Predictive Capabilities of GIS for the Distribution of Specialist Wetland Skippers (Lepidoptera: Hesperiidae) in South Carolina
Smith, Thomas P.
MetadataShow full item record
Little is known about the wetland specialist skippers (Lepidoptera: Hesperiidae) that are indigenous to the lower coastal plain of the southeast United States. The intermittent occurrence of these species across their range may be due to the elusive lifestyles and rarity of many of the species within this group. This is the first attempt to correlate specific wetland types from a GIS layer with known occurrence data of these skipper species within the limits of the lower coastal plain of South Carolina. These data show two groups of wetland specialist skippers: one specific to palustrine emergent wetlands (Group 1) and the other to palustrine forested wetlands (Group 2). These specific wetland types were checked for new records of these species in order to ground-truth initial predictions, and these data were combined to re-analyze the correlations of wetland types with species using a logistic regression. Those wetland types deemed significant due to their correlation with the occurrence data were then tested by randomly splitting the entire dataset into training and testing datasets that provided scores of accuracy based on the predictions of the wetland types. Group 1 species were well predicted based upon this single map layer, but Group 2 species could not be related to the wetland types used to predict their occurrence. The results suggest that the forested wetland category within the map layer is so widespread and too vaguely defined to pinpoint the habitat requirements of Group 2, while the much less common habitat type used to predict Group 1 seemed to be a good indicator of the occurrence of those species. The occurrence of common species also proves to be a valuable indicator of the presence of the less common species. The habitat requirements of each species are discussed individually.