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

The last of the desmatophocid seals: a new species of Allodesmus from the upper Miocene of... Abstract The family Desmatophocidae represents an early radiation of extinct pinnipeds that peaked in diversity during the middle Miocene. Although represented by abundant well-preserved fossils, the taxonomy and evolutionary relationships of this family remain poorly known. Late Miocene desmatophocids have been recorded, although none have been formally described, preventing a thorough appraisal of their decline and extinction. We report the discovery of a new species, Allodesmus demerei sp. nov., represented by a partial skeleton with cranium, mandibles, and axial skeleton, from the upper Miocene Montesano Formation of Washington, prompting reinterpretation of desmatophocid taxonomy, phylogeny, and extinction. Phylogenetic analysis (95 characters, 26 taxa) found strong support for monophyletic Desmatophocidae and Allodesmus. Desmatophocidae was found as sister to Phocidae with poor support. Allodesmus demerei was placed within the Allodesmus as the sister taxon to Allodesmus kernensis. The geochronologically young age (10.5–9.1 Mya) of Al. demerei establishes this species as the last of the desmatophocid seals. The middle Miocene peak in desmatophocid diversity coincides with the middle Miocene climatic optimum, suggesting that declining sea surface temperature played a role in their decline and extinction. Walruses diversified and increased in body size during the mid- to late Miocene as desmatophocids declined, suggesting some form of ecological displacement. Carnivora, Pinnipedia, Desmatophocidae, Allodesmus, Miocene, Washington, Montesano Formation, Mammalia, extinction INTRODUCTION The Desmatophocidae are an extinct monophyletic group of seal-like pinnipeds that diversified during the early and middle Miocene and are the first large-bodied pinnipeds to evolve. They comprise three to five genera: Desmatophoca, Atopotarus, Brachyallodesmus, Megagomphos, and Allodesmus; all are generally characterized by bulbous cheek teeth, large orbits, a mortised jugal–squamosal articulation, and forelimbs similar to extant otariids (fur seals and sea lions) indicative of forelimb-dominated rowing (Mitchell, 1966; Barnes, 1972; Giffin, 1992; Barnes & Hirota, 1995; Bebej, 2009; Pierce et al., 2011; Debey & Pyenson, 2013). Allodesmus in particular is characterized by highly simplified bulbous cheek teeth, enormous orbits, and a prenarial shelf, which has led some to speculate that Allodesmus exhibited an enlarged soft tissue proboscis, like modern elephant seals (Mitchell, 1966; Barnes, 1972; Debey & Pyenson, 2013). Desmatophocids are confined to the North Pacific (Baja California, California, Oregon, Washington, and Japan; Barnes & Hirota, 1995) and were amongst the first fossil pinnipeds to be known from more than fragmentary material, owing to the abundance of their remains at key fossil localities prospected early in the 20th century (Desmatophoca oregonensis, Astoria Formation, OR, USA, Condon, 1906; Allodesmus kernensis, Round Mountain Silt, CA, USA Kellogg, 1922). Their early discovery, combination of features found in Phocidae (earless seals), Otariidae (fur seals and sea lions), and Odobenidae (walruses), and early geochronological age led early workers to consider Allodesmus as an ancestral pinniped (Kellogg, 1922; Scheffer, 1958). Some early researchers found taxa like Allodesmus rather specialized (Kellogg, 1922), and later discoveries led to the consensus that desmatophocids are an early diversification of pinnipeds with no modern descendants (Downs, 1956; Mitchell, 1966; Barnes, 1972). Despite the abundance of material available for study, strong debate still exists regarding the phylogenetic relationships of desmatophocids to other pinniped groups. Originally, desmatophocids were considered closely allied to ‘enaliarctines’, Otariidae, and Odobenidae, forming the clade Otarioidea (Barnes, 1989, 2008; Koretsky et al., 2016); under this arrangement, phocids were considered not to share a common ancestor with otarioids, but rather to have evolved independently from musteloid ancestors. However, this hypothesis has yet to be confirmed in any rigorous phylogenetic analysis of morphology (Berta & Ray, 1990; Berta & Wyss, 1994), and it conflicts with molecular data, which support pinniped monophyly (Arnason et al., 2006; Agnarsson et al., 2010). Alternatively, desmatophocids were considered to be the sister group to phocid seals, with the ‘enaliarctine’ Pinnarctidion, odobenids, and otariids forming successive sister taxa to this clade, forming a monophyletic Pinnipedia (Berta & Ray, 1990; Berta & Wyss, 1994). Although analyses of molecular data support pinniped monophyly, they instead recover a significantly different topology within Pinnipedia, with Otariidae and Odobenidae forming a clade to the exclusion of Phocidae. This calls into question whether Desmatophocids really are the sister group to Phocidae or whether, like odobenids, they might be more closely related to otariids. Relationships within Desmatophocidae are also unsettled. Past studies interpreting the evolution of allodesmine seals pre-dated computer-aided cladistics (Barnes, 1972; Barnes & Mitchell, 1975; Repenning & Tedford, 1977), did not use rigorous phylogenetic methodologies (Barnes, 1989; Barnes & Hirota, 1995), and had limited outgroup sampling (Kohno, 1996) and small character datasets (Kohno, 1996; Deméré & Berta, 2002). This has led to considerable controversy regarding the taxonomy of desmatophocids, with disagreements focusing on the number of genera within Allodesminae (e.g. Barnes, 1970, 1972; Barnes & Hirota, 1995) and number of species in Allodesmus (e.g. Mitchell, 1966; Barnes & Hirota, 1995; Kohno, 1996; Deméré & Berta, 2002; Barnes, 2008). Specific points of contention include the recognition of Atopotarus, Brachyallodesmus, and Megagomphos as distinct genera separate from Allodesmus, as well as the number of species present within the middle Miocene Round Mountain Silt of California (Mitchell, 1966; Barnes, 1970, 1972; Barnes & Hirota, 1995). Uncertainty over the taxonomy of the group has also hindered analyses of diversity of the Desmatophocidae. Past studies have examined the placement of Desmatophocidae within Pinnipedia (Barnes, 1989; Berta & Ray, 1990; Berta & Wyss, 1994; Deméré & Berta, 2002), as well as their paleoecology (Mitchell, 1966; Debey & Pyenson, 2013) and locomotor abilities (Bebej, 2009; Pierce et al., 2011), but little has been published on their diversification and the cause and timing of their eventual extinction. This study reports a new species of Allodesmus from the upper Miocene Montesano Formation of Washington, featured in a brief taphonomic report (Bigelow, 1994) but never properly described or named. This new species is represented by a well-preserved cranium, mandibles, and articulated axial skeleton. A new cladistic analysis of the Desmatophocidae provides the basis of a revision of their problematical taxonomy. This new species represents the geochronologically youngest desmatophocid seal yet reported (10.5–9.1 Ma, Tortonian equivalent), prompting investigation of potential causes for the decline and extinction of the formerly diverse Desmatophocidae. MATERIAL AND METHODS Preparation, anatomical description, measurements, and photography UWBM 75640 was mechanically prepared in the University of Washington Burke Museum paleontology laboratory. Concretionary sandstone matrix was left surrounding some vertebrae and within the thoracic cavity to retain the skeletal articulations; natural breaks were left so that the skeleton could be assembled in a series of five blocks. Anatomical terminology generally follows Deméré & Berta (2002), with modifications by Boessenecker & Churchill (2013, 2015). Measurements were recorded with digital callipers to the nearest tenth of a millimetre. Photographs were taken with a Rebel XS and 80 mm zoom lens. Phylogenetic analysis For our phylogenetic analysis, we examined material for all currently recognized species of desmatophocid seal and closely allied taxa. This included two species of the Desmatophoca (D. oregonensis and Desmatophoca brachycephala), Atopotarus courseni, and all eight species generally placed within the genus Allodesmus (Allodesmus gracilis, Allodesmus kelloggi, Al. kernensis, Allodesmus naorai, Allodesmus packardi, Allodesmus sadoensis, and Allodesmus sinanoensis), along with the new taxon we describe below. Allodesmus taxa from the Round Mountain Silt of Sharktooth Hill were coded as a single taxonomic unit (Al. kernensis), following Barnes (1972; also see our discussion below). We also included a large range of outgroup taxa representing the full extent of fossil pinnipedimorph diversity, including four ‘enaliarctines’ (Enaliarctos spp., Pteronarctos goedertae, Pinnarctidion bishopi, and Pinnarctidion rayi), three phocids (Devinophoca claytoni, Erignathus barbatus, and Monachus monachus), three otariids (Thalassoleon mexicanus, Callorhinus ursinus, and Zalophus californianus), and five odobenids (Proneotherium repenningi, Neotherium mirum, Imagotaria downsi, Gomphotaria pugnax, and Odobenus rosmarus). Enaliarctos was coded from all recognized species within that genus, especially Enaliarctos emlongi and Enaliarctos mealsi. We examined in total 171 specimens (Supporting Information, Appendix S2), which were used to code 95 morphological characters. Three of these characters were new, whereas the remaining characters were modified from those used in prior phylogenetic analyses (Berta & Wyss, 1994; Kohno, 1996, 2006; Deméré & Berta, 2002; Boessenecker & Churchill, 2013, 2015; Furbish, 2015). These include 36 cranial, seven mandibular, 24 dental, and 28 postcranial characters (Supporting Information, Appendix S3). Scaled character coding (Wiens, 1999) was applied to taxa that were polymorphic for a given character, with taxa coded as possessing an intermediate state (state name ‘polymorphic’) between two character states. Characters that included polymorphic states were then run as ordered characters in all analyses, with penalties applied to character state transitions that skip the polymorphic step. Overall, 12 polymorphic characters were included within the matrix. Our character matrix is available in the Supporting Information, Appendix S7. All phylogenetic analyses were carried out in TNT 1.1 using 10000 replicates with sectorial and tree-fusing options checked. Two separate phylogenetic analyses were run. In the first analysis, alternative implied weighting schemes were tested, varying the weighting constant K from K = 2 to K = 5. An implied weight of K = 3 was selected because this resulted in the maximal amount of resolution within trees produced with the least amount of weighting. In the second analysis, no implied weighting was carried out. Bootstrap support values were calculated for both analyses, using symmetrical resampling and 1000 replicates. Evaluating trends in diversification To evaluate plausible factors in desmatophocid extinction, we assembled a dataset including North Pacific pinniped species level diversity, global diatom diversity, eustatic sea level, and sea surface temperature, from the Chattian to Holocene. We focused on the North Pacific for pinniped diversity because desmatophocids are restricted to this region (chiefly USA, Mexico, and Japan), and the faunas are both densely sampled and well described (Miyazaki et al., 1995; Deméré et al., 2003; Barnes, 2008; Velez-Juarbe, 2017). Overall, we were able to assemble a dataset containing 100 occurrences of modern and fossil pinnipeds from this region. These data were sorted by taxonomic group (‘enaliarctine’, Odobenidae, Otariidae, Desmatophocidae, and Phocidae) and binned by stage level, with originations/extinctions brought to stage boundaries. Additionally, odobenids were further binned by size, and sorted into small (total body length <2 m) and large (total body length >2 m) size classes, with body size data from Churchill et al. (2014). Diatom species diversity is a reasonable proxy for primary productivity, and we used the species richness curve from Marx and Uhen (2010) for this proxy. Data for sea surface temperature (δ18O) is from Fedorov et al. (2013) and Rousselle et al. (2013), and eustatic sea level is from Miller et al. (2005). Institutional abbreviations UWBM, University of Washington Burke Museum of Natural History and Culture, University of Washington, Seattle, WA, USA. GEOLOGICAL BACKGROUND The upper Miocene Montesano Formation consists of ~800 m of shallow to deep marine sandstones and siltstones deposited within the Grays Harbor Basin near Grays Harbor in southwestern Washington, USA (Fig. 1; Fowler, 1965; Rau, 1967). The Montesano Formation unconformably overlies the lower to middle Miocene Astoria Formation and is in turn unconformably overlain by the Pleistocene Satsop Formation. The Montesano Formation is divided into a 460-m-thick lower member (units 1–5 of Fowler, 1965) consisting of fine sandstone with lensoidal calcareous nodules with intermittent pebbly horizons and abundant carbonized wood fragments, and a 240-m-thick upper member (units 6–7 of Fowler, 1965) consisting of well-bedded tuffaceous mudrocks and occasional fine sandstone interbeds (Fowler, 1965; Prothero & Lau, 2001). Both members of the Montesano Formation are richly fossiliferous; Addicott (1976) designated much of the lower member (units 1–4 of Fowler, 1965) as the basis for the Wishkahan provincial megainvertebrate stage, and the top of the lower member and the upper member as the stratotype for the Graysian stage (units 5–7 of Fowler, 1965). The holotype (UWBM 75640) and referred specimen (UWBM 109823) of Al. demerei were collected from the lower member of the Montesano Formation; the holotype was collected ~140 m above the base of the unit (Fig. 1; Bigelow, 1994). Paleomagnetic studies by Prothero and Lau (2001) correlated the Montesano Formation with Chrons C4Ar to C5n, indicating an age of 10.5–9.1 Mya for the formation. Although originally correlated with the Empire Formation on the basis of shared Wishkahan stage molluscs, paleomagnetic analysis indicates that the Empire Formation is somewhat younger (6.8–8.3 Mya; Prothero et al., 2001). The position of the type locality (UWBM C0343; 47°15′N, 123°31′W) ~140 m above the base of the formation suggests that the specimen might originate from above Chron C5n and within Chron C4Ar (Fig. 1); if so, this would suggest an even finer age control of 9.8–9.1 Mya (following chron boundary dates from Gradstein et al., 2012). However, paleomagnetism has not been studied at the type locality, and extrapolation from exposures along the Wishkah River (Prothero & Lau, 2001) seems unwise for now, and we adopt the slightly more coarse age of 10.5–9.1 Mya for the formation as a whole (Prothero & Lau, 2001). Isolated teeth of the dogfish Squalus occidentalis were discovered in the matrix associated with the holotype specimen (Bigelow, 1994). Massively bedded, pervasively bioturbated sandstone with occasional climbing ripples, mud drapes, and shell beds suggest lower shoreface deposition just below fair-weather wave base for this part of the Montesano Formation (Bigelow, 1994). Figure 1. View largeDownload slide Geological and stratigraphic context of Allodesmus demerei. A, geological map of Washington (modified from Walsh et al., 1987), showing exposures of the Montesano Formation (Fm.) and location of the type locality (UWBM C0343). B, stratigraphic column of the Montesano Formation exposed along the middle fork of the Wishkah River, after Fowler (1965). Magnetostratigraphy after Prothero and Lau (2001), from west fork of Wishkah River. Figure 1. View largeDownload slide Geological and stratigraphic context of Allodesmus demerei. A, geological map of Washington (modified from Walsh et al., 1987), showing exposures of the Montesano Formation (Fm.) and location of the type locality (UWBM C0343). B, stratigraphic column of the Montesano Formation exposed along the middle fork of the Wishkah River, after Fowler (1965). Magnetostratigraphy after Prothero and Lau (2001), from west fork of Wishkah River. SYSTEMATIC PALAEONTOLOGY Class Mammalia Linneaus, 1758 Order Carnivora Bowditch, 1821 Clade Pinnipedia Illiger, 1811 Familiy Desmatophocidae Hay, 1930 Genus Allodesmsus Kellogg, 1922 Amended diagnosis Desmatophocidae differing from Desmatophoca and Atopotarus in possessing a prenarial shelf, lacking a lateral wall of the alisphenoid canal, preglenoid process, possessing single-rooted postcanine teeth, and posteriorly deepening mandible with posteroventrally expanded digastric insertion; differing further from Desmatophoca in possessing bulbous postcanine teeth lacking accessory cusps or cingula, and differing further from Atopotarus in retaining M2. Type species Allodesmus kernensis. Included species Allodesmus kernensisKellogg, 1922; Al. naoraiKohno, 1996; Al. packardiBarnes, 1972; Al. sadoensisBarnes and Hirota, 1995; Al. sinanoensisNagao, 1941. Allodesmus demerei sp. nov. Allodesmus n. sp. Bigelow, 1994 Allodesmus n. sp. Barnes and Hirota, 1995 Etymology Allodesmus demerei is named in honor of Dr. Thomas A. Deméré for his mentorship, support, and influential contributions to the study of fossil pinnipeds and other marine mammals. Diagnosis of species A large species of Allodesmus similar in adult size to Al. kernensis and Al. sadoensis, and differing from Al. naorai and Al. packardi in possessing a prenarial shelf that is anteriorly transversely expanded, and differing from Al. sinanoensis in lacking tusk-like canines. With the exception of Al. naorai, Al. demerei differs from all Allodesmus in exhibiting proportionally more elongate and triangular nasals that are widest anteriorly. Allodesmus demerei differs from Al. kernensis in exhibiting more strongly developed nuchal crests, which obscure the occipital condyles in dorsal view, a jugal that extends posteriorly to the level of the glenoid fossa, and a proportionally deeper mandible, and from Al. sadoensis in retaining an M2. Allodesmus demerei further differs from all other Allodesmus in possessing a dorsally prominent and sharply triangular postorbital process, lacking postcanine diastemata, exhibiting a posteriorly elongate neural spine of the axis that overhangs C3, and a transversely expanded and dorsoventrally flattened anterior half of the manubrium. Holotype UWBM 75640, a partial articulated skeleton including skull and mandible missing all teeth other than right P1, hyoid bones, vertebral column (C1–T13), ribs, and manubrium. Collected 26–27 September 1984, by P. K. Bigelow and colleagues. Type locality and stratigraphic context University of Washington-Burke Museum locality C0343, 140 m above the base of the lower member of the Montesano Formation exposed in bank of Canyon River near Grays Harbor, WA, USA; upper Miocene, 10.5–9.1 Mya in age based on paleomagnetism (Tortonian equivalent; Prothero & Lau, 2001). More detailed locality information is available on request from UWBM to qualified researchers. Tentatively referred specimen UWBM 109823, left humerus missing the head, collected by M. S. Kelly from the lower Montesano Formation along the bank of the west fork of the Satsop River near Swinging Bridge Park, Grays Harbor County, WA, USA. More detailed locality information is available on request from UWBM to qualified researchers. Description Occurrence and preservation UWBM 75640 consists of a partial articulated skeleton (Fig. 2; Bigelow, 1994: fig. 1) including mandibles, hyoids, vertebral column (C1–T13), ribs (LR1–LR5, RR1–RR13), and manubrium, as well a complete cranium lacking all teeth except the right P1. The skeleton was preserved lying on its right side, with all preserved elements articulated in life position (Bigelow, 1994). Most of the teeth, post-manubrial sternebrae, and forelimbs are absent. The entirety of the skeleton was embedded in a cylindrical concretion enveloping the bones. Fourteen teeth of the dogfish Squalus occidentalis were found concentrated around the skull and cervical vertebrae, and several bite marks are evident on the skull (Bigelow, 1994); aside from those bite marks figured by Bigelow (1994), the extensive surface fracturing precludes us from identifying additional bite marks as it is not possible at this point to distinguish them from damage that occurred during fossil preparation. The posteriormost thoracic vertebrae were exposed in the bank of the Canyon river (Bigelow, 1994), raising the possibility that the entire vertebral column and possibly hindlimb elements were originally preserved and eroded away prior to discovery. Figure 2. View largeDownload slide Allodesmus demerei sp. nov.: skeleton of holotype UWBM 75640 (top), skeletal reconstruction (middle), and life restoration (bottom). Artwork by ©Robert Boessenecker. Figure 2. View largeDownload slide Allodesmus demerei sp. nov.: skeleton of holotype UWBM 75640 (top), skeletal reconstruction (middle), and life restoration (bottom). Artwork by ©Robert Boessenecker. Ontogeny and sex Most of the nine cranial sutures outlined by Sivertsen (1954) could be scored, with the exception of the median maxillary suture on the palate. The basisphenoid–presphenoid suture and frontoparietal sutures are open, the squamoso-parietal suture is partly closed, and the remaining sutures are mostly or completely closed. This results in a suture age of 22, indicative of adult status (Sivertsen, 1954). Further evidence of adult status includes well-developed nuchal and sagittal crests and complete epiphyseal fusion in the vertebral column. Furthermore, the relatively large canine alveoli, alongside well-developed nuchal and sagittal crests indicate that UWBM 75640 is a male; no baculum is preserved, although the entire post-thoracic skeleton is also missing. Rostrum and dentition The rostrum is relatively elongate (134 mm from orbit to anterior tip of premaxilla) and transversely narrow (82 mm wide at canines) and exhibits a broad prenarial shelf, which is smoothly convex dorsally and lacks a distinct prenarial process (Fig. 3; Table 1). In dorsal outline, the rostrum is widest anteriorly at the canines and narrows conspicuously at the level of P3, and widens posteriorly towards the base of the zygomatic arch. In dorsal view, the anterior margin of the rostrum is evenly convex. A small, suboval narial fossa is present, with a thick lateral wall formed by the premaxilla and maxilla. In lateral view, the rostrum is roughly triangular but bears a concave and ‘stepped’ anterodorsal margin owing to the elongate prenarial shelf. Ventrally on the lateral surface, the premaxilla–maxilla suture is indistinct on the left and on the right it is completely fused and obliterated; dorsally on the right side it is visible as a crack. In lateral view, the premaxilla is not continuously visible, and it forms a transversely narrow splint where it contacts the nasal; the nasal process of the premaxilla terminates along the anterior half of the nasal. The nasals are triangular, taper posteriorly, and penetrate between the frontals; the posterior half of the lateral margin is straight, and within the anterior half the nasal widens abruptly, giving the nasals a trumpet-shaped outline in dorsal view. Figure 3. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in dorsal view. Figure 3. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in dorsal view. Table 1. Measurements (in millimetres) of cranium of Allodesmus demerei (UWBM 75640) Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– View Large Table 1. Measurements (in millimetres) of cranium of Allodesmus demerei (UWBM 75640) Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– View Large The palate is not exposed (Fig. 4), but an elongate palate with a ‘pterygoid process’ of the maxilla is present, as in all Pinnipediformes (Berta & Wyss, 1994), that extends at least 15–20 mm posterior to the M2. The toothrows (and by extension the lateral margins of the palate) diverge posteriorly at a 41° angle. With the exception of the right P1, all teeth are missing. Details of the incisor alveoli are unclear owing to incomplete preparation and the close contact between the mandibles and cranium. The canine alveoli are large, circular, and appear to have housed a procumbent canine. The P1 is positioned posteromedial to the canine, single rooted, and bears a bulbous crown lacking a labial cingulum; only the protoconid cusp is present. Wear on the crown is unclear owing to incomplete preparation. The root is somewhat inflated, especially near the base of the crown. The P2 alveolus is oval and indicates a single root. The P3 and P4 alveoli are also oval, but bear a minute vertical ridge laterally, indicating a partial interalveolar septum, suggesting bilobate roots. The M1 alveolus is somewhat smaller than the premolar alveoli and bears a better developed interalveolar septum, suggesting a root that was more strongly bilobate or perhaps double rooted internally. The M2 alveolus is circular, smaller than the M1, and indicates the presence of a single root. Figure 4. