Sunday, May 11, 2014

The best known fossil pinniped, part 2: How many species (...and genera) are there of Allodesmus, anyway?

Last week I summarized the history of research on Allodesmus (and Atopotarus). How many species of Allodesmus exist? And how many genera of desmatophocids are there? On one extreme, one could take the view that there are only two genera – Allodesmus and Desmatophoca – and that Atopotarus, Megagomphos, and Brachyallodesmus are all synonyms of Allodesmus. On the other hand, a more diverse view of the Desmatophocidae would include as many as 10 species in the aforementioned 5 genera. We’ll start with genera – and then look at the species level.

How many genera of desmatophocids are there?

I won’t bother with Allodesmus and Desmatophoca, since these are both clearly “real” and the oldest available names for generic concepts.

AtopotarusAtopotarus is largely diagnosed by the retention of primitive characteristics in comparison to Allodesmus spp. For example, Atopotarus lacks the prenarial shelf of more derived Allodesmus such as Allodesmus kernensis/kelloggi/gracilis, has a less inflated braincase, and retains double-rooted postcanine teeth (premolars and molars). In contrast, the teeth of Allodesmus are single rooted. All these features were listed by Barnes and Hirota (1995) as distinguishing A. courseni from Allodesmus at the generic level. Mitchell (1966) was the first to suggest that Atopotarus courseni was possibly a species of Allodesmus – prior to his description of Allodesmus kelloggi, insufficient cranial and postcranial material was known to really evaluate the generic validity of Atopotarus. This skepticism was echoed by Barnes (1972), Kohno (1996) and Deméré and Berta (2002) – although its generic distinctiveness was endorsed by Barnes and Hirota (1995). I’ve always been on the fence about Atopotarus – it’s barely distinguishable from Allodesmus as it is slightly more plesiomorphic and its morphology – with the exception of the possibly lost second lower molar – could reflect the ancestral morphology of the entire genus. Amongst extant pinnipeds, including phocids – closely related clusters of species (e.g. Zalophus, Arctocephalus, Monachus), with the exception of the Phoca-Pusa-Halichoerus complex – tooth rooting doesn’t vary this much between species, potentially suggesting that Atopotarus might be real. More fossil material of Atopotarus is necessary to further evaluate it, since the holotype is a somewhat flattened articulated specimen exposed in relief in a large slab.

The holotype specimen of Atopotarus courseni, from Downs (1956).

Brachyallodesmus: this genus was erected by Barnes and Hirota (1995) for Allodesmus packardi, the skull from Portola Valley in southern San Mateo County. Features used by these authors to distinguish it from other Allodesmus include canines with an oval cross section (rather than round), an inflated braincase lacking strong nuchal crests, enormous orbits with an extremely narrow intertemporal region, and some primitively retained characteristics such as bilobed premolar roots, a furrow on the side of the braincase marking the position of the pseudosylvian sulcus on the brain, and a less flattened tympanic bulla. This species is arguably much less primitive than Atopotarus courseni, and while I’m on the fence about the generic distinction of Atopotarus, the claim that Allodesmus packardi is generically distinct is spurious in my opinion, and in my preference for taxonomic conservatism (aka lumping), I’ve always held the consideration that this is a species of Allodesmus.
Megagomphos: This genus of desmatophocid was also erected by Barnes and Hirota (1995) for a fragmentary rostrum previously known as Allodesmus sinanoensis (originally erroneously placed in Eumetopias). This specimen is the same size as the largest specimens of Allodesmus from California. Characteristics used by Barnes and Hirota (1995) to elevate it to a separate genus include the lack of a prenarial shelf (like Atopotarus) and the presence of larger canines than other Allodesmus. Kohno (1996), however, identified several features of this specimen suggesting that it was immature, such as unworn teeth that have large pulp cavities and are not completely erupted. In contrast, Barnes and Hirota (1995) thought that this specimen was an adult because of the shape of the tooth roots and purported fusion of the premaxilla-maxilla suture, but didn’t discuss the ontogenetic stage of this specimen any further. In contrast, Kohno (1996) interpreted the specimen to have a large prenarial shelf, eliminating one purported feature excluding it from Allodesmus; similarly, it exhibits single rooted postcanine teeth and no features of the dentition preclude it from being Allodesmus. Furthermore, one Allodesmus synapomorphy – the prenarial shelf – unequivocally links this specimen with Allodesmus to the exclusion of all other pinnipeds. Although Barnes and Hirota (1995) considered this specimen to lack a shelf, it is present but since the specimen is bilaterally compressed and crushed, which has made the shelf appear less apparent than in life. While this specimen has a few minor differences with other described species of Allodesmus such as procumbent, large canines – no features suggest that establishment of Megagomphos was necessary.

In conclusion, evidence supporting the generic distinction of Megagomphos and Brachyallodesmus is limited to non-existent – but Atopotarus may be “real”, although more cranial material is necessary to evaluate Atopotarus.

How many species of Allodesmus?

