Simulation of Stream Wood Source Distance for Small Streams in the Western Cascades, Oregon

The model, STREAMWOOD, is an individual-based stochastic model designed to simulate the dynamics of wood in small streams of the Pacific Northwest. We used STREAMWOOD to examine source distance as a function of tree fall regime and stand age. Our results suggest that source distance increased with stand age for the first 400 years of stand development and then declined. Simulated source distance for mature conifer forests (81 to 200 years old) were consistent with observed data, but simulated source distances for old-growth forests (201 to 1,000 years old) were below observed data. Further information on stand ages for the forests used in the observational study would refine the compassion with simulated data. Introduction One consideration of riparian forest management is the long-term recruitment of wood to the stream. The chance of a riparian tree entering the channel is related to its source distance (Van Sickle and Gregory 1990). We define source distance as the slope distance between the stream bank and the base of the tree perpendicular to the stream channel. Several studies have addressed the relationship between source distance and riparian forest width in the Pacific Northwest. McDade (1987) and McDade and others (1990) surveyed 39 streams adjacent to either mature conifer (80 to 200 years old) or old-growth conifer (> 200 years old) riparian forests in western Oregon and Washington. The source distance for 90 percent of the wood inputs was found to originate within 26 m for mature conifer and 36 m for the old-growth stands. McDade and others (1990) also modeled the source distances for riparian stands composed of uniform heights. The simulated source distance for 90 percent of the trees 40 m and 50 m tall was 32 m and 40 m, respectively. Van Sickle and Gregory (1990) developed a computer model that simulates the recruitment of stream wood from riparian forests with trees of mixed heights. They applied this model to a mixed-hardwood/conifer stand in the Oregon Cascades and found that 90 percent of the pieces originated within 18 m from the bank. The authors 1 An abbreviated version of this paper was presented at the Symposium on the Ecology and Management of Dead Wood in Western Forests, November 2-4, 1999, Reno Nevada. 2 Ph.D. Candidate and Professor, respectively, Fisheries and Wildlife Department, Oregon State University, 104 Nash Hall, Corvallis OR 97331 (e-mail: [email protected] and [email protected]) 3 Professor, Bioresource Engineering Department, Oregon State University, Corvallis OR 97331 (e-mail: [email protected]) Simulation of Stream Wood Source Distance for Small Streams in the Western Cascades, Oregon— Meleason, Gregory, and Bolte USDA Forest Service Gen. Tech. Rep. PSW-GTR-181. 2002. 458 also proved mathematically that, for a forest of uniform height, approximately three times as many trees would enter the channel falling directly towards the stream bank as opposed to falling randomly. These studies provide valuable insights into the relationship between width of the riparian zone and recruitment of wood into a stream. However, these studies give limited insight into how source distance varies with stand age. The purpose of this study was twofold: to compare the results of the simulations reported here with the results from this previous studies and to investigate changes in source distance through time as a function of stand age. Model Description We developed STREAMWOOD, a computer simulation model, to investigate the dynamics of wood in streams. (A brief description of STREAMWOOD is provided here; further information can be found at our Web site: http://www. fsl.orst.edu/lter/research/compplfr.htm). STREAMWOOD is an individual-based stochastic model that operates on an annual time step at the reach scale. Stream systems that can be simulated range from a single reach to a small basin. Stream wood dynamics represented in the model are tree entry, breakage, movement, and decomposition. Riparian forest inputs are generated either from a simplified forest gap model built within STREAMWOOD or from a user-specified input file. The model is run under a Monte Carlo procedure and the results are reported as average conditions per reach. The current version of STREAMWOOD was developed for fifth-order and smaller streams in the coniferous forests of the Pacific Northwest. Species considered include Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco), western hemlock (Tsuga heterophylla [Raf.] Sarg.), western red cedar (Thuja plicata Donn ex D. Don), and red alder (Alnus rubra Bong.). The forest model within STREAMWOOD was based largely on three existing forest–gap models: ZELIG (Pacific Northwest version by Steven Garman, Forest Science Laboratory, Corvallis Oregon), JABOWA (Botkin 1993, Botkin and others 1972), and CLIMACS (Dale and others 1986, Dale and Hemstrom 1984, Hemstrom and Dale 1982). Since the forest model adheres closely to the original design of Botkin and others (1972) and Shugart (1984), it inherits many of their limitations (Schenk 1996). Riparian Zone Definition in STREAMWOOD A simulated stream reach consists of a length, bankfull width, and a riparian forest adjacent to each bank. The dimensions of each riparian forest is the reach length by a width selected between 0 m to 100 m. A tree fall regime is associated with each riparian forest. In a random fall regime, trees have an equal chance of falling in any direction. In a directional fall regime, tree fall angle is normally distributed around a mean (and standard deviation) fall angle. A directional fall regime defined with a mean of 180 degrees and a standard deviation of zero forces all trees to fall directly towards the channel. Simulation of Stream Wood Source Distance for Small Streams in the Western Cascades, Oregon— Meleason, Gregory, and Bolte USDA Forest Service Gen. Tech. Rep. PSW-GTR-181. 2002. 459 Initial Conditions We conducted two simulations with STREAMWOOD using the forest model to grow identical riparian forests. Both simulations were run for 1,000 years and 200 iterations. In the first simulation, trees could fall in any direction (random fall), and in the second all trees were forced to fall directly across the channel (directional fall). Both simulations represented riparian forests in the Oregon Cascades at an elevation of 300 m. Soil moisture content was set such that the growth of red alder was negligible, but supported Douglas-fir, western hemlock, and western red cedar. The reach dimensions were 100 m long and 15 m wide. The thermal growth index was set at 1,500 growing degree-days (5.5oC base) with a standard deviation of 100 growing degree-days. The simulated riparian forests were 75 m wide and started with no trees. The riparian forest was divided into 31 intervals, or source distances, parallel to the stream bank. The first 30 intervals were 2 m wide, and the last interval was for recruitment of all pieces > 60 m from the bank. For each source distance, the mean number of input events that entered the channel during each 10-year period was calculated. An input event included any tree that contributed a piece at least 10 cm top diameter and 1 m in length to the stream. Results and Discussion Comparison with Previous Studies The first purpose of this study was to compare the results from the simulations with the results of previously published studies. Van Sickle and Gregory (1990) proved mathematically that, given a stand composed of trees of the same height, exactly 1/л less input events would result from trees falling randomly as opposed to falling directly across the channel. We extend this proof to stands composed of mixed heights. Consider a riparian forest containing trees of two different heights. The ratio of input events between random and directional fall would be identical (1/л) for both heights. The ratio for the entire stand would also be 1/л since it is the sum of all random fall inputs over the sum of directional fall inputs. Extending this argument, the ratio of 1/л would hold for any number of height classes. The input ratio described by Van Sickle and Gregory (1990) provided a simple test for evaluating STREAMWOOD’s performance. The two simulations reported had identical initial conditions except for the fall regime used. The ratio of the number of input events from the two simulations was determined (fig. 1). The mean ratio between the number of input events from random and directional fall was 0.307 with a standard deviation of 0.018. The mean input ratio from the simulations was 3.6 percent lower than the predicted value and could be due to small differences between the two simulated forests. The coefficient of variation was 5.8 percent and most likely could be reduced with an increase in the number of iterations. Thus, the performance of the model was consistent with the predictions of Van Sickle and Gregory (1990). Simulation of Stream Wood Source Distance for Small Streams in the Western Cascades, Oregon— Meleason, Gregory, and Bolte USDA Forest Service Gen. Tech. Rep. PSW-GTR-181. 2002. 460 0.2 0.25 0.3 0.35 0.4 0 200 400 600 80