16 



The overall goal of this study is to attempt to answer the five outstanding questions 

 regarding phocid phylogeny that we have identified above. This is done via a cladistic 

 analysis (sensu Hennig 1966) based on parsimony, using the outgroup method to determine 

 character polarities. Morphological data are used exclusively. Further assumptions and 

 details concerning this analysis are found in the Methods and Materials section, with the 

 results being presented in the Overall Parsimony section. 



The remainder of this study deals largely with the various means available to judge the 

 robustness of the indicated solution. Attempts to place confidence intervals on 

 phylogenies/taxonomies have been difficult, and thus have only rarely been carried out. 

 One advantage of cladistic analysis in this regard is its ability to roughly indicate the 

 support for a solution (or any portion thereof) by the number of synapomorphies supporting 

 various nodes of the cladogram. However, this measure of support is still somewhat 

 subjective, as it is dependent upon the characteristics of the data set (e.g., the number and 

 type of characters examined) and therefore does not allow for easy comparison between 

 data sets. Cladistics has recently seen the development of statistical and other comparative 

 tools that seemingly allow an even more objective assessment of the quality of a solution, 

 as well as facilitating comparisons between different phylogenetic hypotheses. Again, the 

 tests and assumptions behind them are described in the Methods and Materials section, 

 with the results presented in Statistical Tests and Comparative Tools sections. 

 With the rise of the use of statistics in cladistics, the realization that any cladistic 

 hypothesis is only as good as the data it is based upon seems to have been forgotten. This 

 point becomes even more crucial when one realizes that the outcomes of most of the 

 newly-developed statistical procedures seem to be consistently misinterpreted (see 

 Statistical Tests). Thus, we are left with one real, but increasingly rarely used "test" as 

 to the quality of a solution: an in-depth examination of the characters that were used. This 

 is to be found in the Character Analysis section, in which descriptions and historical 

 notes are presented for all the characters examined in this study, together with a description 

 of the evolutionary pathway implied for each character by the overall solution that was 

 found. 



Finally, this study concludes by examining such broad-ranging topics as potential sources 

 of error, future lines of research, miscellaneous corroborating evidence (biogeography and 

 timing of parturition), and the taxonomic implications of the proposed phylogeny of the 

 phocid seals advocated herein. 



ACKNOWLEDGEMENTS 



This contribution grew from the M.Sc. thesis work of the senior author. For comments 

 and suggestions throughout that process, the assistance of Marc Ereshefsky, Larry Linton, 

 Gordon Pritchard, Herb Rosenberg, and especially Harold Bryant are gratefully 

 acknowledged. Late hints by Andy Purvis helped to round things out. We also thank the 

 many people involved in loaning us specimens, or providing access to collections in their 

 care: Dr. G.G. Musser and W.K.-H. Fuchs (American Museum of Natural History, New 

 York); Dr. Ted Daeschler and Fred Ulmer (Academy of Natural Sciences of Philadelphia): 



