Modeling Riverine Fish Habitat Conditions in the Midwest Region

The quality of fish habitats is strongly influenced by human land use and natural variation in landscape attributes.  With advances in GIS technology and improved landscape databases, there is considerable interest in constructing models that can be used to characterize the quality of fish habitats across a range of spatial scales.  We describe a modeling process that uses Boosted Regression Trees to characterize fish habitat quality at scales ranging from the stream segment, 12-digit HUC, and 8-digit HUC.  This process uses information on mapped landscape attributes and fish populations to: 1- predict fish population status (presence and abundance); 2- identify natural habitat features that influence fish populations; 3- identify anthropogenic stressors that influence fish populations; and 4- produce relative measures of natural habitat quality and underlying anthropogenic stress.  We will demonstrate this process for brook trout populations in the central Appalachians, as well as for brook trout and walleye in the Great Lakes region.  Finally, we will demonstrate how results from statistical modeling can be built into a spatially-explicit interactive system and used to support fish habitat management decisions.