GIS Habitat Modeling: Details of the Boosted Regression Tree Modeling Process for the Midwest Regional Fish Habitat Assessment

Fishery scientists must quantitatively assess habitats to understand the distribution, status, threats, and relative abundance of aquatic resources. Due to the spatially explicit nature of riverine habitats, geographic information systems (GIS) must be used to collate, archive, display, analyze, and model spatial data. Additionally, developing habitat assessment models that are spatially explicit allows for the interpretation of how terrestrial and aquatic factors interact to influence fish habitat conditions, including physical and chemical conditions, fish assemblage organization, habitat diversity, watershed characteristics, and socioeconomic forces. We will present the methods used to extrapolate, analyze, and visualize the results of the boosted regression tree models for the Midwest regional assessment. Specifically, we will present methodology used to evaluate stress and natural condition of aquatic habitats.  Stress and natural habitat quality are defined as the difference between expected current condition values and values under a scenario of zero stress or maximal natural habitat quality. We will also present an analysis of potential habitat loss due to modeled future climate conditions and discuss how aquatic resource managers can utilize the data produced to prioritize conservation actions.