144 
James L. Gooch and Jeffrey S. Wiseman 
previously given and this character scale. A one-to-one conformance of 
ranks would indicate that interdemic variation in these characters is com- 
pletely determined by openness of habitats and their proximity to 
hypogean environments. The scales conform well for six localities, which 
conform from zero to two rank positions. Three large discrepancies do 
occur and require explanation: (1) Cunninghams has the highest 
hypogean character score, very narrowly over Smoke Hole Spring, but 
ranks as only the fourth most hypogean habitat; (2) Greenland Gap 
ranks eighth in character scale, third in habitat; (3) James Creek ranks 
sixth and ninth, respectively. It should be underscored that neither scale 
is an absolute standard to which the other can be compared. A difference 
in rank may mean faulty ranking of habitat, i.e. the habitat is less or more 
epigean to the amphipods than to the observer, or that factors other than 
habitat scale affect the characters and their rank. Our discussion can take 
factors of the second type into account more easily than the first. For ex- 
ample, James Creek is almost certainly the most fully epigean site, yet its 
character rank does not reflect this (note, however, that its mean rank is 
only 0.75 positions higher than that of the ninth ranked locality). Its 
overall rank is higher due simply to the high appendage ratios of its 
females, as was mentioned earlier and is apparent from Tables 3 and 6. 
Presumably slender-limbed females is a genetic characteristic indigenous 
to the James Creek population and is unrelated to the habitat scale. 
The rank discrepancies of Greenland Gap and Cunninghams are at 
first puzzling. Greenland Gap is about 175 km southwest of seven of the 
localities and could obey different ecogeographic rules. However, Smoke 
Hole Spring and most of the localities of Holsinger and Culver are even 
more distant and they match the Pennsylvania populations quite closely 
in character trends. One potential influence on ecotype that has not been 
considered is migration and attendant gene flow. Populations subjected 
to high migration rates from contrasting habitats would not undergo as 
marked ecotypic differentiation as isolated demes. This is especially true 
if the phenotypic expression is largely under genetic control. The 
Greenland Gap site is only 2 m upstream from North Fork Patterson 
Creek, which is an open, typical Ill-habitat harboring a population of G. 
minus. It is probable that gene exchange between the populations has 
prevented the Greenland Gap population from developing the characters 
associated with its habitat type. Cunninghams, on the other hand, feeds 
immediately into Standing Stone Creek, a stream that lacks G. minus. We 
infer that Cunninghams is thus a highly isolated deme that has developed 
strongly habitat specific characters. No other locality appears to be as ex- 
posed to gene flow or as isolated as these two, although this factor has 
probably influenced character scores everywhere to some extent. 
DISCUSSION AND CONCLUSIONS 
A summary analysis of our data falls into three categories. First, a 
