Wednesday, September 24, 2014

The evolutionary history of walruses, part 4: the odobenines and the evolution of the modern walrus

For other entries in this series, see:

Walrus Evolution, part 1
Walrus Evolution, part 2
Walrus Evolution, part 3

Walrus Evolution, part 5

The Walrus (Trichechus rosmarus) is a very fat, clumsy brute, much uglier than his picture, with a coarse, oily skin all wrinkled and scarred; long, protruding tusks; bristly whiskers and scuffling flippers that barely serve to move his bulky body over the land. In the water he is more at home, and though it does not require a high degree of strength and skill to dig clams, that being his daily occupation, yet he is able to keep very fat on the fruits of his industry and has much leisure to swim about or doze on ice floes and sea beaches.” – Dane Coolidge, Birds and Nature 10:2, September 1901

First introductory note: I have varied so far in writing in a chronological order in terms of the history of research or phylogenetic order (e.g. up the cladogram one node at a time). However, because the majority of non-odobenine odobenids (e.g. “Imagotariinae” and Dusignathinae) were not recognized as walruses until the 1970’s, only the odobenines were recognized as walruses during the early history of fossil walrus research. Because of this, the story of odobenines can largely be told in chronological order.

Second introductory note: This is by far and away my longest post ever on coastal paleo, so bear with me – save a half hour, or bookmark the page and come back. More has been published on odobenines than the rest of the odobenids combined, so there is quite a lot to summarize – and I think it’s quite a fascinating story.

The unfortunate taxonomic history of Alachtherium

The first fossil walrus was described in the mid 19th century by Du Bus (1867) and named Alachtherium cretsii based upon a well-preserved mandible from the lower Pliocene Scaldisian sands of Belgium. This mandible shares several features with modern Odobenus including an elongate mandibular symphysis and small coronoid process, a lower canine that is reduced to the same size as the cheek teeth, and incisors that are positioned anterior to the canine and in line with the toothrow (rather than medial to the canine). However, the mandible differs in its much larger size, having an upturned and unfused symphysis, and primitively retaining a fourth lower premolar (lost in Odobenus, which only has p1-3). Van Beneden (1877) later referred a partial braincase and humerus to the species, but Rutten (1907) thought the braincase and mandible were incompatible and erected a new taxon for the braincase, Trichechus antverpiensis*. This braincase is also larger than modern Odobenus rosmarus, and differs principally in having a rectangular dorsal margin in posterior view. Further unnecessary complications arose when Hasse (1910) described some partial skulls and postcrania of several individuals from the slightly younger, upper Pliocene Merxemian sands, which he named Alachtherium antwerpiensis (note: antverpiensis versus antwerpiensis) as he also considered the new material incompatible with the type.

*Note that early workers often included walrus in the genus Trichechus, which is the genus that the manatees belong in; most recent works do not discuss the errors of earlier workers, and a bit of searching on google has failed to enlighten me any further. I assume that superficial similarities such as blubber and a short muzzle as well as bottom feeding contributed to the confusion of earlier workers. However, Linneaus originally got it right by naming the species Phoca rosmarus – obviously not a phocid in the modern sense, but Linneaus placed practically all pinnipeds within the genus Phoca, so at least he recognized the pinniped affinities of the walrus.

Beautiful illustrations of the holotype mandible and referred braincase (in posterior view) of Alachtherium cretsii, (Pliocene Scaldisian Sands, Belgium) from Van Beneden (1877). Thanks to Olivier Lambert for the excellent scan of this work.

Before we return to the complicated taxonomic history of Alachtherium, another discovery was made around the turn of the century in Virginia: a new walrus, also based on a lower jaw, was named by Berry and Gregory (1906) as Prorosmarus alleni. The fossil jaw was collected from the lower Pliocene Yorktown Formation, and is similar to both Odobenus and Alachtherium cretsii in having a less-upturned ramus as in the former, but again having an unfused symphysis as in the latter as well as primitively retaining a lower fourth premolar, which (along with the lower molar) is missing in Odobenus.

The holotype mandible of Prorosmarus alleni, (Pliocene Yorktown Formation, Virginia) photographed at the Smithsonian.
For the rest of the 20th century, work on Pliocene walruses assignable to Alachtherium was monopolized by European researchers. Van der Feen (1968) described some new cranial material which he assigned, without explanation, to T. antverpiensis, which he placed in Odobenus (as Odobenus antverpiensis); this was done without explaining why the fossil was not assignable to either species of Alachtherium. Later work by Erdbrink and Van Bree (1990) figured and described a beautiful, complete, and gigantic skull dredged from the seafloor off the Dutch coast which they similarly identified as Odobenus antverpiensis. Erdbrink and Van Bree (1986, 1990) considered virtually all Pliocene walruses to belong in the genus Odobenus and assigned all specimens from the North Atlantic to O. antverpiensis, identifying other species Trichecodon huxleyi (see below), Trichecodon koninkii (see below), and Alachtherium cretsii as nomina nuda* as well as (rightfully) questioning the generic distinctiveness of Prorosmarus.

The skull of Alachtherium cretsii (see below) reported by Erdbrink and Van Bree (1990, as O. antverpiensis) is large (a bit larger than extant Odobenus), and bears large canine alveoli indicating the presence of tusks that are slightly anteriorly sloping, unlike the vertical tusks of extant Odobenus. The skull also has a somewhat elongate rostrum (as opposed to the blunt and inflated rostrum of Odobenus), and possesses more teeth than Odobenus: the modern walrus lacks incisors and an upper molar, and these teeth are primitively retained in Alachtherium. A single postcranial feature unites Alachtherium with the modern walrus: a deltoid insertion on the humerus that is separated from the deltopectoral crest (in other pinnipeds and non-odobenines, the deltoid insertion is positioned on the crest and not easily identifiable). As it turns out, humeral morphology is fairly diagnostic in walruses (more on this below).

*Nomen nudum means a “naked name”: a name that has been proposed but an insufficient description or diagnosis has been used. In all three cases these names satisfy the minimum requirements for being valid names under ICZN rules and thus this was a bit of a taxonomic faux-pas on behalf of Erdbrink and Van Bree. However, they were right to question the distinctiveness of T. huxleyi and T. koninckii (see below), but Alachtherium cretsii is clearly a good name with a well-preserved, readily diagnosable (and, historically diagnosed and well-figured) type specimen. However, the plot thickens – so read on…

Well-preserved skull of Alachtherium cretsii dredged from the North Sea offshore of the Netherlands (Pliocene), from Post (2004).

In 1994, San Diego Natural History Museum curator Tom Deméré published two papers, one of which described the dusignathine Dusignathus seftoni (see previous post) and also the bizarre odobenine Valenictus chulavistensis (see below). The second paper was a phylogenetic analysis and revision of the Odobenidae, and he summarized much of the prior work on North Atlantic Pliocene walruses. He indicated that the large skull assigned to “Odobenus antverpiensis” by Erdbrink and Van Bree (1990) uniquely shares a rectangular outline of the braincase in posterior view with braincases figured by Van Beneden (1877), Rutten (1907), and Hasse (1910), previously assigned to either T./O. antverpiensis (Rutten, 1907) or Alachtherium antwerpiensis (Hasse, 1910). Deméré (1994) concluded that insufficient evidence existed to distinguish between two (or three, for that matter) species of Alachtherium, and synonymized all with Alachtherium cretsii, noting that Prorosmarus alleni may also fall victim to synonymy. In defense of this lumping, Deméré (1994) noted that all the material is relatively larger than extant Odobenus and the toothrow of the skull and mandible are both sinuous and matching in profile. He further indicated that two species of Alachtherium may be a defensible hypothesis considering that the material reported by Hasse (1910: named as A. antwerpiensis) is late Pliocene in age as opposed to the early Pliocene age of the holotype A. cretsii mandible. Deméré (1994) further noted that the skull reported by Erdbrink and Van Bree (1990) has four postcanine teeth as opposed to five in the A. antwerpiensis skull described by Hasse (1910), and that the former specimen retains a medial incisor whereas it is lost in the latter.

