78. Humerus, deltoid tubercle. Boessenecker and Churchill (2018: character 78), 
Deméré (1994: character 45). 
0 = on deltopectoral crest 
1 = separated from deltopectoral crest 
 
79. Humerus, posterior margin in lateral view. Boessenecker and Churchill (2018: 
character 79). 
0 = straight to concave  
1 = strongly sinuous owing to anteriorly prominent supinator ridge  
 
80. Ulna, olecranon process. Boessenecker and Churchill (2018: character 80), Berta 
and Wyss (1994: character 94). 
0 = knoblike and unexpanded 
1 = transversely flattened and posteriorly expanded 
 
81. Radius, shape in lateral view. Boessenecker and Churchill (2018: character 81), 
Berta and Wyss (1994: character 95). 
0 = convexly arched and distal end similar in anteroposterior width to proximal 
end  
1 = distal end expanded and approximately twice as wide as proximal end 
 
82. Radius, distal end. Boessenecker and Churchill (2018: character 82), Deméré 
(1994: character 48). 
0 = unexpanded 
1 = expanded, with small radial process 
2 = expanded, with large radial process 
 
83. Radius, pronator teres process. Boessenecker and Churchill (2018: character 83), 
Berta and Wyss (1994: character 96). 
0 = positioned at midpoint or on distal 1/2 of shaft 
1 = positioned on proximal 1/2 of shaft 
 
84. Scapholunar, pit for magnum. Boessenecker and Churchill (2018: character 84), 
Deméré (1994: character, 50). 
0 = absent 
1 = present 
 
85. Metacarpal 1, pit or rugosity. Boessenecker and Churchill (2018: character 85), 
Berta and Wyss (1994: character 99). 
0 = absent 
1 = present 
 
86. Ungual phalanges. Boessenecker and Churchill (2018: character 86), Berta and 
Wyss (1994: character 102). 
0=small claw core present 
1=distal end blocky and rugose for cartilaginous extension of digit 
134 
 
87. Innominate, foramen for obturator nerve. Boessenecker and Churchill (2018: 
character 87), Berta and Wyss (1994: character 113). 
0 = obturator foramen absent  
1 = obturator foramen present  
 
88. Innominate, ischial spine. Boessenecker and Churchill (2018: character 88), Berta 
and Wyss (1994: character 114). 
0 = small 
1 = large and triangular 
 
89. Femur, fovea for teres femoris ligament. Boessenecker and Churchill (2018: 
character 89), Berta and Wyss (1994: character 115). 
0 = present  
1 = absent 
 
90. Femur, lesser trochanter. Boessenecker and Churchill (2018: character 90), Berta 
and Wyss (1994: character 116). 
0 = present 
1 = absent 
 
91. Femur, greater trochanter. Boessenecker and Churchill (2018: character 91), Berta 
and Wyss (1994: character 117). 
0 = small and rounded 
1 = large and anteroposteriorly flattened 
 
92. Astragalus, calcaneal process. Boessenecker and Churchill (2018: character 92), 
Berta and Wyss (1994: character 126). 
0 = absent 
1 = present 
2 = elongated 
 
93. Calcaneum, calcaneal tuber. Boessenecker and Churchill (2018: character 93), 
Berta and Wyss (1994: character 125). 
0 = straight  
1 = medially prominent 
 
94.  Calcaneum, secondary sustentacular shelf. Boessenecker and Churchill (2018: 
character 94), Berta and Wyss (1994: character 123). 
0 = absent 
1 = present 
 
95. Tarsals, entocuneiform/mesocuneiform articulation. Boessenecker and Churchill 
(2018: character 95), Deméré (1994: character 53).  
0 = abutting 
1 = overlapping 
 
135 
 
96. Anterior narial opening. Tonomori et al. (2018: character 3). 
0 = large, thin margin, and rounded 
1= large, thick margin, and dorsoventrally elliptical 
 
97. Epitympanic recess. Tonomori et al. (2018: character 28). 
0= small 
1= large 
2= very large 
 
98. Jugal, postorbital process. Tonomori et al. (2018: character 22). 
0 = small 
1 = dorsally projecting 
136 
 
 
APPENDIX B 
 
NEXUS FORMATTED MORPHOLOGICAL CHARACTER MATRIX FOR 
CHAPTER II 
 
#NEXUS 
[written Thu Dec 19 17:47:40 PST 2019 by Mesquite  version 3.51 (build 898) at LAPTOP-
2ALJS32L/10.0.0.48] 
 
BEGIN TAXA; 
 TITLE Taxa; 
 DIMENSIONS NTAX=26; 
 TAXLABELS 
  Enaliarctos_spp. Pteronarctos_goedertae Pinnarctidion_bishopi 
Pinnarctidion_rayi Desmatophoca_oregonensis Desmatophoca_brachycephala 
Allodesmus_kernensis Atopotarus_courseni Allodesmus_packardi Allodesmus_naorai 
Allodesmus_demerei Allodesmus_uraiporensis Allodesmus_sinanoensis 
Thalassoleon_mexicanus Callorhinus_ursinus Zalophus_californianus 
Proneotherium_repenningi Neotherium_mirum Imagotaria_spp. Gomphotaria_pugnax 
Odobenus_rosmarus Devinophoca_claytoni Monachus_monachus Erignathus_barbatus 
USNM_335445 UOMNCH_F_68583  
 ; 
 
END; 
 
 
BEGIN CHARACTERS; 
 TITLE  Character_Matrix; 
 DIMENSIONS  NCHAR=98; 
 FORMAT DATATYPE = STANDARD GAP = - MISSING = ? SYMBOLS = "  0 1 2 3 
4"; 
 MATRIX 
 Enaliarctos_spp.            
0000000000000000000000000000000000000000000000000000020000000000000?1000000000
00000?01000000000000 
 Pteronarctos_goedertae      
0000011000000000000000000000010000000?00011000000201020000000000000?10100??????
????????????????000 
 Pinnarctidion_bishopi       
???0???10000?02?01000000010001000101???????0??00000?02000?????????????????????????
??????????????10 
137 
 
 Pinnarctidion_rayi          
0000010100000001010000000100210?010101102?1000000201020001001010100?1000010000
1?????????001????010 
 Desmatophoca_oregonensis    
1100000100101001010001300000221011001111211110010211131011011010100?10100101?00
1???????????????021 
 Desmatophoca_brachycephala  
??00??010010000101000131010012101100???????1?00?02???4001?????????????????????????
??????????????21 
 Allodesmus_kernensis        
11211011101120311111123010112211111121102211111223121411121122211101101101110001
111001101110101121 
 Atopotarus_courseni         
??0?01????1??0???????2?000??2??0????2110211111????1??????11120?0?121101101110001111
?01?????????1?1 
 Allodesmus_packardi         
11101?010110?13?1100123011002?111101?????11??11?23??040?1??????????????????????????
????????????121 
 Allodesmus_naorai           
1110101?1110?13?1???1?00??00?????????????????10223??04??????????????????????????????
???????????1?? 
 Allodesmus_demerei          
1121101?111???31????123001?1221111112110121?1?1?03??040???????????01????0111000???
?????????????121 
 Allodesmus_uraiporensis     
111?1?1111112001????123?01?????????????????1??1?23???40?????????????1??11???????????
??1010101011?1 
 Allodesmus_sinanoensis      
1?2?101??01120????????3????1221?????????21??111?23???4?????1?2???1??101101110001111
??100?1?010?1?? 
 Thalassoleon_mexicanus      
0200010002100240101000000010011000000010211010000212020111001020112?1012110100
111110011010100000?0 
 Callorhinus_ursinus         
020001000210024110100000001001100000001020101000231203111100121211201012110100
11111001101010010000 
 Zalophus_californianus      
0200010002100241101000000010011000000010211010012312041111001212112?1012110100
11111001101010010000 
 Proneotherium_repenningi    
0000010100100000100000000000011002100010011010000101010001001000100????????????
?????????0?111010?0 
 Neotherium_mirum            
0000?10100100000?00?00000011011002100010011010000201010??1001020110??????101100
?100???11??1110?010 
138 
 
 Imagotaria_spp.             
0000010002100200010000000011011002100110201010011211020?11011110111?10?01101100
112011???1011101110 
 Gomphotaria_pugnax          
01000100021012001100000000110??0?21001100?1111022312042?11012222112?????1101100
112011??11011101111 
 Odobenus_rosmarus           
20000101021102021100000001110110021000101?1?21022?12??2???01222?112010101101110
1120111111011101111 
 Devinophoca_claytoni        
01000?000000012?????00?0?1110322??0220110?10??00020200200?000000?????????????????
??????????????021 
 Monachus_monachus           
110000000210012111020220010013220310211111100001021212211100000010200110001100
11100000011112000021 
 Erignathus_barbatus         
11020000001001111102021?00000322030201011110000002120220110000?01?2?01000011001
1100000011112000021 
 USNM_335445                 
211000010000000000?0020000?0121?11100101210011010212??2??1011020100?????????????
???????????????0?1 
 UOMNCH_F_68583              
2110000101000020?0000200001121111110??????????010212140???????????????????????????
?????????????0?1 
 
; 
 
END; 
ctype ord : 15 23 37 49 50 54 55 62 64 67; 
 
END; 
BEGIN ASSUMPTIONS; 
 TYPESET * UNTITLED   =  unord:  1- 98; 
 
END; 
 
BEGIN MESQUITECHARMODELS; 
 ProbModelSet * UNTITLED   =  'Mk1 (est.)':  1- 98; 
END; 
 
Begin MESQUITE; 
  MESQUITESCRIPTVERSION 2; 
  TITLE AUTO; 
  tell ProjectCoordinator; 
  timeSaved 1576806460323; 
  getEmployee #mesquite.minimal.ManageTaxa.ManageTaxa; 
139 
 
  tell It; 
   setID 0 421770968555457424; 
  endTell; 
  getEmployee #mesquite.charMatrices.ManageCharacters.ManageCharacters; 
  tell It; 
   setID 0 3333749591159525760; 
   mqVersion 351; 
   checksumv 0 3 2067079459 null  getNumChars 98 numChars 98 
getNumTaxa 26 numTaxa 26   short true   bits 31   states 31   sumSquaresStatesOnly 11630.0 
sumSquares 11630.0 longCompressibleToShort false usingShortMatrix true   NumFiles 1 
NumMatrices 1; 
   mqVersion; 
  endTell; 
  getWindow; 
  tell It; 
   suppress; 
   setResourcesState false false 141; 
   setPopoutState 300; 
   setExplanationSize 0; 
   setAnnotationSize 0; 
   setFontIncAnnot 0; 
   setFontIncExp 0; 
   setSize 2091 1174; 
   setLocation 1510 150; 
   setFont SanSerif; 
   setFontSize 10; 
   getToolPalette; 
   tell It; 
   endTell; 
   desuppress; 
  endTell; 
  getEmployee  #mesquite.trees.BasicTreeWindowCoord.BasicTreeWindowCoord; 
  tell It; 
   makeTreeWindow #421770968555457424  
#mesquite.trees.BasicTreeWindowMaker.BasicTreeWindowMaker; 
   tell It; 
    suppressEPCResponse; 
    setTreeSourceEditMode #mesquite.trees.StoredTrees.StoredTrees; 
    tell It; 
     laxMode; 
     setTreeBlock -2147483647; 
     toggleUseWeights off; 
    endTell; 
    setAssignedID 985.1576714518141.3785034216871739372; 
    getTreeWindow; 
    tell It; 
140 
 
     setExplanationSize 30; 
     setAnnotationSize 20; 
     setFontIncAnnot 0; 
     setFontIncExp 0; 
     setSize 1950 1102; 
     setLocation 1510 150; 
     setFont SanSerif; 
     setFontSize 10; 
     getToolPalette; 
     tell It; 
      setTool 
mesquite.trees.BasicTreeWindowMaker.BasicTreeWindow.interchange; 
     endTell; 
     getTreeDrawCoordinator 
#mesquite.trees.BasicTreeDrawCoordinator.BasicTreeDrawCoordinator; 
    tell It; 
     suppress; 
     setTreeDrawer  
#mesquite.trees.SquareLineTree.SquareLineTree; 
     tell It; 
      setNodeLocs 
#mesquite.trees.NodeLocsStandard.NodeLocsStandard; 
      tell It; 
       branchLengthsToggle off; 
       toggleScale on; 
       toggleBroadScale off; 
       toggleCenter on; 
       toggleEven on; 
       setFixedTaxonDistance 0; 
      endTell; 
      setEdgeWidth 4; 
      showEdgeLines on; 
      orientRight; 
     endTell; 
     setBackground White; 
     setBranchColor Black; 
     showNodeNumbers off; 
     showBranchColors on; 
     labelBranchLengths off; 
     centerBrLenLabels on; 
     showBrLensUnspecified on; 
     showBrLenLabelsOnTerminals on; 
     setBrLenLabelColor 0 0 255; 
     setNumBrLenDecimals 6; 
     desuppress; 
141 
 
     getEmployee 
#mesquite.trees.BasicDrawTaxonNames.BasicDrawTaxonNames; 
     tell It; 
      setFontSize 24; 
      setColor Black; 
      toggleColorPartition off; 
      toggleColorAssigned on; 
      toggleShadePartition off; 
      toggleShowFootnotes on; 
      toggleNodeLabels on; 
      toggleCenterNodeNames off; 
      toggleShowNames on; 
      namesAngle ?; 
     endTell; 
    endTell; 
     setTreeNumber 1; 
     setTree 
'(((((((25,26),((5,6),((7,(8,9)),((10,11),(12,13))))),(24,(23,22))),(14,(15,16)),((20,21),19),(18,17)),
(3,4)),2),1);'; 
     setDrawingSizeMode 0; 
     toggleLegendFloat on; 
     scale 0; 
     toggleTextOnTree off; 
     togglePrintName off; 
     showWindow; 
    endTell; 
    setEditMode 'Edited, based on Untitled Tree   [Stored Trees]'; 
    desuppressEPCResponseNORESET; 
    getEmployee #mesquite.trees.ColorBranches.ColorBranches; 
    tell It; 
     setColor Red; 
     removeColor off; 
    endTell; 
    getEmployee #mesquite.ornamental.BranchNotes.BranchNotes; 
    tell It; 
     setAlwaysOn off; 
    endTell; 
    getEmployee 
#mesquite.ornamental.ColorTreeByPartition.ColorTreeByPartition; 
    tell It; 
     colorByPartition off; 
    endTell; 
    getEmployee 
#mesquite.ornamental.DrawTreeAssocDoubles.DrawTreeAssocDoubles; 
    tell It; 
     setOn on; 
142 
 
     toggleShow consensusFrequency; 
     toggleShow posteriorProbability; 
     toggleShow bootstrapFrequency; 
     toggleShow consensusFrequency; 
     toggleShow posteriorProbability; 
     toggleShow bootstrapFrequency; 
     setDigits 4; 
     setThreshold ?; 
     writeAsPercentage off; 
     toggleCentred off; 
     toggleHorizontal on; 
     toggleWhiteEdges on; 
     toggleShowOnTerminals on; 
     setFontSize 10; 
     setOffset 0  0; 
    endTell; 
    getEmployee 
#mesquite.ornamental.DrawTreeAssocStrings.DrawTreeAssocStrings; 
    tell It; 
     setOn on; 
     toggleCentred on; 
     toggleHorizontal on; 
     setFontSize 10; 
     setOffset 0  0; 
     toggleShowOnTerminals on; 
    endTell; 
    getEmployee #mesquite.trees.TreeInfoValues.TreeInfoValues; 
    tell It; 
     panelOpen false; 
    endTell; 
   endTell; 
  endTell; 
  getEmployee  
#mesquite.charMatrices.BasicDataWindowCoord.BasicDataWindowCoord; 
  tell It; 
   showDataWindow #3333749591159525760 
#mesquite.charMatrices.BasicDataWindowMaker.BasicDataWindowMaker; 
   tell It; 
    getWindow; 
    tell It; 
     setExplanationSize 30; 
     setAnnotationSize 20; 
     setFontIncAnnot 0; 
     setFontIncExp 0; 
     setSize 1950 1102; 
     setLocation 1510 150; 
143 
 