Allodesmus from Japan: To date, four species of Allodesmus have been described from Japan. These include the aforementioned Allodesmus sinanoensis, the well-preserved Allodesmus sadoensis, and the more fragmentary Allodesmus naorai (the “Mito seal”) and Allodesmus megallos. Allodesmus sadoensis is known from a nearly complete skull and partial mandibles, and is readily distinguished from all other Allodesmus by having anteriorly crowded and procumbent (forward pointing) postcanine teeth, more vertically oriented mandibular symphysis, and lower postcanine teeth without gaps between them. Allodesmus naorai is similar to Allodesmus packardi, and differs only in a few minor features such as having a slightly wider interorbital region and supraorbital processes positioned further posterior – potentially diagnostic features. The fourth species, Allodesmus megallos, was also named by Barnes and Hirota (1995) and diagnosed based on its enormous size, procumbent tusk-like canines and proportionally larger incisors. However, Kohno (1996) independently referred this specimen to Allodesmus sinanoensis, based upon the shared large size, large procumbent canines, and enlarged incisors, and considered this specimen to be an adult – it is substantially larger than the holotype of A. sinanoensis, now recognized to be a juvenile. Furthermore, Kohno (1996) pointed out that both this specimen and the juvenile A. sinanoensis type specimen were from the same rock unit. This specimen, by the way, represents one of the most gigantic of all pinnipeds – it has an estimated skull length of 55-59 centimeters, just a hair smaller than the giant walrus Pontolis (60 cm), and even larger than the double tusked walrus Gomphotaria (40 cm). I think Kohno (1996) makes a reasonable case that this specimen is an adult of Allodesmus sinanoensis, and therefore that Allodesmus megallos is a junior synonym. In conclusion, three species of Allodesmus appear to be known from Japan.

A size comparison with casts of the partial gigantic rostrum of Allodesmus sinanoensis (=Allodesmus megallos) and the type specimen of Allodesmus kelloggi.

Allodesmus from Sharktooth Hill, California: Five species of Allodesmus are known from the middle Miocene of California, and three of them have all been named from the Round Mountain Silt: Allodesmus kernensis, Allodesmus kelloggi, and Allodesmus gracilis. As stated in the previous post, Allodesmus kernensis was described from a partial mandible with an erect canine, a deep mandibular symphysis, and a somewhat double-rooted lower molar. Mitchell (1966) originally drew attention to the observation that the mandibles of Allodesmus kelloggi and other mandibles from the Sharktooth Hill Bonebed consistently differed from Kellogg’s type specimen in having a shallower mandibular symphysis, having a more erect canine, and having a single rooted lower molar. Mitchell (1966) also pointed out that the locality data for the Allodesmus kernensis holotype actually pointed to two different, mutually exclusive localities, and suggested that the locality data indicated that the Allodesmus kernensis holotype was collected somewhat below the stratigraphic level of the Sharktooth Hill Bonebed, lower down within the Round Mountain Silt. Because of these perceived differences, Mitchell restricted A. kernensis to the type specimen, named his new skeleton A. kelloggi, and referred all known Allodesmus material from the bonebed to his new species.

The holotype skull of Allodesmus kelloggi. Check out the size of that orbit!

            Many of these claims were criticized by Barnes (1970, 1972). He pointed out that in the context of variation of modern Zalophus (California Sea Lion), the features identified by Mitchell as separating Allodesmus kernensis from the rest of the bonebed sample are within the range of variation expected for a pinniped species, with the exception of shape of the mandibular symphysis – which varies less within extant Zalophus than it does within the bonebed sample of Allodesmus. Regardless, Barnes (1970) pointed out that all of the morphologic features used by Mitchell (1966) to separate Allodesmus kernensis actually fell within the range of variation of Allodesmus mandibles from within the bonebed itself. Barnes (1970, 1972) then synonymized A. kelloggi with A. kernensis, and referred all known material from the bonebed to A. kernensis. Barnes further pointed out “no conclusive arguments that the holotype of A. kernensis was collected from some formation other than the Round Mountain Silt”. Although this does not really negate Mitchell’s suggestion that the holotype came from an older horizon, Barnes correctly points out that there is uncertainty regarding the A. kernensis type locality – which in my mind means that Mitchell’s hypothesis is speculative. In the context of known Allodesmus variation and the fact that the A. kernensis holotype falls within the range of variation reported for mandibles from the bonebed – I think the idea that the A. kernensis type is more primitive than the bonebed sample falls on its face, and that the hypothesis that A. kernensis was really collected from a lower, older horizon need not be invoked. After all, the type specimen was (thought to be) collected in 1911 by R.C. Stoner, according to the specimen label, or by Charles Morrice (according to Kellogg, 1922), in 1909-1912; locality data were not always accurate in those days. Given the two sets of contradictory data regarding the provenance of the type specimen, any argument being made about the locality of this specimen should be considered to be speculative at best. I speculate that, given how abundant vertebrate fossils are within the bonebed, and how rare they are outside the bonebed, that it is more likely that the specimen was in fact collected from the bonebed and that the locality data is inexact. This seems more reasonable to me, but I note that it is not falsifiable – and that Mitchell’s (1966) arguments are not falsifiable either (although perhaps less parsimonious than my suggestion).

