SETTING REALISTIC OBJECTIVES:
VEGETATION INVENTORY AND MONITORING
AT SHENANDOAH NATIONAL PARK
Duane R. Diefenbach
National Park Service
Northeast Region, Philadelphia Support Office
Stewardship and Partnerships
U.S. Custom House
200 Chestnut Street
Philadelphia, PA 19106
The National Park Service is committed to inventorying
and monitoring the natural resources under its stewardship. The work
reported here contributes to the refinement of the process of data collection
so that it can provide the specific information needed for the optimal
management of those resources.
In March of 2000, a group of natural resource experts met to develop
objectives for Shenandoah National Parks Vegetation Inventory
and Monitoring Program. This report combines two papers about that project.
Carolyn Mahan, Assistant Professor of Biology at Penn State University,
reported on the workshop itself, which resulted in the identification
of specific management and sampling objectives in three areas of interest:
general forest trends, (e.g., tree species composition and tree growth
rates.), forest health (e.g., trends in hemlock woolly adelgid infestation),
and special and unique ecosystems and species (e.g., trends in abundance
of endangered plant species).
Duane Diefenbach, U.S. Geological Survey, Pennsylvania Cooperative Fish
and Wildlife Research Unit, conducted a statistical evaluation of vegetation
data collected at the park from 1997 through 2000 to determine whether
the objectives stated at the workshop could be met, and recommended
adjustments to the sampling design.
The data used in the evaluation were collected as part of the parks
Long Term Ecological Monitoring System (LTEMS) program from 1987 to
2000 to estimate basal area (m2/ha) of trees (>5 m tall), stem density
(stems/ha) of shrubs and saplings (1-5 m tall), and stem density (stems/ha)
of seedlings (woody vegetation <1 m tall). Also, data collected at
the Big Meadows site were used to estimate changes in shrub coverage
before and after treatment by burning. The coefficient of variation
(CV = standard error/mean x 100%) was used as a measure of the precision
of an estimate. A CV < 10% is generally considered necessary for
research, a CV < 25% is recommended for management, and a CV 50%
is usually sufficient for pilot studies. The species for which basal
area and stem density were calculated were determined in consultation
with park staff. All forest cover types were sampled >2 times during
1987-2000, although they were not sampled during the same year such
that parkwide estimates for any given year could be calculated. These
data provided variances that were incorporated into a power analysis
to assess whether the current LTEMS and Big Meadows sampling designs
could meet stated inventory and monitoring objectives.
The following objectives, established during the workshop, were evaluated:
1. Data collected for the LTEMS program should ensure a 90% probability
of detecting a >50% change in the basal area or stem density of any
woody plant species (in a given size class) within any one forest cover
type over a five-year period ( =0.20). The ability of current sampling
efforts to meet this objective were assessed by calculating power curves
for tree basal area, shrub and sapling stem density, and seedling stem
2. Data collected for the LTEMS program should ensure an 80% probability
of detecting a >20% change in the coverage of a particular exotic
species parkwide over a five-year period ( =0.20). The ability of the
current LTEMS program at Shenandoah NP to meet this objective was assessed
using the power curves calculated above for changes in seedling and
sapling stem density of tree-of-heaven (Ailanthus altissima). Stem density
was used as an indicator of areal coverage for this species.
3. Data collected for the LTEMS program should ensure an 80% probability
of detecting a >20% change parkwide in species affected by disease
or insects over a five-year period ( =0.20). The ability of the current
LTEMS program at Shenandoah NP to meet this objective was assessed using
the power curves calculated above for changes in stem density of flowering
dogwood (Cornus florida) and basal area of all oak species.
4. Monitoring of shrub coverage at Big Meadows should ensure a 95% probability
of detecting a >40% reduction in shrub coverage over a five-year
period ( =0.15). The program TRENDS was used to estimate the statistical
power to detect these changes.