     setFont SanSerif; 
     setFontSize 10; 
     getToolPalette; 
     tell It; 
      setTool 
mesquite.charMatrices.BasicDataWindowMaker.BasicDataWindow.ibeam; 
     endTell; 
     setActive; 
     setTool 
mesquite.charMatrices.BasicDataWindowMaker.BasicDataWindow.ibeam; 
     colorCells  #mesquite.charMatrices.NoColor.NoColor; 
     colorRowNames  
#mesquite.charMatrices.TaxonGroupColor.TaxonGroupColor; 
     colorColumnNames  
#mesquite.charMatrices.CharGroupColor.CharGroupColor; 
     colorText  #mesquite.charMatrices.NoColor.NoColor; 
     setBackground White; 
     toggleShowNames on; 
     toggleShowTaxonNames on; 
     toggleTight off; 
     toggleThinRows off; 
     toggleShowChanges on; 
     toggleSeparateLines off; 
     toggleShowStates on; 
     toggleAutoWCharNames on; 
     toggleAutoTaxonNames off; 
     toggleShowDefaultCharNames off; 
     toggleConstrainCW on; 
     toggleBirdsEye off; 
     toggleShowPaleGrid off; 
     toggleShowPaleCellColors off; 
     toggleShowPaleExcluded off; 
     togglePaleInapplicable on; 
     toggleShowBoldCellText off; 
     toggleAllowAutosize on; 
     toggleColorsPanel off; 
     toggleDiagonal on; 
     setDiagonalHeight 80; 
     toggleLinkedScrolling on; 
     toggleScrollLinkedTables off; 
    endTell; 
    showWindow; 
    getWindow; 
    tell It; 
     forceAutosize; 
    endTell; 
144 
 
    getEmployee #mesquite.charMatrices.AlterData.AlterData; 
    tell It; 
     toggleBySubmenus off; 
    endTell; 
    getEmployee #mesquite.charMatrices.ColorByState.ColorByState; 
    tell It; 
     setStateLimit 9; 
     toggleUniformMaximum on; 
    endTell; 
    getEmployee #mesquite.charMatrices.ColorCells.ColorCells; 
    tell It; 
     setColor Red; 
     removeColor off; 
    endTell; 
    getEmployee #mesquite.categ.StateNamesStrip.StateNamesStrip; 
    tell It; 
     showStrip off; 
    endTell; 
    getEmployee #mesquite.charMatrices.AnnotPanel.AnnotPanel; 
    tell It; 
     togglePanel off; 
    endTell; 
    getEmployee 
#mesquite.charMatrices.CharReferenceStrip.CharReferenceStrip; 
    tell It; 
     showStrip off; 
    endTell; 
    getEmployee 
#mesquite.charMatrices.QuickKeySelector.QuickKeySelector; 
    tell It; 
     autotabOff; 
    endTell; 
    getEmployee 
#mesquite.charMatrices.SelSummaryStrip.SelSummaryStrip; 
    tell It; 
     showStrip off; 
    endTell; 
    getEmployee 
#mesquite.categ.SmallStateNamesEditor.SmallStateNamesEditor; 
    tell It; 
     panelOpen true; 
    endTell; 
   endTell; 
  endTell; 
  endTell; 
end; 
145 
 
 
 
APPENDIX C 
 
FIRST AND LAST APPEARANCE DATA FOR TAXA IN CHAPTER II 
 
 
First appearance datum (FAD) and last appearance datum (LAD) in 
millions of years ago for each of the taxa used to timescale the 
phylogeny in Figure 4 of Chapter II after Boessenecker and Churchill 
( 2018) and Poust and Boessenecker (2018).  
Taxon Name FAD LAD 
Enaliarctos spp. 30.6  16.6 
Pteronarctos goedertae 20.2 16.6 
Pinnarctidion bishopi 24 21.5 
Pinnarctidion rayi 24.7 19.1 
Desmatophoca oregonensis 20.2 16.6 
Desmatophoca brachycephala 22.9 19 
Allodesmus kernensis 15.85 15.15 
Atopotarus courseni 15.7 14 
Allodesmus packardi 14.8 11.6 
Allodesmus naorai 13.5 11.8 
Allodesmus demerei 10.5 9.1 
Allodesmus uraiporensis 16.3 13.5 
Allodesmus sinanoensis 13.5 12.5 
Thalassoleon mexicanus 6 4.9 
Callorhinus ursinus 1.81 0 
Zalophus californianus 0.78 0 
Proneotherium repenningi 17.3 16.6 
Neotherium mirum 15.85 15.15 
Imagotaria downsi 10 7 
Gomphotaria pugnax 7.1 4.9 
Odobenus rosmarus 2.7 0 
Devinophoca claytoni 16.3 12.82 
Monachus monachus 0.98 0 
Erignathus barbatus 1.81 0 
Eodesmus condoni 17.3 16.6 
USNM 335445 17.3 16.6 
 
146 
APPENDIX D 
 
TIME-SCALED PHYLOGENY IN NEWICK FORMAT USED IN 
CHAPTER III 
 
 
 
((((((((Devinophoca_claytoni:3.68,(((Erignathus_barbatus:0.5,Erignathus_barbatus1:0.5):
10.5,((Cystophora_cristata1:0.5,Cystophora_cristata:0.5):6.5,(((Histriophoca_fasc
iata1:0.5,Histriophoca_fasciata:0.5):3.5,(Pagophilus_groenlandicus:0.5,Pagophilu
s_groenlandicus1:0.5):3.5):1.5,((Phoca_vitulina:0.5,Phoca_vitulina1:0.5,Phoca_vi
tulina2:0.5):2.5,(((Pusa_hispida:0.5,Pusa_hispida1:0.5):1,(Pusa_sibirica2:0.5,Pusa
_sibirica3:0.5,Pusa_sibirica1:0.5,Pusa_sibirica:0.5):1):1,(Pusa_caspica:1.75,(Hali
choerus_grypus1:0.5,Halichoerus_grypus:0.5):1.25):0.75):0.5):2.5):1.5):4):5,((Ho
miphoca_capensis:3.504,Piscophoca_pacifica:3.504):2,(((Neomonachus_schauins
landi1:0.5,Neomonachus_schauinslandi2:0.5,Neomonachus_schauinslandi:0.5):8.
5,Monachus_monachus:9):1.5,((Mirounga_angustirostris:0.5,Mirounga_angustiro
stris1:0.5):8.5,((Ommatophoca_rossii:0.5,Ommatophoca_rossii1:0.5):5.5,((Lepton
ychotes_weddellii:0.5,Leptonychotes_weddellii1:0.5):3,(Hydrurga_leptonyx1:0.5,
Hydrurga_leptonyx:0.5):3):2.5):3):1.5):2.25):3.25):1.5):9.44,(((Eodesmus_condo
ni:1.06,Eodesmus_sp.:1.06):3.25,(Allodesmus_kernensis:1.47,(Desmatophoca_br
achycephala:0.97,Desmatophoca_oregonensis:3.56):0.5):0.25):0.17,((Neotherium
_mirum:1.16,(Imagotaria_downsi:6.615,(Pontolis_barroni:4.41,(Dusignathus:2.20
5,(Pliopedia_pacifica:0.367,(Valenictus_imperialensis:0.1835,(Odobenus_rosmar
us:0.5,Odobenus_rosmarus1:0.5,Odobenus_rosmarus2:0.5,Odobenus_rosmarus3:
0.5,Odobenus_rosmarus4:0.5,Odobenus_rosmarus5:0.5,Odobenus_rosmarus8:0.5,
Odobenus_rosmarus9:0.5,Odobenus_rosmarus6:0.5,Odobenus_rosmarus7:0.5):5.
0165):0.1835):8.125):2.205):2.205):3.365):1.16,(((((((((Arctocephalus_australis:0
.5,Arctocephalus_australis1:0.5):0.2,(Arctocephalus_galapagoensis1:0.5,Arctocep
halus_galapagoensis:0.5):0.2):0.4,(Arctocephalus_forsteri:0.5,Arctocephalus_fors
teri1:0.5):0.6):3.4,(Otaria_bryonia:0.5,Otaria_bryonia1:0.5):4):0.7,((Eumetopias_j
ubatus:0.5,Eumetopias_jubatus1:0.5,Eumetopias_jubatus2:0.5):4,(Zalophus_calif
ornianus:0.5,Zalophus_californianus1:0.5):4):0.7):0.9,(Neophoca_cinerea1:0.5,Ne
ophoca_cinerea:0.5):5.6):2.1,(Callorhinus_ursinus:0.5,Callorhinus_ursinus1:0.5):
7.7):0.4,Thalassoleon_mexicanus:1.354):8.59,Pithanotaria_starri:5.57):5.57):2.16
):2.02):1.36,Pinnarctidion_rayi:7.86):1.2,Pteronarctos_goedertae:9.06):2.5,Enaliar
ctos_emlongi:3.9):1,Enaliarctos_tedfordi:4.9):2.5,Enaliarctos_mealsi:7.4):3.91,Po
tamotherium_vallentoni:11.01); 
147 
 
APPENDIX E 
 
PRINCIPAL COMPONENTS USED TO TRAIN THE FLEXIBLE 
DISCRIMINANT ANALYSES IN CHAPTER III 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
148 
 
 
150 
 
 
150 
 
 
APPENDIX F 
 
PLOT OF PRINCIPAL COMPONENTS 1 AND 2 FROM PRINCIPAL 
COMPONENT ANALYSIS OF ENVIRONMENTAL PARAMETERS FROM 
CHAPTER IV  
 
 
 