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in ventral view. Figure 4. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in ventral view. Orbit, intertemporal region, and zygomatic arch The interorbital bar is composed entirely of the frontal, is anteroposteriorly elongate, transversely narrow, and gradually rises posteriorly (Figs 3, 5; Table 1). The supraorbital process of the frontal is reduced to a small bump positioned posteriorly, approximately three-quarters of the distance between the anterior margin of the orbit and the anterior margin of the braincase. The orbital margin of the frontal is gently laterally concave in dorsal view. Posterior to the supraorbital process, the frontal is also laterally concave. A short orbitotemporal crest extends posteromedially from the supraorbital process and converges medially on either side to form the anterior end of the sagittal crest. The sagittal crest is low (~15 mm at the highest) and elongate, extending from the nuchal crest anteriorly to the middle of the intertemporal bar. Figure 5. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A and B, right lateral view. C, left lateral view. Figure 5. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A and B, right lateral view. C, left lateral view. The orbit is enormous and circular; the anterior part of the zygomatic arch slightly descends ventrally to accommodate the large orbit (Fig. 5; Table 1). The maxillary root of the zygomatic arch is pierced anteriorly by a transversely narrow, oval infraorbital foramen ~15 mm high. The maxilla–jugal suture is completely fused and obliterated. The jugal–squamosal articulation on the zygomatic arch is mortised and dorsoventrally expanded. A large, triangular postorbital process is developed and extends dorsally; it is separated from the supraorbital process of the frontal by only 20 mm. The zygomatic process of the squamosal deepens anteriorly and includes a triangular anteroventral corner. Posteriorly, the jugal splits into a dorsal and a posteroventral splint that cradle the anterior margin of the zygomatic process. The anteroposteriorly narrow dorsal splint leads to the postorbital process, whereas the posteroventral splint tapers to a point, curves ventrally, and extends posteriorly to the level of the glenoid fossa. The zygomatic arch forms the transversely widest part of the skull at about the level of the preglenoid process. Braincase The braincase is transversely narrow and ovoid in dorsal view, and gradually widens posteriorly; the anterolateral margin lacks a corner as in ‘enaliarctines’, otariids, and Desmatophoca, and is instead smoothly convex (Fig. 3). The lateral wall of the braincase is fractured, but the frontoparietal and parieto-squamosal sutures can be traced; a pseudosylvian sulcus is not apparent but coded as uncertain (?) owing to fracturing of the braincase and diagenetic compression in the transverse direction, which might obscure the presence of a pseudosylvian sulcus. The frontoparietal suture has a roughly transverse orientation, is mostly straight and positioned at the position of the external acoustic meatus in dorsal view. The parietal extends posteriorly and forms a large posteriorly extending nuchal crest with an evenly convex posterior margin in lateral view; the nuchal crest extends just posterior to the level of the occipital condyles, and the nuchal crest arises dorsally from the mastoid process. Dorsally, the nuchal crests converge anteriorly, and a small triangular wedge of the occipital shield is visible in dorsal view (Fig. 3). The posterior surface of the nuchal crest is dorsoventrally thick (10–17 mm) and rugose for insertion of epaxial muscles (Fig. 6). The occipital shield is nearly vertical, low, and exhibits a blunt external occipital crest dorsally; lateral to the external occipital crest the supraoccipital is concave. The occipital condyles are obscured by the atlas, but judging from exposed portions are transversely wide and define a circular foramen magnum, which is in turn encircled by deeply excavated dorsal condyloid fossae. The exoccipital bears a circular pit for the sternomastoid fossa. The paroccipital process extends ventrolaterally and is separated from the occipital condyles by a deep ventral condyloid fossa. The paroccipital process descends posteroventrally in lateral view. Figure 6. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A, anterior view. B, posterior view. Figure 6. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A, anterior view. B, posterior view. Basicranium The basioccipital bears deeply excavated oval pits for the anterior insertion of the rectus capitis muscle and is separated by a median crest. Large rugose basioccipital crests are present medial to the large circular posterior lacerate foramina (Fig. 4; Table 1). The basioccipital is wide posteriorly but narrows anteriorly and becomes flatter where it nears the basisphenoid. The pterygoid strut is somewhat exposed and bears a small, vertically oriented pterygoid hamulus that is triangular in lateral view. The pterygoid strut is widest at the position of the hamulus, and laterally defines the transversely narrow internal choanae. The alisphenoid is flat anteromedial to the tympanic bulla. The morphology of the anterior bulla is not evident except for the median lacerate foramen, which pierces the anteromedial spur of the bulla. The tympanic bulla is pentagonal in ventral view, mostly flattened but slightly convex ventrally. It bears a rugose tubercle medially, immediately anterior to the opening of the carotid canal. Anteriorly, the bulla is closely appressed to the postglenoid process. The tympanic bulla has a flat lateral wall, which is anteriorly pierced by the external acoustic meatus; the meatus is positioned within an anteroposteriorly narrow and transversely oriented trough between the postglenoid and mastoid processes. The orientation of the meatus in lateral view (e.g. Barnes, 1987) is unclear. The stylomastoid foramen is a dorsoventrally shallow oval and separated from the external acoustic meatus by a low ridge that continues laterally to meet the enlarged mastoid process. The squamosal is ventrally flattened lateral to the bulla, and bears a rugose knob-like mastoid process with a convex, subcircular ventrolaterally facing facet (Fig. 4). The mastoid process is widely separated from the paroccipital process by an incised notch. The paroccipital process is dorsoventrally shallow and somewhat ventrally excavated; a small ridge separates the stylomastoid foramen from a small tubercle-like tympanohyal. A small fossa is developed posterior to the tympanohyal. A large knob is present on the exoccipital posterolateral to the posterior lacerate foramen; posteriorly, the tympanic bulla laps onto this knob. The glenoid fossa is anteroventrally facing and bears a deep postglenoid process; the preglenoid process is not exposed. Mandible The mandible is gracile and subrectangular, with a dorsoventrally shallow triangular symphyseal portion (Fig. 5; Table 2). The mandibular symphysis is relatively elongate (32% of mandibular length) and is posteriorly distinguished by an indistinct genial tuberosity. The mandible transversely widens anteriorly toward the canine root and narrows posteriorly (Fig. 4). The body of the mandible deepens posteriorly towards the digastric insertion. The ventral margin of the mandible is faintly concave ventrally; posteriorly, the digastric insertion is expanded into a posteroventrally extending flange; a small notch is present between the digastric insertion and the angular process, defining the limits of the flange. Unlike Al. kernensis, the angular process is large and extends medially as a horizontal shelf. The mandibular condyle is wide, cylindrical, and positioned at about the level of the toothrow. The masseteric fossa is shallow and extends ventrally from the coronoid process to the dorsoventral midpoint of the body of the mandible; the floor of the fossa is somewhat rugose and possesses a series of shallow dimples up to 5 mm in diameter. The coronoid process is posteriorly positioned and incompletely prepared but high, rising ~100% of the depth of the body above the toothrow; it extends posteriorly to the level of the posterior edge of the digastric flange. Table 2. Measurements (in millimetres) of mandible of Allodesmus demerei (UWBM 75640) Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 View Large Table 2. Measurements (in millimetres) of mandible of Allodesmus demerei (UWBM 75640) Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 View Large Hyoid The hyoid bones were preserved (Bigelow, 1994) but on loan to P. K. Bigelow at the time of his death; their current whereabouts are unknown and presumed lost. Cervical vertebrae The atlas bears a posterolaterally projecting transverse process that is suboval in outline (Fig. 7A, B; Table 3). The neural spine is a low tubercle. The transverse foramen is large and oval anteriorly and has a much smaller posterior opening. The axis bears a finger-like, posteriorly inclined transverse process with a sharp anteroventral ridge. A sharp, deep median ridge is present on the ventral surface of the centrum; the atlantal articular facet is obscured by matrix. The neural spine of the axis (Table 4) is dorsoventrally short but includes an elongate posterior apex that is subrectangular in lateral view and that probably articulated with the low neural spine of C3. The postzygapophyses are positioned anterior to the posterior apex of the neural spine, are subhorizontal and D-shaped in dorsal view, and are more medially placed than on the rest of the post-atlantal cervical vertebrae. The transverse foramina of the axis are relatively small. Figure 7. View largeDownload slide Allodesmus demerei sp. nov.: holotype axial elements of Allodesmus demerei (UWBM 75640). A, cervical vertebrae in dorsal view, with skull in articulation as preserved. B, cervical vertebrae in right lateral view. C, thoracic vertebrae 1–2 with ribs 1–2 in ventral view; manubrium in (E) ventral, (F) dorsal, and (G) right lateral view. D, thoracic vertebrae 6–11 and ribs 4–13 in right lateral view. Figure 7. View largeDownload slide Allodesmus demerei sp. nov.: holotype axial elements of Allodesmus demerei (UWBM 75640). A, cervical vertebrae in dorsal view, with skull in articulation as preserved. B, cervical vertebrae in right lateral view. C, thoracic vertebrae 1–2 with ribs 1–2 in ventral view; manubrium in (E) ventral, (F) dorsal, and (G) right lateral view. D, thoracic vertebrae 6–11 and ribs 4–13 in right lateral view. Table 3. Measurements (in millimetres) of atlas vertebra of Allodesmus demerei (UWBM 75640) Greatest transverse width 90.2 Greatest depth at midline 69.0 Greatest transverse width 90.2 Greatest depth at midline 69.0 View Large Table 3. Measurements (in millimetres) of atlas vertebra of Allodesmus demerei (UWBM 75640) Greatest transverse width 90.2 Greatest depth at midline 69.0 Greatest transverse width 90.2 Greatest depth at midline 69.0 View Large Table 4. Measurements (in millimetres) of axis vertebra of Allodesmus demerei (UWBM 75640) Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 View Large Table 4. Measurements (in millimetres) of axis vertebra of Allodesmus demerei (UWBM 75640) Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 View Large The remaining cervicals (C3–C7; Fig. 7A, B; Table 5) have elongate ventrolaterally projecting transverse processes that are subrectangular in shape with apical tubercles; the transverse processes become increasingly anteroposteriorly broad and hatchet shaped in C5 and C6, with the transverse process longest in C6. The articular surfaces of the centra are subrectangular anteriorly and oval posteriorly. C3 has an elongate shelf-like lamina; the lamina becomes shorter posteriorly. C5 and C6 have a large laterally directed knob on the base of the transverse process lateral to the canal. In C7, the transverse process is shorter and less ventrally deflected, with a more dorsally positioned lateral apex; the process becomes dorsoventrally deeper laterally, giving it a fan shape in anterior/posterior view. The transverse process is posteriorly concave. The transverse foramina increase in diameter posteriorly, culminating in the enlarged canal in C7. The neural spines are short and stout anteriorly and become higher and transversely narrower posteriorly toward the C7. Table 5. Measurements (in millimetres) of cervical vertebrae C3–C7 of Allodesmus demerei (UWBM 75640) Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – View Large Table 5. Measurements (in millimetres) of cervical vertebrae C3–C7 of Allodesmus demerei (UWBM 75640) Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – View Large Thoracic vertebrae Thoracic vertebrae T1–T13 are preserved (Fig. 7C, D; Table 6); T1–T6 are the most complete and best exposed. T1–T2 are relatively similar to C7 but have a more dorsally positioned transverse process with a large facet for the tubercle of the ribs. All thoracic vertebrae except T1 have a posterolaterally facing fossa on the dorsolateral edge of the posterior centrum for the articulation of the rib heads. Thoracics T1–T2 differ from all posterior thoracics in exhibiting more medially placed, narrowly separated prezygapophyses; they are more widely separated in posterior thoracic vertebrae. The neural spine is poorly preserved in many thoracics, but from anterior to posterior within the thoracic series the spine becomes increasingly anteroposteriorly narrow and posterodorsally inclined. A low ventral median ridge is present on T1–T2, but becomes slightly more transversely rounded in T3–T11; in T12–T13, the ventral margin of the centrum is completely rounded, and no median ridge is evident. The postzygapophyses decrease in size posteriorly within the thoracic series and by T4 they are on the base of the neural spine at the medial termination of the lamina; they are strongly reduced in T10–T13. Table 6. Measurements (in millimetres) of thoracic vertebrae of Allodesmus demerei (UWBM 75640) Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – View Large Table 6. Measurements (in millimetres) of thoracic vertebrae of Allodesmus demerei (UWBM 75640) Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – View Large Ribs Few ribs other than LR1 and RR1 are complete, but the proximal ends of LR2–LR5 and the proximal half of RR2–RR13 are preserved (Fig. 7C, D). Left R1 (126.1 mm long, 60.0 mm capitulum to tubercle) and RR1 (128.0 mm long, 61.6 mm capitulum to tubercle) bear a small capitulum, a large tubercle, and a short shaft with a flat and circular distal end bearing a rugose texture. Ribs 2–5 are morphologically similar but with a longer shaft than R1; R6–R8 have a much larger, subspherical head and a longer shaft. Manubrium The manubrium is the only preserved portion of the sternum (Fig. 7E-G; Table 7). It is anteriorly flattened, transversely wide and rectangular with parallel lateral margins. It bears tubercles with posterolaterally facing rugose articular surfaces for the ribs at the midpoint. A ventral keel is present and dissipates anteriorly; the manubrium deepens posteriorly. The posterior portion is subcylindrical and transversely widens posteriorly toward a large transversely compressed, oval and rugose posterior articular surface. The transversely wide anterior part of the manubrium is unique amongst pinnipeds and differs from the narrower manubrium of Al. kernensis, although in Al. kernensis the anterior half is slightly expanded transversely; a somewhat expanded manubrium may also be present in At. courseni, but further preparation is required to be certain. Table 7. Measurements (in millimetres) of sternum of Allodesmus demerei (UWBM 75640) Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 View Large Table 7. Measurements (in millimetres) of sternum of Allodesmus demerei (UWBM 75640) Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 View Large Humerus UWBM 109823 (Fig. 8; Table 8) is missing the head. The shaft is nearly cylindrical and straight in anterior and posterior view and widens slightly proximally, and abruptly widens distally toward the transversely wide distal end. In posterior view, the medial margin is slightly more concave than the lateral margin. The deltopectoral crest is elongate and runs along two-thirds of the length of the humerus. The crest is anteroposteriorly deepest where it abruptly terminates; proximally, the anterior margin is nearly straight. The crest is positioned laterally on the shaft in anterior view, and is transversely widest near the distal terminus. In lateral view, the deltopectoral crest has an abrupt distal terminus. The deltoid insertion is not separated from the deltopectoral crest. The greater and lesser tuberosities are nearly the same height, and both appear to have extended slightly proximally to the head. In proximal view, the tuberosities are both anteromedially slanted. The tuberosities are separated by a narrow intertubercular groove with a V-shaped cross-section. The distal end is much wider than the proximal end. The prominent medial entepicondyle is medially convex and anteroposteriorly flattened; the supinator ridge is poorly developed and lacks a sharp edge. In medial and lateral view, the posterior margin of the humerus is concave. The medial and lateral edges of the distal trochlea are slanted anterolaterally in distal view; the lateral lip of the trochlea is slightly thicker anteroposteriorly than the medial lip. Figure 8. View largeDownload slide Referred humerus of Allodesmus demerei (UWBM 109823). A, anterior view. B, posterior view. C, medial view. D, lateral view. Figure 8. View largeDownload slide Referred humerus of Allodesmus demerei (UWBM 109823). A, anterior view. B, posterior view. C, medial view. D, lateral view. Table 8. Measurements (in millimetres) of referred humerus of Allodesmus demerei (UWBM 109823) Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 View Large Table 8. Measurements (in millimetres) of referred humerus of Allodesmus demerei (UWBM 109823) Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 View Large Referral of UWBM 109823 to Allodesmus demerei The holotype specimen lacks a humerus and thus there are no overlapping elements to refer UWBM 109823 unequivocally to Al. demerei. However, the humerus compares relatively well with Al. kernensis, and differs from otariids in lacking an arcuate anterior margin of the deltopectoral crest and a sinuous posterior margin of the humerus. UWBM 109823 differs from odobenids in exhibiting a concave posterior margin and a distally positioned apex of the deltopectoral crest. UWBM 109823 differs from both odobenids and otariids in possessing a distal trochlea that is slightly anteroposteriorly wider laterally (similar diameter in otariids, wider medial edge in odobenids; Repenning & Tedford, 1977). The distal apex of the deltopectoral crest, large medial entepicondyle, wide lateral portion of the distal trochlea, and relatively large lesser tuberosity support referral to Allodesmus. Given that Al. demerei is the only known Allodesmus from the Montesano Formation, we tentatively assign this specimen to the species; future discoveries of more complete skeletons are needed to confirm this identification. Comparisons Allodesmus demerei differs from Desmatophoca and Atopotarus in possessing a prenarial shelf on the rostrum; the prenarial shelf is anteroposteriorly longer than Al. naorai and Al. packardi but somewhat shorter than Al. kernensis; the prenarial shelf further differs from these taxa by being transversely expanded anteriorly, which Al. demerei shares with Al. kernensis, Al. sadoensis, and Al. sinanoensis. Rather than converging posteriorly as in all other desmatophocids, the lateral margins of the nasals are posteriorly parallel and terminate along the V-shaped fronto-maxillary suture, similar to Al. packardi, dusignathine walruses, and the odobenine walrus Ontocetus (Deméré, 1994; Kohno, 2006) but differing from all other desmatophocids for which the suture is preserved. The nasals are also proportionally longer than most other Allodesmus spp. (62% of rostrum length) except Al. naorai. Allodesmus demerei primitively retains a triangular postorbital process like Desmatophoca and Atopotarus, differing from the dorsally rounded condition in Al. kernensis and Al. packardi. In Al. demerei, the jugal extends posteriorly to the level of the anterior margin of the glenoid fossa, similar to Desmatophoca and perhaps Al. packardi; however, in Al. kernensis the jugal terminates well anterior to the glenoid fossa. Allodesmus demerei further differs from all other species of Allodesmus in exhibiting nuchal crests so large and posteriorly directed that the occipital condyles and supraoccipital are completely hidden in dorsal view; in published specimens of Al. kernensis the condyles are widely visible, although in newly available but unpublished specimens of Al. kernensis the nuchal crest obscures the condyles (T. A. Deméré, personal communication September 2016). In Al. sadoensis, the nuchal crests partly obscure the condyles, and the supraoccipital is not visible. In extant Otariidae, this is related to ontogeny, and perhaps published specimens of Al. kernensis are young adult males rather than old adults; an assessment of ontogenetic and intraspecific variation within the Sharktooth Hill sample of Al. kernensis would be highly instructive. The dentition of Al. demerei differs from Al. kernensis, Al. naorai, Al. packardi, and Al. sinanoensis (and shares with Al. sadoensis) in lacking diastemata between the upper postcanines; retention of an M2 distinguishes Al. demerei from Al. sadoensis. Unlike Desmatophoca, Al. kernensis, and Al. sadoensis, the molar alveoli appear to be of similar size to the premolars (a feature shared with Al. packardi). The mandible of Al. demerei differs from Al. kernensis in being dorsoventrally deeper, and is much deeper than At. courseni; the mandible exhibits a proportionally smaller genial tuberosity than At. courseni and Al. sadoensis. Mandibles of D. oregonensis differ in exhibiting a horizontal ramus that is dorsoventrally deepest at the genial tuberosity rather than the digastric insertion. Well-preserved postcranial skeletons are currently available only for At. courseni and Al. kernensis. The axis differs from these two allodesmines and all other pinnipeds in possessing an elongate posterior extension of the neural spine, which appears to articulate with or closely approximate the tip of the neural spine of the third cervical vertebra. An extension appears present in At. courseni and longer than in Al. kernensis but is not as elongate as Al. demerei. Phylogenetic results In the analysis where implied weighting (IW) was performed, one most parsimonious tree was recovered [Fig. 9; tree length = 29.54, consistency index (CI) = 0.48, retention index (RI) = 0.71]. In the analysis where no implied weighting (NW) was performed, ten most parsimonious trees were recovered (Fig. 10; tree length = 300, CI = 0.48, RI = 0.71). Given the focus of this paper on relationships within Desmatophocidae, synapomorphies are listed in this section only for members of this clade (as character number:state number). Figure 9. View largeDownload slide Time-calibrated strict consensus tree of desmatophocid relationships based on implied weighting analysis (K = 3). Age range of taxa is represented by terminal horizontal bars (see Supporting Information, Appendix S4 for age explanations); black bars represent outgroup taxa, green bars desmatophocines, and blue bars allodesmines. Numbers