For basal area, most CVs were <40% for species in forest cover types
where they were dominant (e.g., northern red oak [Quercus rubra] in
northern red oak cover types). Declines in oaks and the decline of Virginia
pine (Pinus virginiana) and pitch pine (Pinus rigida) were evident from
the changes in estimated basal area between sampling periods. Because
most basal area measurements are >20 m2/ha for species in their primary
forest cover types (e.g., yellow poplar in cove hardwoods, northern
red oak in chestnut oak cover type, etc.), current sampling effort should
have >90% power to detect changes in basal area of 50% for dominant
For stem density of shrubs and saplings, most CVs were >50% (range
31-1,169%). The power analysis suggested that stem density changes of
>2,000 stems/ha had >90% probability of being detected. Because
most stem densities during both sampling periods were <1,000 stems/ha,
current sample sizes are inadequate to detect important changes in stem
density of shrubs and saplings.
Stem density of seedlings was extremely variable, and the power analysis
suggested that only extremely large changes in stem density (>70,000
stems/ha) could be detected under the current sampling effort. Moreover,
large enough sample sizes likely cannot be obtained to meet stated objectives
because of the inherent variability of these data.
Increases in stem density for tree-of-heaven >1m tall (sapling) ranged
from 0-143 stems/ha (Appendix G). The ability to detect such changes
is poor (power < 70%) even if sample sizes were tripled.
Shrub stem densities for flowering dogwood ranged from 15.7 to 536.1
stems/ha. Under current sampling efforts, power was estimated as >80%
for changes >1,000 stems/ha. Consequently, the sampling effort would
have to increase 2-3 times current levels to detect ~100% changes in
The effect of gypsy moth on oak abundance, as measured by changes in
basal area for all oak species, has a good chance of being detected
under current sample sizes. Mean stem densities of oak saplings ranged
from 0 to 871 stems/ha, and thus the ability to detect only large changes
in stem densities (>1,000 stems/ha) for saplings will likely have
The current sampling design for Big Meadows provided estimates of total
shrub coverage (all species combined) and of changes in shrub coverage,
with CVs < 20%. Although estimates of coverage for individual shrub
species were not precise (CVs > 30%), biologically important changes
in overall shrub coverage should be detected under the current sampling
To meet the monitoring objectives developed at the workshop, recommendations
for the most important changes to the LTEMS program at Shenandoah NP
are listed here. Additional recommendations are detailed in the report.
1. A sampling design needs to be implemented that will permit parkwide
estimates of vegetation parameters for a given point in time. Presently,
changes in basal area or stem density can be estimated within each forest
cover type, but cannot be estimated across all cover types for the same
time period because each forest cover type is visited in a different
2. Requirements to monitor the spatial distribution of forest cover
types should be investigated before implementing changes to the sampling
design. Traditional stratified sampling designs cannot incorporate changes
in the distribution of cover types over time.
3. Sample sizes need to be increased such that all strata contain >1
plot. Sample sizes overall may need to be increased, depending on the
selected sampling design, to meet objectives for detecting changes in
stem density of shrubs and saplings.
4. Trees within plots should continue to be permanently marked with
unique identifiers to reduce misidentification and data collection errors.
5. An electronic field-based data entry system should be fully implemented
to speed data collection, reduce data entry errors, and eliminate transcription
errors that may occur with a paper system.
6. The purpose and need to collect seedling stem densities should be
reviewed. It is unlikely that it will be possible to obtain adequate
sample sizes to detect biologically important changes in seedling density
The entire report
is available to view or download in PDF file format. Using PDF files
requires the free Adobe Acrobat Reader. If you do not already have it
installed on your computer, you may download it now. Download
a pdf file, click on
this icon in the toolbar of the pdf window: .
This will allow you to save the file on your computer. If you want to
copy or print only a small part of the saved file, click on this icon
to select the desired text: .
The full report
is contained in one pdf file. When you open this file, you can easily
select any part you wish to view by clicking on the Bookmarks tab to
the left of the text. There you will find a clickable Table of Contents.
If the Bookmarks tab is not showing, you can make it visible by clicking
on this icon
in the pdf toolbar.