32 Michael E. Seidel 



neck are broad whereas paramedian dorsal stripes are very narrow, 

 sometimes convergent or obscure. All seven New River specimens that I 

 collected have a yellow plastron that was lightly tinged with orange in 

 two individuals. Plastral markings typical of P. concinna are present on 

 each turtle, but less prominent in older specimens (Fig. 2). Slight fading 

 of markings and coloration was evident after 2 to 3 months in captivity. 

 The cutting edge (tomium) of the upper jaw in most individuals is 

 weakly emarginate and the lower jaw is moderately serrate (Fig. 2). 

 Four skulls which were analyzed show no evidence of a vomerine shelf 

 contributing to the crushing (alveolar) surface of the upper jaw, a uni- 

 que characteristic of species in the rubriventris series (McDowell 1964). 



Twelve adult New River turtles examined have a shell depth range 

 of 31.1 to 39.4 (ratio calculations following Weaver and Rose 1967), 

 which broadly overlaps the range reported for P. concinna but falls 

 below the ranges of P. floridana and P. nelsoni. Nuchal scute underlap 

 ratios in New River Pseudemys, 16.0 to 23.4, are greater than in P. 

 concinna but are within the combined range of ratios for P. floridana 

 and P. nelsoni. The gular scute overlap ratio was highly variable (7.9- 

 18.8), falling within the combined ratio range reported for all three spe- 

 cies (Weaver and Rose 1967). Because most of the specimens examined 

 by Weaver and Rose were from Florida, shells of adult Pseudemys from 

 northern localities were measured and compared to New River speci- 

 mens. These ratios (Fig. 3) again indicate a shallow shell depth in New 

 River turtles (B) and P. concinna (C,H). Gular scute overlap and nuchal 

 scute underlap are greater in P. rubriventris (R), than in New River 

 specimens (B) and northern subspecies of P. floridana (F) and P. con- 

 cinna (C,H). However, these two characters are highly variable and 

 apparently not effective in separating P. concinna and P. floridana out- 

 side of Florida. 



Results from the discriminant analysis of all Pseudemys skulls are 

 presented in Figure 4. On the first canonical axis (K i) plots of skulls in 

 the rubriventris series (A,N,R) are clearly disjunct from all forms except 

 P. c. texana (T). Broad separation is seen between P. rubriventris (R) 

 and New River Pseudemys (B). Also noteworthy is the separation, on 

 the second canonical axis (K 2 ) of P. alabamensis (A) from P. rubriven- 

 tris (R) and P. nelsoni (N). The first two axes account for 51% and 12% 

 of the total dispersion, respectively. In order of increasing importance, 

 temporal arch width, jugal-quadratojugal length, interorbital width and 

 dentary alveolar width were the most influential characters providing 

 separation on the first axis (Table la). Anterior skull width, dentary 

 alveolar width, and lingual alveolar width contributed most to separa- 

 tion on the second axis (Table la). All individuals were classified into 

 their appropriate taxa, except one P. c. mobilensis that was placed in P. 

 f floridana and one P. f hoyi that was placed in P. c. hieroglyphic a. 

 The P. c. mobilensis specimen (MCZ 1651) was collected within the 