Projections of the ecospace for modern (maroon) and fossil (light blue) occurrences of O. 
rosmarus onto the first two axes of a principal component analysis (PCA). According to the 
broken stick criterion, principal component (PC1) is non-significant and PC2 is significant. The 
arrows indicate how sea surface temperature minimum, sea surface temperature range, and sea 
surface salinity range each influence PCs 1 and 2. 
151 
APPENDIX G 
ECOLOGICAL NICHE MODEL SUMMARY STATISTICS 
Table 1. Evaluation summary statistics for the modern ecological niche models from ENMevaluate object (Kass et 
al., 2021). Tr = training dataset, AUC = area under the curve, CBI = Boyce Index (an index measuring the quality of 
the ecological niche model predictions for the presence of the relevant species; Boyce et al. 2002), Mean AUC.diff = 
the average difference between the training and testing AUC across k bins (indicates degree of model overfitting; 
Warren and Seifert, 2011), SD = standard deviation, AUC.val = AUC calculated on the validation training sets (the 
data withheld during cross-validation), CBI.val = CBI calculated on the validation training sets, ORMTP = omission 
rate with threshold as the minimum suitability value across occurrence records, OR10P = omission rate with threshold 
as the minimum suitability value across occurrence records after removing the lowest 10 percent, AICc = Akaike 
criterion corrected for small sample sizes, AICc.delt = highest AIC value across all models subtracted from this 
model’s AICc, where lower values mean higher performance, NC = number of model coefficients, fc = function 
class, L = linear, LQ = linear-quadratic, LQP = linear-quadratic-product, rm = regularization multiplier. Continued 
onto the subsequent page. 
Run Tuning Arguments AUCTr CBI
Mean SD Mean SD Mean 
Tr AUC.diff AUC.diff AUC.val AUC.val CBI.val 
1 fc.L_rm.0.1 0.8726 0.890 0.00181 0.00110 0.872 0.00088 0.882 
2 fc.LQ_rm.0.1 0.8933 0.942 0.00776 0.00282 0.892 0.00582 0.900 
3 fc.P_rm.0.1 0.8773 0.898 0.00184 0.00078 0.877 0.00157 0.901 
4 fc.LQP_rm.0.1 0.9014 0.970 0.00753 0.00203 0.900 0.00566 0.919 
5 fc.L_rm.0.25 0.8725 0.888 0.00180 0.00110 0.872 0.00089 0.886 
6 fc.LQ_rm.0.25 0.8931 0.948 0.00757 0.00270 0.892 0.00552 0.909 
7 fc.P_rm.0.25 0.8772 0.898 0.00192 0.00076 0.877 0.00162 0.903 
8 fc.LQP_rm.0.25 0.9013 0.971 0.00745 0.00203 0.900 0.00554 0.920 
9 fc.L_rm.0.5 0.8725 0.889 0.00178 0.00110 0.872 0.00090 0.884 
10 fc.LQ_rm.0.5 0.8930 0.950 0.00718 0.00252 0.892 0.00495 0.913 
11 fc.P_rm.0.5 0.8769 0.902 0.00205 0.00067 0.877 0.00165 0.903 
12 fc.LQP_rm.0.5 0.9010 0.973 0.00727 0.00202 0.900 0.00539 0.924 
13 fc.L_rm.1 0.8724 0.890 0.00175 0.00110 0.872 0.00093 0.878 
14 fc.LQ_rm.1 0.8927 0.947 0.00654 0.00217 0.892 0.00422 0.913 
15 fc.P_rm.1 0.8764 0.914 0.00218 0.00056 0.876 0.00157 0.910 
16 fc.LQP_rm.1 0.8997 0.958 0.00700 0.00187 0.899 0.00496 0.927 
17 fc.L_rm.2 0.8723 0.883 0.00169 0.00111 0.872 0.00097 0.882 
18 fc.LQ_rm.2 0.8924 0.929 0.00573 0.00147 0.892 0.00328 0.899 
19 fc.P_rm.2 0.8764 0.911 0.00214 0.00058 0.876 0.00153 0.908 
20 fc.LQP_rm.2 0.8985 0.951 0.00626 0.00161 0.898 0.00374 0.926 
21 fc.L_rm.4 0.8721 0.881 0.00162 0.00113 0.872 0.00107 0.879 
22 fc.LQ_rm.4 0.8918 0.896 0.00462 0.00037 0.892 0.00176 0.883 
23 fc.P_rm.4 0.8764 0.915 0.00207 0.00060 0.876 0.00148 0.912 
24 fc.LQP_rm.4 0.8956 0.932 0.00489 0.00106 0.895 0.00255 0.906 
152 
Table 1. Continued. 
Run SD Tuning Mean SD CBI.val Arguments OR  OR  AICc AICc.delt NC MTP MTP
1 0.0035 fc.L_rm.0.1 0.00245 0.00347 10833 488.65 3 
2 0.0750 fc.LQ_rm.0.1 0.00490 0.00693 10344 0.00 6 
3 0.0120 fc.P_rm.0.1 0.00245 0.00347 10810 466.17 3 
4 0.0566 fc.LQP_rm.0.1 0.00980 0.01386 10473 129.35 8 
5 0.0057 fc.L_rm.0.25 0.00245 0.00347 10833 489.36 3 
6 0.0728 fc.LQ_rm.0.25 0.00490 0.00693 10351 6.76 6 
7 0.0134 fc.P_rm.0.25 0.00245 0.00347 10807 463.40 3 
8 0.0559 fc.LQP_rm.0.25 0.00980 0.01386 10476 132.17 8 
9 0.0000 fc.L_rm.0.5 0.00245 0.00347 10834 490.51 3 
10 0.0686 fc.LQ_rm.0.5 0.00490 0.00693 10363 18.93 5 
11 0.0120 fc.P_rm.0.5 0.00245 0.00347 10807 463.14 3 
12 0.0580 fc.LQP_rm.0.5 0.00980 0.01386 10478 134.41 7 
13 0.0028 fc.L_rm.1 0.00245 0.00347 10837 492.94 3 
14 0.0559 fc.LQ_rm.1 0.00490 0.00693 10390 46.22 5 
15 0.0191 fc.P_rm.1 0.00245 0.00347 10809 464.62 2 
16 0.0566 fc.LQP_rm.1 0.00735 0.01040 10487 142.63 6 
17 0.0092 fc.L_rm.2 0.00245 0.00347 10842 498.07 3 
18 0.0382 fc.LQ_rm.2 0.00490 0.00693 10450 106.26 5 
19 0.0226 fc.P_rm.2 0.00245 0.00347 10816 471.62 2 
20 0.0481 fc.LQP_rm.2 0.00490 0.00693 10531 186.79 6 
21 0.0092 fc.L_rm.4 0.00245 0.00347 10853 509.27 3 
22 0.0601 fc.LQ_rm.4 0.00490 0.00693 10580 236.10 5 
23 0.0247 fc.P_rm.4 0.00245 0.00347 10830 485.56 2 
24 0.0757 fc.LQP_rm.4 0.00245 0.00347 10626 281.84 6 
153 
Table 2. Evaluation summary statistics for the fossil ecological niche models from ENMevaluate object (Kass et al., 
2021). Abbreviations are the same as in Table 1. Continued onto the subsequent page. 
Run Tuning Arguments AUC  CBI  
Mean SD Mean SD Mean SD 
Tr Tr AUC.diff AUC.diff AUC.val AUC.val CBI.val CBI.val 
1 fc.L_rm.0.1 0.8470 0.966 0.00448 0.00221 0.846 0.00129 0.897 0.0163 
2 fc.LQ_rm.0.1 0.8568 0.882 0.01462 0.00312 0.857 0.00859 0.762 0.0460 
3 fc.P_rm.0.1 0.8293 0.959 0.01293 0.00312 0.829 0.00959 0.936 0.0021 
4 fc.LQP_rm.0.1 0.8649 0.929 0.01072 0.00839 0.862 0.00794 0.878 0.0672 
5 fc.L_rm.0.25 0.8470 0.967 0.00453 0.00227 0.846 0.00131 0.891 0.0014 
6 fc.LQ_rm.0.25 0.8571 0.881 0.01538 0.00413 0.856 0.00937 0.771 0.0431 
7 fc.P_rm.0.25 0.8293 0.96 0.01329 0.00264 0.829 0.00986 0.940 0.0035 
8 fc.LQP_rm.0.25 0.8651 0.923 0.01100 0.00733 0.864 0.00735 0.875 0.0516 
9 fc.L_rm.0.5 0.8470 0.967 0.00459 0.00235 0.846 0.00135 0.894 0.0141 
10 fc.LQ_rm.0.5 0.8576 0.891 0.01461 0.00368 0.857 0.00846 0.780 0.0283 
11 fc.P_rm.0.5 0.8292 0.96 0.01391 0.00176 0.830 0.01027 0.937 0.0035 
12 fc.LQP_rm.0.5 0.8655 0.923 0.00902 0.00682 0.865 0.00372 0.872 0.0403 
13 fc.L_rm.1 0.8470 0.966 0.00471 0.00249 0.846 0.00141 0.896 0.0156 
14 fc.LQ_rm.1 0.8581 0.885 0.01432 0.00425 0.858 0.00811 0.794 0.0651 
15 fc.P_rm.1 0.8291 0.964 0.01515 0.00025 0.831 0.01051 0.943 0.0092 
16 fc.LQP_rm.1 0.8663 0.921 0.00760 0.00523 0.862 0.00349 0.897 0.0502 
17 fc.L_rm.2 0.8470 0.969 0.00498 0.00276 0.846 0.00156 0.886 0.0000 
18 fc.LQ_rm.2 0.8584 0.844 0.01089 0.00504 0.856 0.00393 0.746 0.1718 
19 fc.P_rm.2 0.8288 0.957 0.01562 0.00008 0.832 0.01069 0.936 0.0071 
20 fc.LQP_rm.2 0.8670 0.926 0.00584 0.00008 0.861 0.00790 0.921 0.0219 
21 fc.L_rm.4 0.8471 0.962 0.00563 0.00327 0.847 0.00178 0.889 0.0134 
22 fc.LQ_rm.4 0.8588 0.827 0.00622 0.00339 0.847 0.00202 0.880 0.0127 
23 fc.P_rm.4 0.8291 0.962 0.01589 0.00006 0.835 0.01027 0.887 0.0247 
24 fc.LQP_rm.4 0.8671 0.919 0.01076 0.00633 0.863 0.00289 0.906 0.0163 
154 
Table 2. Continued. 
Run Tuning Mean SD Arguments OR  OR AICc AICc.delt NC MTP MTP 
2 fc.LQ_rm.0.1 0.00943 0.01334 2770 0.00 6 
3 fc.P_rm.0.1 0.01087 0.01537 2949 179.13 3 
4 fc.LQP_rm.0.1 0.00943 0.01334 2800 30.27 9 
5 fc.L_rm.0.25 0.02174 0.03074 2857 86.59 3 
6 fc.LQ_rm.0.25 0.02030 0.00203 2771 1.18 6 
7 fc.P_rm.0.25 0.01087 0.01537 2949 179.18 3 
8 fc.LQP_rm.0.25 0.00943 0.01334 2802 31.59 9 
9 fc.L_rm.0.5 0.02174 0.03074 2857 86.90 3 
10 fc.LQ_rm.0.5 0.03117 0.01740 2773 2.88 6 
11 fc.P_rm.0.5 0.01087 0.01537 2949 179.28 3 
12 fc.LQP_rm.0.5 0.00943 0.01334 2804 33.80 9 
13 fc.L_rm.1 0.02174 0.03074 2858 87.52 3 
14 fc.LQ_rm.1 0.02174 0.03074 2777 7.10 6 
15 fc.P_rm.1 0.01087 0.01537 2950 179.52 3 
16 fc.LQP_rm.1 0.02174 0.03074 2809 38.53 9 
17 fc.L_rm.2 0.02174 0.03074 2859 88.71 3 
18 fc.LQ_rm.2 0.02174 0.03074 2787 16.77 6 
19 fc.P_rm.2 0.01087 0.01537 2950 180.14 3 
20 fc.LQP_rm.2 0.02174 0.03074 2811 40.97 7 
21 fc.L_rm.4 0.02174 0.03074 2861 91.26 3 
22 fc.LQ_rm.4 0.02174 0.03074 2809 39.12 5 
23 fc.P_rm.4 0.01087 0.01537 2949 179.03 2 
24 fc.LQP_rm.4 0.02174 0.03074 2816 45.93 5 
 
155 
 
 
REFERENCES CITED 
 
Chapter I 
Barnes, L. G. 1987. A new Miocene pinniped of the genus Desmatophoca (Mammalia: 
Otariidae) from Washington. Contributions in Science, Natural History Museum of Los 
Angeles County 382:1–20. 
Bebej, R. M. 2009. Swimming mode inferred from skeletal proportions in the fossil pinnipeds 
Enaliarctos and Allodesmus (Mammalia, Carnivora). Journal of Mammalian Evolution 
16:77–97. 
Berta, A., M. Churchill, and R. W. Boessenecker. 2018. The origin and evolutionary biology of 
pinnipeds: seals, sea lions, and walruses. Annual Review of Earth and Planetary Sciences 
46:203–228. 
Boessenecker, R. W., and M. Churchill. 2013. A reevaluation of the morphology, paleoecology, 
and phylogenetic relationships of the enigmatic walrus Pelagiarctos. PLoS ONE 8. 
Böhme, M. 2003. The Miocene Climatic Optimum: evidence from ectothermic vertebrates of 
Central Europe. Palaeogeography, Palaeoclimatology, Palaeoecology 195:389–401. 
Bowen, W. D. 1997. Role of marine mammals in aquatic ecosystems. Marine Ecology Progress 
Series 158:267–274. 
Cammen, K. M., D. B. Rasher, and R. S. Steneck. 2019. Predator recovery, shifting baselines, 
and the adaptive management challenges they create. Ecosphere 10. 
Condon, T. A. 1906. A new fossil pinniped from the Miocene of the Oregon coast. University of 
Oregon Bulletin 3:5–14. 
Deméré, T. A., and A. Berta. 2002. The Miocene pinniped Desmatophoca oregonensis Condon, 
1906 (Mammalia: Carnivora), from the Astoria Formation, Oregon. Smithsonian 
Contributions to Paleobiology 93:113–147. 
Diebold, F. X., and G. D. Rudebusch. 2021. Probability assessments of an ice-free Arctic: 
comparing statistical and climate model projections. Journal of Econometrics. 
Dietl, G. P., and K. W. Flessa. 2011. Conservation paleobiology: putting the dead to work. Trends 
in Ecology & Evolution 26:30–37. 
Dietl, G. P., S. M. Kidwell, M. Brenner, D. A. Burney, K. W. Flessa, S. T. Jackson, and P. L. 
Koch. 2015. Conservation paleobiology: leveraging knowledge of the past to inform 
conservation and restoration. Annual Review of Earth and Planetary Sciences 43:79–103. 
156 
 
Estes, J. A., J. Terborgh, J. S. Brashares, M. E. Power, J. Berger, W. J. Bond, S. R. Carpenter, T. 
E. Essington, R. D. Holt, J. B. C. Jackson, R. J. Marquis, L. Oksanen, T. Oksanen, R. T. 
Paine, E. K. Pikitch, W. J. Ripple, S. A. Sandin, M. Scheffer, T. W. Schoener, J. B. 
Shurin, A. R. E. Sinclair, M. E. Soulé, R. Virtanen, and D. A. Wardle. 2011. Trophic 
downgrading of planet Earth. Science 333:301–306. 
Estes, J. A., R. S. Steneck, and D. R. Lindberg. 2013. Exploring the consequences of species 
interactions through the assembly and disassembly of food webs: a Pacific-Atlantic 
comparison. Bulletin of Marine Science 89:11–29. 
Fay, F. H. 1982. Ecology and biology of the Pacific walrus, Odobenus rosmarus divergens 
(Illiger). North American Fauna 74:1–279. 
Flower, B. P., and J. P. Kennett. 1993. Middle Miocene ocean-climate transition: high resolution 
oxygen and carbon isotopic records from Deep Sea Drilling Project Site 588A, southwest 
Pacific. Integration The Vlsi Journal 8:811–843. 
Furbish, R. 2015. Something old, something new, something swimming in the blue: an analysis 
of the pinniped family Desmatophocidae, its phylogenetic position and swimming mode. 
75 pp. 
Hammerschlag, N., R. A. Martin, and C. Fallows. 2006. Effects of environmental conditions on 
predator-prey interactions between white sharks (Carcharodon carcharias) and Cape fur 
seals (Arctocephalus pusillus pusillus) at Seal Island, South Africa. Environmental 
Biology of Fishes 76:341–350. 
Heithaus, M. R., A. Frid, A. J. Wirsing, and B. Worm. 2008. Predicting ecological consequences 
of marine top predator declines. Trends in Ecology and Evolution 23:202–210. 
Hocking, D. P., F. G. Marx, S. Wang, D. Burton, M. Thompson, T. Park, B. Burville, H. L. 
Richards, R. Sattler, J. Robbins, R. P. Miguez, E. M. G. Fitzgerald, D. J. Slip, and A. R. 
Evans. 2021. Convergent evolution of forelimb-propelled swimming in seals. Current 
Biology 31:2404-2409.e2. 
Huntington, H. P., S. L. Danielson, F. K. Wiese, M. Baker, P. Boveng, J. J. Citta, A. De Robertis, 
D. M. S. Dickson, E. Farley, J. C. George, K. Iken, D. G. Kimmel, K. Kuletz, C. Ladd, R. 
Levine, L. Quakenbush, P. Stabeno, K. M. Stafford, D. Stockwell, and C. Wilson. 2020. 
Evidence suggests potential transformation of the Pacific Arctic ecosystem is underway. 
Nature Climate Change 10:342–348. 
IUCN. 2022. The IUCN Red List of Threatened Species. Version 2022-2. 
https://www.iucnredlist.org. Accessed on 4 March 2023. 
Jablonski, D., and J. J. Sepkoski. 1996. Paleobiology, community ecology, and scales of 
ecological pattern. Ecology 77:1367–1378. 
Jacobs, P., N. Lenssen, G. Schmidt, and R. Rohde. 2021. The Arctic is now warming four times 
as fast as the rest of the globe. AGU Fall Meeting. 
157 
 
Jahn, A., J. E. Kay, M. M. Holland, and D. M. Hall. 2016. How predictable is the timing of a 
summer ice-free Arctic? Geophysical Research Letters 43:9113–9120. 
King, J. E. 1983. Seals of the World. Cornell University Press, Ithaca, New York, 240. pp.  
Kohno, N. 1992. An Early Miocene enaliarctine pinniped (Carnivora: Otariidae) from the 
western North Pacific. Bulletin of the Mizunami Fossil Museum 19:273–292. 
Kovacs, K. M., A. Aguilar, D. Aurioles, V. Burkanov, C. Campagna, N. Gales, T. Gelatt, S. D. 
Goldsworthy, S. J. Goodman, G. J. G. Hofmeyr, T. Härkönen, L. Lowry, C. Lydersen, J. 
Schipper, T. Sipilä, C. Southwell, S. Stuart, D. Thompson, and F. Trillmich. 2012. Global 
threats to pinnipeds. Marine Mammal Science. 
Lipps, J. H., and E. Mitchell. 1976. Trophic model for the adaptive radiations and extinctions of 
pelagic marine mammals. Paleontological Society 2:147–155. 
McCauley, D. J., M. L. Pinsky, S. R. Palumbi, J. A. Estes, F. H. Joyce, and R. R. Warner. 2015. 
Marine defaunation: animal loss in the global ocean. Science 347. 
Notz, D., and J. Stroeve. 2018. The trajectory towards a seasonally ice-free Arctic Ocean. 
Current Climate Change Reports 4:407–416. 
Overland, J. E., and M. Wang. 2013. When will the summer Arctic be nearly sea ice free? 
Geophysical Research Letters 40:2097–2101. 
Paterson, R. 2017. Evidence for independent acquisition of aquatic specializations in pinnipeds 
(seals, sea lions, and walruses): insights from study of the phylogenetic position, 
locomotor behaviour and description of the stem pinniped, Puijila darwini. 
Paterson, R. S., N. Rybczynski, N. Kohno, and H. C. Maddin. 2020. A total evidence 
phylogenetic analysis of pinniped phylogeny and the possibility of parallel evolution 
within a monophyletic framework. Frontiers in Ecology and Evolution 7:1–16. 
Pyenson, N. D., R. B. Irmis, J. H. Lipps, L. G. Barnes, E. D. Mitchell, and S. A. McLeod. 2009. 
Origin of a widespread marine bonebed deposited during the middle Miocene Climatic 
Optimum. Geology 37:519–522. 
Rantanen, M., A. Y. Karpechko, A. Lipponen, K. Ruosteenoja, T. Vihma, A. Laaksonen, K. 
Nordling, and O. Hyvärinen. 2021. The Arctic has warmed nearly four times faster than 
the globe since 1979. Communication 3:1–10. 
Ray, G. C., J. McCormick-Ray, P. Berg, and H. E. Epstein. 2006. Pacific walrus: benthic 
bioturbator of Beringia. Journal of Experimental Marine Biology and Ecology 330:403–
419. 
Riedman, M. 1990. The Pinnipeds: Seals, Sea Lions, and Walruses. University of California 
Press, Berkeley. 1990, 149 pp.   
158 
 
Rule, J. P., J. W. Adams, F. G. Marx, A. R. Evans, A. J. D. Tennyson, R. P. Scofield, and E. M. G. 
Fitzgerald. 2020. First monk seal from the Southern Hemisphere rewrites the 
evolutionary history of true seals. Proceedings of Biological Sciences 287:20202318. 
Springer, A. M., J. A. Estes, G. B. van Vliet, T. M. Williams, D. F. Doak, E. M. Danner, K. A. 
Forney, and B. Pfister. 2003. Sequential megafaunal collapse in the North Pacific Ocean: 
an ongoing legacy of industrial whaling? PNAS 100:12223–12228. 
Theobald, M. R., P. D. Crittenden, A. P. Hunt, Y. S. Tang, U. Dragosits. 2006. Ammonia 
emissions from a Cape fur seal colony, Cape Cross, Namibia. Geophysical Research 
Letters 33: 1–4. 
Tonomori, W., and H. Taru. 2022. Tarsals of early pinnipedimorphs (Mammalia: Carnivora) from 
the Lower Miocene, Japan, and their early diversity in the western North Pacific. 
Paleontological Research 26:314–326. 
Tonomori, W., H. Sawamura, T. Sato, and N. Kohno. 2018. A new Miocene pinniped Allodesmus 
(Mammalia: Carnivora) from Hokkaido, Northern Japan. Royal Society Open Science 5. 
Tsuchi, R. 1992. Neogene events in Japan and on the Pacific coast of South America. Revista 
Geológica de Chile: An International Journal on Andean Geology 19:67–73. 
Turner, M. G., W. J. Calder, G. S. Cumming, T. P. Hughes, A. Jentsch, S. L. LaDeau, T. M. 
Lenton, B. N. Shuman, M. R. Turetsky, Z. Ratajczak, J. W. Williams, A. P. Williams, and 
S. R. Carpenter. 2020. Climate change, ecosystems and abrupt change: science priorities. 
Philosophical Transactions of the Royal Society B: Biological Sciences 375. 
Valenzuela-Toro, A., and N. D. Pyenson. 2019. What do we know about the fossil record of 
pinnipeds? A historiographical investigation. Royal Society Open Science 6. 
Yletyinen, J. 2019. Arctic climate resilience. Nature Climate Change 9:805–806. 
 
