Invertebrate Cave Fauna 
125 
The basic units of analysis are the seven drainage basins defined 
above. Area was estimated in three ways, each with somewhat different 
biological interpretations and with its share of technical problems. The 
first measurement is the total area of the drainage basin. While rela- 
tively accurate, the measurement combines karst and non-karst areas. If 
most terrestrial cave species occur in shallow underground compart- 
ments (Juberthie and Delay 1981, and see elsewhere this paper) in karst 
and non-karst areas as well as caves, and if most aquatic cave species 
occur in interstitial habitats as well as caves (Henry 1978, and elsewhere 
this paper), then total area of the drainage basin is the appropriate mea- 
surement of area. The second measurement is that of the area underlain 
by soluble carbonate rock in each drainage basin. This should measure 
the area in which caves potentially occur. The measurement itself was 
obtained by finding the percentage of limestone on a series of randomly 
chosen USGS topographic quadrangles, according to Douglas’s (1964) 
mapping of exposed carbonate rock. The problem is that not all of the 
exposed carbonates are equally likely to have caves, because, as shown 
by Douglas, they include both limestone and dolomite, and the latter 
usually has significantly fewer and smaller caves depending on its com- 
position (see Holsinger 1975). The third measurement estimates the area 
underlain by caves by the total number of caves for each drainage basin. 
This measurement avoids the problem of differential cave development 
in different limestone strata, but adds the problem that the number of 
known caves is correlated with sampling intensity. 
We have limited our analysis to terrestrial troglobites that are 
endemic to a particular drainage basin for several reasons. First and 
most important, the terrestrial troglobite endemics are assumed to be a 
relatively homogeneous group with respect to their time of isolation in 
caves. This allows us to formulate and test the following hypothesis that 
terrestrial species were isolated in caves during the series of Pleistocene 
interglacials (immigrations) with extinctions also occurring. If the endemic 
terrestrial fauna is a more or less perfect record of these events, then the 
power series model should be a better fit than the linear model, and the 
exponent z of equation (1) should be around 0.26. Alternative hy- 
potheses of special interest are three. If the number of endemic terrestrial 
troglobites reflects sampling intensity, then the linear model should pro- 
vide a better fit. If a significant proportion of the terrestrial species 
arose not directly from epigean ancestors, but from subterranean colo- 
nization by other troglobites, then species numbers should depend on the 
amount of fragmentation of the limestone. Finally, if significant extinc- 
tions have occurred since the Pleistocene, the coefficient of area effect z, 
will be much greater than 0.26. 
The second reason for limiting analysis to terrestrial troglobites 
endemic to a basin is that non-endemic terrestrial species either can 
move between basins, unlike the endemics, or are actually several 