            A curious recent development was a popular article by Klaas Post (2004) who agreed with studies by Deméré (1994) and Kohno et al. (1995) in assigning Pliocene North Atlantic walruses to Alachtherium cretsii. Despite this rather sober taxonomic opinion, Post (2004: page 70) explained that “even Americans (Deméré, 1994) and the Japanese (Kohno et al., 1995)” have contributed to the “babylonian confusion” of North Atlantic walrus taxonomy. Those dastardly North Americans and Japanese! As it happens, this was nothing more than an unfortunate accident in word choice by Post (2004; see below) and difficulty in translating Dutch to English. Kohno and Ray (2008) humorously responded to these comments:

“If Post’s (2004:70) characterization of the study of Pliocene odobenines as complete Babylonian confusion is correct, then he deserves some of the confusion for it. He repeated previous generic misspellings, Trichechodon (p.70) and Obdobaenus (p.73), in allusion to Trichecodon huxleyi Lankester, 1865. He did not mention Alachtherium antwerpiensis Hasse, 1909, but did attribute A. antverpiensis (Rutten, 1907) to Hasse, while not citing Rutten at all, even though published in Amsterdam. Although he noted (p.70) that even Americans and Japanese had meddled into discussion of Pliocene walruses, he neglected to cite two of the most prominent recent transgressions in support of his case…Our paper, based on Pliocene fossils from the eastern United States, may well be perceived as yet another transgression into European affairs. The notorious disregard by marine mammals for political boundaries, though an intractable problem in conservation, has made their fossils far more interesting than would extreme provincialism. We do not share Post’s pessimism about the status of knowledge of Pliocene walruses, but feel rather that much progress has been made through the contributions of all who have focused on the fossils, irrespective of nationality and in spite of multiplicity of languages.”

UPDATE: I've been informed by a European colleague that the wording used by Post (2004) was accidental and not meant to convey any nationalistic issues, and instead was intended to simply express the variety of researchers from different regions weighing in on the taxonomy of Alachtherium. I've been informed that Kohno and Post get on quite well and have exchanged casts, and Post has even lived in Japan for a time. Thus it seems abundantly clear to me from my correspondence that no ill-will was intended and it was simply a case of "lost in translation". That all being said, lapses in citations and the like (as pointed out by Kohno and Ray, 2008) are arguably acceptable for a popular article. With this caveat, I've left the unedited quotations above.

The holotype of Pliopedia pacifica (left; latest Miocene Paso Robles Formation) and the referred forelimb and braincase (right; latest Miocene-earliest Pliocene Etchegoin Formation, California), from Kellogg (1921) and Repenning and Tedford (1977).

The Santa Margarita Walrus

In March 1909, Mr. Robert Anderson found some large pinniped bones in a conglomerate about a mile southeast of the dinky town of Santa Margarita in the California coast ranges. Later, this unit would later be named the Paso Robles Formation; in the vicinity of Santa Margarita, it overlies the type section of the Santa Margarita Sandstone, and is correlative in age with the Purisima Formation (latest Miocene and Pliocene). The Paso Robles Formation was deposited on the west side of the proto-coast ranges of California, which during the latest Miocene formed a large island separating the Temblor Sea* to the east (where the Etchegoin/San Joaquin Formations were deposited), which was connected to the Pacific by a straight to the north (where the Purisima Formation was deposited) and another straight to the south (where the Pismo Formation was deposited). Pliopedia pacifica was originally named from a fragmentary forelimb, which Kellogg (1922) tentatively assigned to the Otariidae, but also recognized some walrus features. Repenning and Tedford (1977) reported another partial skeleton including a braincase, complete humerus, radius, and ulna from the Etchegoin Formation (i.e. from within the Temblor Sea). They curiously referred Pliopedia, along with Valenictus imperialensis (keep reading) to the Dusignathinae, despite correctly identifying that Pliopedia had an Odobenus-like braincase and a deltoid tubercle separate from the deltopectoral crest of the humerus. The braincase is similar to Odobenus, Alachtherium, and Valenictus chulavistensis (see below) in lacking a sagittal crest and having a nuchal crest expanded into a crescent-shaped muscle attachment surface. Barnes and Raschke (1991) subsequently removed the Etchegoin Formation specimen from Pliopedia and cited unpublished research on a toothless odobenine from the Purisima Formation, which as of yet is still incomplete. Deméré (1994a) dismissed this removal; I have reservations about why the specimen was removed, and I believe that Repenning and Tedford (1977) were correct in their identification. Because of the derive humeral and cranial morphology, Deméré (1994a) placed Pliopedia within the Odobenini (see below) and indicated that more complete remains would likely show that this walrus bore a pair of enlarged tusks. Pliopedia, despite being poorly known (no additional material has been discovered since the “pinniped bible” was published), demonstrates that a single species of walrus inhabited both east and west shores of the coast range in the Pliocene. I would absolutely love to conduct some fieldwork in the Kettleman Hills (type section of Etchegoin and San Joaquin formations, along the west side of I-5 in California between Lost Hills and Kettleman City) and search for additional walrus material.

*The Temblor Sea takes its name from the Temblor Range (or possibly the Temblor Formation, or perhaps both are derived from the nearby Temblor Range). The Early Miocene age Pyramid Hill marine mammal assemblage from the Jewett Sand and the middle Miocene Sharktooth Hill assemblage from the Round Mountain Silt, both in the vicinity of Bakersfield, California, were deposited along the eastern shore of the Temblor Sea.

The holotype humerus of Valenictus imperialensis from the Pliocene Deguynos Formation of Imperial County, photographed at LACM.

Subtropical walruses? The Isla Cedros and Imperial walruses

In 1961, Ed Mitchell named a new genus and species of walrus from an isolated humerus collected from the early Pliocene Imperial Group of southern California. For the uninitiated, the Imperial Group is exposed in the Imperial desert near El Centro, which is located just north of the US-Mexico border, southwest of the Salton Sea, and east of San Diego. The Imperial Group was deposited in a rapidly subsiding basin and appears to have hosted a subtropical warm-water invertebrate fauna, preserved within the proto-gulf of California. The humerus was collected in 1949 from the Coyote Mountains, and subsequent visits to the locality by Ed Mitchell and others failed to yield any additional fossils; in the subsequent 50 years, only a handful of additional bones have been collected from the same formation. Mitchell (1961) correctly identified this specimen as a walrus, and as such this fossil represented the first explicitly recognized walrus from pre-Pleistocene rocks in the North Pacific. At the time, the absence of pre-Pleistocene walrus remains in the region led other researchers to propose that walruses immigrated to the Atlantic via the Panama seaway during the early Miocene. It seems a bit ridiculous now, but they really were operating in a vacuum of information. Mitchell (1961) was thus the first to identify that walruses did in fact have a North Pacific evolutionary heritage, and suggested that a center of origin may yet be identified in the North Pacific for walruses (as had already been identified for the Otariidae). Mitchell also pointed out the rather robust and strange construction of the humerus, including the huge knob-like medial entepicondyle, and suggested that Valenictus had a powerful flipper stroke and probably did not swim in a manner similar to otariids. The most interesting aspect of Valenictus imperialensis, aside from its strange morphology, is that it – like Pliopedia – was found in sediments deposited within a large embayment. Further making matters interesting is that Repenning and Tedford (1977) reported a somewhat younger partial humerus from the upper Pliocene San Joaquin Formation. The San Joaquin Formation overlies the Etchegoin Formation in the Kettleman Hills, and marks the final phase of marine sedimentation in the Temblor Sea; at the end of the Pliocene, uplift of the Sierra Nevada caused a massive influx of sediment shed westward into the San Joaquin Basin, which in concert with a Plio-Pleistocene fall in sea level, caused the Temblor Sea to dry up, forming the modern day southern San Joaquin valley. A last vestige of the Temblor Sea is the shallow and freshwater Tulare Lake west of Bakersfield, which has been mostly emptied by 20th century irrigation. The San Joaquin Formation is overlain by the estuarine and nonmarine uppermost Pliocene and Pleistocene Tulare Formation (which is fossiliferous and has yielded scrappy terrestrial mammals).

