114 



Michael E. Seidel and William M. Palmer 



HW 



LC 



Fig. 7A. Anterior view of the head illustrating length of the tomial cusp (LC) 

 and head width (HW). 



Barr et al. 1976) was initially applied, thus avoiding assignment of 

 individuals to groups (species). Morphological similarity or divergence 

 was examined by observing clustering of individuals on bivariate plots 

 of their principal component scores. That provided a test to determine if 

 a priori species identifications based on qualitative characters could be 

 corroborated by mensural characters. It also provided a more objective 

 means to determine morphological overlap between species and possible 

 cases of hybridization or intergradation. If the a priori assignment of a 

 specimen had been noted as questionable (based on qualitative 

 characters) and its PCA plot was clearly outside its species cluster but 

 within the range of another species, it was reidentified. Otherwise, 

 taxonomic reassignment was avoided. Principal components analysis 

 was followed by stepwise discriminant analysis (BMDP7M, Dixon 

 1977). Discriminant analysis was applied to test for significant 

 morphometric differences between P. rubriventris, P. concinna, and P. 

 floridana. Sexes were again examined separately and the influence of 

 size (age) was reduced by linear regression. Size-adjusted residuals were 

 obtained from the 30 shell and head-stripe measurements by regressing 

 each character on carapace length. 



Multivariate analysis of variance (MANOVA-SAS) followed by 

 Fisher's protected least significant difference (/-tests) were used to test 

 for utilitarian taxonomic characters that might provide a more objective 



