177 



ancestor to the exclusion of M. monachus; 2) the Phocini are paraphyletic, as are its 

 constituent genera Phoca (both sensu stricto and Burns & Fay 1970), and possibly Pusa; 

 3) the Monachinae are monophyletic; 4) the species level relationships of the phocids are 

 as indicated in Fig.5B; and. 5) the pinnipeds are monophyletic, with lutrine affinities. 



Potential sources of error 



Obviously, the above conclusions must be viewed in the light of a number of qualifying 

 statements. Perhaps the key such qualification is that we have merely presented a summary 

 of one particular data set, using one particular method of tree reconstruction, and whose 

 underlying distribution need not necessarily coincide with the actual phylogeny of the 

 phocids (see also Statistical Tests section). Therefore, any judgment as to the accuracy 

 of the results presented herein must ultimately derive from an assessment of the overall 

 quality of the data set, and whether the characters examined in this study constitute a 

 reasonably random sample from the universe of all possible characters. Given these 

 restrictions, however, we believe that both the size and scope of our data matrix give this 

 study several advantages over previous analyses of phocid phylogeny. To our knowledge, 

 this study is the only species level analysis of the entire phocid family. Thus, no potential 

 pairings of species were prevented by the assumed monophyly of any higher level taxa. 

 As well, multiple specimens were examined for each species to attempt to account for the 

 unusually high amount of intraspecific variation within the pinnipeds (Mivart 1885; Doutt 

 1942; King 1966; Ray 1976b). A wide range of characters, representing most of the 

 osseous skeleton, were also employed, the use of which allowed for a data matrix with 

 very few missing data points. This range of characters, moreover, provided a better overall 

 representation of the suite of all possible characters, the impetus behind the total consensus 

 approach to phylogenetic analysis (see Kluge 1989; Kluge & Wolf 1993). In contrast, the 

 use of sets of characters derived primarily from single anatomical regions yielded solutions 

 even more in conflict with the "traditional wisdom" regarding phocid phylogeny. Finally, 

 simplifying and/or biasing assumptions (e.g., use of only a single outgroup, avoidance of 

 polymorphic data) were avoided, hopefully resulting in a more realistic analysis. 

 Yet, the sources of information employed in this study still possess inherent limitations. 

 The imposed exclusion of any fossil taxa (due to their general unavailability) could prove 

 detrimental due to the possible effects of the exclusion of any taxa in a low level analysis 

 such as this (Arnold 1981). Be that as it may, the extremely poor fossil record of the 

 pinnipeds, and of the monachines in particular (Hendey & Repenning 1972; Ray 1976a), 

 equates to a high proportion of missing (i.e., undiscovered) fossil taxa. As well, the 

 exclusion of fossil pinniped taxa does not appear to alter the basic topology of pinniped 

 relationships as determined from extant forms only (Flynn et al. 1988; Berta & Wyss 

 1994). Nonetheless, it would have been interesting at the very least, and informative in 

 any case (particularly with respect to elucidating the evolutionary pathway of certain 

 characters), to have included various fossil pinnipeds and putative fossil pinniped ancestors 

 (e.g., the lutrine-like Potamotherium). 



Of more concern is the high intraspecific variation characteristic of the pinnipeds 

 mentioned above. This variability is so severe with respect to cranial anatomy in particular. 