A few years later, a curious pinniped was dug out of the badlands of the south end of Isla Cedros off the Vizcaino Peninsula of Baja California by UC Riverside paleontology expeditions led by paleontologist Frank Kilmer. They collected an enormous volume of fossils which eventually led to the naming of various marine mammals like the fur seal Thalassoleon mexicanus, the false killer whale Praekogia cedrosensis, the porpoises Piscolithax boreios, Piscolithax tedfordi, Albireo whistleri, and Parapontoporia pacifica, and the early pilot whale-convergent beluga Denebola brachycephala. These fossils came from deposits of the Almejas Formation – and the strange new pinniped, despite lacking tusks, had several skull and postcranial features that allied it with the modern walrus – and Repenning and Tedford (1977) named it Aivukus cedrosensis (the genus name is Inuit for ‘walrus’). Aivukus has an elongate rostrum, small canines, highly worn teeth, reduced incisors, but an Odobenus-like basicranium and postcrania. The mandible of Aivukus was thought to have some similarities with Prorosmarus alleni, leading Repenning and Tedford (1977) to hypothesize that it was directly ancestral to Prorosmarus. Aivukus represents the most southerly described walrus, at about 28˚ latitude, and like Valenictus imperialensis, demonstrates that walruses inhabited subtropical waters. The southern occurrence of Aivukus led Repenning and Tedford (1977) to hypothesize that it or something similar immigrated to the Atlantic via the still-open Panamanian Seaway to give rise to Prorosmarus, Alachtherium, and eventually Odobenus during the Pliocene and Pleistocene.

Cast of the holotype skull of Aivukus cedrosensis from the late Miocene Almejas Formation of Baja California, photographed at the USNM.

The toothless Chula Vista walrus and a makeover of Valenictus

In the late 1980’s more discoveries were being made in the hills around San Diego, California. The Pliocene San Diego Formation had long been known to host a magnificent invertebrate fossil assemblage but had also produced a fair number of birds including several species of the flightless auk Mancalla, the fur seal Callorhinus gilmorei, the longirostrine dolphin Parapontoporia sternbergi, and baleen whales like Balaenoptera davidsonii. Subdivision-scale housing construction was booming in the 1980’s, and these construction operations often scraped off bedrock, uncovering vertebrate fossils in the process. Paleontological mitigation began in Orange County in the 1970’s and San Diego followed shortly thereafter (I’m not exactly sure on the timing of mitigation in San Diego or LA county, to be perfectly honest). This led to a figurative explosion in the amount of fossil vertebrates collected, and now institutions housing mitigation-derived collections like the Natural History Museum of LA, the San Diego Natural History Museum, and (most significantly) the Cooper Center are packed to the brim with exciting collections of marine vertebrates (99% of which are undescribed!).

The holotype skull of Valenictus chulavistensis (Pliocene, San Diego Formation, California), photographed at the SDNHM.

Valenictus chulavistensis was one of those discoveries. A couple of tusks from the San Diego Formation were originally thought to represent some sort of weird proboscidean; a strange, toothless mandible found later yielded no further clues. A partial skeleton was subsequently discovered, including a fragmentary odobenine walrus skull with the same type of tusk, a humerus with the same strange morphology as Valenictus imperialensis, and the strange mandible, as well as a bunch of other postcranial bones – confirming that Valenictus imperialensis was indeed a tusked odobenine walrus (Deméré 1994b). Weirder yet, the mandible and skull lacked any teeth aside from the upper canine. A nearly complete but smaller (and therefore younger) male skull was found and designated as the paratype for the species; this second skull, along with an isolated juvenile maxilla, demonstrated that the lack of teeth in the adult holotype specimen was not some weird pathologic condition. Since the early 90’s, four additional skulls have been found, and none of them exhibit any non-canine teeth.


The more completely preserved paratype skull of Valenictus chulavistensis; this is the skull figured in Demere (1994b).

Deméré (1994b) explained that studies of walrus feeding show that modern walruses do not use their teeth during feeding, and rather only clack their teeth together as a form of underwater communication. The classic study of walrus biology and behavior by Francis Fay (1982) examined feeding behavior and showed that walruses have a powerful ability to generate oral suction by using their tongue as a piston against the deeply vaulted palate. The poor defenseless clam, after being unearthed (typically by water jetting, using the opposite of suction; walruses do not use their tusks for “digging”) is manipulated into place by the walrus’ fleshy lips, and the suction generated is sufficient enough to suck the soft tissues right off of the shell (other observations by Fay included a walrus feeding on a small phocid seal it had presumably killed, and was just sucking the flesh right off of the bone). Valenictus chulavistensis shares a vaulted palate and was just as well-adapted for suction feeding as Odobenus. Tooth loss in Valenictus is therefore analogous to tooth loss in suction-feeding beaked whales, and is a remarkably derived condition amongst pinnipeds. In fact, it’s also worth pointing out that Odobenus is evolving towards tooth loss: it’s already lost its medial incisors, the fourth premolar, and upper and lower molars.

The holotype mandible of Valenictus chulavistensis...

Another curious feature is the highly dense, pachyosteosclerotic nature of the postcranial bones. The bones have a reduced medullary zone (osteosclerosis) and inflated proportions and cortex relative to other pinnipeds (pachyostosis). Modern Odobenus bones are slightly denser than other pinnipeds, and the skull in particular is extraordinarily dense – but the postcrania of Valenictus are massive, dense, and very heavy (and not just because of fossilization). SDNHM visits can actually be sort of a pain if I’m interested in photographing Valenictus bones because they’re so damn heavy (but chicken scratch compared to baleen whales, I’ll add). Dense bones are thought to act as ballast, and the denser bones of Valenictus suggests it had a unique ecology with respect to other odobenines. Valenictus is known from the proto-gulf of California, the Temblor Sea (Deméré, 1994a,b actually reidentified the San Joaquin Fm. specimen as a specimen of V. chulavistensis), and the Pliocene San Diego embayment, and was a benthic feeder. Deméré (1994b) suggested that benthic feeding in relatively warm waters would have favored increased bone ballast. Barnes (2005) preliminarily reported additional pachyosteosclerotic Valenictus occurrences from southern Baja California, and suggested that the hypersaline environments in some of these embayments would have also fostered adaptations towards overcoming greater buoyancy. Finally, some new specimens from the Purisima Formation near Santa Cruz include a femur and a complete skull (collected recently by high school student and avid amateur paleontologist Forrest Sheperd), both identifiable as Valenictus – demonstrate that Valenictus also existed somewhat further north at the terminus of the northern connection of the Temblor Sea.

