Abstract.-In the southern Appalachian Mountains, the distributions of native brook trout Salvelinus fontinalis and introduced rainbow trout Oncorhynchus mykiss and brown trout Salmo trutta are presently limited by temperature and are expected to be limited further by a warmer climate. To estimate trout habitat in a future, warmer climate, we produced a regional map of wild trout habitat based on information from stream samples, expert knowledge, and suitable land cover. We then developed a quantile regression model of the elevation-latitude boundary for the present distribution of trout; this constitutes a more direct, spatially explicit approach to modeling trout distribution than the use of thermal limits. In combination with a lapse rate model, the boundary model was used to project future wild trout distributions over a range of higher temperatures. If the predictions of the Hadley Centre global circulation model (GCM) are assumed, about 53% of trout habitat would be lost; if the more extreme Canadian Centre GCM is used, 97% would be lost. With increasing temperature, fragmentation would increase, leaving populations in small, isolated patches vulnerable to extirpation because of the decreased likelihood of recolonization. The regional trout habitat map and the models produced here were useful for making these predictions, and the map could be used for assessing the impacts of other regional stressors.

Ahstrcrcr.-Wilderness areas in the Appalachian Mountains of North Carolina are set aside to preserve characteristics of both old-growth and second-growth forests and associated streams. Woody debris loadings, trout habitat, and trout were inventoried in three southern Appalachian wilderness streams in North Carolina by the basin-wide visual estimation tcchniquc. Two strcnms in old-growth wilderness areas contained more large woody debris (LWD, diameter > IO cm) and more and smaller pools and riffles than did a stream in a second-growth area managed as wilderness. Furthermore, the size distribution of woody debris in the second-growth stream was skewed to smaller size-classes than that in the old-growth streams. Brook trout Saivrlirzu.s,fhnfirzolis. rainbow trout 0ncorhynchu.s mykis.s, and brown trout S&no tr~tta in the three streams were always found in habitat units that had large amounts of LWD but were present in only 70-90'51 of the large number of units with little or no LWD. In the absence of high fishing pressure, the stream with large amounts of LWD supported higher trout density and biomass than the stream with little or no LWD. These old-growth streams provide a benchmark against which recovery of previously disturbed streams may be compared. Furthermore, if the goal for restoration of trout habitat is to recreate old-growth stream conditions, these two old-growth wilderness streams provide a basis for selecting appropriate amounts and sizes of LWD.