At left is another reconstruction, this time of a fossil walrus, Valenictus chulavistensis. Valenictus chulavistensis was described by Tom Demere, the Paleontology Curator of the San Diego Natural History Museum.
Valenictus is a very odd pinniped, for several reasons. For one, V. chulavistensis (the only well known species of Valenictus) lacks all of its teeth, save the two elongate canines, which are fairly similar to the tusks of the extant walrus, Odobenus rosmarus. Nevertheless, modern walruses do not use their dentition for feeding, so while the disappearance of the noncanine dentition is unprecendented in other pinnipeds, it is not completely unsurprising as modern walrus noncanine teeth are nearly vestigial, so to speak.
Valenictus also has a number of postcranial features that are highly derived; the fore- and hind-limb bones do not closely resemble those of other fossil or modern Odobenidae (walruses). Also, the skeleton of Valenictus is pachyostotic and osteosclerotic (pachyostotic = thick bones, osteosclerotic = dense bones, more or less). This is possibly a ballast-like adaptation for maintaining bouyancy in a warm water environment.
These adaptations are also seen in sirenians. This is probably an adaptation for feeding in shallow, warm waters: the extant walrus feeds on benthic molluscs in shallow (but very, very cold) water, and (most) sirenians feed on seagrass in the photic zone. Valenictus chulavistensis evidently had many of the same feeding adaptations for molluscivory as does Odobenus, and Valenictus thus far has been found in relatively warm-watter settings, based on mollusc and microfossil assemblages. Additionally, other fossils of Valenictus are known from the proto-gulf of California (i.e. the Salton trough), which was hypersaline - this would have made a large, fat animal even more positively bouyant in the water column, a major hindrance to a critter that relies on benthic invertebrates as a food source.
I'll do some more in-depth posts on this fascinating fossil walrus in the future.