...And the holotype humerus of Valenictus chulavistensis, photographed at the SDNHM. Note the similarities with Valenictus imperialensis (above).

Deméré (1994a) also conducted the first phylogenetic analysis of walruses, as I’ve alluded to in earlier posts. This analysis confirmed the monophyly of the Odobeninae, and recovered Aivukus as the earliest diverging odobenine. Alachtherium was the next diverging odobenine, which in turn was sister to an Odobenus + Valenictus clade. Demere (1994a) importantly noted that amongst walruses, only Alachtheirum, Valenictus, and Odobenus possessed globular dentine, and named the tribe Odobenini to unite the long-tusked odobenines together. In a phylogenetic and morphological context, Deméré (1994a) argued that Valenictus is actually more derived than the extant walrus Odobenus rosmarus – and certainly, given details of the dentition and postcranial skeleton, he makes an excellent case. For whatever reason, Valenictus went extinct at the end of the Pliocene. If you want to read up more on that, I wrote a bit about Plio-Pleistocene marine mammal extinctions in my recent Geodiversitas monograph (Boessenecker, 2013).


One last Valenictus chulavistensis - a composite skeleton (all San Diego Formation material) on display at the SDNHM; the paratype skull can be seen upside-down behind the skeleton.

More records of the Odobeninae from Japan

Another important Pliocene odobenine was published in the 1995 special volume of The Island Arc by Hideo Horikawa, in which he named the small, primitive odobenine Protodobenus japonicus. Protodobenus lacks tusklike canines, and it is unclear if it possessed globular dentine. It did on the other hand possess an Odobenini-like deep, robust rostrum, and retained the primitive number of teeth. Protodobenus lacked extreme dental wear and also had a flattish palate, suggesting it was incapable of effective suction feeding and likely subsisted on fish. In this context, it’s unclear why it evolved such a deep rostrum; damage to the skull shows that the canines have elongate roots, and in the Odobenini, the inflated rostrum accommodates the enormous canine roots. More on this in the next post…

An important contribution towards the evolutionary history of odobenines was published by Kohno et al. (1995), who reported several tusks from the Pliocene of Japan. Following the definition of Deméré (1994), they identified a number of tusks with globular dentine from lower and upper Pliocene localities in Japan, which they identified to the tribe Odobenini. Most of these have an oval-shaped cross section, are more highly curved and tapering than Odobenus rosmarus, and some possess longitudinal fluting – and therefore compare well with tusks of the Alachtherium-Ontocetus-Prorosmarus-Trichecodon morphotype (see below). Critically, they identified a single tusk of Odobenus sp. from the upper Pliocene (see section on Odobenus for more on this). The importance here is that, based on tusks, a minimum of two species of Odobenini appear to have coexisted in Japan during the Pliocene.


The skull and mandible of Protodobenus japonicus - the first deep-snouted walrus, from the early Pliocene of Japan.


The first record of Alachtherium from outside the North Atlantic: a skull of Alachtherium sp. from the early Pliocene of Japan. From Kohno et al. (1998).

Alachtherium from Japan… and Africa?!

A couple of surprising occurrences of the Pliocene walrus Alachtherium were reported in the late 1990’s. In 1997, Denis Geraads (who specializes in African mammal paleontology) described a new species of Alachtherium from the upper Pliocene of Morocco, based on a fragmentary skull, a partial mandible, and a partial humerus. Although the species is founded upon material that is of dubious diagnostic value, it clearly represented Alachtherium and nonetheless demonstrates that walruses formerly inhabited the northwestern shoreline of Africa.

The following year, Kohno et al. (1998) described a fantastically preserved skull from the early Pliocene of Japan they identified as Alachtherium sp. This skull doesn’t quite have the rectangular outline in posterior view like Alachtherium cretsii, but has a similarly short, curved tusk with an oval cross section, a slightly longer rostrum, and a full complement of postcanine teeth (preserved only as alveoli, unfortunately). This discovery indicates that Alachtherium was present in both the Atlantic and Pacific during the Pliocene, and likely used the Arctic portal as a means for dispersal, which had recently opened up at 5 Ma (Marincovich, 2000; see below).

The referred humeri, tibia, and mandible of Alachtherium africanus, from Geraads (1997).

The Lee Creek walruses and a taxonomic solution for Alachtherium

Various researchers including Deméré (1994a), Post (2004), and Kohno and Ray (2008) have preferred a single species assignment for North Atlantic walrus remains. In fact, although using a weird taxonomy, Erdbrink and Van Bree (1999) also preferred a single species, “Odobenus antverpiensis”, which others (Deméré, 1994a; Kohno and Ray, 2008) cogently argued was clearly a junior synonym of Alachtherium cretsii.

In 2008, Naoki Kohno and Clayton Ray published their monograph of walrus remains from the Pliocene Yorktown Formation in the long-awaited Lee Creek IV volume (seriously, if you’re interested in marine mammals from the east coast, do yourself a favor and buy a copy through the Virginia Museum of Natural History). First, they laid out all the prior taxonomic arguments, which are summarized here in bullet format for convenience:

-Ontocetus emmonsi is a Pliocene walrus from the Yorktown Formation, originally named as a cetacean in 1859 by Joseph Leidy, based on a partial tusk
-Trichecodon huxleyi from the upper Pliocene Red Crag (UK), named in 1865 by Lankester, was based on an isolated tusk similar to Ontocetus emmonsi
-Alachtherium cretsii named by Du Bus (1867) from lower Pliocene Scaldisian sands of Belgium based on well-preserved mandible
-Trichechodon koninckii was named from a fragmentary tusks by Van Beneden (1871), also from the lower Pliocene Scaldisian sands of Belgium, later identified by many later authors as non-diagnostic and a nomen nudum or nomen dubium
-Prorosmarus alleni named from incomplete mandible from Yorktown Fm. by Berry and Gregory (1906)
-Trichechus antverpiensis erected by Rutten (1907) for partial skull originally referred to A. cretsii by Van Beneden (1877)
-Alachtherium antwerpiensis named by Hasse (1910) for other cranial material from Pliocene of Belgium
-Trichechus antverpiensis recombined as Odobenus antverpiensis by Van der Feen (1968), followed by Erdbrink and Van Bree (1986, 1990, 1999)
-T./O. antverpiensis and Alachtherium antwerpiensis synonymized with Alachtherium cretsii by Deméré (1994a), followed by Post (2004)
-Alachtherium africanus named by Geraads (1997) from fragmentary cranial elements from the Pliocene of Morocco

The holotype tusk fragment of Ontocetus emmonsi - and the "parent" specimen of virtually all North Atlantic Pliocene walruses (=Alachtherium cretsii, Alachtherium antwerpiensis, Trichecodon huxleyi, Trichecodon/Odobenus antverpiensis, Prorosmarus alleni), photographed at the USNM.

Kohno and Ray (2008) further considered Prorosmarus alleni to be relatively similar to the type mandible of A. cretsii, and figured a new mandible that is intermediate between the two, indicating that the perceived absence of an upturned symphysis is probably an ontogenetic feature. They also pointed out that the holotype tusk of Ontocetus emmonsi is a walrus, and that it and the holotype tusk of Trichecodon huxleyi both share an oval, transversely compressed cross section, longitudinal fluting, greater curvature, and are more tapered in contrast to the more elongate, straighter, smoother tusks of Odobenus rosmarus, which also have more of a circular cross section. Most significantly, they identify that all tusks found in sediments of Pliocene age yielding remains of Alachtherium/Trichecodon/Prorosmarus all conform to this morphology. They showed some bivariate plots of Atlantic Pliocene walrus tusks, and showed conclusively that all of these tusks conform to similar proportions, and cluster together to the exclusion of Odobenus rosmarus (Kohno and Ray 2008: fig. 27). In light of this information, they synonymized all Pliocene Atlantic walruses with Ontocetus emmonsi.