Chapter II 
Armentrout, J. M. 1981. Correlation and ages of Cenozoic stratigraphic units in Oregon and 
Washington; pp. 137–148 in J. M. Armentrout (ed.), Pacific Northwest Cenozoic 
Biostratigraphy. Geological Society of America, Boulder, Colorado. 
Barnes, L. G. 1972. Miocene Desmatophocinae (Mammalia: Carnivora) from California. 
University of California Publications in Geological Sciences 89:1–69. 
Barnes, L. G. 1979. Fossil enaliarctine pinnipeds (Mammalia: Otariidae) from Pyramid Hill, 
Kern County, California. Contributions in Science (Los Angeles) 318:1–41. 
Barnes L. G. 1987. A new Miocene pinniped of the genus Desmatophoca (Mammalia: Otariidae) 
from Washington. Contributions in Science, Natural History Museum of Los Angeles 
County 382:1–20. 
159 
 
Barnes L. G. 1989. A new enaliarctine pinniped from the Astoria Formation, Oregon, and a 
classification of the Otariidae (Mammalia: Carnivora). Contributions in Science, Natural 
History Museum of Los Angeles County 403:1–26. 
Barnes L. G. 1990. A new Miocene enaliarctine pinniped of the genus Pteronarctos (Mammalia: 
Otariidae) from the Astoria Formation, Oregon. Contributions in Science, Natural History 
Museum of Los Angeles County 442:1–20. 
Barnes, L. G. 1992. A new genus and species of Middle Miocene enaliarctine pinniped 
(Mammalia, Carnivora, Otariidae) from the Astoria Formation in coastal Oregon. 
Contributions in Science (Natural History Museum of Los Angeles County) 431:1–27. 
Barnes, L. G., and K. Hirota. 1994. Miocene pinnipeds of the otariid subfamily Allodesminae in 
the North Pacific Ocean: systematics and relationships. The Island Arc 3:329–360. 
Bell, M. A, and G. T. Lloyd. 2015. Strap: an R package for plotting phylogenies against 
stratigraphy and assessing their stratigraphic congruence. Palaeontology 58:379–389. 
Berta, A. 1991. New Enaliarctos (Pinnipedimorpha) from the Oligocene and Miocene of Oregon 
and the role of “enaliarctids” in pinniped phylogeny. Smithsonian Contributions to 
Paleobiology 69:1–33. 
Berta, A. 1994. A new species of phocoid pinniped Pinnarctidion from the Early Miocene of 
Oregon. Journal of Vertebrate Paleontology 14:405–413. 
Berta, A., M. Churchill, and R. W. Boessenecker. 2018. The origin and evolutionary biology of 
pinnipeds: seals, sea lions, and walruses. Annual Review of Earth and Planetary Sciences 
46:203–228. 
Boessenecker, R. W., and M. Churchill. 2018. The last of the desmatophocid seals: a new species 
of Allodesmus from the upper Miocene of Washington, USA, and a revision of the 
taxonomy of Desmatophocidae. Zoological Journal of the Linnaean Society XX:1–25. 
Böhme, M. 2003. The Miocene Climatic Optimum: evidence from ectothermic vertebrates of 
Central Europe. Palaeogeography, Palaeoclimatology, Palaeoecology 195:389–401. 
Bowditch, T.E. 1821. An Analysis of the Natural Classifications of Mammalia for the Use of 
Students and Travelers. J. Smith, Paris, 115 pp. 
Churchill, M., M. T. Clementz, and N. Kohno. 2015. Cope’s rule and the evolution of body size 
in Pinnipedimorpha (Mammalia: Carnivora). Evolution 69:201–215. 
Colbath, S. L. 1985. Gastropod predation and depositional environments of two molluscan 
communities from the Miocene Astoria Formation at Beverly Beach State Park, Oregon. 
Journal of Paleontology 59:849–869. 
Condon, T. A. 1906. A new fossil pinniped from the Miocene of the Oregon coast. University of 
Oregon Bulletin 3:5–14. 
160 
 
Coombs, M. C. 1979. Tylocephalonyx, a new genus of North American dome-skulled 
chalicotheres. Bulletin of the American Museum of Natural History 164:1–68. 
Deméré, T. A., and A. Berta. 2002. The Miocene pinniped Desmatophoca oregonensis Condon, 
1906 (Mammalia: Carnivora), from the Astoria Formation, Oregon. Smithsonian 
Contributions to Paleobiology 93:113–147. 
Dewaele, L., O. Lambert, and S. Louwye. 2017. On Prophoca and Leptophoca (Pinnipedia, 
Phocidae) from the Miocene of the North Atlantic realm: redescription, phylogenetic 
affinities and paleobiogeographic implications. PeerJ 5:e3024. 
Dewaele, L., O. Lambert, and S. Louwye. 2018. A critical revision of the fossil record, 
stratigraphy and diversity of the Neogene seal genus Monotherium (Carnivora, Phocidae). 
Royal Society Open Science 5:171669. 
Downs, T. 1956. A new pinniped from the Miocene of Southern California: with remarks on the 
Otariidae. Journal of Paleontology 30:115–131. 
Drehmer, C. J., D. Sanfelice, and C. Loch. 2015. Dental anomalies in pinnipeds (Carnivora: 
Otariidae and Phocidae): occurrence and evolutionary implications. Zoomorphology 
134:325–338. 
Domning, D. P. 2008. Desmostylia; pp. 640-645 in C. M. Janis, G. F. Gunnell, and M. D. Uhen 
(eds.), Evolution of Tertiary Mammals of North America, Volume II. Cambridge 
University Press, Cambridge, Massachusetts. 
Flower, B. P., and J. P. Kennett. 1993. Middle Miocene ocean-climate transition: high resolution 
oxygen and carbon isotopic records from Deep Sea Drilling Project Site 588A, southwest 
Pacific. Integration The Vlsi Journal 8:811–843. 
Goloboff, P. A., S. A. Catalano. 2016. TNT version 1.5, including a full implementation of 
phylogenetic morphometrics. Cladistics 32:221–238. 
Illiger, C. 1811. Prodromus systematis mammalium et avium. Salfeld, Berlin, 304 pp. 
Kellogg, R. 1922. Pinnipeds from Miocene and Pliocene deposits of California and a résumé of 
current theories regarding the origin of Pinnipedia. University of California Publications 
in Geological Sciences 13:23–132. 
Kohno, N. 1996. Miocene pinniped Allodesmus (Mammalia: Carnivora); with special reference 
to the “Mito seal” from Ibaraki Prefecture, Central Japan. Transactions of the Proceedings 
of the Paleontological Society of Japan 181:388–404. 
Kohno, N., L. G. Barnes, and K. Hirota. 1994. Miocene fossil pinnipeds of the genera Prototaria 
and Neotherium (Carnivora; Otariidae; Imagotariinae) in the North Pacific Ocean: 
evolution, relationships and distribution. The Island Arc 3:285–308. 
161 
 
Koretsky, I. A., and P. Holec. 2002. A primitive seal (Mammalia: Phocidae) from the Early 
Middle Miocene of Central Paratethys. Smithsonian Contributions to Paleobiology 
93:163–178. 
Linnaeus, C. 1758. Systema Naturae per Regna Tria Naturae, secundum Classes, Ordines, 
Genera, Species, cum Characteribus, Differentiis, Synonymis, Locis. Tomus I. Laurentii 
Salvii, Holmiae, Stockholm, 824 pp. 
Lipps, J. H., and E. Mitchell. 1976. Trophic model for the adaptive radiations and extinctions of 
pelagic marine mammals. Paleontological Society 2:147–155. 
Maddison, W.P., and D. R. Maddison. 2018 Mesquite: a modular system for evolutionary 
analysis. Version 3.51. Available at http://www.mesquiteproject.org. Accessed September 
22, 2018. 
Mitchell, E. 1966. The Miocene pinniped Allodesmus. University of California Publications in 
Geological Sciences 61:1–105. 
Mitchell, E. 1968. The Mio-Pliocene pinniped Imagotaria. Journal of Fisheries Research Board 
of Canada 25:1843–1900. 
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 Paleon 53:77–91. 
Nagao, T. 1941. An occurrence of a fossil sea lion in the Miocene deposits of Sinano, Japan. 
Journal of the Faculty of Science, Hokkaido Imperial University 6:75–84. 
Packard, E. L., and R. Kellogg. 1934. A new cetothere from the Miocene Astoria Formation of 
Newport, Oregon. Contributions to Paleontology of the Carnegie Institution of 
Washington 447:1–62. 
Poust, A., and R. W. Boessenecker. 2018. Expanding the geographic and geochronologic range of 
early pinnipeds: new specimens of Enaliarctos from Northern California and Oregon. 
Acta Palaeontologica Polonica 63: 25–40. 
Prothero, D.R., C. Z. Bitboul, G. W. Moore, and E. J. Moore. 2001. Magnetic stratigraphy of the 
lower and middle Miocene Astoria Formation, Lincoln County, Oregon. Pacific Section 
SEPM Special Publication 91: 272–283. 
Pyenson, N. D., R. B. Irmis, J. H. Lipps, L. G. Barnes, E. D. Mitchell, and S. A. McLeod. 2009. 
Origin of a widespread marine bonebed deposited during the middle Miocene Climatic 
Optimum. Geology 37:519–522. 
R Core Team. 2019. R: a language and environment for statistical computing. R Foundation for 
Statistical Computing. Available at http://www.R-project.org/. Accessed December 18, 
2018. 
162 
 
Rohlf, F.J. 2016. tpsDig 2, digitize landmarks and outlines. Version 2.31. Department of Ecology 
and Evolution, State University of New York at Stony Brook. Available at 
http://life.bio.sunysb.edu/morph/index.html. Accessed June 1, 2019. 
Siversten, E. 1954. A survey of the eared seals (Family Otariidae) with remarks on the Antarctic 
seals collected by M/K <<Norvegia>> in 1928-1929. Det Norske Videnskaps-Akademi I 
Oslo 36:5–86. 
Tonomori, W.H., H. Sawamura, T. Sato, and N. Kohno. 2018. A new Miocene pinniped 
Allodesmus (Mammalia: Carnivora) from Hokkaido, Northern Japan. Royal Society Open 
Science 5:172440. 
Tsuchi, R. 1992. Neogene events in Japan and on the Pacific coast of South America. Revista 
Geológica de Chile: An International Journal on Andean Geology 19:67–73. 
Valenzuela-Toro, A., and N. D. Pyenson. 2019. What do we know about the fossil record of 
pinnipeds? A historiographical investigation. Royal Society Open Science 6:1–18. 
Winer, J. N., B. Arzi, D. M. Leale, P. H. Kass, and F. J. M. Verstraete. 2016. Dental and 
temporomandibular joint pathology of the walrus (Odobenus rosmarus). Journal of 
Comparative Pathology 155:242–253. 
 
Chapter III 
Adams, D., M. Collyer, A. Kaliontzopoulou, E. Baken. 2022. Geomorph: Software for geometric 
morphometric analyses. R package version 4.0.4. https://cran.r-
project.org/package=geomorph 
Allen, J. A. 1880. History of North American pinnipeds. US Geological and Geographical 
Survey of the Territories, Miscellaneous Publications 12:1–785 
Amson, E., C. Argot, H. G. McDonald, and C. de Muizon. 2015. Osteology and functional 
morphology of the axial postcranium of the marine sloth Thalassocnus (Mammalia, 
Tardigrada) with paleobiological implications. Journal of Mammalian Evolution 473–518 
pp. 
Andersson, K. 2004. Elbow-joint morphology as a guide to forearm function and foraging 
behaviour in mammalian carnivores. Zoological Journal of the Linnean Society 142:91–
104. 
Arnason, U., A. Gullberg, A. Janke, M. Kullberg, N. Lehman, E. A. Petrov, and R. Väinölä. 
2006. Pinniped phylogeny and a new hypothesis for their origin and dispersal. Molecular 
Phylogenetics and Evolution 41:345–354. 
Backhouse, K. M. 1962. Locomotion of seals with particular reference to the forelimb. Symposia 
of the Zoological Society of London 5:59–75. 
 
