Search

Search Results

• 1. [Article] An examination of several methods of quantifying forest structure in headwater riparian forests of western Oregon

  Headwater streams are generally small, first-order streams that can comprise up to 80% (by stream length) of the drainage network in mountainous areas of the Pacific Northwest. These streams are intimately ...

  Citation × Citation Citation Abstract Copy Title: An examination of several methods of quantifying forest structure in headwater riparian forests of western Oregon Author: Haxton, Zane A. Headwater streams are generally small, first-order streams that can comprise up to 80% (by stream length) of the drainage network in mountainous areas of the Pacific Northwest. These streams are intimately connected with downstream reaches, serving as a source of sediment, woody debris, organic matter and nutrients. The surrounding forests can strongly influence the ecology of headwater streams by regulating microclimate, influencing nutrient and organic matter cycling process, and providing habitat for wildlife species. Central to the provision of ecosystem services by these headwater forests is their structure, the three-dimensional arrangement of trees, shrubs, snags and down wood in space. This thesis explored several methods of quantifying forest structure in headwater riparian forests of western Oregon. N-tree distance sampling, a sampling method where the nearest n trees to a sample point are measured, was compared with fixed plot and variable plot sampling for estimation of density and basal area in a simulation study using stem-mapped data taken from 8 headwater sites across western Oregon. In general, variable plot sampling gave the best statistical performance for estimation of basal area, while fixed plot sampling gave the best statistical performance for estimation of density. While n- tree distance sampling gave reasonable performance on some stem maps, the sampling method performed poorly for estimating density of populations with a clumped spatial pattern. Relative root mean square error of the best n-tree distance sampling estimator was up to 75% higher than for fixed plot sampling under these conditions. Neighborhood-based indices are a set of diversity measures that are based on the relationship between a reference tree and a certain number of nearest neighbors (i.e. trees to which it has the lowest horizontal distance). I compared the bias and relative root mean square error of several different methods of choosing reference trees for neighborhood-based index calculation, using stem-mapped data from the same 8 headwater sites. The selection of a random tree from a fixed plot gave the lowest absolute relative bias, but results with this method were quite variable, with relative root mean square error ranging from 33-67% for the species mingling index and 21-25% for the diameter differentiation index. The nearest-tree (selection of the nearest tree to the sample point) and azimuth (selection of the first tree from north in a fixed plot) methods were biased, but had lower variability, with relative root mean square error ranging from 31-64% for the species mingling index and 18-22% for the diameter differentiation index. The variable plot method gave highly variable results for all species combined, but performed well for some individual species. I close with an exploration of the use of measures of forest structure in characterizing the Trask Watershed, located in the headwaters of the northern Oregon Coast Range. The structure of the Trask Watershed is essentially even-aged, having regenerated in the aftermath of the Tillamook Burn. Douglas-fir and red alder are the most prevalent species, with western hemlock, western redcedar and noble fir locally abundant. While neighborhood-based indices were somewhat informative, they offered an impoverished picture of structural diversity in the Trask Watershed by ignoring the contribution made by tall shrubs such as vine maple and western hazelnut. Title: An examination of several methods of quantifying forest structure in headwater riparian forests of western Oregon Author: Haxton, Zane A. Headwater streams are generally small, first-order streams that can comprise up to 80% (by stream length) of the drainage network in mountainous areas of the Pacific Northwest. These streams are intimately connected with downstream reaches, serving as a source of sediment, woody debris, organic matter and nutrients. The surrounding forests can strongly influence the ecology of headwater streams by regulating microclimate, influencing nutrient and organic matter cycling process, and providing habitat for wildlife species. Central to the provision of ecosystem services by these headwater forests is their structure, the three-dimensional arrangement of trees, shrubs, snags and down wood in space. This thesis explored several methods of quantifying forest structure in headwater riparian forests of western Oregon. N-tree distance sampling, a sampling method where the nearest n trees to a sample point are measured, was compared with fixed plot and variable plot sampling for estimation of density and basal area in a simulation study using stem-mapped data taken from 8 headwater sites across western Oregon. In general, variable plot sampling gave the best statistical performance for estimation of basal area, while fixed plot sampling gave the best statistical performance for estimation of density. While n- tree distance sampling gave reasonable performance on some stem maps, the sampling method performed poorly for estimating density of populations with a clumped spatial pattern. Relative root mean square error of the best n-tree distance sampling estimator was up to 75% higher than for fixed plot sampling under these conditions. Neighborhood-based indices are a set of diversity measures that are based on the relationship between a reference tree and a certain number of nearest neighbors (i.e. trees to which it has the lowest horizontal distance). I compared the bias and relative root mean square error of several different methods of choosing reference trees for neighborhood-based index calculation, using stem-mapped data from the same 8 headwater sites. The selection of a random tree from a fixed plot gave the lowest absolute relative bias, but results with this method were quite variable, with relative root mean square error ranging from 33-67% for the species mingling index and 21-25% for the diameter differentiation index. The nearest-tree (selection of the nearest tree to the sample point) and azimuth (selection of the first tree from north in a fixed plot) methods were biased, but had lower variability, with relative root mean square error ranging from 31-64% for the species mingling index and 18-22% for the diameter differentiation index. The variable plot method gave highly variable results for all species combined, but performed well for some individual species. I close with an exploration of the use of measures of forest structure in characterizing the Trask Watershed, located in the headwaters of the northern Oregon Coast Range. The structure of the Trask Watershed is essentially even-aged, having regenerated in the aftermath of the Tillamook Burn. Douglas-fir and red alder are the most prevalent species, with western hemlock, western redcedar and noble fir locally abundant. While neighborhood-based indices were somewhat informative, they offered an impoverished picture of structural diversity in the Trask Watershed by ignoring the contribution made by tall shrubs such as vine maple and western hazelnut. Close