To compare between overstory types among different slopes and make manipulative comparisons within overstory types, a control and three treatments were randomly assigned to the four plots:
Cut: In a 5 m x 8 m section adjacent to the stream, understory was cut every 2 months for the duration of the study. All plants <10 cm diameter at breast height (dbh) and >12 cm in height were cut at ground level and removed from the site.
5-m fence: An 8 m long fence was constructed from 1.4 mm mesh parallel to the stream azimuth 5 m slope distance from the stream edge to block litter moving down-slope from upslope of the 1 m tall fence.
10-m fence: An 8 m long fence was constructed from 1.4 mm mesh parallel to the stream azimuth 10 m slope distance from the stream edge to block litter moving down-slope from upslope of the 1 m tall fence.
Control: No cutting or fencing occurred in the control.
Traps to catch vertically falling litter (vertical traps) and litter moving laterally (lateral traps) were installed in August of 2003 at each site. Vertical traps, constructed from laundry baskets and suspended mesh (1.4 mm), were set to capture overstory and understory litter falling vertically from a height of >0.33 m above the forest floor. Vertical traps collected from an area of 0.26 m2 each. Lateral traps were set to collect litter moving down the hillslope rather than falling directly from vegetation. Lateral traps were constructed of PVC rectangles (0.33 x 0.5 m) oriented vertically with a mesh (1.4 mm) bag on the downhill side; each sampled from a length of 0.5 m.
The cut, control, and 5-m fence plots each had two vertical and three lateral traps edge in a given trap array deployed at the stream. The 10-m fence treatment had no lateral traps at the stream edge; rather, the 10-m fence and 5-m fence treatments each had an array of one vertical and three lateral traps located 5 m from the stream. These fifteen lateral traps and eight vertical traps, plus two additional vertical traps in the 10-m fence treatment (at 0 m and 10 m from the stream), made a total of 25 traps at each site. Due to logistical problems measuring direct inputs to streams, vertical traps at the stream edge were used to estimate inputs to streams. We also installed four vertical traps directly over the stream for direct determination of vertical litter input at two sites of each overstory type for comparison with streamside verticals.
Within each treatment at each distance from the stream, we systematically installed traps at longitudinal and lateral locations. Stream edge lateral and vertical traps were installed as close to bankfull width as possible. Within each trap array, lateral traps were installed at 2, 4, and 6 m from the downstream edge of the plot (longitudinal direction) and vertical traps were installed at 3 and 5 m from the downstream edge. Downed wood, tip-ups, root-wads, water, and rooted trees or snags were avoided during installation.
Our aim was to install lateral traps with the trap mouth positioned vertically (e.g., like a soccer goal). We measured and recorded (in degrees) the deviation from vertical (trap-angle) for each lateral trap. The deviation from horizontal of the mouth of each vertical trap (vertical trap slope) was also measured (in degrees). The angle of the first 0.5 m of slope (proximal slope) in front of each lateral trap was measured.
To test if understory was related to lateral inputs, we planned to compare the inputs to streamside lateral trap in the control and cut plots. Fewer lateral inputs in control lateral traps would indicate that the presence of understory vegetation was related to reduced lateral inputs. Comparison of streamside lateral traps in the control and 5-m fence plots tested if litter was moving > 5 m. Greater lateral inputs in control plots would suggest that litter moved further than 5 m and was collected in the trap. If litter was found to move more than 5 m, similar comparison of 5-m and 10-m fence plots could be made to establish if litter moved more than 5 m not adjacent to the stream.
Vegetation and Plot Sampling Measurements of vegetation and plot characteristics were conducted between June and September 2003 for each site. Each plot had an area of 240 m2 and was separated into six sections.
Trees and snags (>10 cm dbh) were identified to species and dbh was measured (in cm) for each tree within each section (Table 2). Canopy cover of overstory species (>2 m in height) vegetation was measured using a moosehorn (view angle of 13°) and categorized by vegetation type at seven systematically selected points within each plot, for a total of twenty-eight per site. The number of cross hairs covered by canopy (out of 25) for overstory deciduous, evergreen species and mixed were counted and recorded separately. Understory species canopy cover (>1 m in height) was also measured using a moosehorn at the same twenty-eight locations as overstory cover in each site. Understory woody vegetation cover, number of stems, stem cross-sectional area, basal or stem cover, and density were visually estimated in four randomly located 1-m^2 quadrats within the first 10 m closest to the stream (two in 0-5 m and two in 5-10 m distance) within each plot. The 0-5 m distance upslope from the stream is also referred to as the 0 m section, the 5-10 m distance is the 5 m section, and likewise for further distances. Visual estimates of groundcover (<12 cm in height), dead herbs and shrubs (dead understory), moss, and bare mineral soil percent cover, and large wood coverage were also made and recorded (in percent) in each quadrat to test the uniformity within sites and with increasing distance from the stream. Only wood pieces larger than 10 cm in diameter were included in large wood estimates. Categorical exclusions that required moving the location of each quadrat included areas where tip-ups, root-wads, water, or rooted trees or snags dominated the 1-m^2 at the pre-determined locations.
To characterize potential obstructions to lateral litter movement, we estimated spatial density of a given path along the ground using a visual density board (0.5 m x 0.4 m) held at two distances from the line along which traps were arrayed. The board was held at ground level at both 2 and 5 m distances upslope from each trap location. It had 20 equally spaced 10 cm diameter circles, and the portions of non-visible circles on the top and bottom half of the board were estimated as the index of obstruction for litter traveling that path and distance. Obstructive spatial density, measured as the obstructed area of the total area, was measured at six locations (3 points, 2 distances each) in each section in the 10 m closest to the stream in each plot.
Bankfull width was measured perpendicular to the direction of flow at the mid point of each plot . Slope was measured in 5 m slope-distance increments from the stream to 25 m and a value was recorded for each section. The slope was measured using a clinometer on the downstream boundary of each treatment to sight 5 m upslope.
In addition, large wood (bigger than 10 cm in diameter and longer than 1m in length) was measured and recorded in 5 m increments upslope along the downstream boundary of each plot. Each piece of wood was identified to species and measured for diameter where it crossed the plot boundary. Logs that lay across the plot boundary were also classified by being a) on the ground b) perpendicular to the plot boundary (within 45 degrees) or not and c) of a decay class less than 3. These three categories were combined to become the mean 1) total numbers of logs on ground, 2) obstructive (perpendicular to plot boundary) and on ground, and 3) low decay, obstructive, and on ground.
Control treatments from eight randomly selected sites (4 coniferous and 4 deciduous) were later re-sorted into seven categories: Deciduous leaves, deciduous-other, coniferous needles, coniferous-other, understory non-twig parts, twigs (overstory and understory), and leftover. The re-sorted control litter types were again weighed. Species and specific litter descriptions were recorded for each litter type.
Control resorted samples were composited into seasons by combining like trap types and litter types across months for chemical analysis. Seasons were assigned based on current understanding of different conditions in streams important to structuring food webs. September, autumn (October-December), winter (January-March), and spring/summer (April-August) were the categories used to bulk chemistry samples. For other seasonal comparisons, September was added to the autumn category and only three seasons were used. At the time of compositing, the most abundant types of each litter type in each season were recorded. These seasonal litter type samples were ground to a fine powder using a Wiley mill, ball mill and/or roller grinder. Approximately 3-5 mg of each litter type was weighed into tin capsules and analyzed for total carbon and nitrogen in a Costech ECS4010 elemental analyzer using atropine and acetanilide as standards. At least one-quarter of all samples were run in duplicate; on average, the duplicates varied from one another by 0.01 %N and 0.31 %C.