Predictive Models for Differentiating Habitat Use of Coastal Cutthroat Trout and Steelhead at the Reach and Landscape Scale

Steelhead Oncorhynchus mykiss and Coastal Cutthroat Trout Oncorhynchus clarkii clarkii are closely related species that are difficult to differentiate as juveniles and are commonly sympatric at the watershed scale. If Cutthroat Trout spawning and early rearing occur in small streams, it is often difficult to assess which parts of a stream network are dominated by each species. In this study I used catch data from 649 sites in coastal British Columbia to develop quantitative models of species presence and relative dominance as a function of stream size. An independent data set of 561 streams from the USA was used for the cross validation of models developed with data from British Columbia. The relative dominance of Cutthroat Trout or steelhead was predicted using logistic regression with stream and channel width, stream order, watershed area, unit runoff, ecoregion placement, and long-term mean annual discharge (LT mad) as predictor variables. The LT mad was the best predictor of Cutthroat Trout and steelhead dominance, with a correct classification rate of 98% for the entire species range. Costal Cutthroat Trout dominated in reaches or streams where LT mad was 630 L/s, and steelhead dominated in reaches where LT mad was >1,000 L/s. The models have practical application for predicting stream-bearing length and area used primarily by each species at the landscape scale of productive capacity relative to habitat threats.