124 Michael E. Seidel and William M. Palmer 



The pattern of male melanism observed in all species was a loss of 

 yellow lines on the soft parts and carapace and development of a 

 reticulated (worm-like) pattern of dark speckled markings on the head, 

 carapace, and plastron (Fig. 18). This pattern is quite different from the 

 melanism that develops in populations of Trachemys scripta (Lovich et 

 al. 1990), but similar to that of the Cuban slider, Trachemys decussata 

 decussata (Seidel 1988). A different form of melanism is found in adult 

 female and young male P. rubriventris in northern portions of their 

 range (Pennsylvania and New Jersey). In those areas, the soft parts and 

 carapace become nearly solid black, but the plastron remains bright 

 coral or red. It is interesting that darkening to this extent and loss of 

 yellow lines apparently do not occur in the southern populations in 

 Virginia and North Carolina. 



DISCUSSION 



Morphological divergence between P. concinna and P. floridana is 

 much less than their collective divergence from P. rubriventris. Neither 

 principal components analysis nor discriminant analysis clearly separated 

 the two former taxa from each other (Fig. 8 and 9). Male specimens of 

 P. rubriventris showed no overlap with P. concinna and P. floridana, 

 but some overlap was observed for females. Six female P. rubriventris 

 (NCSM 20166, 28753, 28897, 29278; AMNH 90644; USNM Field Series 

 159366) and two female P. floridana (NCSM 14783; CM Field Series 

 24447) plotted intermediately between species clusters (Fig. 9). Specific 

 identification of all these individuals was originally noted as questionable, 

 and all were collected from the relatively small area of southeastern 

 Virginia and northeastern North Carolina where the ranges of these 

 species overlap. There is little doubt that P. rubriventris in the southern 

 portion of its range is somewhat morphologically convergent with P. 

 floridana. One possible explanation for this is that reproductive isolation 

 is not complete and a limited amount of gene flow occurs between these 

 species. That would support Crenshaw's (1965) proposal of hybridization 

 in the region. Another explanation is that selection pressures are similar 

 in this area of sympatry, resulting in homoplastic (convergent) character 

 states (as suggested for other geographic regions by Carr 1952 and 

 Ward 1984). 



Two of the morphologically intermediate specimens from north- 

 eastern North Carolina (NCSM 29278 from Dare Co. and 14783 from 

 Gates Co.) strongly suggest hybridization of P. rubriventris with other 

 Pseudemys . Skulls were prepared from these two turtles to examine 

 osteological characters that have been used to distinguish P. rubriventris 

 from P. concinna and P. floridana (McDowell 1964, Ward 1984). One 



