Use of a watershed-scale GIS model to predict responses of 50 Wisconsin stream fishes to climate warming

We estimated potential effects of climate warming on the distribution of stream fishes in Wisconsin.  We developed a GIS database of 58 measured geology, geomorphology, stream network position, land cover, and climate variables plus 11 modeled stream flow and water temperature variables at the watershed, riparian, or channel spatial scale for all stream reaches in the state (1:100,000 scale; 86,898 km length) coupled with fish collections from 1,628 locations.  We used Classification Tree analyses to develop predictive models for 50 widespread stream fish species.  Model accuracy ranged from 56 to 93.5% in validation tests.  We then ran the species models under current climate conditions and three warming scenarios for the entire state.  Air and water temperature increases were projected to cause declines in the occurrence of all 3 coldwater and 16 coolwater fishes as well as 4 of 31 warmwater species.  Four warmwater species were predicted to have little change and the remaining 23 to expand.  Although the same number of species declined as increased, total losses in stream length inhabited by declining species were far greater than gains by increasing species.  Our findings suggest that climate warming may reduce fish diversity and eliminate sport fisheries in many Wisconsin streams.