144 Joshua Laerm 



Results indicate that the interaction term was not significant for the 

 characters examined. Therefore, the intercept (the differences between 

 populations) for the covariate was tested under the following model; 



V4 



population. 



The results of this model (Table 1) indicate significant differences 

 between populations for most characters. For these characters the dif- 

 ference in least squares adjusted means for each population was deter- 

 mined by the Scheffe Test (Morrison 1967) (Table 2). 



No significant differences for any of the characters examined can 

 be seen between females of G. cumberlandius and other populations of 

 female G. pinetis. Males of G. cumberlandius can be distinquished from 

 males of other G. pinetis populations on the basis of a single character — 

 total length. 



Results of the Multiple Discriminant Function Analysis (Fig. 2) 

 similarly indicate low levels of morphological distinction between G. 

 cumberlandius and mainland populations of G. pinetis. Two assemb- 

 lages are indicated in both males and females: an upland assemblage 

 consisting of the populations from Adam, Augusta, and Hursman's 

 Lake, and a coastal assemblage consisting of the two Camden County 

 populations and G. cumberlandius.The only inconsistency in clustering 

 in both males and females occurs in the apparent relatedness of the coast- 

 al Camden County assemblage. Female G. cumberlandius and Kings- 

 land G. pinetis appear more closely related than either is to the Scotch- 

 ville G. pinetis population, while males from both Camden County 

 populations appear more closely related to each other than either does 

 to the Cumberland Island population. The important point is, of course, 

 that G. cumberlandius appears more closely related to coastal Camden 

 County G. pinetis than do these G. pinetis to their upland conspecifics. 



DISCUSSION 



Bangs' (1898) description of the insular G. cumberlandius was 

 based on a small series of specimens (N = 13) collected at "Stafford 

 Place." He distinquished it from adjacent mainland Georgia and Florida 

 populations of G. pinetis on the basis of the very large size and slight 

 pelage and cranial differences. I find that his pelage and cranial features 

 are generally unsatisfactory to permit the distinction of G. cumberlan- 

 dius from other populations of southeastern pocket gophers. Pelage in 

 the G. pinetis complex is extremely variable and tends to be correlated 

 with local soil color (Williams and Genoways 1980; Laerm et al., in 

 press). Hence, it has little value in taxonomy. The results of cranial 

 morphometry reported by Williams and Genoways (1980) and herein 

 indicate that cranial differences between G. cumberlandius and main- 

 land populations of G. pinetis are not sufficient to warrant species level 

 recognition for G. cumberlandius. 



