Each study site was classified in terms of stream order (firs, second, or third), stand age (old-growth or mature timber), and average steepness of side slopes (steep slopes >25 degrees, gentle slopes less than 25 degrees). These characteristics were pre-established so that selected sites would fit a 3 X 2 X 2 factorial design with a minimum of three sample sites (replicates) in each cell of the matrix. The choice of 25 degrees for categorizing slope steepness was based on our field observations that conspicuous sliding occurs primarily on slopes steeper than 25-30 degrees. A total of 1258 debris pieces ere sampled: 619 conifer and 2 hardwood pieces in old-growth stands and 551 conifer and 86 hardwood pieces in mature stands.
Dependent variables (Table 1) had highly skewed distributions, and nonparametric methods were used throughout to test for location differences among distributions classified by various stand and geomorphic attributes.
The length of stream sampled at each site ranged from 0.4 to 2 km and was determined by the distance required to locate 30 trees (sources) that had provided pieces of coarse woody debris to the stream; minimum diameters were greater than 10 cm at the small end and lengths were greater than 1 m. Coarse debris includes tree tops, large limbs, fragments of snags, and whole trees. Sample size (30) was a compromise between the need to properly characterize a site and to keep stream length short enough that geomorphic conditions (e.g., stream order, floodplain width) would not vary greatly. The origin of each piece of debris within or straddling the stream was determined. Pieces not identifiable as to source (47.7% of pieces encountered at the study sites) were not included in the study; they were usually short fragments that were quite mobile at high flows. Source distance was measured from the origin to the stream bank along a line perpendicular to the channel. To evaluate the distance moved downslope by debris, we also measured the origin to piece distance (Fig. 1). Separate measurements were made of distances on bench and hillslope areas and then summed to give a total slope distance from origin to stream (source distance) or origin to piece, ass appropriate. A bench was defined as the relatively flat floodplain and terraces adjacent to the stream. Side-slope steepness was measured with a hand-held clinometer. Species diameter, total piece length, and length actually in channel were recorded for pieces with discernible origins. In cases of uprooted trees, the distance from the center of the root pit to the stream bank was measured. In cases where a fallen tree delivered more than one piece of debris to the stream, all of these pieces were inventoried separately, but their source was counted as one. Debris pieces that straddled the stream but were identifiable as to source location were included in the study because they will eventually enter the steam.
Twenty-seven of the Cascade Range streams are in the central western Cascades of Oregon, in the H.J. Andrews Experimental Forest or in the nearby Hagan Block and Middle Santiam research natural areas and Three Sisters Wilderness Area. Nine of the remaining Cascade streams are close t the Wind River Experimental Forest near Carson, Washington, and one is in Mount Rainier National Park. Forests bordering the streams are dominated by Douglas-fir, western hemlock, and western red cedar, which are characteristic of the Tsuga heterophylla Zone.
Valleys in both the Cascade and Coast ranges are typically V-shaped with steep side slopes and narrow valley floors along first through third order streams. Hillslopes adjacent to sampled streams range in steepness from 3 to 40 degrees, with first and second order streams having significantly steeper side slopes than third-order streams.
