Effectiveness of Using Summer Thermal Indices to Classify and Protect Brook Trout Streams in Northern Ontario

We tested five thermal indices for their ability to differentiate streams containing brook trout Salvelinus fontinalis from streams not containing brook trout in forested watersheds of the Precambrian Shield, northern Ontario, with the goal of identifying and protecting riparian areas of thermally sensitive trout streams during timber harvesting. Logistic regression was used to predict brook trout presence and absence, with maximum summer temperature, mean summer temperature, mean sampling temperature, mean maximum summer temperature, and thermal stability as independent variables. Brook trout streams were cooler and thermally more stable than non-brook-trout streams, but temperatures overlapped considerably between the two types of stream. Correct classification of streams ranged from 60.3% for summer temperature stability to 67.1% for maximum summer and mean sampling temperatures. The models yielded correct predictions more often for brook trout absence (∼80%) than for brook trout presence (≤50%) because streams with temperatures above lethal limits clearly precluded brook trout presence, whereas cooler temperatures merely indicated thermal suitability. In cooler streams, other factors, such as suitable spawning and rearing habitat and migration barriers, likely contributed to variation in brook trout presence. The specific prediction probabilities of the models could be used to assign management protection levels or identify additional sampling requirements necessary for determining brook trout distributions in streams with suitable temperatures.