| Abstract |
- Characterization of canopy structure, the horizontal and vertical distribution of the tree crowns in a forest, is important for the management of forests in the Pacific Northwest. The canopy is an important habitat element for many wildlife species, canopy structure affects understory development, and influences various natural processes, such as the intensity of propagation of wildfire. Thus, improving
our understanding of canopy structures and trends can aid forest management. The overall goal of this study was to characterize vertical and horizontal canopy structure for multiple forest groups in western Oregon. The specific objectives were to: 1) characterize vertical and horizontal canopy structure for dominant forest types in western Oregon, 2) evaluate methods for measuring canopy cover and structure, 3) compare methods to predict forest canopy cover and vertical diversity using standard inventory measurements, and 4) predict bird species occurrence with different canopy diversity measures. I evaluated patterns of vertical and horizontal canopy structure and
understory cover along a successional gradient using 934 forested plots in western
Oregon. Observed data were from the USDA Forest Service Forest Inventory and Analysis (FIA) program from the 1995-97 survey on private and non-federal public lands. Patterns were examined for wet-conifer, wet-hardwood, and dry-hardwood forests. The upper tree canopy layer contributed the most to total cover except in the dry-hardwood stands, where the vertical distribution of tree cover was more evenly distributed. However, mean canopy cover rarely exceeded 85%, even in productive young conifer forests. Shade-tolerant species rarely made up more than 20% of canopy cover, even in the lower canopy layers and in stands> 100 yrs old. Contrary to expectations, percent cover of understory shrubs and herbs was not
substantially lower in young closed-canopy stands than in other stands. Ground-based measures of canopy cover on inventory plots were compared to predictions with regression models that regressed canopy cover on standard forest measurements, with estimates from aerial photography, and predictions with the forest vegetation simulator (FVS) program. Model predictions from inventory
measurements were within 15% of measured cover for> 82% of the observations. Standard inventory estimates of cover using 1:40,000 scale aerial photos were poorly correlated with ground-measured cover, especially in wet-hardwood (r=0.58) and dry-hardwood (r0.61) stands. FVS tended to underestimate cover by up to 50% in wet-conifer and wet-hardwood stands. The aerial photos and FVS equations used in this study are not recommended as surrogates for ground-based
measurements of cover. However, the level of accuracies of the predictive models
developed in this study may be adequate for some purposes. I compared fourteen measures of vertical structural diversity and layering
using the inventory plots. I then attempted to predict selected vertical diversity
indices from standard forest variables. Simpson's diversity index on tree heights
best differentiated among the range of vertical structure classes of the inventory
plots. I developed predictive equations for Simpson's height diversity index (SDI),
Foliage Height Diversity (FHD), and Canopy Height Diversity Index (CHDI), which used basal area, standard deviation of dbh, and stem frequency of size classes as the best variables. Predicted SDI values were within 0 15 units of calculated SDI for > 79% of the observations, predicted CHDI values were within 1.5 units for > 91% of the observations, except in the dry-hardwood stands (only 69%), and predicted FHD measures were within 0.2 units for > 85% of the observations among forest groups. The equations for FHD and SDI were applied to a wildlife-habitat database for western Oregon to determine if classification efficiency of existing models using CHDI to predict presence of bird species could be improved. The classification efficiency of bird-habitat association models improved for 33% and 66% of models for the Oregon Coast Range with the FHD and SDI variables, respectively. Models with FHD and SDI had improved classification efficiency for 18% of Cascade Range models. Although improvements in classification efficiency were less than six percentage points, future use of these diversity indices is warranted in place of CHDI when estimates of FHD and/or SDI are available and CHDI estimates are not. Four ground-based techniques for estimating forest overstory cover - line-intercept, spherical densiometer, moosehorn, and hemispherical photography - and estimates generated using FVS were compared across a range of stand structure
types. Canopy cover estimates for the four ground-based methods were not correlated with structure type. Differences among estimates of cover using FVS and the other methods did depend on the forest structure type. Differences among ground-based methods were primarily related to differences in angle of view. Although the line-intercept had the narrowest angle of view, the moosehorn provided the most conservative estimates of overstory cover. Regression equations were derived to allow conversion among canopy cover estimates developed with the four ground-based methods. The FVS calculated cover should not be used as a
substitute for ground-based measures in these forest types given that it was
consistently much lower (up to 70%) than estimates from the ground-based methods
within each forest-structure type. Overall, this study provides researchers and forest managers with new information and tools that can be applied across the forested landscape of Oregon. Models to predict canopy cover and diversity, and bird habitat can be substituted for field studies, assuming the accuracies of predictions are adequate for desired purposes. In field studies where ground-based cover measures are needed, the
moosehorn is recommended as the most conservative estimator of cover. For more
detailed canopy data, the line-intercept method is warranted. Modifying the line-intercept method to use fixed height intervals may be preferable to the use of three relative layers. This adjustment will allow for more direct comparisons of canopy
cover of layers among stands.
|
|---|
