83 



apparently somewhat stronger within the monachines). Differences with the overall 

 solution are limited exclusively to within the phocid subfamilies, which demonstrate how 

 labile the relationships at this level are with respect to which characters are used, and/or 

 how they are used. Finally, as demonstrated by a constraint analysis (Fig.140 and Tab.4), 

 the inversely weighted data matrix can accommodate the "unweighted" solution with a 

 minimal amount of extra homoplasy: 214 extra steps (four extra corrected steps). The 

 bootstrap frequency for the least supported major clade of this solution (estimated at 31%) 

 is also remarkably high compared to those of the remaining constraint analyses. 



Condensed analysis (Fig. 17) 

 The constrained monophyly of four higher level phocid taxa - Mirounga, Monachus, 

 Phoca (sensu Burns & Fay 1970). and Lobodontini - resulted in some rather major 

 topological changes within each phocid subfamily, with respect to the overall solution. 



39 



43 



24 



94 



58 



52 



77 



70 



80 



Halichoerus 

 Phoca spp. 



(sensu Burns & Fay 1970) 



46 



C 



96 



Erignathus 



Cystophora 



Lobodontini 



Mirounga spp. 



Monachus spp. 



Odobenus 



Zalophus 



Lutra 



Enhydra 



Martes 



~ ~ Procyon 



Ursus 

 Canis 



Fig. 17: Cladogram resulting from a parsimony analysis of the inversely weighted data matrix with 

 the taxa Mirounga, Monachus, Phoca (sensu Burns & Fay 1970), and Lobodontini collapsed so as 

 to be monophyletic (length = 49,058 steps, CI = 0.518, HI = 0.709, RI = 0.634, RC = 0.483). Numbers 

 represent bootstrap frequencies supporting each node (1,000 replications). 