above nodes represent bootstrap values. ‘Round Mountain Silt’ Allodesmine composite taxon including Allodesmus kernensis, Al. kelloggi, and Al. gracilis. Tree scaled to time using the Paleotree package in R. 3.3.1 (Bapst, 2012). Outlines of skulls redrawn from Barnes (1972, 1979, 1987, 1989), Barnes and Hirota (1995), Berta (1994), Deméré and Berta (2001, 2002), Kohno (1996), and Repenning and Tedford (1977), and photographs provided by J. Velez-Juarbe. Figure 9. View largeDownload slide Time-calibrated strict consensus tree of desmatophocid relationships based on implied weighting analysis (K = 3). Age range of taxa is represented by terminal horizontal bars (see Supporting Information, Appendix S4 for age explanations); black bars represent outgroup taxa, green bars desmatophocines, and blue bars allodesmines. Numbers above nodes represent bootstrap values. ‘Round Mountain Silt’ Allodesmine composite taxon including Allodesmus kernensis, Al. kelloggi, and Al. gracilis. Tree scaled to time using the Paleotree package in R. 3.3.1 (Bapst, 2012). Outlines of skulls redrawn from Barnes (1972, 1979, 1987, 1989), Barnes and Hirota (1995), Berta (1994), Deméré and Berta (2001, 2002), Kohno (1996), and Repenning and Tedford (1977), and photographs provided by J. Velez-Juarbe. Figure 10. View largeDownload slide Strict consensus tree of desmatophocid relationships based upon equal weighted analysis. Numbers above nodes represent bootstrap values. †Extinct taxa. Figure 10. View largeDownload slide Strict consensus tree of desmatophocid relationships based upon equal weighted analysis. Numbers above nodes represent bootstrap values. †Extinct taxa. Relationships of outgroup taxa to Desmatophocidae largely differ in their levels of resolution between analyses (Figs 9, 10). In both analyses, Pteronarctos is recovered as the sister taxon to a clade comprising Pinnarctidion and Pinnipedia, with good support [bootstrap (BS) values of 100]. A monophyletic Pinnarctidion is recovered as the sister taxon to Pinnipedia, but with only moderate support in the the NW analyses (BS = 61), and was poorly supported in the IW tree. Within the IW tree, a clade comprising Odobenidae and Otariidae is recovered, and is found to be the sister group to Phocoidea (Fig. 9). A monophyletic Otariidae is recovered with good support (BS = 75), but odobenids monophyly is only weakly supported. In contrast, within the NW tree a strongly supported (BS = 73) Otariidae is recovered within a polytomy consisting of Proneotherium, Neotherium, a poorly supported clade comprising the remainder of odobenids, and Phocoidea. Although a monophyletic Phocoidea is recovered in both analyses, it is only poorly supported. Phocidae is recovered with moderate to strong support in both analyses (IW BS = 59; NW BS = 79). Desmatophocidae is recovered with weak support in both analyses (Figs 9, 10). It is supported by nine unequivocal synapomorphies, including reduced premaxilla–nasal contact (1:1), a V-shaped nasal–frontal suture (2:1); a posterolaterally expanded pterygoid process of the maxilla below the orbit (8:1), very large orbits (16:1), presence of an elongate anteroventral splint of the jugal in the jugal–maxillary suture, which extends to the level of the M1 (23:3), pseudosylvian sulcus visible on the braincase (27:0 reversal), squamosal fossa divided and unequal in size (29:2), paroccipital process large but not excavated (30:2), stylomastoid foramen separated from tympanohyal by raised ridge (33:1), and a dorsoventrally and laterally projecting pterygoid strut (34:1). Both analyses recover a monophyletic Desmatophoca, which forms the sister group to a moderately supported Allodesminae (IW BS = 61; NW BS = 64). Monophyly of Desmatophoca in both analyses is poorly supported. Only one unequivocal synapomorphy can be identified for this genus, a jugal mortised with the zygomatic process of the squamosal with little dorsoventral development (22:1). Six unequivocal synapomorphies unite Allodesminae, including a well-developed prenarial shelf (3:1), a mortised squamosal–jugal articulation with dorsoventral development (22:2), loss of the angular process of the mandible (37:2) and posterior and medial carina of the C1 (46:1), and lower postcanines divergent in orientation (59:1) and lacking cingulum (61:2). Within Allodesminae, both analyses (Figs 9, 10) recover Atopotarus as the sister taxon to a moderately supported (IW BS = 60; NW BS = 51) clade comprising the remainder of allodesmine taxa. This clade is supported by six unequivocal synapomorphies [absence of an antorbital process (10:1), a small and posteriorly positioned supraorbital process of the frontal (15:3), absence of the styloid process of the tympanic bullae and retraction of the entotympanic posteriorly (32:1), a transversely expanded posterior lacerate foramen (36:1), and single-rooted P2–3 (49:2) and P4 (50:3)]. Two additional synapomorphies may diagnose this clade, or a more inclusive clade including Atopotarus: presence of an orbital vacuity (17:1) and a retracted dorsal margin of the maxillary root of the zygomatic arch (21:1). Relationships within Allodesmus are completely unresolved in the NW analysis, but the IW analysis shows a fair degree of resolution. Within the the latter analysis, the ‘broad head’ taxa Al. packardi and Al. naorai are recovered as successive sister taxa to a moderately supported (BS = 63) clade comprising the ‘long head’ subgroup. Two unequivocal synapomorphies unite Al. naorai with later diverging allodesmines: posterior termination of premaxilla extends to level of infraorbital foramen (7:1) and the presence of a small infraorbital foramen (9:1). The ‘long head’ clade is supported by six unequivocal synapomorphies [anterior margins of nasal well posterior to P1 (5:2), a transversely arched palate (12:1), reduced incisive foramina (13:2), absence of the lateral wall of the alisphenoid (20:1), a smoothly convex braincase (28:1), and single-rooted P2–4 (62:2)] and one equivocal synapomorphy (slightly divergent upper tooth rows; 14:0). Relationships are largely unresolved within the ‘long head’ clade, although Al. demerei was recovered as the sister taxon to the Round Mountain Silt Allodesmus with poor support. This latter clade is supported by two unequivocal synapomorphies: a distinctively concave ventral edge of the rostrum (4:2) and symphyseal angle of the mandible <40° (42:2). DISCUSSION Phylogeny of the Pinnipedimorpha Although our character sampling was focused on resolving relationships within the Allodesminae, our phylogenetic tree possessed several major differences in topology from past higher-level phylogenies of Pinnipedimorpha (Barnes, 1989; Berta, 1994; Berta & Wyss, 1994; Boessenecker & Churchill, 2013). These differences include the position of Pinnarctidion and recovery of a clade comprising Odobenidae and Otariidae. Prior phylogenetic hypotheses have generally considered Pinnarctidion as closely allied to desmatophocid seals, either as the sister taxon to Allodesminae (Barnes, 1989) or as a stem phocoid (Berta, 1994; Berta & Wyss, 1994), a clade comprising Phocidae and Desmatophocidae. Our phylogenetic tree presents a third alternative, placing Pinnarctidion as pinnipediform and as the sister taxon to Pinnipedia (see also Barnes & Hirota, 1995; Deméré & Berta, 2002). This clade is supported by three unambiguous synapomorphies, including a deep ramus of the mandible at the position of the posteroventral terminus of the symphysis (39:1), P2–M1 teeth which are all the same size (58:1), and a talonid basin that exists as a small concavity or shelf (67:1). In contrast, the only character shared between Pinnarctidion and Allodesminae is possession of a transversely expanded posterior lacerate foramen (36:1). Barnes (1989) lists an additional ten features shared between Pinnarctidion and Allodesminae, but nearly all of these are plesiomorphic and diagnose larger clades, are present in clades other than Allodesminae, or are poorly defined. The only shared feature found exclusively between both Desmatophocidae and Pinnarctidion is a dorsoventrally and laterally projecting pterygoid strut (34:1; character ten of Berta, 1994). Berta (1994) lists a total of 15 characters uniting Pinnarctidion with Phocidae and Desmatophocidae. Of those characters, ten are included within this study, including the formerly mentioned shared character. Of these ten characters, we interpret four of these as plesiomorphic and with a wider character distribution than presented by Berta (1994), including presence of a shelf-like pterygoid process of the maxilla (found in Proneotherium), posteroventral location of the optic foramina and anterior lacerate foramen (also found in most odobenids), enlargement of the paroccipital process (found in Pteronarctos and all later diverging pinnipediforms), and presence of a knob-like acromion process of the scapula (the plesiomorphic condition for pinnipeds). An additional character, lateral elongation and expansion of the posterior lacerate foramen, is shared with Allodesminae but absent in Desmatophoca. The remaining five characters present in our study have been modified, resulting in a different interpretation of the character coding as presented by Berta (1994). Characters defined in a slightly different manner and thus coded differently from Berta (1994) include palate shape (here defined as upper tooth row orientation), squamosal–jugal articulation, and separation of the stylomastoid foramen from the tympanohyal pit. Given the limited sampling of characters related to higher level pinniped phylogeny, we consider the phylogenetic position of Pinnarctidion to be unsettled at this point. Future studies, with more exhaustive sampling of characters pertinent for resolving relationships among early pinnipedimorphs, need to be undertaken to determine whether this genus is indeed more closely related to Desmatophocidae or to other pinniped groups. Non-cladistic studies of pinniped morphology (Barnes, 1989) have generally advocated a close relationship between Otariidae, Desmatophocidae, and Odobenidae. These studies have received additional support from analyses of molecular data (Arnason et al., 2006; Agnarsson et al., 2010), which have found Odobenidae to be the sister taxon to Otariidae. Although our phylogenetic analysis does not recover a clade representing the classical definition of Otarioidea, we do recover within our IW analysis a clade comprising Odobenidae and Otariidae, a relatively novel finding for morphology-based phylogenetic analysis of fossil pinnipeds. This relationship is, however, not recovered in our NW tree, which instead recovers a polytomy consisting of Otariidae, Phocoidea, Proneotherium, Neotherium, and a clade comprising the remainder of Odobenidae. Implied weighting can lead to increased resolution in phylogenetic analyses, but this increased resolution often comes at the price of decreased accuracy (Congreve & Lamsdell, 2016; Puttick et al., 2017). Although the results of IW analyses should be viewed cautiously, their congruence with prior molecular phylogenetic studies suggests that this relationship was correctly inferred. Two unequivocal synapomorphies exist for this clade: greater tuberosity of the humerus extends far proximal to humeral head and lesser tuberosity (73:1) and the presence of a separate foramen for the obturator nerve (87:1). As the focus of this study was the assess the phylogenetic relationships of taxa within Desmatophocidae, our sampling of taxa and character relevant to higher level pinniped phylogeny are limited. Large-scale analyses of pinniped phylogeny incorporating additional characters, taxa, and molecular data are needed to assess the validity of Otarioidea fully, and may change our understanding of what pinniped family desmatophocids are truly most closely related to. Taxonomy of the Allodesminae As many as one to four genera of Allodesminae have been recognized (Table 9), including Allodesmus (Kellogg, 1922), Atopotarus (Downs, 1956), Brachyallodesmus (Barnes & Hirota, 1995), and Megagomphos (Barnes & Hirota, 1995). Of these genera, both phylogenetic analyses suggest the presence of two diagnosable genera within Allodesminae: Atopotarus and Allodesmus. Atopotarus courseni from the Altamira Shale of California was originally recognized as a separate genus from Allodesmus by Downs (1956) on the basis of a nearly complete skeleton preserved in situ on a slab, missing only the hindlimbs. This species has either been recognized as a separate genus (Downs, 1956; Barnes & Hirota, 1995) or as a species of Allodesmus (Mitchell, 1966; Barnes, 1972). More recent studies have included it as a species within Allodesmus (Kohno, 1996; Deméré & Berta, 2002). No explicit reasons were given for the recognition of one genus of Allodesminae by Deméré & Berta (2002), other than a statement that the Allodesminae were ‘over-split’. Table 9. Published records of Desmatophocidae Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) View Large Table 9. Published records of Desmatophocidae Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) View Large Our phylogenetic analysis places At. courseni as the sister taxon to all other allodesmine seals, and this genus can be differentiated by the absence of a prenarial shelf and M2 (also absent in Al. sadoensis), double-rooted cheek teeth, a small, triangular postorbital process, and a mastoid process projecting ventral to the postglenoid process. In light of these differences and placement of this species as the sister taxon to all other Allodesmus, we follow Barnes and Hirota (1995) in retaining a separate genus for Atopotarus. Further discoveries of better preserved cranial material of At. courseni and other archaic allodesmines (Velez-Juarbe, 2017) from the early middle Miocene will help in re-evaluation of the generic distinctiveness of Atopotarus. Allodesmus packardi from the ‘Ladera Sandstone’ of California was originally reported by Packard (1962), who referred the partly prepared skull to Atopotarus sp., cf. At. courseni; Mitchell (1966) agreed with this identification. Barnes (1972) re-described the specimen following additional preparation and erected the new species Al. packardi, noting various craniodental features differentiating it from Al. kernensis (Barnes, 1972: 41–42). Barnes and Hirota (1995) later erected the new genus Brachyallodesmus to contain this species, noting several plesiomorphic differences placing it outside Allodesmus, including canines with oval cross-section, more inflated tympanic bulla, a large pre-glenoid process, complete lateral wall of alisphenoid canal, a well-developed inferior petrosal venous sinus, a less cubic mastoid process, and retention of a pseudosylvian sulcus on the braincase. Barnes and Hirota (1995) further distinguished Brachyallodesmus packardi from Allodesmus spp. on the basis of unique derived features, including the lack of sagittal and nuchal crests, widely diverging tooth rows, cheek teeth with longitudinal sulci, and P1 positioned posteromedial to the C1. Our IW phylogenetic analysis finds Al. packardi to be the sister taxon to a clade including Al. naorai and all later diverging Allodesmus, although with very little support. Of the unique derived features listed by Barnes and Hirota (1995), widely diverging tooth rows are shared with Al. naorai (character 14 of our study). Extent and size of nuchal and sagittal crests are prone to sexual dimorphism and ontogenetic variation, and without a larger sample size we hesitate to consider their absence a valid diagnostic character. A similar problem can be found with using position of the P1, which may be influenced by size of the canine, another sexually dimorphic trait. Finally, presence of cheek teeth with a longitudinal sulcus is related to progression from double-rooted to single-rooted cheek teeth. This can be somewhat variable in pinnipeds, and pending further work on dental variation in pinnipeds we do not treat it as a separate intermediate state between single- and double-rooted cheek teeth. Given the relatively poor support for Al. packardi within our phylogenetic tree, and the lack of resolution within Allodesmus in our NW tree, we consider Brachyallodesmus a junior synonym of Allodesmus and follow Deméré and Berta (2002) in considering this species as Al. packardi as originally reported by Barnes (1972). Megagomphos represents the fourth putative genus of allodesmine. The holotype specimen is a large articulated anterior portion of the rostrum and mandible, and was originally reported by Nagao (1941) as Eumetopias sinanoensis; its desmatophocid affinities were later recognized by Mitchell (1966), who synonymized it with Al. kernensis; Repenning and Tedford (1977) considered it to represent a separate species within Allodesmus on account of its gigantic size. Barnes and Hirota (1995) later placed it outside Allodesmus owing to lack of a prenarial shelf, erecting the new genus Megagomphos. In the same study, Barnes and Hirota (1995) named the new species Allodesmus megallos based on a gigantic rostrum with procumbent incisors and canines. Both holotypes were collected from a narrow stratigraphic interval, the lower Aoki Formation (Barnes & Hirota, 1995; Kohno, 1996). In contrast, Kohno (1996) referred both these specimens to the same taxon, which he recombined as Al. sinanoensis. Kohno’s (1996) referral of both specimens to the same species was based upon the large size of both specimens, presence of a prenarial shelf, and origination from the same stratigraphic unit (Serravallian equivalent Aoki Formation). Despite the fact that both specimens originated from the Aoki Formation, Barnes and Hirota (1995) did not publish any comparisons between them. The holotype of Megagomphos sinanoensis is transversely crushed (Barnes & Hirota, 1995; Kohno, 1996), with damage obscuring the prenarial shelf. However, Kohno (1996) indicated that a prenarial shelf is present given that the bony nares are posteriorly retracted to a degree similar to other Allodesmus spp. He identified several features shared between these, including a prenarial shelf and large, procumbent, tusk-like canines and incisors. Owing to these shared similarities and their origination from a narrow stratigraphic interval, we follow Kohno (1996) in recognizing a single species, Al. sinanoensis, and consider Megagomphos a junior synonym of Allodesmus and Allodesmus megallos a junior synonym of Al. sinanoensis. We note that these specimens may be differentiated from other Japanese desmatophocids: from Al. sadoensis owing to its larger size, proportionally larger I3, proportionally smaller postcanines, and presence of postcanine diastemata; and from Al. naorai in its much larger size, anteriorly widening prenarial shelf, and proportionally shorter and posteriorly retracted nasals. Additional material of Al. sinanoensis was reported by Kohno et al. (2007). Desmatophocid diversity from the Round Mountain Silt Allodesmus kernensis was originally reported by Kellogg (1922) on the basis of the partial holotype mandible and other isolated cranial and postcranial elements from the Round Mountain Silt member of the Temblor Formation in the vicinity of Sharktooth Hill, Kern County, CA, USA. Mitchell (1966) reported a nearly complete skeleton collected from the Sharktooth Hill Bonebed, and noted subtle and unquantified mandibulodental differences with Al. kernensis (narrower symphyseal angle, more erect canine, narrower interalveolar septa, larger alveoli, and I2 positioned posteromedial rather than posterior to I3; Mitchell, 1966). Mitchell (1966) remarked upon problematical locality information for the holotype mandible and speculated that the holotype was collected three miles east of the bonebed and therefore stratigraphically below the Sharktooth Hill Bonebed. Owing to these subtle differences and potential stratigraphic separation, Mitchell (1966) erected the species Al. kelloggi for the new skeleton from the bonebed and referred additional specimens from the bonebed to this new species. Barnes (1970, 1972) reported many new specimens of Allodesmus from the bonebed including mandibles and evaluated morphological variation of Allodesmus mandibles. Barnes (1970, 1972) found that the morphological characters (e.g. symphyseal angle) reported by Mitchell (1966) as distinguishing Al. kelloggi from Al. kernensis fell within the range of variation of modern pinnipeds (Z. californianus) and declared Al. kelloggi a junior synonym of Al. kernensis. Barnes (1972) further indicated that contradictory locality information exists for the Al. kernensis holotype, far from clearly indicating a precise locality or stratigraphic interval. Furthermore, Hanna (in Barnes, 1972) considered that unless prolific collector Charles Morrice collected the specimen himself, the specimen more than likely came from the bonebed rather than a different horizon within the Round Mountain Silt; we infer this to mean that Hanna considered collectors other than Morrice lacked stratigraphic expertise and would have been likely to have collected specimens only from the highly prolific bonebed rather than sparsely fossiliferous strata above and below. A very different approach was later advocated by Barnes and Hirota (1995), who reconsidered the taxonomy of Round Mountain Silt Allodesmus spp. in light of new discoveries from Japan and examination of more desmatophocid fossil material. These authors resurrected Al. kelloggi, restricted Al. kernensis and Al. kelloggi to their respective holotype specimens, and erected the new species Al. gracilis, referring all bonebed specimens aside from the Al. kelloggi type to this new species. Barnes and Hirota (1995) further indicated that Al. kelloggi differs from all Al. gracilis specimens in bearing a deep and rugose masseteric fossa, a transversely expanded flange on the coronoid crest, and a single rooted M1. However, Barnes and Hirota (1995) did not re-address the statistical results which Barnes (1970, 1972) cited while synonymizing Al. kelloggi, nor did they quantify any of these differences. Cranial differences between Al. kelloggi and Al. gracilis cited by Barnes and Hirota (1995: 347) are likewise not quantified and are subjective or difficult to quantify (e.g. ‘differing from Al. gracilis by having zygomatic process of squamosal more curved and bowed laterally, not so straight’). We agree with Barnes (1970, 1972) that mandible and dental differences amongst Round Mountain Silt Allodesmus spp. are attributable to intraspecific variation; furtherore, we note that Barnes (1970, 1972) demonstrated that some aspects of mandibular variation cited earlier by Mitchell (1966) and later by Barnes and Hirota (1995) to separate Al. kernensis from other Sharktooth Hill mandibles are actually less extreme than within extant Z. californianus. Furthermore, even if the holotype mandible of Al. kernensis originated from a stratigraphically lower horizon than the Sharktooth Hill bonebed, extinct species are diagnosed based upon morphology, not age. Geochronologically older specimens falling within a range of variation acceptable for referral to a slightly younger species simply expand the geochronological range of a species and do not necessitate naming of new taxa. We therefore identify Al. gracilis and Al. kelloggi as junior synonyms of Al. kernensis (Table 9), in broad agreement with Deméré and Berta (2002: 140). Although the previously described allodesmine taxa from Sharktooth Hill represent a single taxon, additional allodesmine taxa may have been present in the fauna. One intriguing fossil, referred to by Barnes (1972) as ‘Desmatophocine B’, is known from an isolated mandible with large single-rooted teeth. This specimen notably differs from Al. kernensis in having a deeper symphyseal region (and wider symphyseal angle) and absence of postcanine diastemata and M2 (polymorphic in Al. kernensis; Barnes, 1970, 1972). Barnes (1972) considered this mandible to represent a desmatophocid outside the genus Allodesmus. We also find this specimen interesting and note several distinct features that separate this taxon from Al. kernensis, including loss of the M2 and presence of premolars with large roots crowded into the anterior mandible and with dorsally diverging orientations. These features are all shared with Al. sadoensis to the exclusion of all other Allodesmus spp. Given that ‘Desmatophocine B’ differs in only minor ways from Al. sadoensis (transversely narrower mandible, shallower mandibular angle), we reidentify ‘Desmatophocine B’ as Allodesmus sp., cf. Al. sadoensis. Another desmatophocid taxon recognized by Barnes (1972), ‘Desmatophocine C’, is based on a mandible fragment lacking teeth, which he separated from Al. kernensis based on only the transversely wider symphyseal portion of the mandible. As this difference was not quantified and seems indistinguishable from that preserved in juvenile specimens of Al. kernensis (e.g. UCMP 81704; Barnes, 1972: fig. 13b), we reidentify this specimen as Allodesmus sp., cf. Al. kernensis. Extinction of the Desmatophocidae Allodesmus demerei represents the geochronologically youngest species of Allodesmus and the family Desmatophocidae (Fig. 9), and the discovery of this taxon invites the question, ‘Why did desmatophocids become extinct?’ No obvious mass extinctions amongst marine vertebrates have been recorded during the middle to late Miocene; on the contrary, cetacean diversity peaked during the Tortonian (Uhen & Pyenson, 2007; Marx & Uhen, 2010). Primary productivity appears to have peaked during the Serravalian–Tortonian interval (Fig. 11). We find no correlation between changes in productivity and desmatophocid extinction, and this group of pinnipeds seems to be declining while other pinniped lineages and whales are actually increasing in diversity (Marx & Uhen, 2010; Churchill et al., 2014) Change in sea level could radically change the geography of haul-out areas and extent of rocky versus sandy shores and has been implicated in the extinction of Mio-Pliocene phocids in South America (Valenzuela-Toro et al., 2013), but no obvious mid-late Miocene changes in sea level are apparent (Fig. 11). The peak in desmatophocid diversity does, however, correspond to the mid-Miocene climatic optimum, and diversity begins to decline along with falling temperatures (Fig. 11). Figure 11. View largeDownload slide Trends in North Pacific climate and pinniped diversity. Diatom diversity from Marx and Uhen, (2010), sea level relative to modern from Miller et al., (2005), and δ18O values (proxy for sea surface temperature) from Fedorov et al., (2013) and Rousselle et al., (2013). Data are plotted only from the late Oligocene to Holocene. See Supporting Information, Appendix S5 for a complete list of pinniped fossil occurrences. Figure 11. View largeDownload slide Trends in North Pacific climate and pinniped diversity. Diatom diversity from Marx and Uhen, (2010), sea level relative to modern from Miller et al., (2005), and δ18O values (proxy for sea surface temperature) from Fedorov et al., (2013) and Rousselle et al., (2013). Data are plotted only from the late Oligocene to Holocene. See Supporting Information, Appendix S5 for a complete list of pinniped fossil occurrences. The paraphyletic ‘enaliarctines’ (early diverging Pinnipedimorpha, or stem Pinnipedia) represent the earliest diversification of pinnipeds, originating in the late Chattian, peaking in the Aquitanian (N = 6), and extinct by the end of the Burdigalian (Fig. 11). Desmatophocids appear during the Aquitanian and gradually increase in diversity until the Langhian (N = 6) and decline in diversity until the Tortonian (N = 1; Allodesmus demerei, 10.5–9.1 Mya). Desmatophocids then apparently became extinct by the end of the Tortonian (Figs 9, 11). Otariids appear in the Burdigalian (N = 2; Boessenecker & Churchill, 2015; Velez-Juarbe, 2017) but lack a Langhian record; they continue in low diversity (and small body size) until diversifying in the Messinian (N = 5) and Pliocene (N = 6) and mostly maintaining that diversity until the Holocene (Fig. 11). Walruses were ancestrally small in body size (Churchill et al., 2014). These small walruses, which were morphologically very similar to their ‘enaliarctine’ ancestors, were somewhat diverse in the Burdigalian (N = 4) and remained at low diversity through to the Tortonian, after which all species of odobenids are large bodied (Fig. 11). Large odobenids were low in diversity in the Burdigalian and gradually increase until their peak in the Messinian (N = 9) and Pliocene (N = 8), with their diversity decreasing sharply in the Pleistocene (Fig. 11). Phocids do not have a pre-Pleistocene fossil record in the North Pacific (Boessenecker, 2013) and quickly diversify after invading from the Arctic and South Pacific (Deméré et al., 2002; Arnason et al., 2006; Boessenecker & Churchill, 2016; Fig. 11). One obvious relationship between desmatophocid diversity and trends in other pinniped groups is the gradual increase in large walrus diversity as desmatophocid diversity declines (Fig. 11; see also Velez-Juarbe, 2017), suggesting a possible case of competitive displacement by walruses. On the contrary, otariid diversity increases only after desmatophocids are extinct, perhaps suggesting opportunistic niche filling. These data do not distinguish between generalist ‘imagotariine’ walruses and more specialized molluscivorous odobenines; however, we note that several Tortonian odobenids, including Imagotaria, Pontolis, Gomphotaria, and at least one new genus of ‘imagotariine’, lack obvious specialization for benthic feeding (unlike the Odobenini) and may have been exploiting similar prey resources as desmatophocid seals (Adam & Berta, 2002). Alternatively, odobenids and desmatophocids could also have been in competition for haul-out sites for breeding and moulting. Beyond competition for haul-out sites, comparisons of climate trends with pinniped diversity over time may also suggest an important role for climate change in the faunal turnover between desmatophocids and walruses, with walruses perhaps better adapted in some way to cooler environments. This, over time, might have led to the decline and eventual extinction of desmatophocids, while allowing walruses to proliferate and diversify. What behavioural, ecological, and morphological adaptations walruses may have had that desmatophocids lacked cannot be determined at this point but should be the focus of future studies of pinniped paleoecology and evolution. CONCLUSION A new skeleton of a desmatophocid seal from the upper Miocene Montesano Formation of Washington is described as a new species within the genus Allodesmus, Al. demerei sp. nov. Allodesmus demerei represents the geochronologically youngest desmatophocid, coexisting with imagotariine walruses and early otariid fur seals. Phylogenetic analysis supports placement of this taxon within the genus Allodesmus. The genera Brachyallodesmus and Megagomphos are synonymized with Allodesmus, while the genus Atopotarus is retained. Examination of diversity trends shows peak desmatophocid diversity coinciding with the middle Miocene climatic optimum, and desmatophocids decline as odobenids diversify. This suggests a role for climate change as well as competition with the rapidly diversifying walruses in the decline and extinction of the desmatophocids. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Appendix S1. Institutional acronyms. Appendix S2. List of material examined. Appendix S3. Character list for cladistic analysis. Appendix S4. Explanation of geochronological ranges. Appendix S5. Table of fossil pinniped occurrences from the North Pacific. Appendix S6. References cited. Appendix S7. Cladistic matrix for Allodesmus demerei [Version of Record, published online 03 January 2018; http://zoobank.org/urn:lsid:zoobank.org:pub:4DE5001F-A4BE-4A43-84A4-AEA693B3FA14] ACKNOWLEDGEMENTS First and foremost, we wish to thank the late P. K. Bigelow for discovering, collecting, and initiating study of UWBM 75640. We also thank M. Kelly for collecting, preparing, and donating the referred humerus. Comments from the editor and two anonymous reviewers greatly improved the quality of this paper. This study benefitted from discussions with L. G. Barnes, A. Berta, T. A. Deméré, A. Garibay, N. Kohno, J. F. Parham, and Y. Tanaka. Thanks to the following, who provided access to specimens and collections under their care: L. G. Barnes, S. A. McLeod, and V. Rhue (LACM), M. B. Goodwin, and P. A. Holroyd (UCMP), D. J. Bohaska, C. Potter, and N. D. Pyenson (USNM), R. C. Eng, M. Rivin, and C. A. Sidor (UWBM); additional thanks to M. Rivin for facilitating a loan of UWBM 75640 to R.W.B. This study was generously funded by a University of Washington Burke Museum of Natural History and Culture Vertebrate Paleontology Collection Study Grant to R.W.B.; travel grants supporting collections visits include a University of Otago Doctoral Scholarship to R.W.B. REFERENCES Adam PJ , Berta A . 2002 . Evolution of prey capture strategies and diet in the Pinnipedimorpha (Mammalia: Carnivora) . Oryctos 4: 83 – 107 . Addicott WO . 1976 . Neogene molluscan stages of Oregon and Washington . In: Fritsche AE , TerBest HJ , Wornardt WW , eds. The Neogene Symposium: Society of Economic Paleontologists and Mineralogists, Pacific Section . 95 – 115 . Agnarsson I , Kuntner M , May-Collado LJ . 2010 . Dogs, cats, and kin: a molecular species-level phylogeny of Carnivora . Molecular Phylogenetics and Evolution 54 : 726 – 745 . Google Scholar Crossref Search ADS PubMed Arnason U , Gullberg A , Janke A , Kullberg M , Lehman N , Petrov EA , Väinölä R . 2006 . Pinniped phylogeny and a new hypothesis for their origin and dispersal . Molecular Phylogenetics and Evolution 41 : 345 – 354 . Google Scholar Crossref Search ADS PubMed Bapst DW . 2012 . Paleotree: an R package for paleontological and phylogenetic analyses of evolution . Methods in Ecology and Evolution 3 : 803 – 807 . Google Scholar Crossref Search ADS Barnes LG . 1970 . A re-evaluation of mandibles of Allodesmus (Otariidae, Carnivora) from the Round Mountain Silt, Kern County, California . PaleoBios 10 : 1 – 24 . Barnes LG . 1972 . Miocene Desmatophocinae (Mammalia: Carnivora) from California . University of California Publications in Geological Sciences 89 : 1 – 76 . Barnes LG . 1979 . Fossil enaliarctine pinnpeds (Mammalia: Otariidae) from Pyramid Hill, Kern County, California . Contributions in Science, Natural History Museum of Los Angeles County 318 : 1 – 41 . Barnes LG . 1987 . An early Miocene pinniped of the genus Desmatophoca (Mammalia: Otariidae) from Washington . Contributions in Science, Natural History Museum of Los Angeles County 382 : 1 – 20 . Barnes LG . 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 – 28 . Barnes LG . 2008 . Otarioidea . In: Janis CM , Gunnell GF , Uhen MD , eds. Evolution of Tertiary mammals of North America, Vol. 2: Small mammals, xenarthrans, and marine mammals . Cambridge : Cambridge University Press . 523 – 541 . Barnes LG , Hirota K . 1995 . Miocene pinnipeds of the otariid subfamily Allodesminae in the North Pacific Ocean: systematics and relationships . The Island Arc 3 : 329 – 360 . Google Scholar Crossref Search ADS Barnes LG , Mitchell ED . 1975 . Late Cenozoic northeast Pacific Phocidae . Conseil International pour l’Exploration de la Mer, Rapports et Procèsverbaux des Rèunions 169: 34 – 62 . Barnes LG , Raschke RE . 1991 . Gomphotaria pugnax, a new genus and species of late Miocene dusignathine otariid pinniped (Mammalia: Carnivora) from California . Natural History Museum of Los Angeles County Contributions in Science 426 : 1 – 27 . Bebej RM . 2009 . Swimming mode inferred from skeletal proportions in the fossil pinnipeds Enaliarctos and Allodesmus (Mammalia, Carnivora) . Journal of Mammalian Evolution 16 : 77 – 97 . Google Scholar Crossref Search ADS Berta A . 1994 . A new species of phocoid pinniped Pinnarctidion from the early Miocene of Oregon . Journal of Vertebrate Paleontology 14 : 405 – 413 . Google Scholar Crossref Search ADS Berta A , Ray CE . 1990 . Skeletal morphology and locomotor capabilities of the archaic pinniped Enaliarctos mealsi . Journal of Vertebrate Paleontology 10 : 141 – 157 . Google Scholar Crossref Search ADS Berta A , Ray CE , Wyss AR . 1989 . Skeleton of the oldest known pinniped, Enaliarctos mealsi . Science 244 : 60 – 62 . Google Scholar Crossref Search ADS PubMed Berta A , Wyss AR . 1994 . Pinniped phylogeny . Proceedings of the San Diego Society of Natural History 29 : 33 – 56 . Bianucci G , Di Celma C , Urbina M , Lambert O . 2016 . New beaked whales from the late Miocene of Peru and evidence for convergent evolution in stem and crown Ziphiidae (Cetacea, Odontoceti) . PeerJ 4 : e2479 . Google Scholar Crossref Search ADS PubMed Bigelow PK . 1994 . Occurrence of a squaloid shark (Chondrichthyes: Squaliformes) with the pinniped Allodesmus from the upper Miocene of Washington . Journal of Paleontology 68 : 680 – 684 . Google Scholar Crossref Search ADS Boessenecker RW . 2013 . A new marine vertebrate assemblage from the Late Neogene Purisima Formation in Central California, Part II: Pinnipeds and cetaceans . Geodiversitas 35 : 815 – 940 . Google Scholar Crossref Search ADS Boessenecker RW , Churchill M . 2013 . A reevaluation of the morphology, paleoecology, and phylogenetic relationships of the enigmatic walrus Pelagiarctos . PLoS ONE 8 : e54311 . Google Scholar Crossref Search ADS PubMed Boessenecker RW , Churchill M . 2015 . The oldest known fur seal . Biology Letters 11 : 20140835 . Google Scholar Crossref Search ADS PubMed Boessenecker RW , Churchill M . 2016 . The origin of elephant seals: implications of a fragmentary late Pliocene seal (Phocidae: Miroungini) from New Zealand . New Zealand Journal of Geology 59 : 544 – 550 . Google Scholar Crossref Search ADS Bowditch TE . 1921 . An analysis of the natural classifications of Mammalia for the use of students and travelers . Paris : J. Smith . Churchill M , Boessenecker RW , Clementz MT . 2014 . Colonization of the Southern Hemisphere by fur seals and sea lions (Carnivora: Otariidae), revealed by combined evidence phylogenetic and Bayesian biogeographic analysis . Zoological Journal of the Linnean Society 172 : 200 – 225 . Google Scholar Crossref Search ADS Churchill M , Clementz MT . 2015 . The evolution of aquatic feeding in seals: insights from Enaliarctos (Carnivora: Pinnipedimorpha), the oldest known seal . Journal of Evolutionary Biology 29 : 319 – 334 . Google Scholar Crossref Search ADS PubMed Churchill M , Clementz MT , Kohno N . 2014 . Cope’s rule and the evolution of body size in Pinnipedimorpha (Mammalia: Carnivora) . Evolution 69 : 201 – 215 . Google Scholar Crossref Search ADS PubMed Condon T . 1906 . A new fossil pinniped (Desmatophoca oregonensis) from the Miocene of the Oregon coast . University of Oregon Bulletin 3 : 1 – 14 . Congreve CR , Lamsdell JC . 2016 . Implied weighthing and its utility in paleontological datasets: a study using modeled phylogenetic matrices . Paleontology 59 : 44 – 462 . Google Scholar Crossref Search ADS Debey LB , Pyenson ND . 2013 . Osteological correlates and phylogenetic analysis of deep diving in living and extinct pinnipeds: what good are big eyes ? Marine Mammal Science 29 : 48 – 83 . Google Scholar Crossref Search ADS Deméré TA . 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 TA , Berta A . 2001 . A reevaluation of Proneotherium repenningi from the Miocene Astoria Formation of Oregon and its position as a basal odobenid (Pinnipedia: Mammalia) . Journal of Vertebrate Paleontology 21 : 279 – 310 . Google Scholar Crossref Search ADS Deméré TA , Berta A . 2002 . The Miocene pinniped Desmatophoca oregonensis Condon, 1906 (Mammalia: Carnivora) from the Astoria Formation, Oregon . Smithsonian Contributions to Paleobiology93: 113 – 147 . Deméré TA , Berta A , Adam PJ . 2003 . Pinnipedimorph evolutionary biogeography . Bulletin of the American Museum of Natural History 279 : 32 – 76 . Google Scholar Crossref Search ADS Downs T . 1955 . A fossil sea lion from the Miocene of the San Joaquin Hills, Orange County, California . Bulletin of the Southern California Academy of Science 54 : 49 – 56 . Downs T . 1956 . A new pinniped from the Miocene of southern California: with remarks on the Otariidae . Journal of Paleontology 30 : 115 – 131 . Fedorov AV , Brierley CM , Lawrence KT , Liu Z , Dekens PS , Ravelo AC . 2013 . Patterns and mechanisms of early Pliocene warmth . Nature 496 : 44 – 49 . Google Scholar Crossref Search ADS Fowler GA . 1965 . The stratigraphy, foraminifera, and paleoecology of the Montesano Formation, Grays Harbor County, Washington . Unpublished Ph.D. Thesis , University of Southern California . 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. Thesis , San Diego State University . Giffin EB . 1992 . Functional implications of neural canal anatomy in recent and fossil marine carnivores . Journal of Morphology 214 : 357 – 374 . Google Scholar Crossref Search ADS PubMed Gradstein FM , Ogg JG , Schmitz M , Ogg G . 2012 . The geologic time scale 2012 . Oxford : Elsevier . Hay OP . 1930 . Second bibliography and catalogue of the fossil Vertebrata of North America . Carnegie Institution of Washington 390 : 1 – 1074 . Hunt RM Jr. , Barnes LG . 1994 . Basicranial evidence for ursid affinity of the oldest pinnipeds . Proceedings of the San Diego Society of Natural History 29 : 57 – 67 . Illiger C . 1811 . Prodromus systematis mammalium et avium . Berlin : Salfeld . Kellogg R . 1922 . Pinnipeds from Miocene and Pleistocene deposits of California . University of California Publications in Geological Sciences 13 : 23 – 132 . Kellogg R . 1931 . Pelagic mammals from the Temblor Formation of the Kern River region, California . Proceedings of the California Academy of Sciences 19 : 217 – 397 . Kimura M , Hirota K , Kiyono C . 1997 . Fossil pinniped mandible from the middle Miocene of Haboro-cho, Hokkaido . Bulletin of the Hobetsu Museum 13 : 1 – 7 . Kohno N . 1994 . A new Miocene pinniped in the genus Prototaria (Carnivora: Odobenidae) from the Moniwa Formation, Miyagi, Japan . Journal of Vertebrate Paleontology 14 : 414 – 426 . Google Scholar Crossref Search ADS Kohno N . 1996 . Miocene pinniped Allodesmus (Mammalia: Carnivora) with special reference to the “Mito seal” from Ibaraki Prefecture, central Japan . Transactions and Proceedings of the Palaeontological Society of Japan 181 : 388 – 404 . Kohno N . 2006 . A new Miocene odobenid (Mammalia: Carnivora) from Hokkaido, Japan, and its implications for odobenid phylogeny . Journal of Vertebrate Paleontology 26 : 411 – 421 . Google Scholar Crossref Search ADS Kohno N , Koike H , Narita K . 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 . Koretsky IA , Barnes LG , Rahmat SJ . 2016 . Re-evaluation of morphological characters questions current views of pinniped origins . Vestnik Zoologii 50 : 327 – 354 . Google Scholar Crossref Search ADS Lambert O , Bianucci G , Muizon CD . 2017 . Macroraptorial sperm whales (Cetacea, Odontoceti, Physeteroidea) from the Miocene of Peru . Zoological Journal of the Linnean Society 179 : 404 – 474 . Linnaeus C . 1758 . Systema naturae per regna tria naturae, secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis . Holmiae, Stockholm : Tomus I. Laurentii Salvii . Marx FG , Uhen MD . 2010 . Climate, critters, and cetaceans: Cenozoic drivers of the evolution of modern whales . Science 327 : 993 – 996 . Google Scholar Crossref Search ADS PubMed Miller KG , Kominz MA , Browning JV , Wright JD , Mountain GS , Katz ME , Sugarman PJ , Cramer BS , Christie-Blick N , Pekar SF . 2005 . The Phanerozoic record of global sea-level change . Science 310 : 1293 – 1297 . Google Scholar Crossref Search ADS PubMed Mitchell ED . 1966 . The Miocene pinniped Allodesmus . University of California Publications in Geological Sciences 61 : 1 – 105 . Mitchell ED , Tedford RH . 1973 . The Enaliarctinae: A new group of extinct aquatic carnivora and a consideration of the origin of the otariidae . Bulletin of the American Museum of Natural History 151 : 203 – 284 . Miyazaki S , Horikawa H , Kohno N , Hirota K , Kimura M , Hasegawa Y , Tomida Y , Barnes LG , Ray CE . 1995 . Summary of the fossil record of pinnipeds of Japan, and comparisons with that from the eastern North Pacific . The Island Arc 3 : 361 – 372 . Google Scholar Crossref Search ADS 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, Series 4 6 : 75 – 84 . Packard EL . 1962 . Fossil marine mammals from the vicinity of Stanford University . Journal of Paleontology 36 : 29 – 37 . Pierce SE , Clack JA , Hutchison JH . 2011 . Comparative axial morphology in pinnipeds and its correlation with aquatic locomotory behaviour . Journal of Anatomy 219 : 502 – 514 . Google Scholar Crossref Search ADS PubMed Prothero DR , Lau JN . 2001 . Magnetic stratigraphy of the upper Miocene (type Wishkahan-Graysian) Montesano Formation, Grays Harbor County, Washington . In: Prothero DR , ed. Magnetic stratigraphy of the Pacific Coast Cenozoic . Los Angeles, California : Society of Economic Paleontologists and Mineralogists, Pacific Section . 293 – 301 . Prothero DR , Lau JN , Armentrout JM . 2001 . Magnetic stratigraphy of the upper Miocene (Wishkahan) Empire Formation, Coos County, Oregon . In: Prothero DR , ed. Magnetic stratigraphy of the Pacific Coast Cenozoic . Los Angeles, California : Society of Economic Paleontologists and Mineralogists, Pacific Section . 284 – 292 . Puttick MN , O’Reilly JE , Tanner AR , Fleming JE , Clark J , Holloway L , Lozano-Fernandez J , Parry LA , Tarver JE , Pisani D , Donoghue PCJ . 2017 . Uncertain-tree: discriminating among competing approaches to the phylogenetic analysis of phenotype data . Proceedings of the Royal Society B: Biological Sciences 284 : 20162290 . Google Scholar Crossref Search ADS Rau WW . 1967 . Geology of the Wynoochee Valley quadrangle, Grays Harbor County, Washington . Washington Division of Mines and Geology Bulletin 56 : 1 – 51 . Repenning CA , Tedford RH . 1977 . Otarioid seals of the Neogene . US Geological Survey Professional Paper 992 : 1 – 87 . Rousselle G , Beltran C , Sicre M , Raffi I , De Rafélis M . 2013 . Changes in sea-surface conditions in the equatorial Pacific during the middle Miocene-Pliocene as inferred from coccolith geochemistry . Earth and Planetary Science Letters 361 : 412 – 421 . Google Scholar Crossref Search ADS Scheffer VB . 1958 . Seals, sea lions, and walruses . Stanford : Stanford University Press . Sivertsen E . 1954 . A survey of eared seals (family Otariidae) with remarks on the Antarctic seals collected by M/K “Norvegia” in 1928–1929 . Det Vorske Videnkaps-Akademii Oslo 36 : 1 – 76 . Tanaka Y , Kohno N . 2015 . A new late Miocene odobenid (Mammalia: Carnivora) from Hokkaido, Japan suggests rapid diversification of basal Miocene odobenids . PLoS ONE 10 : e0131856 . Google Scholar Crossref Search ADS PubMed Tonomori W , Sawamura H , Sato T , Kohno N . 2016 . New skeletal material of the extinct Miocene pinniped Allodesmus from Japan indicates their new systematics and locomotory adaptation . Journal of Vertebrate Paleontology, Program and Abstracts 2016 : 237 . Uhen MD , Pyenson ND . 2007 . Diversity estimates, biases, and historiographic effects: resolving cetacean diversity in the Tertiary . Palaeontologia Electronica 10: 1 – 22 . Valenzuela-Toro AM , Gutstein CS , Varas-Malca RM , Suarez ME , Pyenson ND . 2013 . Pinniped turnover in the South Pacific Ocean: new evidence from the Plio-Pleistocene of the Atacama Desert, Chile . Journal of Vertebrate Paleontology 33 : 216 – 223 . Google Scholar Crossref Search ADS Velez-Juarbe J . 2017 . Eotaria citrica, sp. nov., a new stem otariid from the “Topanga” formation of Southern California . PeerJ 5 : e3022 . Google Scholar Crossref Search ADS PubMed Walsh TJ , Korosec MA , Phillips WM , Logan RL , Schasse HW . 1987 . Geologic map of Washington - southwest quadrant . Washington Division of Geology and Earth Resources Geologic Map GM-34 : 1 – 28 . Wiens JJ . 1999 . Polymorphism in systematics and conservation biology . Annual Review of Ecology and Systematics 30 : 327 – 362 . Google Scholar Crossref Search ADS Wyss AR . 1987 . The walrus auditory region and the monophyly of pinnipeds . American Museum Novitates 2871 : 1 – 31 . © 2018 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Zoological Journal of the Linnean Society Oxford University Press

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

Loading next page...
 
/lp/ou_press/the-last-of-the-desmatophocid-seals-a-new-species-of-allodesmus-from-JHRpqYR8rn
Publisher
Oxford University Press
Copyright
© 2018 The Linnean Society of London, Zoological Journal of the Linnean Society
ISSN
0024-4082
eISSN
1096-3642
D.O.I.
