Landscape analysis using GIS methodology for the evaluation and prediction of relationships of amphibian health in Missouri wetlands

It is important for state departments of transportation (DOTs) to make decisions regarding transportation route locations that minimize negative impacts to wetland fauna. Here a new methodology to quantify wetland health is developed using landscape characteristics for 49 wetlands in northern Missouri and relating them to wetland health. Wetland health was defined in this project as the presence of sensitive or rare amphibian species, such as tiger salamanders or northern crayfish frogs, which are very sensitive to habitat disturbance. The biology assessment involved in this project was performed as part of other research. Crops, forest, grass, and herbaceous land covers, length of stream, length of roads, length of flowpath from wetland to stream, change in elevation between wetland and nearest drainage channel are landscape characteristics that are considered because they are assumed to affect positively or negatively wetland habitat for amphibians. A commercial geographic information system (GIS) ArcGIS 9.2 was used to quantify the landscape characteristics. Two types of wetlands were identified: nonpairs and pairs. Non-pair wetlands were defined as those wetlands in which the 600 m buffer zone and beyond did not overlap with the buffer zones of other wetland. A pair wetland was defined as one in which the concentric rings of the buffer zones overlapped with the concentric rings of another wetland. Five buffers were generated around each wetland encompassing 0--300 meters, 0--600 meters, 0--900 meters, 0--1500 meters, and 0--2100 meters. A distance parameter was incorporated in the variable values because it was hypothesized that distance from the landscape feature to the wetlands could be a factor for the prediction of the health of the wetlands. The variables were then normalized which allows for the addition and subtraction of dimensionless landscape characteristics with different units (e.g., length of roads, grass area). Multiple linear regression analyses were performed at the different spatial scales mentioned above, to test the relationship between wetland health and landscape variables. It was found that proximity to roads negatively affects wetland health up to a distance of 900 meters, while crops can have a negative effect up to a distance up to 2100 meters.