166 



features (e.g., small versus medium versus large)]. However, another source for any 

 discrepancies might be that we have recorded data for each individual species, whereas 

 many previous studies attempted to describe features that were presumed to apply 

 throughout, or felt to be primitive for, some higher level taxon (e.g., King 1966; Wyss 

 1988a; Wozencraft 1989; Berta 1991). By examining all phocid species, we believe that 

 we have shown that some of these generalities of phocid (or phocine, or monachine, ...) 

 morphology do not necessarily hold absolutely among all of the concerned species. Finally, 

 some inconsistencies between the inferred evolutionary pathways can also be noted. In 

 most cases, this is due to the pathways being derived from different cladograms. However, 

 in a fair number of other cases [most notably, those arising from Wyss (1988a)], the 

 historical evolutionary pathway was shown to be only one of two equally parsimonious 

 possibilities, corresponding to the singular use of only one optimization criterion available 

 for character reconstruction. 



Yet, the most disconcerting contradiction arises when the characters are viewed 

 individually, as opposed to collectively. Together, all of the 168 included characters 

 produced a very clean solution, with a somewhat surprisingly low number of equally most 

 parsimonious (and slightly less than most parsimonious) solutions (see Overall Parsimony 

 Analysis and Statistical Tests). However, when viewed individually, as in this section, 

 very few characters directly and unequivocally indicate the overall solution that they do 

 as a group (i.e., Fig.5B). This is even more apparent when the entire data matrix is divided 

 into process partitions (sensu Bull et al. 1993; i.e., the characters were roughly divided 

 into distinct anatomical regions) and analyzed separately (Fig. 25). (Note that in order to 

 generate character sets of sufficient size to yield reasonable resolution, characters 

 originating from the forelimb, pelvis, hind limb, and atlas were grouped as "post-cranial" 

 characters, while those from the bony falx and tentorium, dorsal braincase, and mandible, 

 and all soft- anatomical features were grouped as "miscellaneous" characters.) Although 

 the resolution is surprisingly good given the very reduced number of characters per taxon 

 (due, in part, to the unusually high number of multistate characters, which can support 

 more putative synapomorphies per character), none of the indicated cladograms really 

 supports a solution comparable to the overall solution. Monophyly of the phocids, one of 

 the strongest nodes in the overall solution, is only indicated in the consensus solutions of 

 the character sets from the basicranial, teeth (which is to some degree an artifact of our 

 coding many teeth characters as being inapplicable for the outgroup taxa, thereby forcing 

 a monophyletic Phocidae), and miscellaneous regions (Fig.25D, E, and G). Monophyly of 

 the phocid subfamilies is even rarer, being indicated only by the basicranial (Phocinae 

 only) and post-cranial (both Monachinae and Phocinae) character sets (Fig.25D and F). 

 Of the individual results worth noting, the now abandoned Cystophorinae (= Cystophora 

 plus Mirounga spp.) is clearly supported by both snout and teeth characters (Fig.25A and 

 E). This distribution of support corresponds quite nicely with the major features used to 

 define the Cystophorinae - a 2/1 incisor formula and some form of inflatable nasal 

 proboscis in the adult males, plus some additional minor characteristics from the same 

 regions (see King 1966; Ridgway 1972) - which were based on feeding specializations 

 and sexual selection (McLaren 1975). Additionally, it appears that even the distinctive 



