Nerodia cyclopion and N. floridana 
89 
on ventral coloration and scale counts, fell within the predicted 
group based on head scale data alone. 
I tested the holdout group against the classification criteria 
established in the calibration group analysis. Again, 100% of the 
males of both taxa were classified correctly. All female floridana 
were also correctly classified, but two female cyclopion were misclassified 
as floridanda. The first, (Auburn University Museum 22559, Ala- 
bama: Baldwin County, 6 miles NNW town of Stockton, Douglas 
Lake) has a typical cyclopion ventral color pattern. Its dorsal scale 
row formula (27-25-21) resembles that found in 70.4% of the fe- 
male cyclopion I examined. This dorsal scale formula was found in 
none of the female floridana. The snout of this individual is some- 
what shorter and narrower than is typical for cyclopion , and this 
feature caused the discriminant analysis to classify it as floridana. 
However, variation from the normal cyclopion pattern is slight; this 
individual was identified as floridana with an estimated 50.2% probability 
of being correct. Considering all the data available, as well as its 
geographic location, I conclude that this individual is a cyclopion. 
The second misclassification (University of South Alabama 1930, 
Alabama: Baldwin County, Negro Lake) also has typical cyclopion 
ventral coloration and dorsal scale row counts (27-25-21). It is the 
longest specimen of cyclopion I examined (1,035-mm snout-vent length, 
1,322-mm total length), which exceeds by 52 mm the record for 
cyclopion as listed by Conant and Collins (1991). The muzzle width 
is less than that of other cyclopion of similar length, which may be 
an allometric consequence of its large size. This variation probably 
exceeded the bounds of the classification criteria, which were estab- 
lished with data from smaller specimens. Again, the data suggest 
that this animal is also a cyclopion. 
If gene flow has occurred between the two taxa in the zone 
of contact, the distributions of the discriminant functions of the 
holdout groups would be skewed away from those of the calibra- 
tion groups. This was tested (for cyclopion only due to the small 
sample sizes in floridana) by comparing sample means of the dis- 
criminant functions of the groups with /-tests. No significant differ- 
ence in sample means was found in female cyclopion (P = 0.183, 
t = 1.34, df = 76), but in the male population the scores of the 
holdout group did appear skewed away from the distribution of the 
calibration group scores. In addition, the /-test indicated a signifi- 
cant difference between the two groups (P = 0.012, t = 2.61, df = 
56). An obvious explanation for these contradictory data is not ap- 
parent. Since gene exchange, if present, must affect both sexes, . I 
suggest that the relatively small sample size for male cyclopion 