163 
 
Baken, E., M. Collyer, A. Kaliontzopoulou, D. Adams. 2021. geomorph v4.0 and gmShiny: 
enhanced analytics and a new graphical interface for a comprehensive morphometric 
experience. 
Bapst, D. W. 2012. Paleotree: An R package for paleontological and phylogenetic analyses of 
evolution. Methods in Ecology and Evolution 3:803–807. 
Barnes, L. G. 1989. A new enaliarctine pinniped from the Astoria Formation, Oregon, and a 
classification of the Otariidae (Mammalia: Carnivora). Contributions in Science 403:1–
26. 
Barnes, L. G., and K. Hirota. 1994. Miocene pinnipeds of the otariid subfamily Allodesminae in 
the North Pacific Ocean: systematics and relationships. The Island Arc 3:329–360. 
Bebej, R. M. 2009. Swimming mode inferred from skeletal proportions in the fossil pinnipeds 
Enaliarctos and Allodesmus (Mammalia, Carnivora). Journal of Mammalian Evolution 
16:77–97. 
Beentjes, M. P. 1990. Comparative terrestrial locomotion of the Hooker’s sea lion (Phocarctos 
hookeri) and the New Zealand fur seal (Arctocephalus forsteri) and the evolutionary and 
ecological implications. Zoological Journal of the Linnean Society 98:307–325. 
Bell, M. A., and G. T. Lloyd. 2015. Strap: an R package for plotting phylogenies against 
stratigraphy and assessing their stratigraphic congruence. 
Berta, A. 1994. A new species of phocoid pinniped Pinnarctidion from the Early Miocene of 
Oregon. Journal of Vertebrate Paleontology 14:405–413. 
Berta, A., and C. E. Ray. 1990. Skeletal morphology and locomotor capabilities of the archaic 
pinniped Enaliarctos mealsi. Journal of Vertebrate Paleontology 10:141–157. 
Berta, A., and A. R. Wyss. 1994. Pinniped phylogeny. Contributions in Marine Mammal 
Paleontology Honoring Frank C. Whitmore, Jr.: Proceedings of the San Diego Society of 
Natural History 29:33–56. 
Berta, A., and P. J. Adam. 2001. Evolutionary biology of pinnipeds; pp. 235–260 in J.-M. Mazin 
and V. de Buffrénil (eds.), Secondary Adaptation of Tetrapods to Life in Water. Verlag Dr. 
Friedrich Pfeil, Munich, Germany. 
Berta, A., C. E. Ray, and A. R. Wyss. 1989. Skeleton of the oldest known pinniped, Enaliarctos 
mealsi. Science 244:60–92. 
Berta, A., J. L. Sumich, K. M. Kovacs. 2015. Marine Mammals: Evolutionary Biology, 3rd ed. 
Academic Press, San Diego, California. 
Berta, A., M. Churchill, and R. W. Boessenecker. 2018. The origin and evolutionary biology of 
pinnipeds: seals, sea lions, and walruses. Annual Review of Earth and Planetary Sciences 
46:203–228. 
164 
 
Biewer, J. N., J. Velez-Juarbe, and J. F. Parham. 2020. Insights on the dental evolution of 
walruses based on new fossil specimens from California. Journal of Vertebrate 
Paleontology 40:1–42. 
Boessenecker, R. W., and M. Churchill. 2015. The oldest known fur seal. Biology Letters 11. 
Boessenecker, R. W., and M. Churchill. 2018. The last of the desmatophocid seals: a new species 
of Allodesmus from the upper Miocene of Washington, USA, and a revision of the 
taxonomy of Desmatophocidae. Zoological Journal of the Linnean Society XX:1–25. 
Botton-Divet, L., R. Cornette, A. C. Fabre, A. Herrel, and A. Houssaye. 2016. Morphological 
analysis of long bones in semi-aquatic mustelids and their terrestrial relatives. Integrative 
and Comparative Biology 56:1298–1309. 
Brusatte, S. L., M. J. Benton, M. Ruta, and G. T. Lloyd. 2008. Superiority, competition, and 
opportunism in the evolutionary radiation of dinosaurs. Science 321:1485–1488. 
Campagna, C., R. Harcourt, R. Gentry, D. Costa, A. Valenzuela-Toro, M. Sepúlveda, T. 
Jeanniard-du-Dot, C. Guinet, D. Hocking, T. Park, J. Rule, F. Marx, M. Cassini, M. 
Kiyota, B. Le Boeuf, C. Bisioli, S. Ryazanov, C. Bonin, V. Franco-Trecu, J. Kingston 
Sinclair, S. Goldsworthy, I. Charrier, F. Hanke, C. Reichmuth, P. Cook, Z. Schakner, D. 
Blumstein, E. DeRango, J. Schwarz, C. Llamazares-Martín, E. Palagi, D. Rodríguez, G. 
Giardino, M. Mandiola, J. Gana, M. De León, J. Bastida, S. Morón, P. Denuncio, R. 
Bastida, E. Miller, A. Kochnev, A. Trites, E. Crespo, B. L. Chilvers, R. McIntosh, B. 
Pitcher, R. Kirkwood, F. Elorriaga-Verplancken, K. Acevedo-Whitehouse, T. Norris, A. 
Gutiérrez-Gutiérrez, M. José Amador-Capitanachi, A. Juárez-Ruiz, J. Sandoval-Sierra, C. 
Gálvez, X. Moreno-Sánchez, D. Naya, P. Inchausti, M. Riofrío-Lazo, and D. Páez-Rosas. 
2021. Ethology and Behavioral Ecology of Otariids and the Odobenid. 667 pp. 
Churchill, M., R. W. Boessenecker, and M. T. Clementz. 2014. Colonization of the Southern 
Hemisphere by fur seals and sea lions (Carnivora: Otariidae) revealed by combined 
evidence phylogenetic and Bayesian biogeographical analysis. Zoological Journal of the 
Linnean Society 172:200–225. 
De Muizon, C. 1982. Phocid phylogeny and dispersal. Annals of the South African Museum 89: 
175–213. 
Deméré, T. A. 1994. Two new species of fossil walruses (Pinnipedia: Odobenidae) from the 
Upper Pliocene San Diego Formation, California. Proceedings of the San Diego Society 
of Natural History 29:77–98. 
Deméré, T. A., and A. Berta. 2002. The Miocene pinniped Desmatophoca oregonensis Condon, 
1906 (Mammalia: Carnivora), from the Astoria Formation, Oregon. Smithsonian 
Contributions to Paleobiology 93:113–147. 
Dewaele, L., O. Lambert, and S. Louwye. 2017a. On Prophoca and Leptophoca (Pinnipedia, 
Phocidae) from the Miocene of the North Atlantic realm: Redescription, phylogenetic 
affinities and paleobiogeographic implications. PeerJ 2017:1–56. 
165 
 
Dewaele, L., E. Amson, O. Lambert, and S. Louwye. 2017b. Reappraisal of the extinct seal 
“Phoca” vitulinoides from the Neogene of the North Sea Basin, with bearing on its 
geological age, phylogenetic affinities, and locomotion. PeerJ 5:e3316:1–79. 
Domning, D. P. 2002. The terrestrial posture of desmostylians; pp. 99–111 in R. J. Emry (ed.), 
Cenozoic Mammals of Land and Sea: Tributes to the Career of Clayon E. Ray. 
Smithsonian Contributions to Paleobiology 93. 
English, A. W. 1976. Limb movements and locomotor function in the California sea lion 
(Zalophus californianus). Journal of Zoology 178:341–364. 
English, A. W. M. 1977. Structural correlates of forelimb function in fur seals and sea lions. 
Journal of Morphology 151:325–352. 
Fabre, A. C., A. Goswami, S. Peigné, and R. Cornette. 2014. Morphological integration in the 
forelimb of musteloid carnivorans. Journal of Anatomy 225:19–30. 
Fabre, A. C., R. Cornette, A. Goswami, and S. Peigné. 2015. Do constraints associated with the 
locomotor habitat drive the evolution of forelimb shape? A case study in musteloid 
carnivorans. Journal of Anatomy 226:596–610. 
Fay, F. H. 1982. Ecology and biology of the Pacific walrus, Odobenus rosmarus divergens Illiger. 
North American Fauna 74:1–279. 
Feldkamp, S. D. 1987a. Foreflipper propulsion in the California sea lion, Zalophus californianus. 
Journal of Zoology 212:43–57. 
Feldkamp, S. D. 1987b. Swimming in the California sea lion: morphometrics, drag and 
energetics. The Journal of Experimental Biology 131:117–135. 
Finarelli, J. A. 2008. A total evidence phylogeny of the Arctoidea (Carnivora: Mammalia): 
Relationships among basal taxa. Journal of Mammalian Evolution 15:231–259. 
Fish, F. E. 1994. Associatin of propulsive swimming mode with behavior in river otters (Lutra 
canadensis). Journal of Mammalogy 75:989–997. 
Fish, F. E. 1996. Transitions from drag-based to lift-based propulsion in mammalian swimming. 
American Zoologist 36:628–641. 
Fish, F. E. 2016. Secondary evolution of aquatic propulsion in higher vertebrates: validation and 
prospect. Integrative and Comparative Biology 56:1285–1297. 
Fulton, T. L., C. Strobeck, and A. Crame. 2010. Multiple fossil calibrations, nuclear loci and 
mitochondrial genomes provide new insight into biogeography and divergence timing for 
true seals (Phocidae, Pinnipedia). Journal of Biogeography 37:814–829. 
Furbish, R. 2015. Something old, something new, something swimming in the blue: an analysis 
of the pinniped family Desmatophocidae, its phylogenetic position and swimming mode. 
75 pp. 
166 
 
Giffin, E. B. 1992. Functional implications of neural canal anatomy in recent and fossil marine 
carnivores. Journal of Morphology 214:357–374. 
Gingerich, P. D. 2003. Land-to-sea transition in early whales: evolution of Eocene Archaeoceti 
(Cetacea) in relation to skeletal proportions and locomotion of living semiaquatic 
mammals. Paleobiology 29:429–454. 
Gingerich, P. D. 2005. Aquatic adaptation and swimming mode inferred from skeletal 
proportions in the Miocene desmostylian Desmostylus. Journal of Mammalian Evolution 
12:183–194. 
Gordon, K. R. 1979. The analysis of locomotor behavior and functional morphology of the limbs 
of the walrus. University of California Press, Davis, California.  
Gordon, K. R. 1981. Locomotor behaviour of the walrus. Journal of Zoology 195:349–367. 
Gordon, K. R. 1983. Mechanics of the limbs of the walrus (Odobenus rosmarus) and the 
California sea lion (Zalophus californianus). Journal of Morphology 175:73–90. 
Gray, J. E. 1821. On the natural arrangement of vertebrose animals. The London Medical 
Repository Monthly Journal and Review 15:296–310. 
Gray, J. E. 1825. An outline of an attempt at the disposition of Mammalia into tribes and 
families, with a list of genera apparently appertaining to each tribe. Annals of 
Philosophy: New Series 10:337–344. 
Gutarra, S., and I. A. Rahman. 2022. The locomotion of extinct secondarily aquatic tetrapods. 
Biological Reviews 97:67–98. 
Hastie, T., R. Tibshirani, and A. Buja. 1994. Flexible discriminant analysis by optimal scoring. 
Journal of the American Statistical Association 89:1255. 
Hay, O. P. 1930. Second bibliography and catalogue of the fossil Vertebrata of North America. 
Carnegie Institute of Washington Publication 390:1–1074. 
Helle, E., H. Hyvärinen, and T. Sipilä. 1984. Breeding habitat and lair structure of the Saimaa 
ringed seal Phoca hispida saimensis in Finland. Acta Zoologica Fennica, 172:125–127. 
Higdon, J. W., O. R. P. Bininda-Emonds, R. M. D. Beck, and S. H. Ferguson. 2007. Phylogeny 
and divergence of the pinnipeds (Carnivora: Mammalia) assessed using a multigene 
dataset. BMC Evolutionary Biology 7. 
Hocking, D. P., F. G. Marx, S. Wang, D. Burton, M. Thompson, T. Park, B. Burville, H. L. 
Richards, R. Sattler, J. Robbins, R. P. Miguez, E. M. G. Fitzgerald, D. J. Slip, and A. R. 
Evans. 2021. Convergent evolution of forelimb-propelled swimming in seals. Current 
Biology 31:2404-2409.e2. 
Hocking, D. P., M. A. Ladds, D. J. Slip, E. M. G. Fitzgerald, and A. R. Evans. 2017. Chew, 
shake, and tear: prey processing in Australian sea lions (Neophoca cinerea). Marine 
Mammal Science 33:541–557. 
167 
 
Hocking, D. P., B. Burville, W. M. G. Parker, A. R. Evans, T. Park, and F. G. Marx. 2020. 
Percussive underwater signaling in wild gray seals. Marine Mammal Science 36:728–
732. 
Hocking, D. P., F. G. Marx, R. Sattler, R. N. Harris, T. I. Pollock, K. J. Sorrell, E. M. G. 
Fitzgerald, M. R. McCurry, and A. R. Evans. 2018. Clawed forelimbs allow northern 
seals to eat like their ancient ancestors. Royal Society Open Science 5. 
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 73:1–
142. 
Illiger, C. 1811. Prodromus systematis mammalium et avium. Salfeld, Berlin, 304 pp. 
Jones, B., A. Martín-Serra, E. J. Rayfield, and C. M. Janis. 2022. Distal humeral morphology 
indicates locomotory divergence in extinct giant kangaroos. Journal of Mammalian 
Evolution 29:27–41. 
Kellogg, A. R. 1922. Pinnipeds from Miocene and Pleistocene deposits of California. Bulletin of 
the Department of Geological Sciences, 13(4):23–132. 
King, J. E. 1983. Seals of the World. Cornell University Press, Ithaca, New York, 240. pp.  
Koretsky, I. A., and P. Holec. 2002. A primitive seal (Mammalia: Phocidae) from the Early 
Middle Miocene of Central Paratethys. Smithsonian Contributions to Paleobiology 
93:163–178. 
Koretsky, I. A., L. G. Barnes, and S. J. Rahmat. 2016. Re-evaluation of morphological characters 
questions current views of pinniped origins. Vestnik Zoologii 50:327–354. 
Kuhn, C., and E. Frey. 2012. Walking like caterpillars, flying like bats--pinniped locomotion. 
Palaeobiodiversity and Palaeoenvironments 92:197–210. 
Laakkonen, J., and H. Nihtilä. 2021. Comparison of the musculoskeletal forelimb anatomy of the 
Saimaa (Pusa hispida saimensis) and Baltic ringed seals (Pusa hispida botnica). Marine 
Mammal Science 37:1552–1563. 
Linnaeus, C. 1758. Systema Naturae per Regna Tria Naturae, secundum Classes, Ordines, 
Genera, Species, cum Characteribus, Differentiis, Synonymis, Locis. Tomus I. Laurentii 
Salvii, Holmiae, Stockholm, 824 pp 
Lone, K., K. M. Kovacs, C. Lydersen, M. Fedak, M. Andersen, P. Lovell, and J. Aars. 2018. 
Aquatic behaviour of polar bears (Ursus maritimus) in an increasingly ice-free Arctic. 
Scientific Reports 8:1–12. 
Luan, P., O. A. Ryder, H. Davis, Y. Zhang, and L. Yu. 2013. Incorporating indels as phylogenetic 
characters: Impact for interfamilial relationships within Arctoidea (Mammalia: 
Carnivora). Molecular Phylogenetics and Evolution 66:748–756. 
168 
 
Lydersen, C., and I. Gjertz. 1986. Studies of the ringed seal (Phoca hispida Schreber 1775) in its 
breeding habitat in Kongsfjorden, Svalbard. Polar Research 4:57–63. 
MacLaren, J. A., and S. Nauwelaerts. 2016. A three-dimensional morphometric analysis of upper 
forelimb morphology in the enigmatic tapir (Perissodactyla: Tapirus) hints at subtle 
variations in locomotor ecology. Journal of Morphology 277:1469–1485. 
Magallanes, I., J. F. Parham, G. Santos, and J. Velez-Juarbe. 2018. A new tuskless walrus from 
the Miocene of Orange County, California, with comments on the diversity and taxonomy 
of odobenids. PeerJ 6:e5708:1–52. 
Marramà, G., and J. Kriwet. 2017. Principal component and discriminant analyses as powerful 
tools to support taxonomic identification and their use for functional and phylogenetic 
signal detection of isolated fossil shark teeth. PLoS ONE 12:1–22. 
Milne, N., S. F. Vizcaíno, and J. C. Fernicola. 2009. A 3D geometric morphometric analysis of 
digging ability in the extant and fossil cingulate humerus. Journal of Zoology 278:48–56. 
Mitchell, E. 1966. The Miocene pinniped Allodesmus. University of California Publications in 
Geological Sciences 61:1–105. 
Mitchell, E. 1968. The Mio-Pliocene pinniped Imagotaria. Journal of Fisheries Research Board 
of Canada 25:1843–1900. 
Mongiardino Koch, N., R. J. Garwood, and L. A. Parry. 2021. Fossils improve phylogenetic 
analyses of morphological characters. Proceedings of the Royal Society B: Biological 
Sciences 288. 
Morales-García, N. M., P. G. Gill, C. M. Janis, and E. J. Rayfield. 2021. Jaw shape and 
mechanical advantage are indicative of diet in Mesozoic mammals. Communications 
Biology 4:1–14. 
Mori, K., S. Suzuki, D. Koyabu, J. Kimura, S. Y. Han, and H. Endo. 2015. Comparative 
functional anatomy of hindlimb muscles and bones with reference to aquatic adaptation 
of the sea otter. Journal of Veterinary Medical Science 77:571–578. 
Motani, R., and L. Schmitz. 2011. Phylogenetic versus functional signals in the evolution of 
form-function relationships in terrestrial vision. Evolution 65:2245–2257. 
Muir, S. F., D. K. A. Barnes, and K. Reid. 2006. Interactions between humans and leopard seals. 
Antarctic Science 18:61–74. 
Muñoz, N. A. 2021. Locomotion in rodents and small carnivorans: are they so different? Journal 
of Mammalian Evolution 28:87–98. 
Murie, J. 1870. Researches upon the anatomy of the Pinnipedia--Part I. On the walrus 
(Trichechus rosmarus, Linn.). The Transactions of the Zoological Society of London 
7:411–463. 
169 
 
Nyakatura, K., and O. R. P. Bininda-Emonds. 2012. Updating the evolutionary history of 
Carnivora (Mammalia): A new species-level supertree complete with divergence time 
estimates. BMC Biology 10. 
O’Reilly, J. E., M. N. Puttick, D. Pisani, and P. C. J. Donoghue. 2018. Probabilistic methods 
surpass parsimony when assessing clade support in phylogenetic analyses of discrete 
morphological data. Palaeontology 61:105–118. 
Pagel, M. 1999. Inferring the historical patterns of biological evolution. Nature 401:877–884. 
Paterson, R. 2017. Evidence for independent acquisition of aquatic specializations in pinnipeds 
(seals, sea lions, and walruses): insights from study of the phylogenetic position, 
locomotor behaviour and description of the stem pinniped, Puijila darwini. 
Paterson, R. S., N. Rybczynski, N. Kohno, and H. C. Maddin. 2020. A total evidence 
phylogenetic analysis of pinniped phylogeny and the possibility of parallel evolution 
within a monophyletic framework. Frontiers in Ecology and Evolution 7:1–16. 
Puttick, M. N., J. E. O’Reilly, D. Pisani, and P. C. J. Donoghue. 2019. Probabilistic methods 
outperform parsimony in the phylogenetic analysis of data simulated without a 
probabilistic model. Palaeontology 62:1–17. 
R Core Team. 2021. R: a language and environment for statistical computing. R Foundation for 
Statistical Computing. Available at www.R- project.org. Accessed December 19, 2022. 
R Project Team. 2017. mda: Mixture and flexible discriminant analysis. v.0.5-2.  
Ray, G. C. 1963. Locomotion in pinnipeds. Natural History. 72:10–21. 
Revell, L. J. 2012. phytools: an R package for phylogenetic comparative biology (and other 
things). Methods in Ecology and Evolution 3:217–223. 
Richards, H. L., D. S. Rovinsky, J. W. Adams, and A. R. Evans. 2022. Inferring the 
palaeobiology of palorchestid marsupials through analysis of mammalian humeral and 
femoral shape. Journal of Mammalian Evolution. 
Rohlf, F. J., and D. Slice. 1990. Extensions of the Procrustes method for the optimal 
superimposition of landmarks. Systematic Zoology 39:40–59. 
Rule, J. P., J. W. Adams, F. G. Marx, A. R. Evans, A. J. D. Tennyson, R. P. Scofield, and E. M. G. 
Fitzgerald. 2020. First monk seal from the Southern Hemisphere rewrites the 
evolutionary history of true seals. Proceedings. Biological Sciences 287:20202318. 
Samuels, J. X., J. A. Meachen, and S. A. Sakai. 2013. Postcranial morphology and the locomotor 
habits of living and extinct carnivorans. Journal of Morphology 274:121–146. 
Savage, R. J. G. 1957. The anatomy of Potamotherium an Oligocene lutrine. Journal of Zoology 
129:151–244. 
170 
 
Schutz, H., and R. P. Guralnick. 2007. Postcranial element shape and function: assessing 
locomotor mode in extant and extinct mustelid carnivorans. Zoological Journal of the 
Linnean Society 150:895–914. 
Simpson, G. G. 1945. The principles of classification and a classification of mammals. American 
Museum of Natural History, 85. 
Smith, M. R. 2019. Bayesian and parsimony approaches reconstruct informative trees from 
simulated morphological datasets. Biology Letters 15. 
Stirling, I. 1977. Adaptations of Weddell and ringed seals to exploit the polar fast ice habitat in 
the absence or presence of surface predators. In G. A. Llano (ed.), Adaptations within 
Antarctic ecosystems (pp. 741–748). Proceedings of the 3rd Scientific Committee on 
Antarctic Research Symposium on Antarctic Biology, Washington, DC. 
Tanaka, Y., and N. Kohno. 2015. A new late Miocene odobenid (Mammalia: Carnivora) from 
Hokkaido, Japan suggests rapid diversification of basal Miocene odobenids. PLoS ONE 
10. 
Tarasoff, F. J., A. Bisaillon, J. Piérard, and A. P. Whitt. 2008. Locomotory patterns and external 
morphology of the river otter, sea otter, and harp seal (Mammalia). Canadian Journal of 
Zoology 50:915–929. 
Tate-Jones, M. K., C. M. Peredo, C. D. Marshall, and S. S. B. Hopkins. 2020. The dawn of 
Desmatophocidae: a new species of basal desmatophocid seal (Mammalia, Carnivora) 
from the Miocene of Oregon, U.S.A. Journal of Vertebrate Paleontology 40:1–11. 
Tonomori, W., H. Sawamura, T. Sato, and N. Kohno. 2018. A new Miocene pinniped Allodesmus 
(Mammalia: Carnivora) from Hokkaido, Northern Japan. Royal Society Open Science 5. 
Valenzuela-Toro, A., N. D. Pyenson, and P. Koch. 2018. How good is the fossil record of 
pinnipeds? A historiographic evaluation of its biases and modes. Society of Vertebrate 
Paleontology Program and Abstract Book. 
Velez-Juarbe, J. 2017. Eotaria citrica, sp. nov., a new stem otariid from the “Topanga” formation 
of Southern California. PeerJ 2017:1–25. 
Velez-Juarbe, J. 2018. New data on the early odobenid Neotherium mirum Kellogg, 1931, and 
other pinniped remains from the Sharktooth Hill Bonebed, California. Journal of 
Vertebrate Paleontology 38:1–14. 
Wickham, H. 2016. ggplot2: elegant graphics for data analysis. Spring-Verlag, New York, New 
York. 
Wyss, A. R. 1988. Evidence from flipper structure for a single origin of pinnipeds. Nature 
334:427–428. 
Zelditch, M. L., D. L. Swiderski, and H. D. Sheets. 2004. Geometric morphometrics for 
biologicals: a primer. Elsevier Academic Press, London, United Kingdom. 
171 
 
Chapter IV 
Aiello-Lammens, M.E., Boria, R.A., Radosavljevic, A., Vilela, B. & Anderson, R.P. (2015) 
spThin: An R package for spatial thinning of species occurrence records for use in 
ecological niche models. Ecography 38, 541–545. https://doi.org/10.1111/ecog.01132 
Alonso-Garcia, M., Sierro, F.J., Kucera, M., Flores, J.A., Cacho, I. & Andersen, N. (2011) Ocean 
circulation, ice sheet growth and interhemispheric coupling of millennial climate 
variability during the mid-Pleistocene (ca 800-400ka). Quaternary Science Reviews 30, 
3234–3247. https://doi.org/10.1016/j.quascirev.2011.08.005 
Amiraux, R., Rontani, J.-F., Armougom, F., Frouin, E., Babin, M., Artigue, L. & Bonin, P. (2021) 
Bacterial diversity and lipid biomarkers in sea ice and sinking particulate organic material 
during the melt season in the Canadian Arctic. Elementa: Science of the Anthropocene 9. 
https://doi.org/10.1525/elementa.2019.040 
Anderson, M.J. (2001) A new method for non-parametric multivariate analysis of variance. 
Austral Ecology 26, 32–46. https://doi.org/10.1046/j.1442-9993.2001.01070.x 
Anderson, M.J. (2006) Distance-based tests for homogeneity of multivariate dispersions. 
Biometrics 62, 245–253. https://doi.org/10.1111/j.1541-0420.2005.00440.x 
Anderson, M.J. & Walsh, D.C.I. (2013) PERMANOVA, ANOSIM, and the Mantel test in the 
face of heterogeneous dispersions: What null hypothesis are you testing? Ecological 
Monographs 83, 557–574. https://doi.org/10.1890/12-2010.1 
Årthun, M., Onarheim, I.H., Dörr, J. & Eldevik, T. (2021) The seasonal and regional transition to 
an ice-free Arctic. Geophysical Research Letters 48. 
https://doi.org/10.1029/2020GL090825 
Assis, J., Tyberghein, L., Bosch, S., Verbruggen, H., Serrão, E.A. & De Clerck, O. (2018) Bio-
ORACLE v2.0: Extending marine data layers for bioclimatic modelling. Global Ecology 
and Biogeography 27, 277–284. https://doi.org/10.1111/geb.12693 
Batchelor, C.L., Margold, M., Krapp, M., Murton, D.K., Dalton, A.S., Gibbard, P.L., Stokes, 
C.R., Murton, J.B. & Manica, A. (2019) The configuration of Northern Hemisphere ice 
sheets through the Quaternary. Nature Communications 10, 1–10. 
https://doi.org/10.1038/s41467-019-11601-2 
Beatty, W.S., Jay, C. V., Fischbach, A.S., Grebmeier, J.M., Taylor, R.L., Blanchard, A.L. & 
Jewett, S.C. (2016) Space use of a dominant Arctic vertebrate: Effects of prey, sea ice, 
and land on Pacific walrus resource selection. Biological Conservation 203, 25–32. 
https://doi.org/10.1016/j.biocon.2016.08.035 
Berends, C.J., De Boer, B. & Van De Wal, R.S.W. (2021) Reconstructing the evolution of ice 
sheets, sea level, and atmospheric CO2during the past 3.6 million years. Climate of the 
Past 17, 361–377. https://doi.org/10.5194/cp-17-361-2021 
172 
 
Bloxam, Q.M.C. & Tonge, S.J. (1995) Amphibians: suitable candidates for breeding-release 
programmes. Review of Industrial Organization 4, 636–644. 
https://doi.org/10.1007/BF00222519 
Boessenecker, R.W. & Churchill, M. (2021) The surprising evolutionary heritage of the Atlantic 
walrus as chronicled by the fossil record. INC. https://doi.org/10.1016/b978-0-12-
817430-2.00006-6 
Bonan, D.B., Lehner, F. & Holland, M.M. (2021) Partitioning uncertainty in projections of Arctic 
sea ice. Environmental Research Letters 16. https://doi.org/10.1088/1748-9326/abe0ec 
Boria, R.A., Olson, L.E., Goodman, S.M. & Anderson, R.P. (2014) Spatial filtering to reduce 
sampling bias can improve the performance of ecological niche models. Ecological 
Modelling 275, 73–77. https://doi.org/10.1016/j.ecolmodel.2013.12.012 
Borisiak, A.A. (1930) A fossil walrus from the Okhotsk coast. Ezhegodnik Russian Paleontology 
8. 
Born, E.W., Acquarone, M., Knutsen, L.Ø. & Toudal, L. (2005) Homing behaviour in an Atlantic 
walrus (Odobenus rosmarus rosmarus). Aquatic Mammals 31, 23–33. 
https://doi.org/10.1578/am.31.1.2005.23 
Born, E.W., Gjertz, I. & Reeves, R.R. (1995) Population assessment of Atlantic walrus 
(Odobenus rosmarus rosmarus L.). 
Bosch, S., Tyberghein, L. & De Clerck, O. (2017) sdmpredictors: an R package for species 
distribution modelling predictor datasets. 
Campagna, C., R. Harcourt, R. Gentry, D. Costa, A. Valenzuela-Toro, M. Sepúlveda, T. 
Jeanniard-du-Dot, C. Guinet, D. Hocking, T. Park, J. Rule, F. Marx, M. Cassini, M. 
Kiyota, B. Le Boeuf, C. Bisioli, S. Ryazanov, C. Bonin, V. Franco-Trecu, J. Kingston 
Sinclair, S. Goldsworthy, I. Charrier, F. Hanke, C. Reichmuth, P. Cook, Z. Schakner, D. 
Blumstein, E. DeRango, J. Schwarz, C. Llamazares-Martín, E. Palagi, D. Rodríguez, G. 
Giardino, M. Mandiola, J. Gana, M. De León, J. Bastida, S. Morón, P. Denuncio, R. 
Bastida, E. Miller, A. Kochnev, A. Trites, E. Crespo, B. L. Chilvers, R. McIntosh, B. 
Pitcher, R. Kirkwood, F. Elorriaga-Verplancken, K. Acevedo-Whitehouse, T. Norris, A. 
Gutiérrez-Gutiérrez, M. José Amador-Capitanachi, A. Juárez-Ruiz, J. Sandoval-Sierra, C. 
Gálvez, X. Moreno-Sánchez, D. Naya, P. Inchausti, M. Riofrío-Lazo, and D. Páez-Rosas. 
2021. Ethology and Behavioral Ecology of Otariids and the Odobenid. 667 pp. 
Clarke, A. & Peck, L.S. (1991) The physiology of polar marine zooplankton. Polar Research 10, 
355–370. https://doi.org/10.3402/polar.v10i2.6752 
Davis, E.B., Mcguire, J.L. & Orcutt, J.D. (2014) Ecological niche models of mammalian glacial 
refugia show consistent bias. Ecography 37, 1133–1138. 
https://doi.org/10.1111/ecog.01294 
173 
 
Dennis, R.L.H. & Thomas, C.D. (2000) Bias in butterfly distribution maps the influence of hot 
spots and recorder’s home range. Journal of Insect Conservation 4, 73–77. 
Diebold, F.X. & Rudebusch, G.D. (2021) Probability assessments of an ice-free Arctic: 
comparing statistical and climate model projections. Journal of Econometrics. 
https://doi.org/10.1016/j.jeconom.2020.12.007 
Dietl, G.P., Kidwell, S.M., Brenner, M., Burney, D.A., Flessa, K.W., Jackson, S.T. & Koch, P.L. 
(2015) Conservation paleobiology: leveraging knowledge of the past to inform 
conservation and restoration. Annual Review of Earth and Planetary Sciences 43, 79–103. 
https://doi.org/10.1146/annurev-earth-040610-133349 
Divoky, G.J., Lukacs, P.M. & Druckenmiller, M.L. (2015) Effects of recent decreases in Arctic 
sea ice on an ice-associated marine bird. Progress in Oceanography 136, 151–161. 
https://doi.org/10.1016/j.pocean.2015.05.010 
Van Dolah, R.F., Hyland, J.L., Holland, A.F., Rosen, J.S. & Snoots, T.R. (1999) A benthic index 
of biological integrity for assessing habitat quality in estuaries of the southeastern USA. 
Marine Environmental Research 48, 269–283. https://doi.org/10.1016/S0141-
1136(99)00056-2 
Dormann, C.F., Elith, J., Bacher, S., et al. (2013) Collinearity: a review of methods to deal with it 
and a simulation study evaluating their performance. Ecography 36, 27–46. 
https://doi.org/10.1111/j.1600-0587.2012.07348.x 
Elith, J., Graham, C.H., Anderson, R.P., et al. (2006) Novel methods improve prediction of 
species’ distributions from occurrence data. Ecography 29, 129–151. 
Elith, J., Kearney, M. & Phillips, S. (2010) The art of modelling range-shifting species. Methods 
in Ecology and Evolution 1, 330–342. https://doi.org/10.1111/j.2041-210x.2010.00036.x 
Fay, F.H. (1982) Ecology and biology of the Pacific walrus, Odobenus rosmarus divergens 
Illiger. North American Fauna 74, 1–279. https://doi.org/10.3996/nafa.74.0001 
Fay, F.H. (1985) Odobenus rosmarus. Mammalian Species, 1–7. 
Feng, X. & Papeş, M. (2017) Physiological limits in an ecological niche modeling framework: A 
case study of water temperature and salinity constraints of freshwater bivalves invasive in 
USA. Ecological Modelling 346, 48–57. https://doi.org/10.1016/j.ecolmodel.2016.11.008 
Fischbach, A.S., Monson, D.H., Jay, C. V & Interior, U.S.D. of the (2009) Enumeration of 
Pacific walrus carcasses on beaches of the Chukchi Sea in Alaska following a mortality 
event, September 2009. Open-File Report, 10. 
Fourcade, Y., Engler, J.O., Rödder, D. & Secondi, J. (2014) Mapping species distributions with 
MAXENT using a geographically biased sample of presence data: A performance 
assessment of methods for correcting sampling bias. PLoS ONE 9, 1–13. 
https://doi.org/10.1371/journal.pone.0097122 
174 
 
