118 
John R. Holsinger and David C. Culver 
both taxa, almost all species are restricted to a single faunal unit, and 
a majority of species in both groups are single-cave isolates. There is 
also a high percentage of regional faunal endemics among lumbriculid 
worms, snails, amphipods, and isopods; and more than half of the spe- 
cies in each of these groups are recorded from a single faunal unit. On 
the whole, comparatively fewer endemics are noted among flatworms, 
mites, millipeds, and diplurans; and no endemics are recorded for spiders 
and collembolans. 
Although the cave-limited fauna contains many highly localized 
endemics, there are a number of species — especially the few, select troglo- 
philes, troglobitic spiders, and collembolans, and some of the troglobitic 
amphipods, isopods, millipeds, and diplurans — that have relatively exten- 
sive ranges within the study area and are apparently good dispersers. 
These species are found in two or more of the seven faunal units, and at 
least two species, the highly vagile spiders Phanetta subterranea and 
Porrhomma cavernicolum , occur in all seven. 
The extent to which drainage basins or faunal units share cave- 
limited species is indicated by the data compiled in Table 12. Theoreti- 
cally, dispersal between basins could take place by some of the more 
vagile troglobites through caves and solution channels developed in 
parts of drainage divides and interfluves composed of carbonate rock, 
and through endogean habitats (e.g., deep ground litter and shallow 
underground compartments) and groundwater habitats (e.g., interstitial, 
hypotelminorheic) outside karst areas. At least three of the troglo- 
philes we have included in the cave-limited fauna ( Euhadenoecus fragi- 
lis , Erebomaster acanthina , and Nesticus carteri ) should be able to 
undergo limited dispersal through ecologically suitable epigean habitats. 
The frequency of species exchange between basins would be influenced 
by proximity of the basins as well as by the geological structure and 
geographic extent of drainage divides. The farther removed two basins 
are from each other, the fewer species they would be expected to share, 
and, by the same analogy, the closer they are, the more species they 
would be expected to share. Moreover, long common divides or inter- 
fluves, especially those containing carbonate rock, would be expected to 
facilitate more dispersal than do short divides or divides composed 
entirely of non-carbonate, clastic rock. 
Our data (Table 12) strongly support these assumptions and indi- 
cate clearly that, with few exceptions, adjacent basins have more species 
in common than far removed ones, and, furthermore, that basins on 
opposite sides of divides or interfluves containing carbonate rock gener- 
ally share more species than those separated by divides composed 
entirely of non-carbonate rock. 
Diversity-Area Relationships 
In the course of the present investigation it became increasingly 
obvious that some drainage basins in the study area had signifi- 