10.1093/zoolinnean/zlx098
Publisher site
See Article on Publisher Site

Abstract

Abstract The family Desmatophocidae represents an early radiation of extinct pinnipeds that peaked in diversity during the middle Miocene. Although represented by abundant well-preserved fossils, the taxonomy and evolutionary relationships of this family remain poorly known. Late Miocene desmatophocids have been recorded, although none have been formally described, preventing a thorough appraisal of their decline and extinction. We report the discovery of a new species, Allodesmus demerei sp. nov., represented by a partial skeleton with cranium, mandibles, and axial skeleton, from the upper Miocene Montesano Formation of Washington, prompting reinterpretation of desmatophocid taxonomy, phylogeny, and extinction. Phylogenetic analysis (95 characters, 26 taxa) found strong support for monophyletic Desmatophocidae and Allodesmus. Desmatophocidae was found as sister to Phocidae with poor support. Allodesmus demerei was placed within the Allodesmus as the sister taxon to Allodesmus kernensis. The geochronologically young age (10.5–9.1 Mya) of Al. demerei establishes this species as the last of the desmatophocid seals. The middle Miocene peak in desmatophocid diversity coincides with the middle Miocene climatic optimum, suggesting that declining sea surface temperature played a role in their decline and extinction. Walruses diversified and increased in body size during the mid- to late Miocene as desmatophocids declined, suggesting some form of ecological displacement. Carnivora, Pinnipedia, Desmatophocidae, Allodesmus, Miocene, Washington, Montesano Formation, Mammalia, extinction INTRODUCTION The Desmatophocidae are an extinct monophyletic group of seal-like pinnipeds that diversified during the early and middle Miocene and are the first large-bodied pinnipeds to evolve. They comprise three to five genera: Desmatophoca, Atopotarus, Brachyallodesmus, Megagomphos, and Allodesmus; all are generally characterized by bulbous cheek teeth, large orbits, a mortised jugal–squamosal articulation, and forelimbs similar to extant otariids (fur seals and sea lions) indicative of forelimb-dominated rowing (Mitchell, 1966; Barnes, 1972; Giffin, 1992; Barnes & Hirota, 1995; Bebej, 2009; Pierce et al., 2011; Debey & Pyenson, 2013). Allodesmus in particular is characterized by highly simplified bulbous cheek teeth, enormous orbits, and a prenarial shelf, which has led some to speculate that Allodesmus exhibited an enlarged soft tissue proboscis, like modern elephant seals (Mitchell, 1966; Barnes, 1972; Debey & Pyenson, 2013). Desmatophocids are confined to the North Pacific (Baja California, California, Oregon, Washington, and Japan; Barnes & Hirota, 1995) and were amongst the first fossil pinnipeds to be known from more than fragmentary material, owing to the abundance of their remains at key fossil localities prospected early in the 20th century (Desmatophoca oregonensis, Astoria Formation, OR, USA, Condon, 1906; Allodesmus kernensis, Round Mountain Silt, CA, USA Kellogg, 1922). Their early discovery, combination of features found in Phocidae (earless seals), Otariidae (fur seals and sea lions), and Odobenidae (walruses), and early geochronological age led early workers to consider Allodesmus as an ancestral pinniped (Kellogg, 1922; Scheffer, 1958). Some early researchers found taxa like Allodesmus rather specialized (Kellogg, 1922), and later discoveries led to the consensus that desmatophocids are an early diversification of pinnipeds with no modern descendants (Downs, 1956; Mitchell, 1966; Barnes, 1972). Despite the abundance of material available for study, strong debate still exists regarding the phylogenetic relationships of desmatophocids to other pinniped groups. Originally, desmatophocids were considered closely allied to ‘enaliarctines’, Otariidae, and Odobenidae, forming the clade Otarioidea (Barnes, 1989, 2008; Koretsky et al., 2016); under this arrangement, phocids were considered not to share a common ancestor with otarioids, but rather to have evolved independently from musteloid ancestors. However, this hypothesis has yet to be confirmed in any rigorous phylogenetic analysis of morphology (Berta & Ray, 1990; Berta & Wyss, 1994), and it conflicts with molecular data, which support pinniped monophyly (Arnason et al., 2006; Agnarsson et al., 2010). Alternatively, desmatophocids were considered to be the sister group to phocid seals, with the ‘enaliarctine’ Pinnarctidion, odobenids, and otariids forming successive sister taxa to this clade, forming a monophyletic Pinnipedia (Berta & Ray, 1990; Berta & Wyss, 1994). Although analyses of molecular data support pinniped monophyly, they instead recover a significantly different topology within Pinnipedia, with Otariidae and Odobenidae forming a clade to the exclusion of Phocidae. This calls into question whether Desmatophocids really are the sister group to Phocidae or whether, like odobenids, they might be more closely related to otariids. Relationships within Desmatophocidae are also unsettled. Past studies interpreting the evolution of allodesmine seals pre-dated computer-aided cladistics (Barnes, 1972; Barnes & Mitchell, 1975; Repenning & Tedford, 1977), did not use rigorous phylogenetic methodologies (Barnes, 1989; Barnes & Hirota, 1995), and had limited outgroup sampling (Kohno, 1996) and small character datasets (Kohno, 1996; Deméré & Berta, 2002). This has led to considerable controversy regarding the taxonomy of desmatophocids, with disagreements focusing on the number of genera within Allodesminae (e.g. Barnes, 1970, 1972; Barnes & Hirota, 1995) and number of species in Allodesmus (e.g. Mitchell, 1966; Barnes & Hirota, 1995; Kohno, 1996; Deméré & Berta, 2002; Barnes, 2008). Specific points of contention include the recognition of Atopotarus, Brachyallodesmus, and Megagomphos as distinct genera separate from Allodesmus, as well as the number of species present within the middle Miocene Round Mountain Silt of California (Mitchell, 1966; Barnes, 1970, 1972; Barnes & Hirota, 1995). Uncertainty over the taxonomy of the group has also hindered analyses of diversity of the Desmatophocidae. Past studies have examined the placement of Desmatophocidae within Pinnipedia (Barnes, 1989; Berta & Ray, 1990; Berta & Wyss, 1994; Deméré & Berta, 2002), as well as their paleoecology (Mitchell, 1966; Debey & Pyenson, 2013) and locomotor abilities (Bebej, 2009; Pierce et al., 2011), but little has been published on their diversification and the cause and timing of their eventual extinction. This study reports a new species of Allodesmus from the upper Miocene Montesano Formation of Washington, featured in a brief taphonomic report (Bigelow, 1994) but never properly described or named. This new species is represented by a well-preserved cranium, mandibles, and articulated axial skeleton. A new cladistic analysis of the Desmatophocidae provides the basis of a revision of their problematical taxonomy. This new species represents the geochronologically youngest desmatophocid seal yet reported (10.5–9.1 Ma, Tortonian equivalent), prompting investigation of potential causes for the decline and extinction of the formerly diverse Desmatophocidae. MATERIAL AND METHODS Preparation, anatomical description, measurements, and photography UWBM 75640 was mechanically prepared in the University of Washington Burke Museum paleontology laboratory. Concretionary sandstone matrix was left surrounding some vertebrae and within the thoracic cavity to retain the skeletal articulations; natural breaks were left so that the skeleton could be assembled in a series of five blocks. Anatomical terminology generally follows Deméré & Berta (2002), with modifications by Boessenecker & Churchill (2013, 2015). Measurements were recorded with digital callipers to the nearest tenth of a millimetre. Photographs were taken with a Rebel XS and 80 mm zoom lens. Phylogenetic analysis For our phylogenetic analysis, we examined material for all currently recognized species of desmatophocid seal and closely allied taxa. This included two species of the Desmatophoca (D. oregonensis and Desmatophoca brachycephala), Atopotarus courseni, and all eight species generally placed within the genus Allodesmus (Allodesmus gracilis, Allodesmus kelloggi, Al. kernensis, Allodesmus naorai, Allodesmus packardi, Allodesmus sadoensis, and Allodesmus sinanoensis), along with the new taxon we describe below. Allodesmus taxa from the Round Mountain Silt of Sharktooth Hill were coded as a single taxonomic unit (Al. kernensis), following Barnes (1972; also see our discussion below). We also included a large range of outgroup taxa representing the full extent of fossil pinnipedimorph diversity, including four ‘enaliarctines’ (Enaliarctos spp., Pteronarctos goedertae, Pinnarctidion bishopi, and Pinnarctidion rayi), three phocids (Devinophoca claytoni, Erignathus barbatus, and Monachus monachus), three otariids (Thalassoleon mexicanus, Callorhinus ursinus, and Zalophus californianus), and five odobenids (Proneotherium repenningi, Neotherium mirum, Imagotaria downsi, Gomphotaria pugnax, and Odobenus rosmarus). Enaliarctos was coded from all recognized species within that genus, especially Enaliarctos emlongi and Enaliarctos mealsi. We examined in total 171 specimens (Supporting Information, Appendix S2), which were used to code 95 morphological characters. Three of these characters were new, whereas the remaining characters were modified from those used in prior phylogenetic analyses (Berta & Wyss, 1994; Kohno, 1996, 2006; Deméré & Berta, 2002; Boessenecker & Churchill, 2013, 2015; Furbish, 2015). These include 36 cranial, seven mandibular, 24 dental, and 28 postcranial characters (Supporting Information, Appendix S3). Scaled character coding (Wiens, 1999) was applied to taxa that were polymorphic for a given character, with taxa coded as possessing an intermediate state (state name ‘polymorphic’) between two character states. Characters that included polymorphic states were then run as ordered characters in all analyses, with penalties applied to character state transitions that skip the polymorphic step. Overall, 12 polymorphic characters were included within the matrix. Our character matrix is available in the Supporting Information, Appendix S7. All phylogenetic analyses were carried out in TNT 1.1 using 10000 replicates with sectorial and tree-fusing options checked. Two separate phylogenetic analyses were run. In the first analysis, alternative implied weighting schemes were tested, varying the weighting constant K from K = 2 to K = 5. An implied weight of K = 3 was selected because this resulted in the maximal amount of resolution within trees produced with the least amount of weighting. In the second analysis, no implied weighting was carried out. Bootstrap support values were calculated for both analyses, using symmetrical resampling and 1000 replicates. Evaluating trends in diversification To evaluate plausible factors in desmatophocid extinction, we assembled a dataset including North Pacific pinniped species level diversity, global diatom diversity, eustatic sea level, and sea surface temperature, from the Chattian to Holocene. We focused on the North Pacific for pinniped diversity because desmatophocids are restricted to this region (chiefly USA, Mexico, and Japan), and the faunas are both densely sampled and well described (Miyazaki et al., 1995; Deméré et al., 2003; Barnes, 2008; Velez-Juarbe, 2017). Overall, we were able to assemble a dataset containing 100 occurrences of modern and fossil pinnipeds from this region. These data were sorted by taxonomic group (‘enaliarctine’, Odobenidae, Otariidae, Desmatophocidae, and Phocidae) and binned by stage level, with originations/extinctions brought to stage boundaries. Additionally, odobenids were further binned by size, and sorted into small (total body length <2 m) and large (total body length >2 m) size classes, with body size data from Churchill et al. (2014). Diatom species diversity is a reasonable proxy for primary productivity, and we used the species richness curve from Marx and Uhen (2010) for this proxy. Data for sea surface temperature (δ18O) is from Fedorov et al. (2013) and Rousselle et al. (2013), and eustatic sea level is from Miller et al. (2005). Institutional abbreviations UWBM, University of Washington Burke Museum of Natural History and Culture, University of Washington, Seattle, WA, USA. GEOLOGICAL BACKGROUND The upper Miocene Montesano Formation consists of ~800 m of shallow to deep marine sandstones and siltstones deposited within the Grays Harbor Basin near Grays Harbor in southwestern Washington, USA (Fig. 1; Fowler, 1965; Rau, 1967). The Montesano Formation unconformably overlies the lower to middle Miocene Astoria Formation and is in turn unconformably overlain by the Pleistocene Satsop Formation. The Montesano Formation is divided into a 460-m-thick lower member (units 1–5 of Fowler, 1965) consisting of fine sandstone with lensoidal calcareous nodules with intermittent pebbly horizons and abundant carbonized wood fragments, and a 240-m-thick upper member (units 6–7 of Fowler, 1965) consisting of well-bedded tuffaceous mudrocks and occasional fine sandstone interbeds (Fowler, 1965; Prothero & Lau, 2001). Both members of the Montesano Formation are richly fossiliferous; Addicott (1976) designated much of the lower member (units 1–4 of Fowler, 1965) as the basis for the Wishkahan provincial megainvertebrate stage, and the top of the lower member and the upper member as the stratotype for the Graysian stage (units 5–7 of Fowler, 1965). The holotype (UWBM 75640) and referred specimen (UWBM 109823) of Al. demerei were collected from the lower member of the Montesano Formation; the holotype was collected ~140 m above the base of the unit (Fig. 1; Bigelow, 1994). Paleomagnetic studies by Prothero and Lau (2001) correlated the Montesano Formation with Chrons C4Ar to C5n, indicating an age of 10.5–9.1 Mya for the formation. Although originally correlated with the Empire Formation on the basis of shared Wishkahan stage molluscs, paleomagnetic analysis indicates that the Empire Formation is somewhat younger (6.8–8.3 Mya; Prothero et al., 2001). The position of the type locality (UWBM C0343; 47°15′N, 123°31′W) ~140 m above the base of the formation suggests that the specimen might originate from above Chron C5n and within Chron C4Ar (Fig. 1); if so, this would suggest an even finer age control of 9.8–9.1 Mya (following chron boundary dates from Gradstein et al., 2012). However, paleomagnetism has not been studied at the type locality, and extrapolation from exposures along the Wishkah River (Prothero & Lau, 2001) seems unwise for now, and we adopt the slightly more coarse age of 10.5–9.1 Mya for the formation as a whole (Prothero & Lau, 2001). Isolated teeth of the dogfish Squalus occidentalis were discovered in the matrix associated with the holotype specimen (Bigelow, 1994). Massively bedded, pervasively bioturbated sandstone with occasional climbing ripples, mud drapes, and shell beds suggest lower shoreface deposition just below fair-weather wave base for this part of the Montesano Formation (Bigelow, 1994). Figure 1. View largeDownload slide Geological and stratigraphic context of Allodesmus demerei. A, geological map of Washington (modified from Walsh et al., 1987), showing exposures of the Montesano Formation (Fm.) and location of the type locality (UWBM C0343). B, stratigraphic column of the Montesano Formation exposed along the middle fork of the Wishkah River, after Fowler (1965). Magnetostratigraphy after Prothero and Lau (2001), from west fork of Wishkah River. Figure 1. View largeDownload slide Geological and stratigraphic context of Allodesmus demerei. A, geological map of Washington (modified from Walsh et al., 1987), showing exposures of the Montesano Formation (Fm.) and location of the type locality (UWBM C0343). B, stratigraphic column of the Montesano Formation exposed along the middle fork of the Wishkah River, after Fowler (1965). Magnetostratigraphy after Prothero and Lau (2001), from west fork of Wishkah River. SYSTEMATIC PALAEONTOLOGY Class Mammalia Linneaus, 1758 Order Carnivora Bowditch, 1821 Clade Pinnipedia Illiger, 1811 Familiy Desmatophocidae Hay, 1930 Genus Allodesmsus Kellogg, 1922 Amended diagnosis Desmatophocidae differing from Desmatophoca and Atopotarus in possessing a prenarial shelf, lacking a lateral wall of the alisphenoid canal, preglenoid process, possessing single-rooted postcanine teeth, and posteriorly deepening mandible with posteroventrally expanded digastric insertion; differing further from Desmatophoca in possessing bulbous postcanine teeth lacking accessory cusps or cingula, and differing further from Atopotarus in retaining M2. Type species Allodesmus kernensis. Included species Allodesmus kernensisKellogg, 1922; Al. naoraiKohno, 1996; Al. packardiBarnes, 1972; Al. sadoensisBarnes and Hirota, 1995; Al. sinanoensisNagao, 1941. Allodesmus demerei sp. nov. Allodesmus n. sp. Bigelow, 1994 Allodesmus n. sp. Barnes and Hirota, 1995 Etymology Allodesmus demerei is named in honor of Dr. Thomas A. Deméré for his mentorship, support, and influential contributions to the study of fossil pinnipeds and other marine mammals. Diagnosis of species A large species of Allodesmus similar in adult size to Al. kernensis and Al. sadoensis, and differing from Al. naorai and Al. packardi in possessing a prenarial shelf that is anteriorly transversely expanded, and differing from Al. sinanoensis in lacking tusk-like canines. With the exception of Al. naorai, Al. demerei differs from all Allodesmus in exhibiting proportionally more elongate and triangular nasals that are widest anteriorly. Allodesmus demerei differs from Al. kernensis in exhibiting more strongly developed nuchal crests, which obscure the occipital condyles in dorsal view, a jugal that extends posteriorly to the level of the glenoid fossa, and a proportionally deeper mandible, and from Al. sadoensis in retaining an M2. Allodesmus demerei further differs from all other Allodesmus in possessing a dorsally prominent and sharply triangular postorbital process, lacking postcanine diastemata, exhibiting a posteriorly elongate neural spine of the axis that overhangs C3, and a transversely expanded and dorsoventrally flattened anterior half of the manubrium. Holotype UWBM 75640, a partial articulated skeleton including skull and mandible missing all teeth other than right P1, hyoid bones, vertebral column (C1–T13), ribs, and manubrium. Collected 26–27 September 1984, by P. K. Bigelow and colleagues. Type locality and stratigraphic context University of Washington-Burke Museum locality C0343, 140 m above the base of the lower member of the Montesano Formation exposed in bank of Canyon River near Grays Harbor, WA, USA; upper Miocene, 10.5–9.1 Mya in age based on paleomagnetism (Tortonian equivalent; Prothero & Lau, 2001). More detailed locality information is available on request from UWBM to qualified researchers. Tentatively referred specimen UWBM 109823, left humerus missing the head, collected by M. S. Kelly from the lower Montesano Formation along the bank of the west fork of the Satsop River near Swinging Bridge Park, Grays Harbor County, WA, USA. More detailed locality information is available on request from UWBM to qualified researchers. Description Occurrence and preservation UWBM 75640 consists of a partial articulated skeleton (Fig. 2; Bigelow, 1994: fig. 1) including mandibles, hyoids, vertebral column (C1–T13), ribs (LR1–LR5, RR1–RR13), and manubrium, as well a complete cranium lacking all teeth except the right P1. The skeleton was preserved lying on its right side, with all preserved elements articulated in life position (Bigelow, 1994). Most of the teeth, post-manubrial sternebrae, and forelimbs are absent. The entirety of the skeleton was embedded in a cylindrical concretion enveloping the bones. Fourteen teeth of the dogfish Squalus occidentalis were found concentrated around the skull and cervical vertebrae, and several bite marks are evident on the skull (Bigelow, 1994); aside from those bite marks figured by Bigelow (1994), the extensive surface fracturing precludes us from identifying additional bite marks as it is not possible at this point to distinguish them from damage that occurred during fossil preparation. The posteriormost thoracic vertebrae were exposed in the bank of the Canyon river (Bigelow, 1994), raising the possibility that the entire vertebral column and possibly hindlimb elements were originally preserved and eroded away prior to discovery. Figure 2. View largeDownload slide Allodesmus demerei sp. nov.: skeleton of holotype UWBM 75640 (top), skeletal reconstruction (middle), and life restoration (bottom). Artwork by ©Robert Boessenecker. Figure 2. View largeDownload slide Allodesmus demerei sp. nov.: skeleton of holotype UWBM 75640 (top), skeletal reconstruction (middle), and life restoration (bottom). Artwork by ©Robert Boessenecker. Ontogeny and sex Most of the nine cranial sutures outlined by Sivertsen (1954) could be scored, with the exception of the median maxillary suture on the palate. The basisphenoid–presphenoid suture and frontoparietal sutures are open, the squamoso-parietal suture is partly closed, and the remaining sutures are mostly or completely closed. This results in a suture age of 22, indicative of adult status (Sivertsen, 1954). Further evidence of adult status includes well-developed nuchal and sagittal crests and complete epiphyseal fusion in the vertebral column. Furthermore, the relatively large canine alveoli, alongside well-developed nuchal and sagittal crests indicate that UWBM 75640 is a male; no baculum is preserved, although the entire post-thoracic skeleton is also missing. Rostrum and dentition The rostrum is relatively elongate (134 mm from orbit to anterior tip of premaxilla) and transversely narrow (82 mm wide at canines) and exhibits a broad prenarial shelf, which is smoothly convex dorsally and lacks a distinct prenarial process (Fig. 3; Table 1). In dorsal outline, the rostrum is widest anteriorly at the canines and narrows conspicuously at the level of P3, and widens posteriorly towards the base of the zygomatic arch. In dorsal view, the anterior margin of the rostrum is evenly convex. A small, suboval narial fossa is present, with a thick lateral wall formed by the premaxilla and maxilla. In lateral view, the rostrum is roughly triangular but bears a concave and ‘stepped’ anterodorsal margin owing to the elongate prenarial shelf. Ventrally on the lateral surface, the premaxilla–maxilla suture is indistinct on the left and on the right it is completely fused and obliterated; dorsally on the right side it is visible as a crack. In lateral view, the premaxilla is not continuously visible, and it forms a transversely narrow splint where it contacts the nasal; the nasal process of the premaxilla terminates along the anterior half of the nasal. The nasals are triangular, taper posteriorly, and penetrate between the frontals; the posterior half of the lateral margin is straight, and within the anterior half the nasal widens abruptly, giving the nasals a trumpet-shaped outline in dorsal view. Figure 3. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in dorsal view. Figure 3. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in dorsal view. Table 1. Measurements (in millimetres) of cranium of Allodesmus demerei (UWBM 75640) Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– View Large Table 1. Measurements (in millimetres) of cranium of Allodesmus demerei (UWBM 75640) Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– Total length, premaxilla to intercondylar notch 372 Facial length 256.6 Orbit length 91.6 Temporal fossa length 60 C1–M2 toothrow length 124.4 Braincase depth 131.7 Nares depth 38 Transverse width of nares 30.8 Nasals length 72.8 Nasals transverse width 32.4 Transverse width of rostrum at C1 89.2 Narrowest width of rostrum 70.6 Transverse width of palate at M2 100 Bizygomatic width 215 Transverse width at mastoid 181.0 Transverse width across tympanic bullae 116.9 Transverse width at paroccipital process 156.5 Transverse width of condyles 100.6 Foramen magnum depth 32.2 Foramen magnum width 41.5 Anteroposterior length, mastoid and paroccipital process 68.1 Width of internal choanae between pterygoid hamuli 20.8 Infraorbital foramen, depth/transverse width 12.7/11.3 Least interorbital width 35.3 Interorbital width at supraorbital process 46.2 Transverse width of braincase 127 Anteroposterior length of sagittal crest 124 Greatest depth of zygomatic arch 72.5 P1 crown height/anteroposterior length 8.2/11.5 C1 alveolus anteroposterior length/ transverse width 28.1/24.6 P2 alveolus anteroposterior length/ transverse width 18.9/– P3 alveolus anteroposterior length/ transverse width 14.6/– P4 alveolus anteroposterior length/ transverse width 16.3/– M1 alveolus anteroposterior length/ transverse width 14.1/– M2 alveolus anteroposterior length/ transverse width 8.6/– View Large The palate is not exposed (Fig. 4), but an elongate palate with a ‘pterygoid process’ of the maxilla is present, as in all Pinnipediformes (Berta & Wyss, 1994), that extends at least 15–20 mm posterior to the M2. The toothrows (and by extension the lateral margins of the palate) diverge posteriorly at a 41° angle. With the exception of the right P1, all teeth are missing. Details of the incisor alveoli are unclear owing to incomplete preparation and the close contact between the mandibles and cranium. The canine alveoli are large, circular, and appear to have housed a procumbent canine. The P1 is positioned posteromedial to the canine, single rooted, and bears a bulbous crown lacking a labial cingulum; only the protoconid cusp is present. Wear on the crown is unclear owing to incomplete preparation. The root is somewhat inflated, especially near the base of the crown. The P2 alveolus is oval and indicates a single root. The P3 and P4 alveoli are also oval, but bear a minute vertical ridge laterally, indicating a partial interalveolar septum, suggesting bilobate roots. The M1 alveolus is somewhat smaller than the premolar alveoli and bears a better developed interalveolar septum, suggesting a root that was more strongly bilobate or perhaps double rooted internally. The M2 alveolus is circular, smaller than the M1, and indicates the presence of a single root. Figure 4. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in ventral view. Figure 4. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640) in ventral view. Orbit, intertemporal region, and zygomatic arch The interorbital bar is composed entirely of the frontal, is anteroposteriorly elongate, transversely narrow, and gradually rises posteriorly (Figs 3, 5; Table 1). The supraorbital process of the frontal is reduced to a small bump positioned posteriorly, approximately three-quarters of the distance between the anterior margin of the orbit and the anterior margin of the braincase. The orbital margin of the frontal is gently laterally concave in dorsal view. Posterior to the supraorbital process, the frontal is also laterally concave. A short orbitotemporal crest extends posteromedially from the supraorbital process and converges medially on either side to form the anterior end of the sagittal crest. The sagittal crest is low (~15 mm at the highest) and elongate, extending from the nuchal crest anteriorly to the middle of the intertemporal bar. Figure 5. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A and B, right lateral view. C, left lateral view. Figure 5. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A and B, right lateral view. C, left lateral view. The orbit is enormous and circular; the anterior part of the zygomatic arch slightly descends ventrally to accommodate the large orbit (Fig. 5; Table 1). The maxillary root of the zygomatic arch is pierced anteriorly by a transversely narrow, oval infraorbital foramen ~15 mm high. The maxilla–jugal suture is completely fused and obliterated. The jugal–squamosal articulation on the zygomatic arch is mortised and dorsoventrally expanded. A large, triangular postorbital process is developed and extends dorsally; it is separated from the supraorbital process of the frontal by only 20 mm. The zygomatic process of the squamosal deepens anteriorly and includes a triangular anteroventral corner. Posteriorly, the jugal splits into a dorsal and a posteroventral splint that cradle the anterior margin of the zygomatic process. The anteroposteriorly narrow dorsal splint leads to the postorbital process, whereas the posteroventral splint tapers to a point, curves ventrally, and extends posteriorly to the level of the glenoid fossa. The zygomatic arch forms the transversely widest part of the skull at about the level of the preglenoid process. Braincase The braincase is transversely narrow and ovoid in dorsal view, and gradually widens posteriorly; the anterolateral margin lacks a corner as in ‘enaliarctines’, otariids, and Desmatophoca, and is instead smoothly convex (Fig. 3). The lateral wall of the braincase is fractured, but the frontoparietal and parieto-squamosal sutures can be traced; a pseudosylvian sulcus is not apparent but coded as uncertain (?) owing to fracturing of the braincase and diagenetic compression in the transverse direction, which might obscure the presence of a pseudosylvian sulcus. The frontoparietal suture has a roughly transverse orientation, is mostly straight and positioned at the position of the external acoustic meatus in dorsal view. The parietal extends posteriorly and forms a large posteriorly extending nuchal crest with an evenly convex posterior margin in lateral view; the nuchal crest extends just posterior to the level of the occipital condyles, and the nuchal crest arises dorsally from the mastoid process. Dorsally, the nuchal crests converge anteriorly, and a small triangular wedge of the occipital shield is visible in dorsal view (Fig. 3). The posterior surface of the nuchal crest is dorsoventrally thick (10–17 mm) and rugose for insertion of epaxial muscles (Fig. 6). The occipital shield is nearly vertical, low, and exhibits a blunt external occipital crest dorsally; lateral to the external occipital crest the supraoccipital is concave. The occipital condyles are obscured by the atlas, but judging from exposed portions are transversely wide and define a circular foramen magnum, which is in turn encircled by deeply excavated dorsal condyloid fossae. The exoccipital bears a circular pit for the sternomastoid fossa. The paroccipital process extends ventrolaterally and is separated from the occipital condyles by a deep ventral condyloid fossa. The paroccipital process descends posteroventrally in lateral view. Figure 6. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A, anterior view. B, posterior view. Figure 6. View largeDownload slide Allodesmus demerei sp. nov.: holotype cranium and mandibles (UWBM 75640). A, anterior view. B, posterior view. Basicranium The basioccipital bears deeply excavated oval pits for the anterior insertion of the rectus capitis muscle and is separated by a median crest. Large rugose basioccipital crests are present medial to the large circular posterior lacerate foramina (Fig. 4; Table 1). The basioccipital is wide posteriorly but narrows anteriorly and becomes flatter where it nears the basisphenoid. The pterygoid strut is somewhat exposed and bears a small, vertically oriented pterygoid hamulus that is triangular in lateral view. The pterygoid strut is widest at the position of the hamulus, and laterally defines the transversely narrow internal choanae. The alisphenoid is flat anteromedial to the tympanic bulla. The morphology of the anterior bulla is not evident except for the median lacerate foramen, which pierces the anteromedial spur of the bulla. The tympanic bulla is pentagonal in ventral view, mostly flattened but slightly convex ventrally. It bears a rugose tubercle medially, immediately anterior to the opening of the carotid canal. Anteriorly, the bulla is closely appressed to the postglenoid process. The tympanic bulla has a flat lateral wall, which is anteriorly pierced by the external acoustic meatus; the meatus is positioned within an anteroposteriorly narrow and transversely oriented trough between the postglenoid and mastoid processes. The orientation of the meatus in lateral view (e.g. Barnes, 1987) is unclear. The stylomastoid foramen is a dorsoventrally shallow oval and separated from the external acoustic meatus by a low ridge that continues laterally to meet the enlarged mastoid process. The squamosal is ventrally flattened lateral to the bulla, and bears a rugose knob-like mastoid process with a convex, subcircular ventrolaterally facing facet (Fig. 4). The mastoid process is widely separated from the paroccipital process by an incised notch. The paroccipital process is dorsoventrally shallow and somewhat ventrally excavated; a small ridge separates the stylomastoid foramen from a small tubercle-like tympanohyal. A small fossa is developed posterior to the tympanohyal. A large knob is present on the exoccipital posterolateral to the posterior lacerate foramen; posteriorly, the tympanic bulla laps onto this knob. The glenoid fossa is anteroventrally facing and bears a deep postglenoid process; the preglenoid process is not exposed. Mandible The mandible is gracile and subrectangular, with a dorsoventrally shallow triangular symphyseal portion (Fig. 5; Table 2). The mandibular symphysis is relatively elongate (32% of mandibular length) and is posteriorly distinguished by an indistinct genial tuberosity. The mandible transversely widens anteriorly toward the canine root and narrows posteriorly (Fig. 4). The body of the mandible deepens posteriorly towards the digastric insertion. The ventral margin of the mandible is faintly concave ventrally; posteriorly, the digastric insertion is expanded into a posteroventrally extending flange; a small notch is present between the digastric insertion and the angular process, defining the limits of the flange. Unlike Al. kernensis, the angular process is large and extends medially as a horizontal shelf. The mandibular condyle is wide, cylindrical, and positioned at about the level of the toothrow. The masseteric fossa is shallow and extends ventrally from the coronoid process to the dorsoventral midpoint of the body of the mandible; the floor of the fossa is somewhat rugose and possesses a series of shallow dimples up to 5 mm in diameter. The coronoid process is posteriorly positioned and incompletely prepared but high, rising ~100% of the depth of the body above the toothrow; it extends posteriorly to the level of the posterior edge of the digastric flange. Table 2. Measurements (in millimetres) of mandible of Allodesmus demerei (UWBM 75640) Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 View Large Table 2. Measurements (in millimetres) of mandible of Allodesmus demerei (UWBM 75640) Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 Total length 283.6 Length of toothrow, C1–M2 125.0 Depth at genial tuberosity 67.6 Greatest length of symphysis 98.6 Greatest depth at coronoid process >130 Shallowest depth of horizontal ramus 61.8 Anteroposterior length of C1 alveolus 23.6 Transverse width of mandibular condyle 54.0 View Large Hyoid The hyoid bones were preserved (Bigelow, 1994) but on loan to P. K. Bigelow at the time of his death; their current whereabouts are unknown and presumed lost. Cervical vertebrae The atlas bears a posterolaterally projecting transverse process that is suboval in outline (Fig. 7A, B; Table 3). The neural spine is a low tubercle. The transverse foramen is large and oval anteriorly and has a much smaller posterior opening. The axis bears a finger-like, posteriorly inclined transverse process with a sharp anteroventral ridge. A sharp, deep median ridge is present on the ventral surface of the centrum; the atlantal articular facet is obscured by matrix. The neural spine of the axis (Table 4) is dorsoventrally short but includes an elongate posterior apex that is subrectangular in lateral view and that probably articulated with the low neural spine of C3. The postzygapophyses are positioned anterior to the posterior apex of the neural spine, are subhorizontal and D-shaped in dorsal view, and are more medially placed than on the rest of the post-atlantal cervical vertebrae. The transverse foramina of the axis are relatively small. Figure 7. View largeDownload slide Allodesmus demerei sp. nov.: holotype axial elements of Allodesmus demerei (UWBM 75640). A, cervical vertebrae in dorsal view, with skull in articulation as preserved. B, cervical vertebrae in right lateral view. C, thoracic vertebrae 1–2 with ribs 1–2 in ventral view; manubrium in (E) ventral, (F) dorsal, and (G) right lateral view. D, thoracic vertebrae 6–11 and ribs 4–13 in right lateral view. Figure 7. View largeDownload slide Allodesmus demerei sp. nov.: holotype axial elements of Allodesmus demerei (UWBM 75640). A, cervical vertebrae in dorsal view, with skull in articulation as preserved. B, cervical vertebrae in right lateral view. C, thoracic vertebrae 1–2 with ribs 1–2 in ventral view; manubrium in (E) ventral, (F) dorsal, and (G) right lateral view. D, thoracic vertebrae 6–11 and ribs 4–13 in right lateral view. Table 3. Measurements (in millimetres) of atlas vertebra of Allodesmus demerei (UWBM 75640) Greatest transverse width 90.2 Greatest depth at midline 69.0 Greatest transverse width 90.2 Greatest depth at midline 69.0 View Large Table 3. Measurements (in millimetres) of atlas vertebra of Allodesmus demerei (UWBM 75640) Greatest transverse width 90.2 Greatest depth at midline 69.0 Greatest transverse width 90.2 Greatest depth at midline 69.0 View Large Table 4. Measurements (in millimetres) of axis vertebra of Allodesmus demerei (UWBM 75640) Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 View Large Table 4. Measurements (in millimetres) of axis vertebra of Allodesmus demerei (UWBM 75640) Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 Greatest depth 108.0 Anteroposterior length of neural spine 143.1 Dorsoventral depth of centrum 46.3 Anteroposterior length of centrum (not including odontoid process) 81.6 Transverse width of centrum 57.6 Greatest width at transverse process 119.1 View Large The remaining cervicals (C3–C7; Fig. 7A, B; Table 5) have elongate ventrolaterally projecting transverse processes that are subrectangular in shape with apical tubercles; the transverse processes become increasingly anteroposteriorly broad and hatchet shaped in C5 and C6, with the transverse process longest in C6. The articular surfaces of the centra are subrectangular anteriorly and oval posteriorly. C3 has an elongate shelf-like lamina; the lamina becomes shorter posteriorly. C5 and C6 have a large laterally directed knob on the base of the transverse process lateral to the canal. In C7, the transverse process is shorter and less ventrally deflected, with a more dorsally positioned lateral apex; the process becomes dorsoventrally deeper laterally, giving it a fan shape in anterior/posterior view. The transverse process is posteriorly concave. The transverse foramina increase in diameter posteriorly, culminating in the enlarged canal in C7. The neural spines are short and stout anteriorly and become higher and transversely narrower posteriorly toward the C7. Table 5. Measurements (in millimetres) of cervical vertebrae C3–C7 of Allodesmus demerei (UWBM 75640) Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – View Large Table 5. Measurements (in millimetres) of cervical vertebrae C3–C7 of Allodesmus demerei (UWBM 75640) Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – Measurement C3 C4 C5 C6 C7 Anteroposterior length of centrum 76.2 73.9 77.9 86.3 88.9 Greatest width at transverse process 164.4 161.5 149.7 150 153.4 Transverse width of centrum 54.1 58.1 – – 61.9 Dorsoventral depth of centrum – 53.2 – – 53.9 Greatest dorsoventral depth 80 94.8 – – – Neural foramen, depth – – 26.6 – – Neural foramen, transverse width – – 59.5 – – View Large Thoracic vertebrae Thoracic vertebrae T1–T13 are preserved (Fig. 7C, D; Table 6); T1–T6 are the most complete and best exposed. T1–T2 are relatively similar to C7 but have a more dorsally positioned transverse process with a large facet for the tubercle of the ribs. All thoracic vertebrae except T1 have a posterolaterally facing fossa on the dorsolateral edge of the posterior centrum for the articulation of the rib heads. Thoracics T1–T2 differ from all posterior thoracics in exhibiting more medially placed, narrowly separated prezygapophyses; they are more widely separated in posterior thoracic vertebrae. The neural spine is poorly preserved in many thoracics, but from anterior to posterior within the thoracic series the spine becomes increasingly anteroposteriorly narrow and posterodorsally inclined. A low ventral median ridge is present on T1–T2, but becomes slightly more transversely rounded in T3–T11; in T12–T13, the ventral margin of the centrum is completely rounded, and no median ridge is evident. The postzygapophyses decrease in size posteriorly within the thoracic series and by T4 they are on the base of the neural spine at the medial termination of the lamina; they are strongly reduced in T10–T13. Table 6. Measurements (in millimetres) of thoracic vertebrae of Allodesmus demerei (UWBM 75640) Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – View Large Table 6. Measurements (in millimetres) of thoracic vertebrae of Allodesmus demerei (UWBM 75640) Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – Anteroposterior length of centrum Greatest width at transverse process Transverse width of centrum Dorsoventral depth of centrum Greatest dorsoventral depth T1 63.0 163.6 61.9 – >130 T2 61.5 143.5 – – >112 T3 61.8 141.8 73.4 47.0 >115 T4 63.7 142.0 71.8 57.9 >115 T5 61.5 128.7 73.2 48.6 >94 T6 63.1 119.4 52.9 41.8 >102 T7 66.2 105.6 87.6 44.4 110.4 T8 66.0 104.6 60.7 47 96.0 T9 61.79 99.0 52.1 43 >83 T10 63.1 99.0 27.9 – – T11 63.4 63.4 – – – T12 56.9 – – – – T13 55.8 – – – – View Large Ribs Few ribs other than LR1 and RR1 are complete, but the proximal ends of LR2–LR5 and the proximal half of RR2–RR13 are preserved (Fig. 7C, D). Left R1 (126.1 mm long, 60.0 mm capitulum to tubercle) and RR1 (128.0 mm long, 61.6 mm capitulum to tubercle) bear a small capitulum, a large tubercle, and a short shaft with a flat and circular distal end bearing a rugose texture. Ribs 2–5 are morphologically similar but with a longer shaft than R1; R6–R8 have a much larger, subspherical head and a longer shaft. Manubrium The manubrium is the only preserved portion of the sternum (Fig. 7E-G; Table 7). It is anteriorly flattened, transversely wide and rectangular with parallel lateral margins. It bears tubercles with posterolaterally facing rugose articular surfaces for the ribs at the midpoint. A ventral keel is present and dissipates anteriorly; the manubrium deepens posteriorly. The posterior portion is subcylindrical and transversely widens posteriorly toward a large transversely compressed, oval and rugose posterior articular surface. The transversely wide anterior part of the manubrium is unique amongst pinnipeds and differs from the narrower manubrium of Al. kernensis, although in Al. kernensis the anterior half is slightly expanded transversely; a somewhat expanded manubrium may also be present in At. courseni, but further preparation is required to be certain. Table 7. Measurements (in millimetres) of sternum of Allodesmus demerei (UWBM 75640) Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 View Large Table 7. Measurements (in millimetres) of sternum of Allodesmus demerei (UWBM 75640) Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 Total length 165.49 Transverse width of anterior end 68.44 Transverse width at rib 1 tubercles 69.81 Transverse width of posterior end 27.26 Dorsoventral depth at rib 1 tubercles 31.67 Dorsoventral depth at posterior end 34.6 View Large Humerus UWBM 109823 (Fig. 8; Table 8) is missing the head. The shaft is nearly cylindrical and straight in anterior and posterior view and widens slightly proximally, and abruptly widens distally toward the transversely wide distal end. In posterior view, the medial margin is slightly more concave than the lateral margin. The deltopectoral crest is elongate and runs along two-thirds of the length of the humerus. The crest is anteroposteriorly deepest where it abruptly terminates; proximally, the anterior margin is nearly straight. The crest is positioned laterally on the shaft in anterior view, and is transversely widest near the distal terminus. In lateral view, the deltopectoral crest has an abrupt distal terminus. The deltoid insertion is not separated from the deltopectoral crest. The greater and lesser tuberosities are nearly the same height, and both appear to have extended slightly proximally to the head. In proximal view, the tuberosities are both anteromedially slanted. The tuberosities are separated by a narrow intertubercular groove with a V-shaped cross-section. The distal end is much wider than the proximal end. The prominent medial entepicondyle is medially convex and anteroposteriorly flattened; the supinator ridge is poorly developed and lacks a sharp edge. In medial and lateral view, the posterior margin of the humerus is concave. The medial and lateral edges of the distal trochlea are slanted anterolaterally in distal view; the lateral lip of the trochlea is slightly thicker anteroposteriorly than the medial lip. Figure 8. View largeDownload slide Referred humerus of Allodesmus demerei (UWBM 109823). A, anterior view. B, posterior view. C, medial view. D, lateral view. Figure 8. View largeDownload slide Referred humerus of Allodesmus demerei (UWBM 109823). A, anterior view. B, posterior view. C, medial view. D, lateral view. Table 8. Measurements (in millimetres) of referred humerus of Allodesmus demerei (UWBM 109823) Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 View Large Table 8. Measurements (in millimetres) of referred humerus of Allodesmus demerei (UWBM 109823) Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 Total length as preserved 247.4 Transverse width, greater tuberosity 30.1 Transverse width, lesser tuberosity 35.4 Least transverse width at midshaft 37.4 Anteroposterior depth at midshaft 78.4 Transverse width of deltopectoral crest at apex 27.3 Greatest transverse width, distal end 99.0 Anteroposterior diameter of distal trochlea, medial 48.0 Anteroposterior diameter of distal trochlea, lateral 52.4 Greatest length of deltopectoral crest 175.5 View Large Referral of UWBM 109823 to Allodesmus demerei The holotype specimen lacks a humerus and thus there are no overlapping elements to refer UWBM 109823 unequivocally to Al. demerei. However, the humerus compares relatively well with Al. kernensis, and differs from otariids in lacking an arcuate anterior margin of the deltopectoral crest and a sinuous posterior margin of the humerus. UWBM 109823 differs from odobenids in exhibiting a concave posterior margin and a distally positioned apex of the deltopectoral crest. UWBM 109823 differs from both odobenids and otariids in possessing a distal trochlea that is slightly anteroposteriorly wider laterally (similar diameter in otariids, wider medial edge in odobenids; Repenning & Tedford, 1977). The distal apex of the deltopectoral crest, large medial entepicondyle, wide lateral portion of the distal trochlea, and relatively large lesser tuberosity support referral to Allodesmus. Given that Al. demerei is the only known Allodesmus from the Montesano Formation, we tentatively assign this specimen to the species; future discoveries of more complete skeletons are needed to confirm this identification. Comparisons Allodesmus demerei differs from Desmatophoca and Atopotarus in possessing a prenarial shelf on the rostrum; the prenarial shelf is anteroposteriorly longer than Al. naorai and Al. packardi but somewhat shorter than Al. kernensis; the prenarial shelf further differs from these taxa by being transversely expanded anteriorly, which Al. demerei shares with Al. kernensis, Al. sadoensis, and Al. sinanoensis. Rather than converging posteriorly as in all other desmatophocids, the lateral margins of the nasals are posteriorly parallel and terminate along the V-shaped fronto-maxillary suture, similar to Al. packardi, dusignathine walruses, and the odobenine walrus Ontocetus (Deméré, 1994; Kohno, 2006) but differing from all other desmatophocids for which the suture is preserved. The nasals are also proportionally longer than most other Allodesmus spp. (62% of rostrum length) except Al. naorai. Allodesmus demerei primitively retains a triangular postorbital process like Desmatophoca and Atopotarus, differing from the dorsally rounded condition in Al. kernensis and Al. packardi. In Al. demerei, the jugal extends posteriorly to the level of the anterior margin of the glenoid fossa, similar to Desmatophoca and perhaps Al. packardi; however, in Al. kernensis the jugal terminates well anterior to the glenoid fossa. Allodesmus demerei further differs from all other species of Allodesmus in exhibiting nuchal crests so large and posteriorly directed that the occipital condyles and supraoccipital are completely hidden in dorsal view; in published specimens of Al. kernensis the condyles are widely visible, although in newly available but unpublished specimens of Al. kernensis the nuchal crest obscures the condyles (T. A. Deméré, personal communication September 2016). In Al. sadoensis, the nuchal crests partly obscure the condyles, and the supraoccipital is not visible. In extant Otariidae, this is related to ontogeny, and perhaps published specimens of Al. kernensis are young adult males rather than old adults; an assessment of ontogenetic and intraspecific variation within the Sharktooth Hill sample of Al. kernensis would be highly instructive. The dentition of Al. demerei differs from Al. kernensis, Al. naorai, Al. packardi, and Al. sinanoensis (and shares with Al. sadoensis) in lacking diastemata between the upper postcanines; retention of an M2 distinguishes Al. demerei from Al. sadoensis. Unlike Desmatophoca, Al. kernensis, and Al. sadoensis, the molar alveoli appear to be of similar size to the premolars (a feature shared with Al. packardi). The mandible of Al. demerei differs from Al. kernensis in being dorsoventrally deeper, and is much deeper than At. courseni; the mandible exhibits a proportionally smaller genial tuberosity than At. courseni and Al. sadoensis. Mandibles of D. oregonensis differ in exhibiting a horizontal ramus that is dorsoventrally deepest at the genial tuberosity rather than the digastric insertion. Well-preserved postcranial skeletons are currently available only for At. courseni and Al. kernensis. The axis differs from these two allodesmines and all other pinnipeds in possessing an elongate posterior extension of the neural spine, which appears to articulate with or closely approximate the tip of the neural spine of the third cervical vertebra. An extension appears present in At. courseni and longer than in Al. kernensis but is not as elongate as Al. demerei. Phylogenetic results In the analysis where implied weighting (IW) was performed, one most parsimonious