Fifty years of taxonomic disagreement contained in a single photograph. How many species of Allodesmus are represented here? One? Two? or Three? From top to bottom, holotype mandible of Allodesmus kelloggi, cast of holotype of Allodesmus kernensis, and cast of referred mandible of Allodesmus gracilis.
Despite the rather sober arguments made earlier, Barnes and Hirota (1995) indicated that the wider variation within the “Allodesminae” resulted in the recognition that the seemingly minor differences between Allodesmus kernensis and the bonebed sample were in fact meaningful. In the absence of quantitative study (as Barnes, 1970, had done) they resurrected Allodesmus kelloggi – a name which they restricted to Mitchell’s type specimen – and went so far as to establish a third species, Allodesmus gracilis, for the remainder of material from the bonebed. Barnes and Hirota (1995) echoed many of Mitchell’s (1966) arguments and casually dismissed Barnes (1970, 1972) earlier, and in my opinion well-reasoned observations. They further pointed out that the holotype mandibles of Allodesmus kelloggi have a strange, widened flange of bone on the anterior part of the coronoid process, in addition to having a single-rooted lower molar, apparently unique amongst specimens from Sharktooth Hill. However, it is well known that modern pinnipeds have tooth roots that vary quite a bit – the short version is that arctoids primitively have triple or double-rooted premolars and molars, and that in the transition to aquatic life, many pinnipeds have lost their carnassial teeth in favor of simplified teeth for catching fish, roughly similar to what happened with dolphins. Dental simplification, in concert with crowding the teeth in the front of the jaw, has resulted in double rooted teeth of archaic pinnipeds fusing together, generally starting from the front of the jaw. In my 2011 paper on fossil Callorhinus, I noted that there is a fair amount of variation in the tooth roots of extant Callorhinus ursinus (Northern Fur Seal). In the context of extant pinniped dental variation (which is considerably more than terrestrial carnivores, probably because pinnipeds do not chew and thus do not require precise occlusion), this observation on Allodesmus kelloggi does not preclude it from being a member of Allodesmus kernensis.

The holotype skull of Allodesmus gracilis.

Deméré and Berta (2002), although not focused on Allodesmus taxonomy, noted that the occurrence of three species of Allodesmus within a narrow stratigraphic interval in the Round Mountain Silt is a bit hard to swallow, and I agree. I can understand the rationale behind restricting names to seemingly oddball specimens like the holotypes of Allodesmus kelloggi and Allodesmus kernensis – these are decisions done in spirit of taxonomic stability. However, any argument that these three perceived species are distinctive enough to be recognized as separate taxonomic entities needs to be done quantitatively – which has been done exactly once - by Barnes (1970). There is a huge discrepancy between the rather taxonomically conservative conclusions of Barnes (1970) and the rather optimistic splitting of Allodesmus into several genera.

The holotype mandible of Allodesmus gracilis - or a referred specimen of Allodesmus kernensis?

So, how many species and genera are there? I tend to be on the taxonomic lumping side of things, as modern pinnipeds show quite a bit of variation – and keeping this variation in mind, I think a bit of skepticism is in order regarding new taxonomic names for the densely sampled and well published record of pinnipeds from the eastern North Pacific. I think the following list best summarizes my opinions on which genera and species are real:

Atopotarus? Allodesmus? courseni
Allodesmus packardi (syn: Brachyallodesmus)
Allodesmus kernensis (syn: A. gracilis, A. kelloggi)
Allodesmus sadoensis
Allodesmus sinanoensis (syn: Megagomphos sinanoensis, Allodesmus megallos)
Allodesmus naorai

Next up: we still have at least two more parts to this, including the paleoecology of Allodesmus and the phylogenetic position of the family Desmatophocidae.


L. G. Barnes. 1970. A re-evaluation of mandibles of Allodesmus (Otariidae, Carnivora) from the Round Mountain Silt, Kern County, California. PaleoBios 10:1-24.

L. G. Barnes. 1972. Miocene Desmatophocinae (Mammalia: Carnivora) from California. University of California Publications in Geological Sciences 89:1-76.

L. G. Barnes and K. Hirota. 1995. Miocene pinnipeds of the otariid subfamily Allodesminae in the North Pacific Ocean: Systematics and relationships. The Island Arc 3:329-360.

Demere, T. A., and A. Berta. 2002. The Miocene pinniped Desmatophoca oregonensis Condon 1906 (Mammalia: Carnivora), from the Astoria Formation of Oregon; pp. 113–147 in R. J. Emry (ed.), Cenozoic
Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithsonian Contributions to Paleobiology 93.

T. Downs. 1956. A new pinniped from the Miocene of southern California: With remarks on the Otariidae. Journal of Paleontology 30(1):115-131.

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

N. Kohno. 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, New Series 181:388-404.

E. D. Mitchell. 1966. The Miocene pinniped Allodesmus. University of California Publications in Geological Sciences 61:1-105.

No comments: