76 



increase in length). Low bootstrap frequencies (all below 32% and most below 7%) again 

 denote weak support for these alternative groupings, while the small number of most 

 parsimonious solutions likewise hints at good resolving power in the data set. 

 Of more interest than the magnitude of the increases in length, however, are the topologies 

 resulting from the various constraint conditions (Fig. 14). Disruption of a monophyletic 

 Phocidae resulted in the otarioids forming a monophyletic sister group to the monachines, 

 reiterating the late convergence of the latter group on the former (Repenning 1990). Of 

 the two subfamilies, paraphyly of the Monachinae was easier to achieve, as might be 

 expected with the slightly weaker support noted earlier for this entire subfamily (see 

 Overall Parsimony Analysis and Statistical Tests sections). The constraint of a 

 paraphyletic Phocinae or a monophyletic Cystophorinae, meanwhile, again demonstrates 

 the strong tendency of Cystophora to join the monachines. The enforced paraphyly of 

 Monachus produced a topology much like that advocated by Wyss (1988a), again 

 demonstrating that a paraphyletic Monachinae is dependent upon a paraphyletic Monachus 

 to some degree (Berta & Wyss 1994). 



Additional observations support some of the more contentious, non-traditional relation- 

 ships indicated by the overall solution. Monophyly of Monachus, as well as its terminal 

 position within the lobodontines, was extremely robust and was only disrupted when 

 specifically forced to do so. Likewise, paraphyly for the lobodontines was always 

 indicated, even when Monachus was disrupted. A more terminal position for Erignathus 

 (or, equivalently, a basal position for Cystophora) within the phocines was also always 

 observed, when not specifically constrained otherwise. Erignathus was typically embedded 

 within the Phocini, but it always clustered internal to Cystophora in any case. 

 Finally, one curious phenomenon was observed in this portion of the analysis. Changes 

 forced within the monachines altered not only the topology elsewhere within this subfam- 

 ily, as would be expected, but often within the phocines as well. These were largely 

 localized within the Phocini (plus Erignathus), and typically amounted to a basal shift of 

 Histriophoca and Pagophilus to form successive sister taxa to the remaining Phocini (plus 

 Erignathus). However, the complete absence of the equivalent reciprocal situation again 

 hints at the comparatively weaker support for the Phocini (plus Erignathus) within the 

 phocids (see Statistical Tests section). Monachine interrelationships, although comparably 

 weak with respect to a bootstrap analysis (see Statistical Tests section), appear to be 

 exceptionally robust by all other indications. Only changes forced directly within the 

 Monachinae seem to be able to disrupt the interrelationships of this subfamily indicated 

 in the overall solution. 



Overall conclusions - possible effects of polymorphic data 



The fact that no constrained solutions were substantially longer than the overall solution 

 might derive from the high amount of polymorphic data in this analysis (see Overall 

 Parsimony Analysis). The flexibility allowed by the alternative states of the polymor- 

 phisms likely permitted the very different competing topological hypotheses to be satisfied 

 with a minimal amount of extra homoplasy. In contrast, those data matrices with less 

 polymorphic data (e.g., Berta & Wyss 1994) would presumably be more rigid, and meeting 



