US-NATIONAL-WETLAND-INVENTORY CLASSIFICATIONS AS PREDICTORS OF THE OCCURRENCE OF COLUMBIA SPOTTED FROGS (RANA-LUTEIVENTRIS) AND PACIFIC TREEFROGS (HYLA-REGILLA)

In the Owyhee Mountains of southwestern Idaho, we compared the habitat characteristics of sites at which we captured Columbia spotted frogs (Rana luteiventris) and Pacific treefrogs (Hyla regilla) to habitat ch aracteristics of sites without frogs. Our primary objective was to det ermine if National Wetland Inventory classifications can be used to pr edict the presence of these species. Adult spotted frogs tended to be at palustrine, shrubscrub, seasonally flooded sites or at intermittent riverine, streambed seasonally flooded sites; they tended not to be a t palustrine, emergent, seasonally flooded or at intermittent riverine , streambed, temporarily flooded sites. Spotted frog sites also tended to have more submerged vegetation and algae and less grass and sagebr ush; they were more likely to be located at oxbows, pools, or ponds; a nd they, were more likely to have obvious hiding places than were site s without spotted frogs. Sites where treefrog adults were found tended to be lower in willow and higher in grass and emergent and submerged vegetation; they were more likely to be at a pond or pool than were si tes without treefrog adults. Treefrog larvae tended to be found at int ermittent riverine, streambed, seasonally flooded sites or at palustri ne, emergent, seasonally flooded sites they tended not to be at palust rine, shrubscrub seasonally flooded sites. Treefrog larval sites also were higher in silt, emergent vegetation, and algae and lower in sageb rush and willow, and they were more likely be located at a pond, oxbow , or pool than were sites without treefrog larvae. Although certain Na tional Wetland Inventory classifications were associated with frog pre sence, none could be used to predict with complete assurance the prese nce or absence of either species. Logistic regression models using hab itat measures were better at predicting the presence of amphibian spec ies than were models using National Wetland Inventory classifications; models using a combination of habitat measures and National Wetland I nventory classifications performed best Because of their ready availab ility in geographic information system data bases however, National We tland Inventory classifications may in some circumstances provide a va luable indicator of the likelihood of finding certain amphibian specie s. National Wetland Inventory classifications should be most useful fo r highly aquatic species in arid environments.