14 



an unresolvable polytomy. This was one option put forth by Flynn et al. (1988). the other 

 being an ursid sister group to the pinnipeds. Taken together, this discussion demonstrates 

 that the question of phocid affinities (and those of the remaining pinnipeds) within the 

 Arctoidea should still be regarded as being uncertain, if they are even resolvable to begin 

 with. 



Yet, the question of phocid ancestry still bears critical importance to the determination of 

 the internal relationships of the phocids. In cladistic analysis, the accepted method of 

 determining character polarities is through outgroup analysis (Hennig 1966; Arnold 1981; 

 Wiley 1981; Maddison et al. 1984). Normally, this procedure is reasonably straightforward, 

 with the most closely related taxon to the ingroup designated as the outgroup, and thus 

 serving to identify the primitive states for the various characters examined. However, with 

 respect to phocid phylogeny, the uncertainty regarding the ancestral affinities of the 

 phocids complicates the question of designating an outgroup taxon. Yet, rather than 

 employing multiple outgroups and allowing the analysis to dictate the most closely related 

 outgroup taxon (and thus character polarities), most studies examining internal phocid 

 relationships have, to date, either not stated an explicit outgroup, or have assumed either 

 an ursid or mustelid (and occasionally lutrine) outgroup, thus potentially biasing the 

 resultant character polarities [see Maddison et al. (1984) for some of the errors inherent 

 in selecting outgroups and how they can, in turn, affect an analysis using outgroups]. 



Goals of this project 



Currently, most of our knowledge concerning phocid phylogeny derives from the studies 

 of de Muizon (1982a), Wyss (1988a), and, recently, Arnason et al. (1995) (Fig.l). However, 

 each study possesses important shortcomings. All three phylogenies are dependent upon 

 the supposition of a particular arctoid outgroup (lutrine, ursid, and ursid respectively). In 

 the case of Wyss (1988a), we feel that such an assumption was not adequately tested in 

 a prior analysis (Wyss 1987; see also Wozencraft 1989). In de Muizon's (1982a) study, 

 some of the characters used deserve closer scrutiny (e.g., aquatic, high snout, "important" " 

 sexual dimorphism), several clades are supported by only a single character, and 

 conflicting (i.e., homoplasious) characters are not mentioned. Yet, despite their limitations 

 and conflicts with each other, these three studies provide the best resolved cladograms of 

 the phocids to date. 



Using these three studies as a guide, and bearing the five outstanding problems we have 

 identified above in mind, we present the current state of knowledge regarding phocid 

 phylogeny in Fig. 2. The cladogram is characterized by large regions of uncertainty and 

 poor resolution, primarily within the Lobodontini and Phocini. within the polytypic genera, 

 and for the ancestral affinities of the phocids as a whole. One area of strong, almost 

 universal, agreement concerns the most primitive members of each phocid subfamily: 

 Erignathus for the phocines (Chapskii 1955a; King 1966. 1983: Burns & Fay 1970: 

 McLaren 1975; Ray 1976a: Wyss 1988a; Berta & Wyss 1994: Arnason et al. 1995: 

 Mouchaty et al. 1995; Perry et al. 1995) and Monachus spp. for the monachines (Hendey 

 1972; Repenning and Ray 1977: Repenning et al. 1979: de Muizon 1982a; King 1983: 

 Wyss 1988a; Arnason et al. 1995; Lento et al. 1995). 



