110 



Joshua Laerm and James L. Boone 



P. maniculatus 



Axis 1 



P. polionotus 



P. gossypinus 



P. leucopus 



Axis 2 





Fig. 2. Distribution of principal component scores generated from external 

 measurements (body, tail, and hind foot lengths) illustrating overlap in 

 the measurements of these characters. 



A priori identifications were based on specimen tag information. 

 We used only specimens we believed were correctly identified. We 

 wanted to create a robust generalized model, but we also wanted to 

 build the model based on, as much as possible, animals that we felt 

 were correctly identified. The selection procedure resulted in using 

 110 P. gossypinus, 108 P. leucopus, 112 P. maniculatus, and 110 P. 

 polionotus. The Appendix lists specimens examined. We used five 

 additional specimens of each species, generally selected from 

 locations not included in the model building process, to test the model. 



One of us (JL) measured 14 cranial characters to the nearest 0.1 

 mm with dial calipers and recorded three external measurements from 

 specimen tags. We estimated age from pelage characters (no juvenile 

 gray), tooth wear (significant wear on all major cusps), and degree of 

 cranial suture fusion. We measured only adults (in age classes 4-6 of 

 Schmidly 1973) and excluded specimens with missing data from all 

 analyses. Mensural characters (Choate et al. 1973, DeBlase and Martin 



