Testing Methods for Challenging the National Wetland Plant List: Using Tsuga canadensis (L.) Carr. (Eastern Hemlock) as a Case Study

This study explored methods for collecting and analyzing data during challenges to wetland ratings on the National Wetland Plant List (NWPL) to determine if study area size and landscape type affect wetland frequency and ratings. Data were collected in three different-sized study areas with different types of landscapes. Wetland frequency was calculated by using the original and an adjusted formula; and wetland ratings were predicted by using a Bayesian model. The original formula produced fewer hydrophytic ratings than the adjusted formula and the Bayesian model. In the smallest study areas (100 km2), wetland ratings varied with landscape characteristics. The same wetland frequencies and ratings were produced in moderately large (20,000 km2) and large (742,800 km2) study areas provided sample size was adequate. These results suggest that a wetland determination should be made for each sample unit based on the presence or absence of wetland indicators. Sample size should be large enough to achieve a confidence interval of 95 percent and a 3 percent - 5 percent margin of error. When wetland frequency is close to 33 percent, Bayesian models could provide support for wetland rating determinations. The National Technical Committee for Wetland Vegetation and the National Panel of the NWPL will work with challengers to create a study design appropriate for each species.