García-Roselló, E., Guisande, C., Manjarrés-Hernández, A., et al. (2015) Can we derive 
macroecological patterns from primary Global Biodiversity Information Facility data? 
Global Ecology and Biogeography 24, 335–347. https://doi.org/10.1111/geb.12260 
Garde, E., Jung-Madsen, S., Ditlevsen, S., Hansen, R.G., Zinglersen, K.B. & Heide-Jørgensen, 
M.P. (2018) Diving behavior of the Atlantic walrus in high Arctic Greenland and Canada. 
Journal of Experimental Marine Biology and Ecology 500, 89–99. 
https://doi.org/10.1016/j.jembe.2017.12.009 
Garlich-Miller, J., MacCracken, J.G., Snyder, J., Meehan, R., Myers, M., Wilder, J.M., Lance, E. 
& Matz, A. (2011) Status review of the Pacific walrus (Odobenus rosmarus divergens). 
GBIF.org (30 July 2021) GBIF Occurrence Download 
https://www.gbif.org/occurrence/download/0334982-200613084148143 
Gebruk, A., Mikhaylyukova, P., Mardashova, M., Semenova, V., Henry, L.A., Shabalin, N., 
Narayanaswamy, B.E. & Mokievsky, V. (2021) Integrated study of benthic foraging 
resources for Atlantic walrus (Odobenus rosmarus rosmarus) in the Pechora Sea, south-
eastern Barents Sea. Aquatic Conservation: Marine and Freshwater Ecosystems 31, 112–
125. https://doi.org/10.1002/aqc.3418 
Gjertz, I., Henriksen, G., Øritsland, T. & Wiig, Ø. (1993) Observations of walruses along the 
Norwegian coast 1967–1992. Polar Research 12, 27–31. 
https://doi.org/10.3402/polar.v12i1.6700 
Glazov, D.M., Shpak, O. V., Kuznetsova, D.M., Solovyev, B.A., Udovik, D.A., Platonov, N.G., 
Mordvintsev, I.N., Ivanov, D.I. & Rozhnov, V. V. (2013) Observations of the walrus 
(Odobenus rosmarus) in the Barents, Kara, and Laptev seas in 2010-2012. Biology 
Bulletin 40, 783–789. https://doi.org/10.1134/S1062359013090057 
Goertz, C.E.C., Polasek, L., Burek, K., Suydam, R. & Sformo, T. (2017) Demography and 
pathology of a Pacific walrus (Odobenus rosmarus divergens) mass-mortality event at Icy 
Cape, Alaska, September 2009. Polar Biology 40, 989–996. 
https://doi.org/10.1007/s00300-016-2023-x 
Green, R.H. (1973) Growth and mortality in an Arctic intertidal population of Macoma balthica 
(Pelecypoda, Tellinidae). Journal of the Fisheries Research Board of Canada 30, 1345–
1348. https://doi.org/10.1139/f73-215 
De Groeve, J., Kusumoto, B., Koene, E., et al. (2022) Global raster dataset on historical coastline 
positions and shelf sea extents since the Last Glacial Maximum. Global Ecology and 
Biogeography 31, 2162–2171. https://doi.org/10.1111/geb.13573 
Gross, M. (2018) Arctic shipping threatens wildlife. Current Biology 28, R803–R805. 
https://doi.org/10.1016/j.cub.2018.07.053 
175 
 
Guarino, M.V., Sime, L.C., Schröeder, D., et al. (2020) Sea-ice-free Arctic during the Last 
Interglacial supports fast future loss. Nature Climate Change 10, 928–932. 
https://doi.org/10.1038/s41558-020-0865-2 
Harington, C.R. (2008) The evolution of Arctic marine mammals. Ecological Applications 18, 
23–40. https://doi.org/10.1890/06-0624.1 
Hernandez, P.A., Graham, C.H., Master, L.L. & Albert, D.L. (2006) The effect of sample size 
and species characteristics on performance of different species distribution modeling 
methods. Ecography 29, 773–785. https://doi.org/10.1111/j.0906-7590.2006.04700.x 
Higdon, J.W. & Stewart, D.B. (2018) State of circumpolar walrus (Odobenus rosmarus) 
populations. 
Hijmans, R.J. (2020) raster: geographic data analysis and modeling. R package version 3.3-7. 
https://CRAN.R-project.org/package=raster 
Hijmans, A.R.J., Phillips, S., Leathwick, J., Elith, J. & Hijmans, M.R.J. (2017) Package ‘dismo.’ 
Circles 9, 1–68. 
Hocking, D.P., Park, T., Rule, J.P. & Marx, F.G. (2021) Prey Capture and processing in fur seals, 
sea lions and the walrus. In: Ethology and Behavioral Ecology of Marine Mammals. (ed. 
by C. Campagna and R. Harcourt), pp. 101–121, Springer Nature Switzerland AG. 
https://doi.org/10.1007/978-3-030-59184-7_5 
Huntington, H.P., Danielson, S.L., Wiese, F.K., et al. (2020) Evidence suggests potential 
transformation of the Pacific Arctic ecosystem is underway. Nature Climate Change 10, 
342–348. https://doi.org/10.1038/s41558-020-0695-2 
Jacobs, P., Lenssen, N., Schmidt, G. & Rohde, R. (2021) The Arctic is now warming four times 
as fast as the rest of the globe. In: AGU Fall Meeting. 
Jahn, A., Kay, J.E., Holland, M.M. & Hall, D.M. (2016) How predictable is the timing of a 
summer ice-free Arctic? Geophysical Research Letters 43, 9113–9120. 
https://doi.org/10.1002/2016GL070067 
Jarvie, S., Worthy, T.H., Saltré, F., Scofield, R.P., Seddon, P.J. & Cree, A. (2021) Using Holocene 
fossils to model the future: Distribution of climate suitability for tuatara, the last 
rhynchocephalian. Journal of Biogeography 48, 1489–1502. 
https://doi.org/10.1111/jbi.14092 
Jay, C. V. & Fischbach, A.S. (2010) Pacific walrus response to Arctic sea ice losses. 
Jay, C. V., Fischbach, A.S. & Kochnev, A.A. (2012) Walrus areas of use in the Chukchi Sea 
during sparse sea ice cover. Marine Ecology Progress Series 468, 1–13. 
https://doi.org/10.3354/meps10057 
176 
 
Jay, C. V., Marcot, B.G. & Douglas, D.C. (2011) Projected status of the Pacific walrus 
(Odobenus rosmarus divergens) in the twenty-first century. Polar Biology 34, 1065–1084. 
https://doi.org/10.1007/s00300-011-0967-4 
Jay, C. V., Taylor, R.L., Fischbach, A.S., Udevitz, M.S. & Beatty, W.S. (2017) Walrus haul-out 
and in water activity levels relative to sea ice availability in the Chukchi Sea. Journal of 
Mammalogy 98, 386–396. https://doi.org/10.1093/jmammal/gyw195 
Jefferson, G.T. (1991) A catalogue of late Quaternary vertebrates from California: Part one, 
nonmarine lower vertebrate and avian taxa. Natural History Museum of Los Angeles 
County: Technical Reports 5, 60. 
Kadmon, R., Farber, O. & Danin, A. (2004) Effect of roadside bias on the accuracy of predictive 
maps produced by bioclimatic models. Ecological Applications 14, 401–413. 
https://doi.org/10.1890/02-5364 
Kaky, E., Nolan, V., Alatawi, A. & Gilbert, F. (2020) A comparison between Ensemble and 
MaxEnt species distribution modelling approaches for conservation: a case study with 
Egyptian medicinal plants. Ecological Informatics 60, 101150. 
https://doi.org/10.1016/j.ecoinf.2020.101150 
Kass, J.M., Muscarella, R., Galante, P.J., Bohl, C.L., Pinilla-Buitrago, G.E., Boria, R.A., Soley-
Guardia, M. & Anderson, R.P. (2021) ENMeval 2.0: Redesigned for customizable and 
reproducible modeling of species’ niches and distributions. Methods in Ecology and 
Evolution 12, 1602–1608. https://doi.org/10.1111/2041-210X.13628 
Katz, T.S. & Zellmer, A.J. (2018) Comparison of model selection technique performance in 
predicting the spread of newly invasive species: a case study with Batrachochytrium 
salamandrivorans. Biological Invasions 20, 2107–2119. https://doi.org/10.1007/s10530-
018-1690-7 
Keighley, X., Pálsson, S., Einarsson, B.F., Petersen, A., Fernández-Coll, M., Jordan, P., Olsen, 
M.T. & Malmquist, H.J. (2019) Disappearance of Icelandic walruses coincided with 
Norse settlement. Molecular Biology and Evolution 36, 2656–2667. 
https://doi.org/10.1093/molbev/msz196 
Kovacs, K.M., Aars, J. & Lydersen, C. (2014) Walruses recovering after 60+ years of protection 
in Svalbard, Norway. Polar Research 33. https://doi.org/10.3402/polar.v33.26034 
Kramer-Schadt, S., Niedballa, J., Pilgrim, J.D., et al. (2013) The importance of correcting for 
sampling bias in MaxEnt species distribution models. Diversity and Distributions 19, 
1366–1379. https://doi.org/10.1111/ddi.12096 
Kryukova, N. V., Kruchenkova, E.P. & Ivanov, D.I. (2012) Killer whales (Orcinus orca) hunting 
for walruses (Odobenus rosmarus divergens) near Retkyn Spit, Chukotka. Biology 
Bulletin 39, 768–778. https://doi.org/10.1134/S106235901209004X 
177 
 
Laidre, K.L., Stern, H., Kovacs, K.M., et al. (2015) Arctic marine mammal population status, sea 
ice habitat loss, and conservation recommendations for the 21st century. Conservation 
Biology 29, 724–737. https://doi.org/10.1111/cobi.12474 
Legendre, P., & Legendre, L. (1998) Numerical Ecology. Elsevier. 
Lefebvre, K.A., Fachon, E., Bowers, E.K., et al. (2022) Paralytic shellfish toxins in Alaskan 
Arctic food webs during the anomalously warm ocean conditions of 2019 and estimated 
toxin doses to Pacific walruses and bowhead whales. Harmful Algae 114, 102205. 
https://doi.org/10.1016/j.hal.2022.102205 
Lima-Ribeiro, M.S., Moreno, A.K.M., Terribile, L.C., Caten, C.T., Loyola, R., Rangel, T.F. & 
Diniz-Filho, J.A.F. (2017) Fossil record improves biodiversity risk assessment under 
future climate change scenarios. Diversity and Distributions 23, 922–933. 
https://doi.org/10.1111/ddi.12575 
Lindqvist, C., Bachmann, L., Andersen, L.W., Born, E.W., Arnason, U., Kovacs, K.M., Lydersen, 
C., Abramov, A. V. & Wiig, Ø. (2009) The Laptev Sea walrus Odobenus rosmarus 
laptevi: An enigma revisited. Zoologica Scripta 38, 113–127. 
https://doi.org/10.1111/j.1463-6409.2008.00364.x 
Linnaeus, C. (1758) Systema Naturae per Regna Tria Naturae, secundum Classes, Ordines, 
Genera, Species, cum Characteribus, Differentiis, Synonymis, Locis. Tomus I. Laurentii 
Salvii, Holmiae, Stockholm, Sweden. 
Lisiecki, L.E. & Raymo, M.E. (2005) A Pliocene-Pleistocene stack of 57 globally distributed 
benthic δ 18O records. Paleoceanography 20, 1–17. 
https://doi.org/10.1029/2004PA001071 
Lowther, A.D., Kovacs, K.M., Griffiths, D. & Lydersen, C. (2015) Identification of motivational 
state in adult male Atlantic walruses inferred from changes in movement and diving 
behavior. Marine Mammal Science 31, 1291–1313. https://doi.org/10.1111/mms.12224 
Lydersen, C., Chernook, V.I., Glazov, D.M., Trukhanova, I.S. & Kovacs, K.M. (2012) Aerial 
survey of Atlantic walruses (Odobenus rosmarus rosmarus) in the Pechora Sea, August 
2011. Polar Biology 35, 1555–1562. https://doi.org/10.1007/s00300-012-1195-2 
MacCracken, J.G. (2012) Pacific walrus and climate change: observations and predictions. 
Ecology and Evolution 2, 2072–2090. https://doi.org/10.1002/ece3.317 
Maguire, K.C., Nieto-Lugilde, D., Fitzpatrick, M.C., Williams, J.W. & Blois, J.L. (2015) 
Modeling species and community responses to past, present, and future episodes of 
climatic and ecological change. Annual Review of Ecology, Evolution, and Systematics 
46, 343–368. https://doi.org/10.1146/annurev-ecolsys-112414-054441 
Mäkinen, J. & Vanhatalo, J. (2018) Hierarchical Bayesian model reveals the distributional shifts 
of Arctic marine mammals. Diversity and Distributions 24, 1381–1394. 
https://doi.org/10.1111/ddi.12776 
178 
 
McLeod, B.A., Frasier, T.R. & Lucas, Z. (2014) Assessment of the extirpated Maritimes walrus 
using morphological and ancient DNA analysis. PLoS ONE 9, 1–14. 
https://doi.org/10.1371/journal.pone.0099569 
Meehl, G.A., Chung, C.T.Y., Arblaster, J.M., Holland, M.M. & Bitz, C.M. (2018) Tropical 
decadal variability and the rate of Arctic sea ice decrease. Geophysical Research Letters 
45, 11,326-11,333. https://doi.org/10.1029/2018GL079989 
Metcalf, V. & Robards, M. (2008) Sustaining a healthy human-walrus relationship in a dynamic 
environment: Challenges for comanagement. Ecological Applications 18, 148–156. 
https://doi.org/10.1890/06-0642.1 
Moreno-Amat, E., Mateo, R.G., Nieto-Lugilde, D., Morueta-Holme, N., Svenning, J.C. & 
García-Amorena, I. (2015) Impact of model complexity on cross-temporal transferability 
in Maxent species distribution models: An assessment using paleobotanical data. 
Ecological Modelling 312, 308–317. https://doi.org/10.1016/j.ecolmodel.2015.05.035 
Muscarella, R., Galante, P.J., Soley-Guardia, M., Boria, R.A., Kass, J.M., Uriarte, M. & 
Anderson, R.P. (2014) ENMeval: An R package for conducting spatially independent 
evaluations and estimating optimal model complexity for Maxent ecological niche 
models. Methods in Ecology and Evolution 5, 1198–1205. https://doi.org/10.1111/2041-
210x.12261 
Myers, C.E., Stigall, A.L. & Lieberman, B.S. (2015) PaleoENM: Applying ecological niche 
modeling to the fossil record. Paleobiology 41, 226–244. 
https://doi.org/10.1017/pab.2014.19 
Naimi, B. (2015) usdm: Uncertainty analysis for species distribution models. R package version 
1.1. 
Notz, D. & Stroeve, J. (2018) The trajectory towards a seasonally ice-free Arctic Ocean. Current 
Climate Change Reports 4, 407–416. https://doi.org/10.1007/s40641-018-0113-2 
Ocean Biodiversity Information System (30 November 2021) www.iobis.org 
Oksanen, A.J., Blanchet, F.G., Friendly, M., et al. (2018) Package ‘vegan’ , 1–286. 
Øren, K., Kovacs, K.M., Yoccoz, N.G. & Lydersen, C. (2018) Assessing site-use and sources of 
disturbance at walrus haul-outs using monitoring cameras. Polar Biology 41, 1737–1750. 
https://doi.org/10.1007/s00300-018-2313-6 
Osman, M.B., Tierney, J.E., Zhu, J., Tardif, R., Hakim, G.J., King, J. & Poulsen, C.J. (2021) 
Globally resolved surface temperatures since the Last Glacial Maximum. Nature 599, 
239–244. https://doi.org/10.1038/s41586-021-03984-4 
Overland, J.E. & Wang, M. (2013) When will the summer Arctic be nearly sea ice free? 
Geophysical Research Letters 40, 2097–2101. https://doi.org/10.1002/grl.50316 
179 
 
