85 



more homoplasious ingroup topology of the condensed solution is puzzling. Beyond the 

 major limitation that the bootstrap can only indicate signal strength within a data matrix 

 (see Statistical Tests), it may also be connected with the reduced number of taxa present 

 in the condensed solution, resulting in fewer reasonable alternative groupings. For instance, 

 one would expect the members of each subfamily to associate with one another whether 

 the subfamily is paraphyletic or not. Thus, the collapsing of the Monachinae to only three 

 taxa in the condensed analysis dramatically reduces the number of alternative pairings, 

 possibly artificially inflating the apparent support for those possibilities that remain. 

 Finally, it should be stressed that the differences in topology that were obtained in this 

 analysis arose solely from improper assumptions of monophyly, and not because of large 

 scale differences between the data matrices. The consensus character states for each higher 

 level taxon are directly based on the same set of observations that led to the overall 

 solution. Therefore, essentially the same matrix was used in both cases. As well, the altered 

 findings are not dependent on either solution being the correct estimate of the actual 

 (phocid) phylogeny. However, by improperly assuming the monophyly of higher taxa, we 

 may actually be hindering our efforts in systematic biology to uncover the true phylogeny 

 of a group of organisms. 



CHARACTER ANALYSIS 



The previous two sections dealt largely with the various methods devised to assess the 

 degree of confidence one may have in a specific cladogram (cladistic hypothesis) versus 

 other rival hypotheses. Together with the general misinterpretation of these tests (see 

 Statistical Tests), what continually appears to be lost in all this is the realization that any 

 cladistic hypothesis is only as good as the set of characters it is based upon and the 

 underlying hierarchical pattern they indicate. In this regard, this section presents an in- 

 depth analysis of all the characters examined in this study (also listed in Appendix B), 

 with an emphasis on the 168 that were included in the cladistic analysis (excluded 

 characters are marked with an asterisk). Historical notes and descriptions are initially 

 provided for each character, followed by a prose equivalent to the information found in 

 the apomorphy list (Appendix E). This latter feature is limited to the included characters, 

 but reconstructions for the excluded characters may be derived from Appendix E. All 

 reconstructions (including those of the excluded characters) are based on the topology 

 displayed in the overall solution (Fig.5B). Goodness-of-fit statistics (but now based on the 

 consensus and not the optimal solutions) for all individual characters are presented in 

 Appendix F. 



In the following, the citation(s) following each character refer(s) to our source for that 

 character. In many cases, they may not correspond to the initial mention of the feature in 

 the literature, but to the first use of the character in a systematic fashion. Descriptions of 

 characters or character states were often modified from their indicated sources. This was 

 typically done to accommodate the larger range of variation we observed in applying 

 characters initially used to distinguish between a limited distribution of taxa over the fuller 

 set examined here. In other cases, characters were derived from those in the literature 