An adorable Ontocetus emmonsi juvenile maxilla (complete with mini-tusk!) from the Pliocene Yorktown Formation, described and figured by Kohno and Ray (2008), photographed at the USNM. Seriously, that is a cute fossil.

It’s admittedly a controversial decision, and some other pinniped workers have expressed the notion that Alachtherium cretsii is a better name as it is founded upon a complete mandible that inherently preserves more morphological information and is thus certainly more diagnostic than a tusk. I’m on the fence; on one hand, if the tusk morphology proves in the long run to really be that distinctive (and so far, it seems to), then Ontocetus emmonsi works; on the other hand, an isolated fragmentary tusk may not be diagnostic and in cetacean paleontology most species named off of isolated teeth were shit-canned a long time ago. My mind isn’t completely made up, and I think both camps in favor of Ontocetus emmonsi or Alachtherium cretsii have decent arguments.

The fossil record and biogeography of Odobenus

Pleistocene fossils of the modern walrus Odobenus have been widely reported from coastlines and the sea floor from both sides of the North Pacific (e.g. California, British Columbia, Japan) and North Atlantic (Maritime provinces of Canada, eastern USA – New Jersey to Georgia, and the UK and Netherlands) as well as the Arctic (Canada). Historically, many of the Pleistocene tusks, crania, mandibles, and other remains were assigned to Trichecodon huxleyi by earlier workers (e.g. Rutten, 1907); Demere (1994a) recombined it as Odobenus huxleyi, remarking that it was possibly diagnosable based on possessing a thin cementum layer in the holotype tusk; however, Kohno and Ray (2008) indicated that the tusk is identical to Ontocetus emmonsi (regardless, most material referred to T. huxleyi does appear to represent Odobenus rather than Alachtherium/Ontocetus). Fossils assignable to the extant genus Odobenus are widely reported from Pleistocene deposits in the Northern Hemisphere; in North America, fossils of Odobenus rosmarus have been dredged from as far south as the San Francisco Bay in California (Harington, 1984) and Georgia (Sanders 2002). These southerly records likely reflect southward latitudinal expansion of the natural range of Odobenus rosmarus during cold glacial periods. Furthermore, trace fossil evidence has recently been identified from the Olympic Peninsula in Washington, USA, indicating the presence of suction/jet-feeding walruses during the late Pleistocene (Gingras et al. 2007).

A Pleistocene skull figured and referred to Odobenus huxleyi (=Odobenus rosmarus), and the braincase of Alachtherium cretsii/Ontocetus emmonsi named as the new species Trichecodon antverpiensis by Rutten (1907).

That’s all neat, but not very surprising: modern Odobenus rosmarus is predominantly Arctic in distribution but occurs at the fringes of the North Pacific and extensively in the Northernmost Atlantic, and we know it was pretty damn cold during parts of the Pleistocene, facilitating southward migration during cold periods. But how old is the Odobenus lineage? And where the hell did it come from?

Cross-sections of Pleistocene and modern Odobenus tusks, with the illustration of the broken cross-section of the type specimen of Hemicaulodon effodiens, a junior synonym of Odobenus rosmarus. From Ray (1975). Note the distinctive core of globular dentine. This is a typical record of isolated Odobenus tusks from Pleistocene sediments.

At the time of writing the pinniped bible, Repenning and Tedford (1977) were sort of at a loss for the more recent evolutionary history of the modern walrus. Most of the remains were Pleistocene in age, and more or less confined to the above described regions: mostly in the North Atlantic and fringes of the Arctic. Based upon their discovery of the tuskless odobenine Aivukus cedrosensis from Baja California (which they also presumed was phylogenetically close to Prorosmarus alleni based on mandibular similarities; of course we now know that Aivukus and “Prorosmarus” had widely disparate skull morphology), they hypothesized that the ancestor of all tusked walruses (=Odobenini of modern usage) dispersed to the North Atlantic prior to the closure of the Panamanian isthmus. Following this, the extant Pacific walrus (Odobenus rosmarus divergens) reinvaded the northernmost Pacific late in the Pleistocene. This hypothesis began to unravel upon the discovery of tusked walruses like Valenictus chulavistensis from California (confirming that the genus Valenictus was assignable to the Odobenini rather than Dusignathinae), Protodobenus from Japan, and another toothless walrus from the Purisima Formation of California, all indicating that tusked walruses persisted in the North Pacific long after the disappearance of Aivukus (Kohno et al., 1995). A new extinct species of Odobenus was named by Tomida (1989) which he named Odobenus mandanoensis, from the middle Pleistocene of Japan. It proportionally differs from extant Odobenus and appears to have been slightly larger; although fragmentary, it appears to genuinely reflect a separate species (Deméré, 1994a). 

An isolated tusk fragment from the Purisima Formation (this specimen may be seen on display at the Santa Cruz Museum of Natural History), with globular dentine; the distinctive dental tissue identifies this tusk to the Odobenini. This and another specimen are from approximately the Miocene-Pliocene boundary, and therefore constitute some of the oldest records of the Odobenini.

One of the most fascinating advances was the discovery tusks and crania assignable to Odobenus from the upper Pliocene of Japan. Aside from the aforementioned tusk described by Kohno et al. (1995), a nearly complete skull of Odobenus sp. with tusk dredged from the Sea of Okhotsk was reported by Miyazaki et al. (1992), who found that it was associated with late Pliocene microfossils. These finds indicate that while Valenictus was hanging out in warm waters along the California and Baja California margin, and while Ontocetus/Alachtherium was proliferating across virtually the entire North Atlantic, the modern walrus had already evolved in the western North Pacific.

The curious referred mandible of Odobenus "koninckii", identified here as Odobenus sp., from the Pliocene Scaldisian Sands of Belgium, from Van Beneden (1877).

A single commonly overlooked fossil from the early Pliocene of Belgium indicates that perhaps neither of these two possibilities are likely. Although the name Trichechodon koninckii is defunct and useless, a single mandible apparently from the Pliocene Scaldisian sands of Belgium referred to T. koninckii and figured by Van Beneden (1877) bears a non-upturned symphyseal region and a fused symphysis, two features unique (amongst the Odobenini) to Odobenus. Deméré (1994a) pointed out that this specimen reflects a primitive Odobenus that retains a fourth lower premolar, a canine that is slightly larger than the premolars, as well as a sinuous outline of the mandible in dorsal view; the specimen is of apparent Pliocene age, and appears to indicate that Odobenus can be tracked to the early Pliocene in the North Atlantic, a bit older than the late Pliocene of the western North Pacific. What could this suggest? Perhaps it suggests a third option, that ancestral Odobenus had, like today, a circum Arctic distribution that extended as far south as Japan and Belgium during the Pliocene, facilitated by the lack of extensive ice sheets. Such a distribution may have pre-adapted Odobenus for Pleistocene glaciation. It’s possible, but other hypotheses are equally likely, and we need more walrus fossils with better dates to get a more complete picture.

Whatever happened, we know the following take-home points: 1) until the Pleistocene, tusked walruses (Odobenini) enjoyed a much wider variety of habitats and happily existed as far south as Baja California, Florida, and Morocco, and Pliocene fossils of the genus Odobenus are found at latitudes that would have been temperate during the Pliocene; 2) Sometime during the past 2 million years, a lineage within the genus Odobenus transformed from a temperate species (as was typical of Pliocene Odobenini) into the Arctic glacially-adapted specialist we know today.

