142 
James L. Gooch and Jeffrey S. Wiseman 
is nonsystematic and unrelated to the epigean-hypogean scale as shown 
by the fact that many similarly ranked sites differ significantly in ratios. 
Tables 4 and 5 show this qualitatively: ratio variation due purely to 
habitat difference would produce a pattern of significant differences 
(asterisks) in comparisons of widely rank separated sites and non- 
significant differences (dashes) in comparisons of similar rank. However, 
asterisks and dashes are so interspersed as to indicate numerous signifi- 
cant differences between like ranked localities. 
Systematic variation lies in the association of ratios with habitat 
rank. If the eye and appendage size differences between ecotypes are also 
distinguishable within the single Form III ecotype, eye length ratio 
should increase and appendage length ratios should decrease with in- 
creasing (more epigean) rank. The predicted trends are roughly con- 
firmed in Table 3. Some ratios, however, do not conform to expectation 
for their rank. This is especially so for eye length in both sexes and 
pereopod 7 length in females, which reveal no discernible trends. 
Holsinger and Culver obtained appendage ratios in males of the 
three ecotypes. Their mean values, followed by ours in parentheses, are: 
antenna 1, I .720, II .644, III .570, (.576); pereopod 7, I .496, II .463, III 
.442 (.483); uropod 3, I .21 1, II .190, III .184 (.169). Our figures approx- 
imate the published ones for Form III amphipods except for pereopod 7, 
which has almost the relative length of Form I. Agreement is good con- 
sidering the wide range of ratios found in both investigations and the 
small sample sizes used in the earlier study. 
In 24 of 27 comparisons male appendage ratios are higher than 
female ratios. The sex differences are significant at the 0.05 level by the 
Mann-Whitney U test for all 3 appendages. Ruling out allometry, these 
data indicate generally greater appendage elongation in males than 
females. The effect does not extend to eye length, which does not differ 
significantly between sexes. Curiously, the three instances of higher ap- 
pendage ratio in females belong to the James Creek population. Unless 
parasitism or an unusual environmental factor was responsible (neither 
were evident), this suggests that relative appendage elongation between 
sexes is genetically labile and occasionally prone to variation in semi- 
isolated populations. 
The findings on all habitat associated characters are summarized in 
Table 6, which presents each locality rank ordered by its score or ratio, 
progressing to epigean typical values to the right. For each site there is 
considerable scatter in rank by both sex and character. Nevertheless the 
pattern of low rank for borderline type II habitats and high rank for fully 
open type III habitats does emerge. Emma Spring and Smoke Hole 
Spring, for example, rank near the hypogean pole for most characters 
and James Creek and Petersburg I and II usually rank near the epigean. 
At the foot of Table 6 overall rank order is given and each locality name 
is provided below in parentheses by its mean rank over all characters. 
This provides us with two scales of rank order— the habitat scale 
