A GIS framework for fish habitat prediction at the river basin scale

We present a geographic information system (GIS) framework to classify stream habitats and provide fish distribution predictions comprehensively at the landscape scale. Stream segments were classified into one of eighteen habitat types using three landscape attributes: stream size (three categories), stream quality (three categories), and water quality (two categories). An extensive literature search was undertaken to classify fish species into the same eighteen habitat types based on preferences for the three landscape attributes. We tested our framework in 39 sites throughout the upper Allegheny River basin in western New York. No difference was detected between observed and predicted numbers of fish species among stream habitats. Further, field collected bankfull width measurements, stream quality ratings, and water quality sampling results were largely consistent with predicted values. The habitat type expected to have the greatest fish species richness was large streams or small rivers with intact stream quality and suitable water quality. Our framework is rapidly applied, comprehensive, inexpensive, and built on widely available data thereby offering an efficient alternative to traditional field-based efforts for regional habitat classification and fish distribution prediction.