The diversity of tusked walruses (Odobeninae); note the much smaller size of Aivukus and Protodobenus, and the gigantic size of Ontocetus emmonsi/Alachtherium cretsii.

L.G. Barnes. 2005. Dense boned late Miocene and Pliocene fossil walruses of the Imperial Desert and Baja California: possible buoyancy-control mechanisms for feeding on benthic marine invertebrates in the proto-Gulf of California. Abstracts of Proceedings, 2005 Desert Symposium.

E. W. Berry and W. K. Gregory. 1906. Prorosmarus alleni, a new genus and species of walrus from the upper Miocene of Yorktown, Virginia. American Journal of Science 21:444-450

R.W. Boessenecker. 2013. A new marine vertebrate assemblage from the late Neogene Purisima Formation in Central California, part II: pinnipeds and cetaceans. Geodiversitas 35:815-940.

T. A. Demere. 1994a. The Family Odobenidae: A phylogenetic analysis of fossil and living taxa. Proceedings of the San Diego Society of Natural History 29:99-123

T. A. Demere. 1994b. Two new species of fossil walruses (Pinnipedia: Odobenidae) from the upper Pliocene San Diego Formation. Proceedings of the San Diego Society of Natural History 29:77-98

B. Du Bus. 1867. Sure quelques mammiferes du Crag d'Anvers. Bulletin de l'Academie royale des sciences, des lettres et des beaux-arts de Belgique 24:562-577.

D. P. B. Erdbrink and P. J. H. Van Bree. 1986. Fossil Odobenidae in some Dutch collections (Mammalia, Carnivora). Beaufortia 36(2):13-33

D. P. B. Erdbrink and P. J. H. Van Bree. 1990. Further observations on fossil and subfossil odobenid material (Mammalia, Carnivora) from the North Sea. Beaufortia 40(5):85-101.

D. P. B. Erdbrink and P. J. H. Van Bree. 1986. Fossil cranial walrus material from the North Sea and estuary of the Schelde (Mammalia, Carnivora). Beaufortia 49(1):1-9
D. Geraads. 1997. Carnivores du Pliocene terminal de Ahl al Oughlam (Casablanca, Maroc). Géobios 30(1):127-164

M. K. Gingras, I.A. Armitage, S.G. Pemberton, and H.E. Clifton. 2007. Pleistocene walrus herds in the Olympic Peninsula area: trace-fossil evidence of predation by hydraulic jetting. Palaios 22:539-545.

C. R. Harington. 1984. Quaternary marine and land mammals and their paleoenvironmental implications - examples from Northern North America. Special publication of the Carnegie Museum of Natural History 8:511-525.

G. Hasse. 1909. Les Morses Pliocene poederlien a Anvers. Bulletin de la Societe Belge de Geologie de Paleontologie et D'Hydrologie (Bruxelles) 23:293-322

H. Horikawa. 1995. A primitive odobenine walrus of Early Pliocene age from Japan. The Island Arc 3:309-328

R. Kellogg. 1922. Pinnipeds from Miocene and Pleistocene deposits of California. University of California Publications in Geological Sciences 13(4):23-132

N. Kohno and C. E. Ray. 2008. Pliocene walruses from the Yorktown Formation of Virginia and North Carolina, and a systematic revision of the North Atlantic Pliocene walruses. Virginia Museum of Natural History Special Publication 14:39-80

N. Kohno, K. Narita, and H. Koike. 1998. An early Pliocene odobenid (Mammalia: Carnivora) from the Joshita Formation, Nagano Prefecture, central Japan. Research Reports of the Shinshushinmachi Fossil Museum 1:1-7.

N. Kohno, Y. Tomida, Y. Hasegawa, and H. Furusawa. 1995. Pliocene tusked odobenids (Mammalia: Carnivora) in the western North Pacific, and their paleobiogeography. Bulletin of the National Science Museum 21:111-131.
E. R. Lankester. 1865. On the sources of the mammalian fossils of the Red Crag, and the discovery of a new mammal in that deposit, allied to the walrus. Quarterly Journal of the Geological Society of London 21:221-232

J. Leidy. 1859. [Remarks on Dromatherium sylvestre and Ontocetus emmonsi]. Proceedings of the Academy of Natural Sciences of Philadelphia 1859:162

L. Marincovich. 2000. Central American paleogeography controlled Pliocene Arctic Ocean molluscan migrations. Geology 28:551-554.

E. D. Mitchell. 1961. A new walrus from the imperial Pliocene of Southern California: with notes on odobenid and otariid humeri. Los Angeles County Museum Contributions in Science 44:1-28

S. Miyazaki, M. Kimura, and H. Ishiguri. 1992. On a Pliocene walrus (Odobenus sp.) discovered in the northern Pacific Ocean. Journal of the Geological Society of Japan 98:723-740.
K. Post. 2004. What's in a name: Alachtherium cretsii, de Pliocene van de Nordzee. Grundboor & Hammer 58:70-74.

C. E. Ray. 1975. The relationshiops of Hemicaulodon effodiens Cope 1869 (Mammalia; Odobenidae). Proceedings of the Biological Society of Washington 88(26):282-304

C. A. Repenning and R. H. Tedford. 1977. Otarioid seals of the Neogene. Geological Survey Professional Paper 992:1-93

L. Rutten. 1907. On fossil trichechids from Zeeland and Belgium. Proceedings of the Royal Netherlands Academy of Arts and Sciences 10(1):2-14 

A.E. Sanders. 2002. Additions to the Pleistocene mammal faunas of South Carolina, North Carolina, and Georgia. Transactions of the American Philosophical Society 92:1-152.

P.J. Van Beneden. 1871. Les phoques de la mer Scaldisienne. Bulletin du Academie Royale de Belgique 32:5-19.

P.J. Van Beneden. 1877. Description des ossements fossiles des environs d'Anvers. Annales du Musee Royal d'Histoire Naturelle de Belgique, Tome 1. Premiere Partie, Pinnipedes ou Amphitheriens: 1-88.

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

Wednesday, September 10, 2014

The evolutionary history of walruses, part 3: double tusked walruses - the dusignathines

The fossil that started it all - the weird holotype mandible of Dusignathus santacruzensis.

In the 1920’s, a rather strange pinniped fossil was collected from a cliff near the Santa Cruz wharf. In 1927, the fossil – including a fragmentary skull and a well preserved pair of mandibles – was described by preeminent marine mammal paleontologist Remington Kellogg (the same guy who described Allodesmus, Neotherium, and a ton of fossil cetaceans) as Dusignathus santacruzensis. The name Dusignathus means “jaw of the setting sun”, referring to the occurrence of this fossil on the west coast. Dusignathus santacruzensis was originally thought to represent a strange sea lion; the fragmentary skull shows that Dusignathus had a low sagittal crest, procumbent but gracile upper canines (later suggesting to other workers that the type specimen is a female), single rooted teeth with simple bulbous crowns, and a robust mandible with a procumbent canine, short lower toothrow, and an enlarged flange for the insertion of the digastric muscle. Furthermore, the mandible of Dusignathus santacruzensis is unique amongst pinnipeds in having the left and right canines nearly contacting medially, and having the left and right mandibles diverge from the chin at a relatively wide (60˚) angle. Because of its incompleteness, Dusignathus continued to be enigmatic for another 50 years until Repenning and Tedford (1977) published the pinniped bible. Thanks to the realization that other fossil pinnipeds like Neotherium and Imagotaria were early sea lion-like walruses, Repenning and Tedford (1977) identified Dusignathus as another walrus, albeit a stranger one. However, in 1962, Ed Mitchell described a partial forelimb (from the type horizon of Dusignathus) with very robust ulna and radius which he identified as an unknown walrus; this specimen was later identified by Repenning and Tedford (1977) as possibly representing Dusignathus santacruzensis. They also referred some isolated Imagotaria-like bones from the Purisima Formation to Dusignathus, and subsequent discoveries of associated postcrania from the Purisima Formation and the St. George Formation near Crescent City have confirmed the referral of these postcranial elements (more on that in the future). Repenning and Tedford (1977) erected the subfamily Dusignathinae, in which they included Dusignathus, Imagotaria, and Valenictus.