Ovsyanikov, N.G., Menyushina, I.E., and Bezrukov, A.V. (2008) Collection of Scientific Papers 
from the Marine Mammals of the Holarctic V Conference, Odessa, Ukraine. 
Pearson, R.G., Raxworthy, C.J., Nakamura, M. & Townsend Peterson, A. (2007) Predicting 
species distributions from small numbers of occurrence records: a test case using cryptic 
geckos in Madagascar. Journal of Biogeography 34, 102–117. 
https://doi.org/10.1111/j.1365-2699.2006.01594.x 
Peters, S.E. & McClennen, M. (2015) The Paleobiology Database application programming 
interface. Paleobiology 42, 1–7. https://doi.org/10.1017/pab.2015.39 
Phillips, S.J., Anderson, R.P., Dudík, M., Schapire, R.E. & Blair, M.E. (2017) Opening the black 
box: an open-source release of Maxent. Ecography 40, 887–893. 
https://doi.org/10.1111/ecog.03049 
Phillips, S.J., Dudík, M. & Schapire, R.E. (2004) A maximum entropy approach to species 
distribution modeling. Proceedings, Twenty-First International Conference on Machine 
Learning, ICML 2004, 655–662. https://doi.org/10.1145/1015330.1015412 
Post, E., Bhatt, U.S., Bitz, C.M., Brodie, J.F., Fulton, T.L., Hebblewhite, M., Kerby, J., Kutz, 
S.J., Stirling, I. & Walker, D.A. (2013) Ecological consequences of sea-ice decline. 
Science 341, 519–525. 
Pourmozaffar, S., Tamadoni Jahromi, S., Rameshi, H., Sadeghi, A., Bagheri, T., Behzadi, S., 
Gozari, M., Zahedi, M.R. & Abrari Lazarjani, S. (2020) The role of salinity in 
physiological responses of bivalves. Reviews in Aquaculture 12, 1548–1566. 
https://doi.org/10.1111/raq.12397 
QGIS Development Team. (2022) QGIS Geographic Information System. Open Source 
Geospatial Foundation Project. http://qgis.osgeo.org 
R Core Team. (2021) R: a language and environment for statistical computing, R Foundation for 
Statistical Computing, Vienna, Austria. http://www.R-project.org/ 
Radosavljevic, A. & Anderson, R.P. (2014) Making better MAXENT models of species 
distributions: complexity, overfitting and evaluation. Journal of Biogeography 41, 629–
643. https://doi.org/10.1111/jbi.12227 
Rantanen, M., Karpechko, A.Y., Lipponen, A., Ruosteenoja, K., Vihma, T., Laaksonen, A., 
Nordling, K. & Hyvärinen, O. (2021) The Arctic has warmed nearly four times faster 
than the globe since 1979. Communication 3, 1–10. https://doi.org/10.1038/s43247-022-
00498-3 
Ray, G.C., Hufford, G.L., Overland, J.E., Krupnik, I., McCormick-Ray, J., Frey, K. & Labunski, 
E. (2016) Decadal Bering Sea seascape change: consequences for Pacific walruses and 
indigenous hunters. Ecological Applications 26, 24–41. https://doi.org/10.1890/15-0430 
180 
 
Ray, G.C., McCormick-Ray, J., Berg, P. & Epstein, H.E. (2006) Pacific walrus: benthic 
bioturbator of Beringia. Journal of Experimental Marine Biology and Ecology 330, 403–
419. https://doi.org/10.1016/j.jembe.2005.12.043 
Reeves, R. R., B. S. Stewart, and S. Leatherwood. 1992. The Sierra Club Handbook of Seals and 
Sirenians. Sierra Club Books, San Francisco, California. xvi + 359 pp. 
Riahi, K., Rao, S., Krey, V., Cho, C., Chirkov, V., Fischer, G., Kindermann, G., Nakicenovic, N. 
& Rafaj, P. (2011) RCP 8.5—A scenario of comparatively high greenhouse gas 
emissions. Climatic Change 109, 33–57. https://doi.org/10.1007/s10584-011-0149-y 
Rode, K.D., Regehr, E. V., Douglas, D.C., Durner, G., Derocher, A.E., Thiemann, G.W. & 
Budge, S.M. (2014) Variation in the response of an Arctic top predator experiencing 
habitat loss: Feeding and reproductive ecology of two polar bear populations. Global 
Change Biology 20, 76–88. https://doi.org/10.1111/gcb.12339 
Roth, J.A. and Laerm, J. (1980) A late Pleistocene vertebrate assemblage from Edisto Island, 
South Carolina, USA. Brimleyana 3, 1-30. 
Sbrocco, E.J. (2014) Paleo-MARSPEC: gridded ocean climate layers for the mid-Holocene and 
Last Glacial Maximum. Ecology 95, 1710–1710. https://doi.org/10.1890/14-0443.1 
Shin, Y., Min, M.S. & Borzée, A. (2021) Driven to the edge: species distribution modeling of a 
clawed salamander (Hynobiidae: Onychodactylus koreanus) predicts range shifts and 
drastic decrease of suitable habitats in response to climate change. Ecology and Evolution 
11, 14669–14688. https://doi.org/10.1002/ece3.8155 
Shoemaker, K.T., Heffelfinger, L.J., Jackson, N.J., Blum, M.E., Wasley, T. & Stewart, K.M. 
(2018) A machine-learning approach for extending classical wildlife resource selection 
analyses. Ecology and Evolution 8, 3556–3569. https://doi.org/10.1002/ece3.3936 
Simoes, M., Romero-Alvarez, D., Nuñez-Penichet, C., Jiménez, L. & E. Cobos, M. (2020) 
General theory and good practices in ecological niche modeling: a basic guide. 
Biodiversity Informatics 15, 67–68. https://doi.org/10.17161/bi.v15i2.13376 
Slobodin, S.B. (2010) About walruses in the Sea of Okhotsk (according to historical sources). 
Bulletin of the Far Eastern Branch of the Russian Academy of Sciences 2, pp. 15-24. 
Spencer, A.M., Embry, A.F., Gautier, D.L., Stoupakova, A. V. & Sørensen, K. (2011) Chapter 1: 
An overview of the petroleum geology of the Arctic. Geological Society Memoir 35, 1–
15. https://doi.org/10.1144/M35.1 
Svenning, J.C., Fløjgaard, C., Marske, K.A., Nógues-Bravo, D. & Normand, S. (2011) 
Applications of species distribution modeling to paleobiology. Quaternary Science 
Reviews 30, 2930–2947. https://doi.org/10.1016/j.quascirev.2011.06.012 
Theocharis, D., Pettit, S., Rodrigues, V.S. & Haider, J. (2018) Arctic shipping: A systematic 
literature review of comparative studies. Journal of Transport Geography 69, 112–128. 
https://doi.org/10.1016/j.jtrangeo.2018.04.010 
181 
 
Turner, M.G., Calder, W.J., Cumming, G.S., et al. (2020) Climate change, ecosystems and abrupt 
change: science priorities. Philosophical Transactions of the Royal Society B: Biological 
Sciences 375. https://doi.org/10.1098/rstb.2019.0105 
Tyberghein, L., Verbruggen, H., Pauly, K., Troupin, C., Mineur, F. & De Clerck, O. (2012) Bio-
ORACLE: a global environmental dataset for marine species distribution modelling. 
Global Ecology and Biogeography 21, 272–281. https://doi.org/10.1111/j.1466-
8238.2011.00656.x 
Udevitz, M.S., Jay, C. V., Taylor, R.L., Fischbach, A.S., Beatty, W.S. & Noren, S.R. (2017) 
Forecasting consequences of changing sea ice availability for Pacific walruses. Ecosphere 
8. https://doi.org/10.1002/ecs2.2014 
Udevitz, M.S., Taylor, R.L., Garlich-Miller, J.L., Quakenbush, L.T. & Snyder, J.A. (2013) 
Potential population-level effects of increased haulout-related mortality of Pacific walrus 
calves. Polar Biology 36, 291–298. https://doi.org/10.1007/s00300-012-1259-3 
Vasilevskii, R.S. (1971) Origin and Ancient Culture of the Koryaks. Institut Istorii, Novosibirsk, 
Siberia. 
Veloz, S.D. (2009) Spatially autocorrelated sampling falsely inflates measures of accuracy for 
presence-only niche models. Journal of Biogeography 36, 2290–2299. 
https://doi.org/10.1111/j.1365-2699.2009.02174.x 
Verbitsky, M.Y., Crucifix, M. & Volobuev, D.M. (2018) A theory of Pleistocene glacial 
rhythmicity. Earth System Dynamics 9, 1025–1043. https://doi.org/10.5194/esd-9-1025-
2018 
van Vuuren, D.P., Edmonds, J., Kainuma, M., et al. (2011) The representative concentration 
pathways: An overview. Climatic Change 109, 5–31. https://doi.org/10.1007/s10584-011-
0148-z 
Warren, D.L. & Seifert, S.N. (2011) Ecological niche modeling in Maxent: the importance of 
model complexity and the performance of model selection criteria. Ecological 
Applications 21, 335–342. https://doi.org/10.1890/10-1171.1 
Wassmann, P., Duarte, C.M., Agustí, S. & Sejr, M.K. (2011) Footprints of climate change in the 
Arctic marine ecosystem. Global Change Biology 17, 1235–1249. 
https://doi.org/10.1111/j.1365-2486.2010.02311.x 
Werner, I. (1997) Grazing of Arctic under-ice amphipods on sea-ice algae. Marine Ecology 
Progress Series 160, 93–99. https://doi.org/10.3354/meps160093 
Wilson, S.C., Crawford, I., Trukhanova, I., Dmitrieva, L. & Goodman, S.J. (2020) Estimating 
risk to ice-breeding pinnipeds from shipping in Arctic and sub-Arctic seas. Marine Policy 
111, 103694. https://doi.org/10.1016/j.marpol.2019.103694 
Winsor, P. & Björk, G. (2000) Polynya activity in the Arctic Ocean from 1958 to 1997. Journal 
of Geophysical Research: Oceans 105, 8789–8803. https://doi.org/10.1029/1999jc900305 
182 
 
Yletyinen, J. (2019) Arctic climate resilience. Nature Climate Change 9, 805–806. 
https://doi.org/10.1038/s41558-019-0616-4 
Zachos, J.C., Dickens, G.R. & Zeebe, R.E. (2008) An early Cenozoic perspective on greenhouse 
warming and carbon-cycle dynamics. Nature 451, 279–283. 
https://doi.org/10.1038/nature06588 
Zagrebelny, S.V., and Kochnev, A.A. (2017) Influence of climate change on summer-fall 
distribution of Pacific walrus in the western Bering Sea: analysis of reasons and 
consequences. Izvestiya TINRO 190, 3. 
Zyryanov, S. V. & Vorontsov, A. V. (2009) Observations of the Atlantic walrus Odobenus 
rosmarus rosmarus in spring of 1997 in the Kara Sea and southeastern part of the Barents 
Sea. Zoological Journal 78, 1254–1256. 
 
Appendix A 
Berta, A., and A. Wyss. 1994. Pinniped phylogeny; pp. 33–56 in T. A. Deméré and A. Berta 
(eds.), Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr.: 
Proceedings of the San Diego Society of Natural History 29. 
 
Boessenecker RW, Churchill M. 2013. A reevaluation of the morphology, paleoecology, and 
phylogenetic relationships of the enigmatic walrus Pelagiarctos. PLoS ONE 8:e54311. 
 
Boessenecker, R. W., and M. Churchill. 2015. The oldest known fur seal. Biology Letters 
11:21040835. 
 
Boessenecker, R. W., and M. Churchill. 2018. The last of the desmatophocid seals: a new species 
of Allodesmus from the upper Miocene of Washington, USA, and a revision of the 
taxonomy of Desmatophocidae. Zoological Journal of the Linnaean Society XX:1–25. 
 
Deméré, T. A. 1994. The family Odobenidae: a phylogenetic analysis of fossil and living taxa. 
Proceedings of the San Diego Society of Natural History 29:99–123. 
 
Deméré, T. A., and A. Berta. 2002. The Miocene pinniped Desmatophoca oregonensis Condon, 
1906 (Mammalia: Carnivora), from the Astoria Formation, Oregon. Smithsonian 
Contributions to Paleobiology 93:113–147. 
 
Furbish, R. 2015. Something old, something new, something swimming in the blue: an analysis 
of the pinniped family Desmatophocidae, its phylogenetic position, and swimming mode. 
Unpublished M.S., San Diego State University, San Diego, California, 75 pp. 
 
Kohno, N. 1996. Miocene pinniped Allodesmus (Mammalia: Carnivora); with special reference 
to the “Mito seal” from Ibaraki Prefecture, Central Japan. Transactions of the 
Proceedings of the Paleontological Society of Japan 181:388–404. 
 
183 
 
Kohno, N., H. Koike, and K. Narita. 2007. Outline of fossil marine mammals from the middle 
Miocene Bessho and Aoki Formations, Nagano Prefecture, Japan. Research Reports of 
the Shinshushinmachi Fossil Museum 10:1–45. 
 
Tonomori, W.H., H. Sawamura, T. Sato, and N. Kohno. 2018. A new Miocene pinniped 
Allodesmus (Mammalia: Carnivora) from Hokkaido, Northern Japan. Royal Society Open 
Science 5:172440. 
 
Appendix C 
Boessenecker, R.W. and M. Churchill. 2018. The last of the desmatophocid seals: a new species 
of Allodesmus from the upper Miocene of Washington, USA, and a revision of the 
taxonomy of Desmatophocidae. Zoological Journal of the Linnean Society XX:1–25. 
https://doi.org/10.1093/zoolinnean/zlx098/4783131 
Poust, A. and R. W. Boessenecker. 2018. Expanding the geographic and geochronologic range of 
early pinnipeds: new specimens of Enaliarctos from Northern California and Oregon. 
Acta Palaeontologica Polonica 63:25–40. https://doi.org/10.4202/app.00399.2017 
 
Appendix E 
Boyce, M. S., P. R. Vernier, S. E. Nielsen, and F. K. A. Schmiegelow. 2002. Evaluating resource 
selection functions. Ecological Modelling 157:281–300. 
Kass, J. M., R. Muscarella, P. J. Galante, C. L. Bohl, G. E. Pinilla-Buitrago, R. A. Boria, M. 
Soley-Guardia, and R. P. Anderson. 2021. ENMeval 2.0: Redesigned for customizable 
and reproducible modeling of species’ niches and distributions. Methods in Ecology and 
Evolution 12:1602–1608. 
184