Fig 1.
Areas of karst are shaded and overlain by a 20 x 20 km grid, with karst grid cells in the Appalachian Mountains and in the Interior Low Plateau outlined in green. Black dots show caves with known troglobionts. Adapted from [31].
Table 1.
Factor loadings of soil characteristics in the 20 X 20 km grid for the study of the first three principle components.
Fig 2.
Frequency of occupancy of different groups in cells with at least one troglobiotic species.
Values for the Interior Low Plateau are shown in blue, and values for the Appalachian Mountains are shown in red.
Table 2.
Explanatory variables in the final model for each combination of taxonomic group and region (ILP = Interior Low Plateau, App = Appalachians Mounntains).
Table 3.
Cross-classification results from each model.
Prob level = value of the cutoff of the probability of observing presence used to classify each prediction into either an event (presence) or non-event (absence). Correct = number of events and non-events properly classified when comparing predicted to observed. Incorrect = number of non-events improperly classified as events and vice versa. Percentages Correct: overall percentage of observations correctly classified. Sensitivity: percentage of actual events correctly identified as such and is complementary to the false negative rate. Specificity: percentage of non-events which are correctly identified as such and is complementary to the false positive rate. False Positive Percentage: percent of predicted event responses that were observed as nonevents. False Negative Percent: percent of predicted nonevent responses that were observed as events.
Fig 3.
Maps of observed and predicted distribution of troglobiotic amphipods (largely the genera Stygobromus and Crangonyx) in the study area.
A. Observed distribution of troglobiotic amphipods in 20 x 20 km grid. B. Predicted probabilities of occurrence of troglobiotic amphipods in those grid cells that have observed troglobionts. C. Predicted probabilities of occurrence of troglobiotic amphipods in all grid cells with karst. See Table 2 for details of the model and Table 3 for goodness of fit.
Fig 4.
Maps of observed and predicted distribution of troglobiotic ground beetles (largely the genus Pseudanophthalmus) in the study area.
A. Observed distribution of troglobiotic ground beetles in 20 x 20 km grid. B. Predicted probabilities of occurrence of troglobiotic ground beetles in those grid cells that have been observed beetles. C. Predicted probabilities of occurrence of troglobiotic ground beetles in all grid cells with karst. See Table 2 for details of the model and Table 3 for goodness of fit.
Fig 5.
Maps of observed and predicted distribution of troglobiotic pseudoscorpions in the study area.
A. Observed distribution of troglobiotic pseudoscorpions in 20 x 20 km grid. B. Predicted probabilities of occurrence of troglobiotic pseudoscorpions in those grid cells that have been sampled. Because of poorness of fit (Table 3), predicted probabilities of occurrence of troglobiotic pseudoscorpions in all grid cells with karst are not shown. See Table 2 for details of the model.