Cave Isopod Ecology 
109 
washout (and less mortality) in large gravels. Thus, minimization of mor- 
tality (whether through natural selection or developmental adjustment of 
body size) should result in a concordance between body and gravel size 
distributions. Supporting this, previous work (Culver 1971, 1973) in- 
dicated that washout is due to encounters between two individuals, and 
thus is density-dependent. The second hypothesis is that interspecific 
competition is size-dependent. If the first hypothesis is true, then the ex- 
tension to interspecific competition should result in character displace- 
ment, and at least a partial discordance between body and gravel size 
distributions. 
Both laboratory and field data strongly support the first hypothesis. 
The artificial stream experiments (Table 2) clearly show that in the ab- 
sence of other factors gravel size and isopod size should be correlated. 
Furthermore, field data support this hypothesis (Fig. 1 , Table 5). In Bow- 
den Cave, both distributions are bimodal. In Glady Cave, large gravels 
are less frequent (Table 3), and C. cannulus is smaller than in Bowden 
Cave. In Harman Cave, small gravels predominate, and C. holsingeri is 
small. The low frequency of large gravels in Harman Cave may be the 
reason for the absence of C. cannulus. Only in Linwood Cave is there a 
discordance difficult to explain by the first hypothesis (Fig. 1, Table 5). 
Larger isopods should be present. However, Linwood Cave is separated 
from the range of C. cannulus by a major drainage divide. Thus, the most 
reasonable explanation for the absence of large C. cannulus is historical. 
The evidence for the second hypothesis is mostly negative. Artificial 
stream experiments suggest that size differences reduce interspecific com- 
petition (Table 3), at least for the larger species. On the other hand, field 
evidence provides no support for the character displacement hypothesis. 
The allotopic population of C. cannulus in Cave Hollow Cave is slightly 
but not significantly smaller than syntopic populations. Unfortunately 
for the testing of the hypothesis, C. cannulus in Cave Hollow Cave does 
not occur in a gravel-bottom stream area. Only four individuals were 
collected in three trips to the cave, and all of these were collected on 
bedrock in the stream (Holsinger, pers. comm.). Gravel-bottom riffles 
had large populations of the amphipod Gammarus minus. The size of C. 
cannulus in water flowing over limestone bedrock may be subject to op- 
timization, but we have no idea what that optimal size is. Thus, even if in- 
dividuals in Cave Hollow Cave were significantly smaller, its significance 
would be hard to interpret. Caecidotea holsingeri sizes also provide no 
support for character displacement. One allotopic population (Linwood 
Cave) is larger than all syntopic populations, and one (Harman Cave) is 
smaller than most syntopic populations. In addition, gravel size distribu- 
tions provide no support for character displacement. If character dis- 
placement were occurring, then discordances between gravel and isopod 
distributions should result in greater separation of isopod sizes. But, in 
the case of Glady Cave, sizes of the two isopod species are unimodal 
while gravel sizes are bimodal (Fig. 1, Table 5). 