tree was recovered [Fig. 9; tree length = 29.54, consistency index (CI) = 0.48, retention index (RI) = 0.71]. In the analysis where no implied weighting (NW) was performed, ten most parsimonious trees were recovered (Fig. 10; tree length = 300, CI = 0.48, RI = 0.71). Given the focus of this paper on relationships within Desmatophocidae, synapomorphies are listed in this section only for members of this clade (as character number:state number). Figure 9. View largeDownload slide Time-calibrated strict consensus tree of desmatophocid relationships based on implied weighting analysis (K = 3). Age range of taxa is represented by terminal horizontal bars (see Supporting Information, Appendix S4 for age explanations); black bars represent outgroup taxa, green bars desmatophocines, and blue bars allodesmines. Numbers above nodes represent bootstrap values. ‘Round Mountain Silt’ Allodesmine composite taxon including Allodesmus kernensis, Al. kelloggi, and Al. gracilis. Tree scaled to time using the Paleotree package in R. 3.3.1 (Bapst, 2012). Outlines of skulls redrawn from Barnes (1972, 1979, 1987, 1989), Barnes and Hirota (1995), Berta (1994), Deméré and Berta (2001, 2002), Kohno (1996), and Repenning and Tedford (1977), and photographs provided by J. Velez-Juarbe. Figure 9. View largeDownload slide Time-calibrated strict consensus tree of desmatophocid relationships based on implied weighting analysis (K = 3). Age range of taxa is represented by terminal horizontal bars (see Supporting Information, Appendix S4 for age explanations); black bars represent outgroup taxa, green bars desmatophocines, and blue bars allodesmines. Numbers above nodes represent bootstrap values. ‘Round Mountain Silt’ Allodesmine composite taxon including Allodesmus kernensis, Al. kelloggi, and Al. gracilis. Tree scaled to time using the Paleotree package in R. 3.3.1 (Bapst, 2012). Outlines of skulls redrawn from Barnes (1972, 1979, 1987, 1989), Barnes and Hirota (1995), Berta (1994), Deméré and Berta (2001, 2002), Kohno (1996), and Repenning and Tedford (1977), and photographs provided by J. Velez-Juarbe. Figure 10. View largeDownload slide Strict consensus tree of desmatophocid relationships based upon equal weighted analysis. Numbers above nodes represent bootstrap values. †Extinct taxa. Figure 10. View largeDownload slide Strict consensus tree of desmatophocid relationships based upon equal weighted analysis. Numbers above nodes represent bootstrap values. †Extinct taxa. Relationships of outgroup taxa to Desmatophocidae largely differ in their levels of resolution between analyses (Figs 9, 10). In both analyses, Pteronarctos is recovered as the sister taxon to a clade comprising Pinnarctidion and Pinnipedia, with good support [bootstrap (BS) values of 100]. A monophyletic Pinnarctidion is recovered as the sister taxon to Pinnipedia, but with only moderate support in the the NW analyses (BS = 61), and was poorly supported in the IW tree. Within the IW tree, a clade comprising Odobenidae and Otariidae is recovered, and is found to be the sister group to Phocoidea (Fig. 9). A monophyletic Otariidae is recovered with good support (BS = 75), but odobenids monophyly is only weakly supported. In contrast, within the NW tree a strongly supported (BS = 73) Otariidae is recovered within a polytomy consisting of Proneotherium, Neotherium, a poorly supported clade comprising the remainder of odobenids, and Phocoidea. Although a monophyletic Phocoidea is recovered in both analyses, it is only poorly supported. Phocidae is recovered with moderate to strong support in both analyses (IW BS = 59; NW BS = 79). Desmatophocidae is recovered with weak support in both analyses (Figs 9, 10). It is supported by nine unequivocal synapomorphies, including reduced premaxilla–nasal contact (1:1), a V-shaped nasal–frontal suture (2:1); a posterolaterally expanded pterygoid process of the maxilla below the orbit (8:1), very large orbits (16:1), presence of an elongate anteroventral splint of the jugal in the jugal–maxillary suture, which extends to the level of the M1 (23:3), pseudosylvian sulcus visible on the braincase (27:0 reversal), squamosal fossa divided and unequal in size (29:2), paroccipital process large but not excavated (30:2), stylomastoid foramen separated from tympanohyal by raised ridge (33:1), and a dorsoventrally and laterally projecting pterygoid strut (34:1). Both analyses recover a monophyletic Desmatophoca, which forms the sister group to a moderately supported Allodesminae (IW BS = 61; NW BS = 64). Monophyly of Desmatophoca in both analyses is poorly supported. Only one unequivocal synapomorphy can be identified for this genus, a jugal mortised with the zygomatic process of the squamosal with little dorsoventral development (22:1). Six unequivocal synapomorphies unite Allodesminae, including a well-developed prenarial shelf (3:1), a mortised squamosal–jugal articulation with dorsoventral development (22:2), loss of the angular process of the mandible (37:2) and posterior and medial carina of the C1 (46:1), and lower postcanines divergent in orientation (59:1) and lacking cingulum (61:2). Within Allodesminae, both analyses (Figs 9, 10) recover Atopotarus as the sister taxon to a moderately supported (IW BS = 60; NW BS = 51) clade comprising the remainder of allodesmine taxa. This clade is supported by six unequivocal synapomorphies [absence of an antorbital process (10:1), a small and posteriorly positioned supraorbital process of the frontal (15:3), absence of the styloid process of the tympanic bullae and retraction of the entotympanic posteriorly (32:1), a transversely expanded posterior lacerate foramen (36:1), and single-rooted P2–3 (49:2) and P4 (50:3)]. Two additional synapomorphies may diagnose this clade, or a more inclusive clade including Atopotarus: presence of an orbital vacuity (17:1) and a retracted dorsal margin of the maxillary root of the zygomatic arch (21:1). Relationships within Allodesmus are completely unresolved in the NW analysis, but the IW analysis shows a fair degree of resolution. Within the the latter analysis, the ‘broad head’ taxa Al. packardi and Al. naorai are recovered as successive sister taxa to a moderately supported (BS = 63) clade comprising the ‘long head’ subgroup. Two unequivocal synapomorphies unite Al. naorai with later diverging allodesmines: posterior termination of premaxilla extends to level of infraorbital foramen (7:1) and the presence of a small infraorbital foramen (9:1). The ‘long head’ clade is supported by six unequivocal synapomorphies [anterior margins of nasal well posterior to P1 (5:2), a transversely arched palate (12:1), reduced incisive foramina (13:2), absence of the lateral wall of the alisphenoid (20:1), a smoothly convex braincase (28:1), and single-rooted P2–4 (62:2)] and one equivocal synapomorphy (slightly divergent upper tooth rows; 14:0). Relationships are largely unresolved within the ‘long head’ clade, although Al. demerei was recovered as the sister taxon to the Round Mountain Silt Allodesmus with poor support. This latter clade is supported by two unequivocal synapomorphies: a distinctively concave ventral edge of the rostrum (4:2) and symphyseal angle of the mandible <40° (42:2). DISCUSSION Phylogeny of the Pinnipedimorpha Although our character sampling was focused on resolving relationships within the Allodesminae, our phylogenetic tree possessed several major differences in topology from past higher-level phylogenies of Pinnipedimorpha (Barnes, 1989; Berta, 1994; Berta & Wyss, 1994; Boessenecker & Churchill, 2013). These differences include the position of Pinnarctidion and recovery of a clade comprising Odobenidae and Otariidae. Prior phylogenetic hypotheses have generally considered Pinnarctidion as closely allied to desmatophocid seals, either as the sister taxon to Allodesminae (Barnes, 1989) or as a stem phocoid (Berta, 1994; Berta & Wyss, 1994), a clade comprising Phocidae and Desmatophocidae. Our phylogenetic tree presents a third alternative, placing Pinnarctidion as pinnipediform and as the sister taxon to Pinnipedia (see also Barnes & Hirota, 1995; Deméré & Berta, 2002). This clade is supported by three unambiguous synapomorphies, including a deep ramus of the mandible at the position of the posteroventral terminus of the symphysis (39:1), P2–M1 teeth which are all the same size (58:1), and a talonid basin that exists as a small concavity or shelf (67:1). In contrast, the only character shared between Pinnarctidion and Allodesminae is possession of a transversely expanded posterior lacerate foramen (36:1). Barnes (1989) lists an additional ten features shared between Pinnarctidion and Allodesminae, but nearly all of these are plesiomorphic and diagnose larger clades, are present in clades other than Allodesminae, or are poorly defined. The only shared feature found exclusively between both Desmatophocidae and Pinnarctidion is a dorsoventrally and laterally projecting pterygoid strut (34:1; character ten of Berta, 1994). Berta (1994) lists a total of 15 characters uniting Pinnarctidion with Phocidae and Desmatophocidae. Of those characters, ten are included within this study, including the formerly mentioned shared character. Of these ten characters, we interpret four of these as plesiomorphic and with a wider character distribution than presented by Berta (1994), including presence of a shelf-like pterygoid process of the maxilla (found in Proneotherium), posteroventral location of the optic foramina and anterior lacerate foramen (also found in most odobenids), enlargement of the paroccipital process (found in Pteronarctos and all later diverging pinnipediforms), and presence of a knob-like acromion process of the scapula (the plesiomorphic condition for pinnipeds). An additional character, lateral elongation and expansion of the posterior lacerate foramen, is shared with Allodesminae but absent in Desmatophoca. The remaining five characters present in our study have been modified, resulting in a different interpretation of the character coding as presented by Berta (1994). Characters defined in a slightly different manner and thus coded differently from Berta (1994) include palate shape (here defined as upper tooth row orientation), squamosal–jugal articulation, and separation of the stylomastoid foramen from the tympanohyal pit. Given the limited sampling of characters related to higher level pinniped phylogeny, we consider the phylogenetic position of Pinnarctidion to be unsettled at this point. Future studies, with more exhaustive sampling of characters pertinent for resolving relationships among early pinnipedimorphs, need to be undertaken to determine whether this genus is indeed more closely related to Desmatophocidae or to other pinniped groups. Non-cladistic studies of pinniped morphology (Barnes, 1989) have generally advocated a close relationship between Otariidae, Desmatophocidae, and Odobenidae. These studies have received additional support from analyses of molecular data (Arnason et al., 2006; Agnarsson et al., 2010), which have found Odobenidae to be the sister taxon to Otariidae. Although our phylogenetic analysis does not recover a clade representing the classical definition of Otarioidea, we do recover within our IW analysis a clade comprising Odobenidae and Otariidae, a relatively novel finding for morphology-based phylogenetic analysis of fossil pinnipeds. This relationship is, however, not recovered in our NW tree, which instead recovers a polytomy consisting of Otariidae, Phocoidea, Proneotherium, Neotherium, and a clade comprising the remainder of Odobenidae. Implied weighting can lead to increased resolution in phylogenetic analyses, but this increased resolution often comes at the price of decreased accuracy (Congreve & Lamsdell, 2016; Puttick et al., 2017). Although the results of IW analyses should be viewed cautiously, their congruence with prior molecular phylogenetic studies suggests that this relationship was correctly inferred. Two unequivocal synapomorphies exist for this clade: greater tuberosity of the humerus extends far proximal to humeral head and lesser tuberosity (73:1) and the presence of a separate foramen for the obturator nerve (87:1). As the focus of this study was the assess the phylogenetic relationships of taxa within Desmatophocidae, our sampling of taxa and character relevant to higher level pinniped phylogeny are limited. Large-scale analyses of pinniped phylogeny incorporating additional characters, taxa, and molecular data are needed to assess the validity of Otarioidea fully, and may change our understanding of what pinniped family desmatophocids are truly most closely related to. Taxonomy of the Allodesminae As many as one to four genera of Allodesminae have been recognized (Table 9), including Allodesmus (Kellogg, 1922), Atopotarus (Downs, 1956), Brachyallodesmus (Barnes & Hirota, 1995), and Megagomphos (Barnes & Hirota, 1995). Of these genera, both phylogenetic analyses suggest the presence of two diagnosable genera within Allodesminae: Atopotarus and Allodesmus. Atopotarus courseni from the Altamira Shale of California was originally recognized as a separate genus from Allodesmus by Downs (1956) on the basis of a nearly complete skeleton preserved in situ on a slab, missing only the hindlimbs. This species has either been recognized as a separate genus (Downs, 1956; Barnes & Hirota, 1995) or as a species of Allodesmus (Mitchell, 1966; Barnes, 1972). More recent studies have included it as a species within Allodesmus (Kohno, 1996; Deméré & Berta, 2002). No explicit reasons were given for the recognition of one genus of Allodesminae by Deméré & Berta (2002), other than a statement that the Allodesminae were ‘over-split’. Table 9. Published records of Desmatophocidae Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) View Large Table 9. Published records of Desmatophocidae Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) Proposed nomenclature Prior nomenclature Locality Age Reference Desmatophoca oregonensis Desmatophoca oregonensis Astoria Formation, OR, USA Burdigalian Condon, (1906), Mitchell, (1966), Barnes, (1987), Deméré & Berta, (2002) Desmatophoca brachycephala Desmatophoca brachycephala Astoria Formation, WA, USA Aquitanian Barnes, (1987), Deméré & Berta, (2002) Atopotarus courseni Atopotarus courseni Altamira Shale, CA, USA Langhian Downs, (1956) Atopotarus sp. Atopotarus sp. Chikubetsu Formation, Japan Langhian Kimura et al. (1997) cf. Allodesmus Allodesminae indet. Astoria Formation, CA, USA Burdigalian Hunt and Barnes (1994) Allodesmus n. sp. 1 Allodesmus n. sp. ‘Topanga’ Formation, CA, USA Burdigalian Velez-Juarbe, (2017) Allodesmus n. sp. 2 Allodesmus n. sp. Okoppezawa Formation, Japan Langhian Tonomori et al. (2016) Allodesmus kernensis Allodesmus gracilis, Allodesmus kelloggi, Desmatophocine C Round Mountain Silt, CA, USA Langhian Kellogg (1922), Kellogg (1931), Barnes, (1970), Barnes, (1972), Mitchell, (1966), Barnes & Hirota, (1995) Allodesmus packardi Brachyallodesmus packardi ‘Ladera Sandstone’, CA, USA Serravallian Barnes, (1972) Allodesmus naorai Allodesmus naorai Mito Formation, Japan Serravallian Kohno (1996) Allodesmus sadoensis Allodesmus sadoensis Tsurushi Formation Serravallian Barnes & Hirota, (1995) Allodesmus sp., cf. A. sadoensis Desmatophocine B Round Mountain Silt Langhian Barnes, (1972) Allodesmus sinanoensis Megagomphos sinanoensis, Allodesmus megallos Aoki Formation, Japan Serravallian Barnes and Hirota (1995), Kohno (1996), Kohno et al. (2007) Allodesmus demerei Allodesmus n. sp. Montesano Formation, WA, USA Tortonian This study Allodesmus sp. Allodesmus sp. Monterey Formation, CA, USA Langhian- Serravallian Barnes & Hirota, (1995) Allodesmus sp. Allodesmus sp. Rosarito Beach Formation, Mexico Langhian Downs (1955) Allodesmus sp. Allodesmus sp. Santa Margarita Sandstone, CA, USA Tortonian Repenning and Tedford (1977) Imagotaria n. sp. Desmatophocine A Round Mountain Silt, CA, USA Langhian Barnes, (1972) View Large Our phylogenetic analysis places At. courseni as the sister taxon to all other allodesmine seals, and this genus can be differentiated by the absence of a prenarial shelf and M2 (also absent in Al. sadoensis), double-rooted cheek teeth, a small, triangular postorbital process, and a mastoid process projecting ventral to the postglenoid process. In light of these differences and placement of this species as the sister taxon to all other Allodesmus, we follow Barnes and Hirota (1995) in retaining a separate genus for Atopotarus. Further discoveries of better preserved cranial material of At. courseni and other archaic allodesmines (Velez-Juarbe, 2017) from the early middle Miocene will help in re-evaluation of the generic distinctiveness of Atopotarus. Allodesmus packardi from the ‘Ladera Sandstone’ of California was originally reported by Packard (1962), who referred the partly prepared skull to Atopotarus sp., cf. At. courseni; Mitchell (1966) agreed with this identification. Barnes (1972) re-described the specimen following additional preparation and erected the new species Al. packardi, noting various craniodental features differentiating it from Al. kernensis (Barnes, 1972: 41–42). Barnes and Hirota (1995) later erected the new genus Brachyallodesmus to contain this species, noting several plesiomorphic differences placing it outside Allodesmus, including canines with oval cross-section, more inflated tympanic bulla, a large pre-glenoid process, complete lateral wall of alisphenoid canal, a well-developed inferior petrosal venous sinus, a less cubic mastoid process, and retention of a pseudosylvian sulcus on the braincase. Barnes and Hirota (1995) further distinguished Brachyallodesmus packardi from Allodesmus spp. on the basis of unique derived features, including the lack of sagittal and nuchal crests, widely diverging tooth rows, cheek teeth with longitudinal sulci, and P1 positioned posteromedial to the C1. Our IW phylogenetic analysis finds Al. packardi to be the sister taxon to a clade including Al. naorai and all later diverging Allodesmus, although with very little support. Of the unique derived features listed by Barnes and Hirota (1995), widely diverging tooth rows are shared with Al. naorai (character 14 of our study). Extent and size of nuchal and sagittal crests are prone to sexual dimorphism and ontogenetic variation, and without a larger sample size we hesitate to consider their absence a valid diagnostic character. A similar problem can be found with using position of the P1, which may be influenced by size of the canine, another sexually dimorphic trait. Finally, presence of cheek teeth with a longitudinal sulcus is related to progression from double-rooted to single-rooted cheek teeth. This can be somewhat variable in pinnipeds, and pending further work on dental variation in pinnipeds we do not treat it as a separate intermediate state between single- and double-rooted cheek teeth. Given the relatively poor support for Al. packardi within our phylogenetic tree, and the lack of resolution within Allodesmus in our NW tree, we consider Brachyallodesmus a junior synonym of Allodesmus and follow Deméré and Berta (2002) in considering this species as Al. packardi as originally reported by Barnes (1972). Megagomphos represents the fourth putative genus of allodesmine. The holotype specimen is a large articulated anterior portion of the rostrum and mandible, and was originally reported by Nagao (1941) as Eumetopias sinanoensis; its desmatophocid affinities were later recognized by Mitchell (1966), who synonymized it with Al. kernensis; Repenning and Tedford (1977) considered it to represent a separate species within Allodesmus on account of its gigantic size. Barnes and Hirota (1995) later placed it outside Allodesmus owing to lack of a prenarial shelf, erecting the new genus Megagomphos. In the same study, Barnes and Hirota (1995) named the new species Allodesmus megallos based on a gigantic rostrum with procumbent incisors and canines. Both holotypes were collected from a narrow stratigraphic interval, the lower Aoki Formation (Barnes & Hirota, 1995; Kohno, 1996). In contrast, Kohno (1996) referred both these specimens to the same taxon, which he recombined as Al. sinanoensis. Kohno’s (1996) referral of both specimens to the same species was based upon the large size of both specimens, presence of a prenarial shelf, and origination from the same stratigraphic unit (Serravallian equivalent Aoki Formation). Despite the fact that both specimens originated from the Aoki Formation, Barnes and Hirota (1995) did not publish any comparisons between them. The holotype of Megagomphos sinanoensis is transversely crushed (Barnes & Hirota, 1995; Kohno, 1996), with damage obscuring the prenarial shelf. However, Kohno (1996) indicated that a prenarial shelf is present given that the bony nares are posteriorly retracted to a degree similar to other Allodesmus spp. He identified several features shared between these, including a prenarial shelf and large, procumbent, tusk-like canines and incisors. Owing to these shared similarities and their origination from a narrow stratigraphic interval, we follow Kohno (1996) in recognizing a single species, Al. sinanoensis, and consider Megagomphos a junior synonym of Allodesmus and Allodesmus megallos a junior synonym of Al. sinanoensis. We note that these specimens may be differentiated from other Japanese desmatophocids: from Al. sadoensis owing to its larger size, proportionally larger I3, proportionally smaller postcanines, and presence of postcanine diastemata; and from Al. naorai in its much larger size, anteriorly widening prenarial shelf, and proportionally shorter and posteriorly retracted nasals. Additional material of Al. sinanoensis was reported by Kohno et al. (2007). Desmatophocid diversity from the Round Mountain Silt Allodesmus kernensis was originally reported by Kellogg (1922) on the basis of the partial holotype mandible and other isolated cranial and postcranial elements from the Round Mountain Silt member of the Temblor Formation in the vicinity of Sharktooth Hill, Kern County, CA, USA. Mitchell (1966) reported a nearly complete skeleton collected from the Sharktooth Hill Bonebed, and noted subtle and unquantified mandibulodental differences with Al. kernensis (narrower symphyseal angle, more erect canine, narrower interalveolar septa, larger alveoli, and I2 positioned posteromedial rather than posterior to I3; Mitchell, 1966). Mitchell (1966) remarked upon problematical locality information for the holotype mandible and speculated that the holotype was collected three miles east of the bonebed and therefore stratigraphically below the Sharktooth Hill Bonebed. Owing to these subtle differences and potential stratigraphic separation, Mitchell (1966) erected the species Al. kelloggi for the new skeleton from the bonebed and referred additional specimens from the bonebed to this new species. Barnes (1970, 1972) reported many new specimens of Allodesmus from the bonebed including mandibles and evaluated morphological variation of Allodesmus mandibles. Barnes (1970, 1972) found that the morphological characters (e.g. symphyseal angle) reported by Mitchell (1966) as distinguishing Al. kelloggi from Al. kernensis fell within the range of variation of modern pinnipeds (Z. californianus) and declared Al. kelloggi a junior synonym of Al. kernensis. Barnes (1972) further indicated that contradictory locality information exists for the Al. kernensis holotype, far from clearly indicating a precise locality or stratigraphic interval. Furthermore, Hanna (in Barnes, 1972) considered that unless prolific collector Charles Morrice collected the specimen himself, the specimen more than likely came from the bonebed rather than a different horizon within the Round Mountain Silt; we infer this to mean that Hanna considered collectors other than Morrice lacked stratigraphic expertise and would have been likely to have collected specimens only from the highly prolific bonebed rather than sparsely fossiliferous strata above and below. A very different approach was later advocated by Barnes and Hirota (1995), who reconsidered the taxonomy of Round Mountain Silt Allodesmus spp. in light of new discoveries from Japan and examination of more desmatophocid fossil material. These authors resurrected Al. kelloggi, restricted Al. kernensis and Al. kelloggi to their respective holotype specimens, and erected the new species Al. gracilis, referring all bonebed specimens aside from the Al. kelloggi type to this new species. Barnes and Hirota (1995) further indicated that Al. kelloggi differs from all Al. gracilis specimens in bearing a deep and rugose masseteric fossa, a transversely expanded flange on the coronoid crest, and a single rooted M1. However, Barnes and Hirota (1995) did not re-address the statistical results which Barnes (1970, 1972) cited while synonymizing Al. kelloggi, nor did they quantify any of these differences. Cranial differences between Al. kelloggi and Al. gracilis cited by Barnes and Hirota (1995: 347) are likewise not quantified and are subjective or difficult to quantify (e.g. ‘differing from Al. gracilis by having zygomatic process of squamosal more curved and bowed laterally, not so straight’). We agree with Barnes (1970, 1972) that mandible and dental differences amongst Round Mountain Silt Allodesmus spp. are attributable to intraspecific variation; furtherore, we note that Barnes (1970, 1972) demonstrated that some aspects of mandibular variation cited earlier by Mitchell (1966) and later by Barnes and Hirota (1995) to separate Al. kernensis from other Sharktooth Hill mandibles are actually less extreme than within extant Z. californianus. Furthermore, even if the holotype mandible of Al. kernensis originated from a stratigraphically lower horizon than the Sharktooth Hill bonebed, extinct species are diagnosed based upon morphology, not age. Geochronologically older specimens falling within a range of variation acceptable for referral to a slightly younger species simply expand the geochronological range of a species and do not necessitate naming of new taxa. We therefore identify Al. gracilis and Al. kelloggi as junior synonyms of Al. kernensis (Table 9), in broad agreement with Deméré and Berta (2002: 140). Although the previously described allodesmine taxa from Sharktooth Hill represent a single taxon, additional allodesmine taxa may have been present in the fauna. One intriguing fossil, referred to by Barnes (1972) as ‘Desmatophocine B’, is known from an isolated mandible with large single-rooted teeth. This specimen notably differs from Al. kernensis in having a deeper symphyseal region (and wider symphyseal angle) and absence of postcanine diastemata and M2 (polymorphic in Al. kernensis; Barnes, 1970, 1972). Barnes (1972) considered this mandible to represent a desmatophocid outside the genus Allodesmus. We also find this specimen interesting and note several distinct features that separate this taxon from Al. kernensis, including loss of the M2 and presence of premolars with large roots crowded into the anterior mandible and with dorsally diverging orientations. These features are all shared with Al. sadoensis to the exclusion of all other Allodesmus spp. Given that ‘Desmatophocine B’ differs in only minor ways from Al. sadoensis (transversely narrower mandible, shallower mandibular angle), we reidentify ‘Desmatophocine B’ as Allodesmus sp., cf. Al. sadoensis. Another desmatophocid taxon recognized by Barnes (1972), ‘Desmatophocine C’, is based on a mandible fragment lacking teeth, which he separated from Al. kernensis based on only the transversely wider symphyseal portion of the mandible. As this difference was not quantified and seems indistinguishable from that preserved in juvenile specimens of Al. kernensis (e.g. UCMP 81704; Barnes, 1972: fig. 13b), we reidentify this specimen as Allodesmus sp., cf. Al. kernensis. Extinction of the Desmatophocidae Allodesmus demerei represents the geochronologically youngest species of Allodesmus and the family Desmatophocidae (Fig. 9), and the discovery of this taxon invites the question, ‘Why did desmatophocids become extinct?’ No obvious mass extinctions amongst marine vertebrates have been recorded during the middle to late Miocene; on the contrary, cetacean diversity peaked during the Tortonian (Uhen & Pyenson, 2007; Marx & Uhen, 2010). Primary productivity appears to have peaked during the Serravalian–Tortonian interval (Fig. 11). We find no correlation between changes in productivity and desmatophocid extinction, and this group of pinnipeds seems to be declining while other pinniped lineages and whales are actually increasing in diversity (Marx & Uhen, 2010; Churchill et al., 2014) Change in sea level could radically change the geography of haul-out areas and extent of rocky versus sandy shores and has been implicated in the extinction of Mio-Pliocene phocids in South America (Valenzuela-Toro et al., 2013), but no obvious mid-late Miocene changes in sea level are apparent (Fig. 11). The peak in desmatophocid diversity does, however, correspond to the mid-Miocene climatic optimum, and diversity begins to decline along with falling temperatures (Fig. 11). Figure 11. View largeDownload slide Trends in North Pacific climate and pinniped diversity. Diatom diversity from Marx and Uhen, (2010), sea level relative to modern from Miller et al., (2005), and δ18O values (proxy for sea surface temperature) from Fedorov et al., (2013) and Rousselle et al., (2013). Data are plotted only from the late Oligocene to Holocene. See Supporting Information, Appendix S5 for a complete list of pinniped fossil occurrences. Figure 11. View largeDownload slide Trends in North Pacific climate and pinniped diversity. Diatom diversity from Marx and Uhen, (2010), sea level relative to modern from Miller et al., (2005), and δ18O values (proxy for sea surface temperature) from Fedorov et al., (2013) and Rousselle et al., (2013). Data are plotted only from the late Oligocene to Holocene. See Supporting Information, Appendix S5 for a complete list of pinniped fossil occurrences. The paraphyletic ‘enaliarctines’ (early diverging Pinnipedimorpha, or stem Pinnipedia) represent the earliest diversification of pinnipeds, originating in the late Chattian, peaking in the Aquitanian (N = 6), and extinct by the end of the Burdigalian (Fig. 11). Desmatophocids appear during the Aquitanian and gradually increase in diversity until the Langhian (N = 6) and decline in diversity until the Tortonian (N = 1; Allodesmus demerei, 10.5–9.1 Mya). Desmatophocids then apparently became extinct by the end of the Tortonian (Figs 9, 11). Otariids appear in the Burdigalian (N = 2; Boessenecker & Churchill, 2015; Velez-Juarbe, 2017) but lack a Langhian record; they continue in low diversity (and small body size) until diversifying in the Messinian (N = 5) and Pliocene (N = 6) and mostly maintaining that diversity until the Holocene (Fig. 11). Walruses were ancestrally small in body size (Churchill et al., 2014). These small walruses, which were morphologically very similar to their ‘enaliarctine’ ancestors, were somewhat diverse in the Burdigalian (N = 4) and remained at low diversity through to the Tortonian, after which all species of odobenids are large bodied (Fig. 11). Large odobenids were low in diversity in the Burdigalian and gradually increase until their peak in the Messinian (N = 9) and Pliocene (N = 8), with their diversity decreasing sharply in the Pleistocene (Fig. 11). Phocids do not have a pre-Pleistocene fossil record in the North Pacific (Boessenecker, 2013) and quickly diversify after invading from the Arctic and South Pacific (Deméré et al., 2002; Arnason et al., 2006; Boessenecker & Churchill, 2016; Fig. 11). One obvious relationship between desmatophocid diversity and trends in other pinniped groups is the gradual increase in large walrus diversity as desmatophocid diversity declines (Fig. 11; see also Velez-Juarbe, 2017), suggesting a possible case of competitive displacement by walruses. On the contrary, otariid diversity increases only after desmatophocids are extinct, perhaps suggesting opportunistic niche filling. These data do not distinguish between generalist ‘imagotariine’ walruses and more specialized molluscivorous odobenines; however, we note that several Tortonian odobenids, including Imagotaria, Pontolis, Gomphotaria, and at least one new genus of ‘imagotariine’, lack obvious specialization for benthic feeding (unlike the Odobenini) and may have been exploiting similar prey resources as desmatophocid seals (Adam & Berta, 2002). Alternatively, odobenids and desmatophocids could also have been in competition for haul-out sites for breeding and moulting. Beyond competition for haul-out sites, comparisons of climate trends with pinniped diversity over time may also suggest an important role for climate change in the faunal turnover between desmatophocids and walruses, with walruses perhaps better adapted in some way to cooler environments. This, over time, might have led to the decline and eventual extinction of desmatophocids, while allowing walruses to proliferate and diversify. What behavioural, ecological, and morphological adaptations walruses may have had that desmatophocids lacked cannot be determined at this point but should be the focus of future studies of pinniped paleoecology and evolution. CONCLUSION A new skeleton of a desmatophocid seal from the upper Miocene Montesano Formation of Washington is described as a new species within the genus Allodesmus, Al. demerei sp. nov. Allodesmus demerei represents the geochronologically youngest desmatophocid, coexisting with imagotariine walruses and early otariid fur seals. Phylogenetic analysis supports placement of this taxon within the genus Allodesmus. The genera Brachyallodesmus and Megagomphos are synonymized with Allodesmus, while the genus Atopotarus is retained. Examination of diversity trends shows peak desmatophocid diversity coinciding with the middle Miocene climatic optimum, and desmatophocids decline as odobenids diversify. This suggests a role for climate change as well as competition with the rapidly diversifying walruses in the decline and extinction of the desmatophocids. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Appendix S1. Institutional acronyms. Appendix S2. List of material examined. Appendix S3. Character list for cladistic analysis. Appendix S4. Explanation of geochronological ranges. Appendix S5. Table of fossil pinniped occurrences from the North Pacific. Appendix S6. References cited. Appendix S7. Cladistic matrix for Allodesmus demerei [Version of Record, published online 03 January 2018; http://zoobank.org/urn:lsid:zoobank.org:pub:4DE5001F-A4BE-4A43-84A4-AEA693B3FA14] ACKNOWLEDGEMENTS First and foremost, we wish to thank the late P. K. Bigelow for discovering, collecting, and initiating study of UWBM 75640. We also thank M. Kelly for collecting, preparing, and donating the referred humerus. Comments from the editor and two anonymous reviewers greatly improved the quality of this paper. This study benefitted from discussions with L. G. Barnes, A. Berta, T. A. Deméré, A. Garibay, N. Kohno, J. F. Parham, and Y. Tanaka. Thanks to the following, who provided access to specimens and collections under their care: L. G. Barnes, S. A. McLeod, and V. Rhue (LACM), M. B. Goodwin, and P. A. Holroyd (UCMP), D. J. Bohaska, C. Potter, and N. D. Pyenson (USNM), R. C. Eng, M. Rivin, and C. A. Sidor (UWBM); additional thanks to M. Rivin for facilitating a loan of UWBM 75640 to R.W.B. This study was generously funded by a University of Washington Burke Museum of Natural History and Culture Vertebrate Paleontology Collection Study Grant to R.W.B.; travel grants supporting collections visits include a University of Otago Doctoral Scholarship to R.W.B. REFERENCES Adam PJ , Berta A . 2002 . Evolution of prey capture strategies and diet in the Pinnipedimorpha (Mammalia: Carnivora) . Oryctos 4: 83 – 107 . Addicott WO . 1976 . Neogene molluscan stages of Oregon and Washington . In: Fritsche AE , TerBest HJ , Wornardt WW , eds. The Neogene Symposium: Society of Economic Paleontologists and Mineralogists, Pacific Section . 95 – 115 . Agnarsson I , Kuntner M , May-Collado LJ . 2010 . Dogs, cats, and kin: a molecular species-level phylogeny of Carnivora . Molecular Phylogenetics and Evolution 54 : 726 – 745 . Google Scholar Crossref Search ADS PubMed Arnason U , Gullberg A , Janke A , Kullberg M , Lehman N , Petrov EA , Väinölä R . 2006 . Pinniped phylogeny and a new hypothesis for their origin and dispersal . Molecular Phylogenetics and Evolution 41 : 345 – 354 . Google Scholar Crossref Search ADS PubMed Bapst DW . 2012 . Paleotree: an R package for paleontological and phylogenetic analyses of evolution . Methods in Ecology and Evolution 3 : 803 – 807 . Google Scholar Crossref Search ADS Barnes LG . 1970 . A re-evaluation of mandibles of Allodesmus (Otariidae, Carnivora) from the Round Mountain Silt, Kern County, California . PaleoBios 10 : 1 – 24 . Barnes LG . 1972 . Miocene Desmatophocinae (Mammalia: Carnivora) from California . University of California Publications in Geological Sciences 89 : 1 – 76 . Barnes LG . 1979 . Fossil enaliarctine pinnpeds (Mammalia: Otariidae) from Pyramid Hill, Kern County, California . Contributions in Science, Natural History Museum of Los Angeles County 318 : 1 – 41 . Barnes LG . 1987 . An early Miocene pinniped of the genus Desmatophoca (Mammalia: Otariidae) from Washington . Contributions in Science, Natural History Museum of Los Angeles County 382 : 1 – 20 . Barnes LG . 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 – 28 . Barnes LG . 2008 . Otarioidea . In: Janis CM , Gunnell GF , Uhen MD , eds. Evolution of Tertiary mammals of North America, Vol. 2: Small mammals, xenarthrans, and marine mammals . Cambridge : Cambridge University Press . 523 – 541 . Barnes LG , Hirota K . 1995 . Miocene pinnipeds of the otariid subfamily Allodesminae in the North Pacific Ocean: systematics and relationships . The Island Arc 3 : 329 – 360 . Google Scholar Crossref Search ADS Barnes LG , Mitchell ED . 1975 . Late Cenozoic northeast Pacific Phocidae . Conseil International pour l’Exploration de la Mer, Rapports et Procèsverbaux des Rèunions 169: 34 – 62 . Barnes LG , Raschke RE . 1991 . Gomphotaria pugnax, a new genus and species of late Miocene dusignathine otariid pinniped (Mammalia: Carnivora) from California . Natural History Museum of Los Angeles County Contributions in Science 426 : 1 – 27 . Bebej RM . 2009 . Swimming mode inferred from skeletal proportions in the fossil pinnipeds Enaliarctos and Allodesmus (Mammalia, Carnivora) . Journal of Mammalian Evolution 16 : 77 – 97 . Google Scholar Crossref Search ADS Berta A . 1994 . A new species of phocoid pinniped Pinnarctidion from the early Miocene of Oregon . Journal of Vertebrate Paleontology 14 : 405 – 413 . Google Scholar Crossref Search ADS Berta A , Ray CE . 1990 . Skeletal morphology and locomotor capabilities of the archaic pinniped Enaliarctos mealsi . Journal of Vertebrate Paleontology 10 : 141 – 157 . Google Scholar Crossref Search ADS Berta A , Ray CE , Wyss AR . 1989 . Skeleton of the oldest known pinniped, Enaliarctos mealsi . Science 244 : 60 – 62 . Google Scholar Crossref Search ADS PubMed Berta A , Wyss AR . 1994 . Pinniped phylogeny . Proceedings of the San Diego Society of Natural History 29 : 33 – 56 . Bianucci G , Di Celma C , Urbina M , Lambert O . 2016 . New beaked whales from the late Miocene of Peru and evidence for convergent evolution in stem and crown Ziphiidae (Cetacea, Odontoceti) . PeerJ 4 : e2479 . Google Scholar Crossref Search ADS PubMed Bigelow PK . 1994 . Occurrence of a squaloid shark (Chondrichthyes: Squaliformes) with the pinniped Allodesmus from the upper Miocene of Washington . Journal of Paleontology 68 : 680 – 684 . Google Scholar Crossref Search ADS Boessenecker RW . 2013 . A new marine vertebrate assemblage from the Late Neogene Purisima Formation in Central California, Part II: Pinnipeds and cetaceans . Geodiversitas 35 : 815 – 940 . Google Scholar Crossref Search ADS Boessenecker RW , Churchill M . 2013 . A reevaluation of the morphology, paleoecology, and phylogenetic relationships of the enigmatic walrus Pelagiarctos . PLoS ONE 8 : e54311 . Google Scholar Crossref Search ADS PubMed Boessenecker RW , Churchill M . 2015 . The oldest known fur seal . Biology Letters 11 : 20140835 . Google Scholar Crossref Search ADS PubMed Boessenecker RW , Churchill M . 2016 . The origin of elephant seals: implications of a fragmentary late Pliocene seal (Phocidae: Miroungini) from New Zealand . New Zealand Journal of Geology 59 : 544 – 550 . Google Scholar Crossref Search ADS Bowditch TE . 1921 . An analysis of the natural classifications of Mammalia for the use of students and travelers . Paris : J. Smith . Churchill M , Boessenecker RW , Clementz MT . 2014 . Colonization of the Southern Hemisphere by fur seals and sea lions (Carnivora: Otariidae), revealed by combined evidence phylogenetic and Bayesian biogeographic analysis . Zoological Journal of the Linnean Society 172 : 200 – 225 . Google Scholar Crossref Search ADS Churchill M , Clementz MT . 2015 . The evolution of aquatic feeding in seals: insights from Enaliarctos (Carnivora: Pinnipedimorpha), the oldest known seal . Journal of Evolutionary Biology 29 : 319 – 334 . Google Scholar Crossref Search ADS PubMed Churchill M , Clementz MT , Kohno N . 2014 . Cope’s rule and the evolution of body size in Pinnipedimorpha (Mammalia: Carnivora) . Evolution 69 : 201 – 215 . Google Scholar Crossref Search ADS PubMed Condon T . 1906 . A new fossil pinniped (Desmatophoca oregonensis) from the Miocene of the Oregon coast . University of Oregon Bulletin 3 : 1 – 14 . Congreve CR , Lamsdell JC . 2016 . Implied weighthing and its utility in paleontological datasets: a study using modeled phylogenetic matrices . Paleontology 59 : 44 – 462 . Google Scholar Crossref Search ADS Debey LB , Pyenson ND . 2013 . Osteological correlates and phylogenetic analysis of deep diving in living and extinct pinnipeds: what good are big eyes ? Marine Mammal Science 29 : 48 – 83 . Google Scholar Crossref Search ADS Deméré TA . 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 TA , Berta A . 2001 . A reevaluation of Proneotherium repenningi from the Miocene Astoria Formation of Oregon and its position as a basal odobenid (Pinnipedia: Mammalia) . Journal of Vertebrate Paleontology 21 : 279 – 310 . Google Scholar Crossref Search ADS Deméré TA , Berta A . 2002 . The Miocene pinniped Desmatophoca oregonensis Condon, 1906 (Mammalia: Carnivora) from the Astoria Formation, Oregon . Smithsonian Contributions to Paleobiology93: 113 – 147 . Deméré TA , Berta A , Adam PJ . 2003 . Pinnipedimorph evolutionary biogeography . Bulletin of the American Museum of Natural History 279 : 32 – 76 . Google Scholar Crossref Search ADS Downs T . 1955 . A fossil sea lion from the Miocene of the San Joaquin Hills, Orange County, California . Bulletin of the Southern California Academy of Science 54 : 49 – 56 . Downs T . 1956 . A new pinniped from the Miocene of southern California: with remarks on the Otariidae . Journal of Paleontology 30 : 115 – 131 . Fedorov AV , Brierley CM , Lawrence KT , Liu Z , Dekens PS , Ravelo AC . 2013 . Patterns and mechanisms of early Pliocene warmth . Nature 496 : 44 – 49 . Google Scholar Crossref Search ADS Fowler GA . 1965 . The stratigraphy, foraminifera, and paleoecology of the Montesano Formation, Grays Harbor County, Washington . Unpublished Ph.D. Thesis , University of Southern California . 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. Thesis , San Diego State University . Giffin EB . 1992 . Functional implications of neural canal anatomy in recent and fossil marine carnivores . Journal of Morphology 214 : 357 – 374 . Google Scholar Crossref Search ADS PubMed Gradstein FM , Ogg JG , Schmitz M , Ogg G . 2012 . The geologic time scale 2012 . Oxford : Elsevier . Hay OP . 1930 . Second bibliography and catalogue of the fossil Vertebrata of North America . Carnegie Institution of Washington 390 : 1 – 1074 . Hunt RM Jr. , Barnes LG . 1994 . Basicranial evidence for ursid affinity of the oldest pinnipeds . Proceedings of the San Diego Society of Natural History 29 : 57 – 67 . Illiger C . 1811 . Prodromus systematis mammalium et avium . Berlin : Salfeld . Kellogg R . 1922 . Pinnipeds from Miocene and Pleistocene deposits of California . University of California Publications in Geological Sciences 13 : 23 – 132 . Kellogg R . 1931 . Pelagic mammals from the Temblor Formation of the Kern River region, California . Proceedings of the California Academy of Sciences 19 : 217 – 397 . Kimura M , Hirota K , Kiyono C . 1997 . Fossil pinniped mandible from the middle Miocene of Haboro-cho, Hokkaido . Bulletin of the Hobetsu Museum 13 : 1 – 7 . Kohno N . 1994 . A new Miocene pinniped in the genus Prototaria (Carnivora: Odobenidae) from the Moniwa Formation, Miyagi, Japan . Journal of Vertebrate Paleontology 14 : 414 – 426 . Google Scholar Crossref Search ADS Kohno N . 1996 . Miocene pinniped Allodesmus (Mammalia: Carnivora) with special reference to the “Mito seal” from Ibaraki Prefecture, central Japan . Transactions and Proceedings of the Palaeontological Society of Japan 181 : 388 – 404 . Kohno N . 2006 . A new Miocene odobenid (Mammalia: Carnivora) from Hokkaido, Japan, and its implications for odobenid phylogeny . Journal of Vertebrate Paleontology 26 : 411 – 421 . Google Scholar Crossref Search ADS Kohno N , Koike H , Narita K . 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 . Koretsky IA , Barnes LG , Rahmat SJ . 2016 . Re-evaluation of morphological characters questions current views of pinniped origins . Vestnik Zoologii 50 : 327 – 354 . Google Scholar Crossref Search ADS Lambert O , Bianucci G , Muizon CD . 2017 . Macroraptorial sperm whales (Cetacea, Odontoceti, Physeteroidea) from the Miocene of Peru . Zoological Journal of the Linnean Society 179 : 404 – 474 . Linnaeus C . 1758 . Systema naturae per regna tria naturae, secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis . Holmiae, Stockholm : Tomus I. Laurentii Salvii . Marx FG , Uhen MD . 2010 . Climate, critters, and cetaceans: Cenozoic drivers of the evolution of modern whales . Science 327 : 993 – 996 . Google Scholar Crossref Search ADS PubMed Miller KG , Kominz MA , Browning JV , Wright JD , Mountain GS , Katz ME , Sugarman PJ , Cramer BS , Christie-Blick N , Pekar SF . 2005 . The Phanerozoic record of global sea-level change . Science 310 : 1293 – 1297 . Google Scholar Crossref Search ADS PubMed Mitchell ED . 1966 . The Miocene pinniped Allodesmus . University of California Publications in Geological Sciences 61 : 1 – 105 . Mitchell ED , Tedford RH . 1973 . The Enaliarctinae: A new group of extinct aquatic carnivora and a consideration of the origin of the otariidae . Bulletin of the American Museum of Natural History 151 : 203 – 284 . Miyazaki S , Horikawa H , Kohno N , Hirota K , Kimura M , Hasegawa Y , Tomida Y , Barnes LG , Ray CE . 1995 . Summary of the fossil record of pinnipeds of Japan, and comparisons with that from the eastern North Pacific . The Island Arc 3 : 361 – 372 . Google Scholar Crossref Search ADS 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, Series 4 6 : 75 – 84 . Packard EL . 1962 . Fossil marine mammals from the vicinity of Stanford University . Journal of Paleontology 36 : 29 – 37 . Pierce SE , Clack JA , Hutchison JH . 2011 . Comparative axial morphology in pinnipeds and its correlation with aquatic locomotory behaviour . Journal of Anatomy 219 : 502 – 514 . Google Scholar Crossref Search ADS PubMed Prothero DR , Lau JN . 2001 . Magnetic stratigraphy of the upper Miocene (type Wishkahan-Graysian) Montesano Formation, Grays Harbor County, Washington . In: Prothero DR , ed. Magnetic stratigraphy of the Pacific Coast Cenozoic . Los Angeles, California : Society of Economic Paleontologists and Mineralogists, Pacific Section . 293 – 301 . Prothero DR , Lau JN , Armentrout JM . 2001 . Magnetic stratigraphy of the upper Miocene (Wishkahan) Empire Formation, Coos County, Oregon . In: Prothero DR , ed. Magnetic stratigraphy of the Pacific Coast Cenozoic . Los Angeles, California : Society of Economic Paleontologists and Mineralogists, Pacific Section . 284 – 292 . Puttick MN , O’Reilly JE , Tanner AR , Fleming JE , Clark J , Holloway L , Lozano-Fernandez J , Parry LA , Tarver JE , Pisani D , Donoghue PCJ . 2017 . Uncertain-tree: discriminating among competing approaches to the phylogenetic analysis of phenotype data . Proceedings of the Royal Society B: Biological Sciences 284 : 20162290 . Google Scholar Crossref Search ADS Rau WW . 1967 . Geology of the Wynoochee Valley quadrangle, Grays Harbor County, Washington . Washington Division of Mines and Geology Bulletin 56 : 1 – 51 . Repenning CA , Tedford RH . 1977 . Otarioid seals of the Neogene . US Geological Survey Professional Paper 992 : 1 – 87 . Rousselle G , Beltran C , Sicre M , Raffi I , De Rafélis M . 2013 . Changes in sea-surface conditions in the equatorial Pacific during the middle Miocene-Pliocene as inferred from coccolith geochemistry . Earth and Planetary Science Letters 361 : 412 – 421 . Google Scholar Crossref Search ADS Scheffer VB . 1958 . Seals, sea lions, and walruses . Stanford : Stanford University Press . Sivertsen E . 1954 . A survey of eared seals (family Otariidae) with remarks on the Antarctic seals collected by M/K “Norvegia” in 1928–1929 . Det Vorske Videnkaps-Akademii Oslo 36 : 1 – 76 . Tanaka Y , Kohno N . 2015 . A new late Miocene odobenid (Mammalia: Carnivora) from Hokkaido, Japan suggests rapid diversification of basal Miocene odobenids . PLoS ONE 10 : e0131856 . Google Scholar Crossref Search ADS PubMed Tonomori W , Sawamura H , Sato T , Kohno N . 2016 . New skeletal material of the extinct Miocene pinniped Allodesmus from Japan indicates their new systematics and locomotory adaptation . Journal of Vertebrate Paleontology, Program and Abstracts 2016 : 237 . Uhen MD , Pyenson ND . 2007 . Diversity estimates, biases, and historiographic effects: resolving cetacean diversity in the Tertiary . Palaeontologia Electronica 10: 1 – 22 . Valenzuela-Toro AM , Gutstein CS , Varas-Malca RM , Suarez ME , Pyenson ND . 2013 . Pinniped turnover in the South Pacific Ocean: new evidence from the Plio-Pleistocene of the Atacama Desert, Chile . Journal of Vertebrate Paleontology 33 : 216 – 223 . Google Scholar Crossref Search ADS Velez-Juarbe J . 2017 . Eotaria citrica, sp. nov., a new stem otariid from the “Topanga” formation of Southern California . PeerJ 5 : e3022 . Google Scholar Crossref Search ADS PubMed Walsh TJ , Korosec MA , Phillips WM , Logan RL , Schasse HW . 1987 . Geologic map of Washington - southwest quadrant . Washington Division of Geology and Earth Resources Geologic Map GM-34 : 1 – 28 . Wiens JJ . 1999 . Polymorphism in systematics and conservation biology . Annual Review of Ecology and Systematics 30 : 327 – 362 . Google Scholar Crossref Search ADS Wyss AR . 1987 . The walrus auditory region and the monophyly of pinnipeds . American Museum Novitates 2871 : 1 – 31 . © 2018 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Zoological Journal of the Linnean SocietyOxford University Press

Published: Sep 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off