The first cranial restoration of Dusignathus santacruzensis, from Mitchell (1975). Compare this with the updated reconstruction provided below, in which I based missing skull parts and proportions on more complete material of Dusignathus seftoni.

Mitchell's (1962) walrus forelimb from the Purisima Formation at Santa Cruz, which now appears to actually represent Dusignathus santacruzensis (based on unpublished studies by yours truly).

In 1980, a gigantic walrus skeleton was discovered in exposures of the Capistrano Formation in the city of San Clemente in Orange County, California. This large walrus was named Gomphotaria pugnax in 1991 by Larry Barnes and Rodney Raschke – it roughly means “pugnacious bolt-toothed sea lion”. The type specimen of Gomphotaria includes a nearly complete, 40 cm long skull (2/3 the size of Pontolis magnus) with a large, California sea lion-like sagittal crest that would’ve given the animal a domed forehead in life, a pair of enormous but short and procumbent (e.g. nearly horizontal) tusks, single rooted and highly abraded cheek teeth, and a robust lower jaw with enlarged, procumbent lower tusks. That’s right – Gomphotaria had upper and lower tusks! In addition to the upper and lower tusks, the upper third incisor was also tusk-like and procumbent, giving Gomphotaria a third pair of smaller tusk-like teeth. Further unlike the modern walrus, Gomphotaria had a mostly flat palate – which indicates that Gomphotaria was unlikely to have been an effective suction feeder like modern Odobenus. Modern Odobenus has a highly vaulted palate which allows the tongue to act as a piston and generate tremendous oral suction. Using this piston, the modern walrus will position a clam shell in its large fleshy lips and suck the meat right out of the shell, without the shell ever entering the oral cavity. Gomphotaria, on the other hand, was probably not capable of this extreme style of suction feeding (although it must be stated that most generalized pinnipeds without arched palates, like seals and sea lions, are capable of limited suction feeding when capturing fish). Because of the rather extreme wear on the postcanine teeth and tusks of Gomphotaria, Barnes and Raschke (1991) hypothesized that it was similarly a molluskivore, but rather than sucking the meat right out of poor little clam shells – Gomphotaria simply crushed the clams to bits, shell and all. The forelimb of Gomphotaria is relatively short and robust, similar to Imagotaria, and a partial forelimb from the Purisima Formation (originally described by Mitchell, 1962) tentatively referred to Dusignathus santacruzensis by Repenning and Tedford (1977). Barnes and Raschke (1991) pointed out that several problematic walrus specimens – including Mitchell’s (1962) Purisima Fm. forelimb, and the poorly known odobenine Pliopedia pacifica (more in the next post!) ought to be reevaluated in the context of (then) newly discovered walruses like Gomphotaria.

The behemoth of a skull of the holotype specimen of Gomphotaria pugnax, photographed at LACM. The tusk is at least as big around as a beer bottle.

The holotype right forelimb of Gomphotaria, which is pathologically fused at the elbow joint.

Morgan photographing the monstrous holotype mandible at LACM in January 2012, as part of our Pelagiarctos study.

Some years after Gomphotaria was dug up in Orange County, a similar find was made further south in San Diego. A handful of specimens including an immature skull, a gigantic mandible, and a couple of humeri were discovered during a similar paleontological mitigation project. These fossils were given the name Dusignathus seftoni in a paper published by Dr. Tom Deméré in 1994, and were collected from the Pliocene San Diego Formation (4.5-2 Ma). The skull and mandible share several similarities with the somewhat older (late Miocene) Dusignathus santacruzensis, including somewhat procumbent upper and lower canines, a low sagittal crest, and a mandible with a sinuous ventral margin, medially appressed lower canines, and widely divergent mandibles. The skull of Dusignathus seftoni preserves much of the skull missing in the species from Santa Cruz, such as a relatively short and wide rostrum, a V-shaped frontal/maxillary suture, and a skull that is otherwise reminiscent of Gomphotaria and Imagotaria in general proportions. Unfortunately, the teeth of the type skull and referred mandible are poorly preserved, precluding Deméré (1994) from making any specific interpretations about feeding adaptations of Dusignathus seftoni, although general cranial similarities with Imagotaria suggested it may have been piscivorous (Demere, 1994). However, subsequently discovered skulls and mandibles with complete canines demonstrate that Dusignathus seftoni had upper and lower tusks just like Gomphotaria – albeit somewhat less procumbent and shorter. Furthermore, these teeth are highly worn, also similar with Gomphotaria. In my opinion, the dentition and toothwear of these two dusignathines are similar enough to confer a similar shared feeding ecology to both species. Extreme toothwear in Dusignathus seftoni therefore may also suggest benthic molluskivory like Gomphotaria. Additional specimens of Dusignathus seftoni now show that skull size of adult males ranged up to 40 cm (just slightly smaller than Gomphotaria and 2/3 the size of Pontolis) and that the species was (unsurprisingly) sexually dimorphic. The humerus of Dusignathus seftoni is relatively similar to Imagotaria but slightly more gracile than the isolated humerus referred to Dusignathus santacruzensis by Repenning and Tedford (1977).

The beautifully preserved holotype subadult male skull of Dusignathus seftoni in anterior view, photographed at the San Diego Natural History Museum in January 2012.

An even more beautifully preserved referred skull of Dusignathus seftoni, which is currently on display in the fossil gallery at the San Diego Natural History Museum.

As discussed in the previous post, Deméré (1994) considered Pontolis magnus to represent a dusignathine walrus (see part 2 for a thorough discussion of Pontolis). This issue brings up the question of dusignathine phylogeny. Prior to the discovery of Dusignathus seftoni and Gomphotaria, Dusignathus was considered by Kellogg (1927) to preserve sea lion and walrus-like features; later, the Dusignathinae was considered by Mitchell (1968, 1975) to only include Dusignathus, and although previously recognizing walrus-like features, interpreted Dusignathus to be more closely related to desmatophocid seals. Repenning and Tedford (1977) later included within the subfamily Imagotaria downsi (based on similarly sized upper and lower canines) as well as the poorly known walruses Valenictus imperialensis and Pliopedia (based on shared robust proportions of forelimb bones). As it turns out, Valenictus and Pliopedia were later recognized to represent tusked walruses (Odobenines; Deméré, 1994).

The early phylogenetic hypothesis of otarioid phylogeny from Mitchell (1975).

One of the first computer-aided cladistic analyses of walrus phylogeny - from Deméré (1994).

Early studies presented phylogenetic hypotheses that were not rigorously or systematically conducted. Prior to the advent of computer-aided cladistics, the majority of phylogenetic hypotheses were formulated in a subjective manner and cladograms were hand-drawn – often with all of the intellectual baggage from the author’s own preconceived notions of relationships intact and unchecked. Fortunately, those dark days are now (mostly) over, and rather than risk making dubiously informed claims of phylogeny based on subjective interpretations of a handful of anatomical characters (e.g. using primarily earbone characters over features of the rostrum and palate for whales and dolphins). Now that we have computers, we can statistically evaluate phylogenetic hypotheses using hundreds (or more) of anatomical characters (my own dissertation on baleen whales features a large dataset with over 300 morphological characters) scored for many species. The first computer-aided analyses of walrus phylogeny were published by Tom Deméré (1994) and Naoki Kohno (1994). These early studies used relatively small datasets (53 and 25, respectively); a later study by Deméré and Berta (2001) on Proneotherium featured only 24 characters. The cladistic analysis of Deméré (1994) confirmed the monophyly of the dusignathine subfamily, and also – using newly recognized USNM specimens that Doug Emlong himself collected in the 1960’s and 1970’s from the Empire Formation – recovered the giant walrus Pontolis magnus as another dusignathine. Kohno (2006) pointed out that the limited dataset used by Deméré and Berta (2001) did not necessarily result in a more accurate phylogeny, and assembled a new cladistic matrix for walrus phylogeny using 60 characters. When Morgan and I started work on Pelagiarctos back in 2012, we used Naoki Kohno’s 2006 matrix as a starting point, and added a ton of mandibular characters so that we could place Pelagiarctos – still only known from a mandible – into a phylogenetic context (our matrix had 90 characters). Interestingly, my labmate Yoshi Tanaka (and his master’s adviser, Dr. Kohno) is working on a new fossil “imagotariine” from the same locality as Pseudotaria muramotoi – and has used our modified matrix and added a new character and modified several others. So, the walrus matrix has been batted around full circle now (or, soon will be).

An improved, larger cladistic analysis by Kohno (2006).

So what happened when additional characters were added? Dusignathines remained monophyletic but Pontolis dropped out and appeared as a sister species to Imagotaria in Kohno’s (2006) study; this is perhaps unsurprising, since Pontolis has an Imagotaria-like basicranium and fossils of the two have been frequently mistaken for one another since the 1940’s (e.g. Lyon, 1941; Mitchell, 1968). Our own study found a compromise of sorts between Pontolis-as-a-dusignathine and Pontolis-as-sister-to-Imagotaria: Pontolis in our study was recovered between the Pelagiarctos-Imagotaria clade and the Dusignathinae (Boessenecker and Churchill, 2013). Furthermore, support for the Dusignathinae decreased a bit. Some of the features which Deméré (1994) noted as dusignathine synapomorphies present in Pontolis include a V-shaped frontomaxillary suture – however, this condition characterizes some fossil odobenines as well; a tusk-like lower canine may genuinely support such a relationship, but a sinuous ventral margin of the mandible (as in Dusignathus spp.) is present in some specimens of Imagotaria.

An even larger study, from our own paper on Pelagiarctos (Boessenecker and Churchill 2013).

Skull diversity amongst the dusignathines, with Pontolis included as it may yet represent a dusignathine, and Neotherium for comparison. These bigger, freakier walruses make Neotherium look generalized, small, and sort of adorable in comparison.

Lastly, to follow up this rather dry phylogenetic talk – what is the evolutionary significance of the dusignathines? Why are they my favorite of all walruses? To sum that second question up, in my mind they are a relatively strange and highly derived group with a strange dentition (lower tusks, and procumbent upper tusks!), in some cases large sagittal crests paralleling some sea lions, and strange postcranial features such as unusually robust, shortened forelimbs. They are the least studied and most neglected group of walruses by far, and also geographically limited: the entirety of the dusignathine fossil record is from California, Oregon, and Baja California. These pinnipeds evolved during a time of relative faunal isolation: warm equatorial waters barred dispersal to the southern hemisphere or dispersal through the Panama Seaway. The Bering strait had not yet opened until the latest Pliocene, barring dispersal to the Atlantic – and for some strange reason, dusignathines never made it to the western North Pacific (as evidenced by their conspicuous absence in Japanese marine mammal fossil assemblages), unlike virtually every other North Pacific pinniped group. This is not to say that I am any less passionate about odobenines, which certainly had some serious weirdos in their own right during the Pliocene – but these dusignathines evolved somewhat in parallel but in a different direction, and represent an extinct group of highly derived walruses most people are unfamiliar with but dominated over 1000 miles of the western coastline of North America until only 2 million years ago. In a way, dusignathines can be viewed as another mini radiation of specialized walruses that evolved in parallel with the tusked odobenines, snuffed out during the end-Pliocene marine mammal extinction for some unknown cause. What questions and research has yet to be done? For starters, at least two new genera of dusignathines (...or possible imagotariines) have yet to be described, and referred material of Pontolis magnus is desperately needing description (this may be the next joint project Morgan and I tackle). The postcranial skeleton of Gomphotaria begs a complete description – and enough new material of Dusignathus santacruzensis now exists to make sense of a wide variety of seemingly unrelated postcranial bits scattered between various collections and publications. Dusignathus seftoni is now the most completely known dusignathine, now known from three or four skulls, almost as many mandibles, a couple of partial skeletons, and a host of postcrania. We still don’t really know what these guys were feeding upon (at least, no more than the idle speculation provided above), and toothwear, morphometric studies, and isotopic analysis could resolve that question. Also, why were their foreflippers so short and stubby? As you can see, there’s a lot left to do.

Next up: the fossil record and evolutionary biogeography of true “tusked” walruses (Odobeninae).


Barnes, L.G. and R. E. Raschke. 1991. Gomphotaria pugnax, a new genus and species of late Miocene dusignathine otariid pinniped (Mammalia: Carnivora) from California. Contributions in Science 426:1-16

Boessenecker, R.W., and M. Churchill. 2013. A Reevaluation of the Morphology, Paleoecology, and Phylogenetic Relationships of the Enigmatic Walrus Pelagiarctos. PLoS One 8(1):e5411.

Deméré, T.A. 1994. The Family Odobenidae: A phylogenetic analysis of fossil and living taxa. Proceedings of the San Diego Society of Natural History 29:99-123.

Deméré, T.A. 1994. Two new species of fossil walruses (Pinnipedia: Odobenidae) from the Upper Pliocene San Diego Formation, California. Proceedings of the San Diego Society of Natural History 29:77-98.

Kellogg, R. 1927. Fossil Pinnipeds from California. Contributions to Palaeontology from the Carnegie Institution of Washington 27-37.

Kohno, N. 1994. A new miocene pinniped in the genus Prototaria (Carnivora: Odobenidae) from the Moniwa Formation, Miyagi, Japan. Journal of Vertebrate Paleontology 14(3):414-426.

Kohno, N. 2006. A new Miocene odobenid (Mammalia: Carnivora) from Hokkaido, Japan, and its implications for odobenid phylogeny. Journal of Vertebrate Paleontology 26(2):411-421.

Lyon, G. M. 1941. A Miocene sea lion from Lomita, California. University of California Publications in Zoology 47:23-41.

Mitchell, E.D. 1962. A walrus and a sea lion from the Pliocene Purisima Formation at Santa Cruz, California: with remarks on the type locality and geologic age of the sea lion Dusignathus santacruzensis Kellogg. Los Angeles County Museum Contributions in Science 56:1-24.

Mitchell, E.D. 1968. The Mio-Pliocene pinniped Imagotaria. Journal of the Fisheries Research Board of Canada 25(9):1843-1900.

Mitchell, E.D. 1975. Parallelism and convergence in the evolution of the Otariidae and Phocidae. Rapports et Proces-verbaux des Reunions 169:12-26.

Repenning, C.A. and R. H. Tedford. 1977. Otarioid seals of the Neogene. Geological Survey Professional Paper 992:1-93.