FOREST HEALTH MONITORING PLOT DESIGN
AND LOGISTICS STUDY
August 1991
Edited by
Kurt Rjitters
ManTech Environmental Technology, Inc.
Research Triangle Park, North Carolina
Michael Papp
Lockheed Engineering & Sciences Co.
Las Vegas, Nevada
David Cassell
ManTech Environmental Technology, Inc.
Corvallis, Oregon
John Hazard
Statistical Consulting Service
Bend, Oregon
Contract 68-DO-0106
Project Officer
Barry E Martin
Atmospheric Research and Exposure Assessment Laboratory
U.S Environmental Protection Agency
Research Triangle Park, North Carolina
OFFICE OF RESEARCH AND DEVELOPMENT
U S. ENVIRONMENTAL PROTECTION AGENCY
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
RESEARCH TRIANGLE PARK, NORTH CAROLINA
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
LAS VEGAS, NEVADA
ENVIRONMENTAL RESEARCH LABORATORY
CORVALLIS, OREGON
Research sponsored by the U.S Environmental Protection Agency under Contract No. 68-02-4444, 68-DO-0106,
and 68-C8-0006 to ManTech Environmental Technology, Inc., Contract No. 68-CO-0049 to Lockheed Engineering
and Sciences Co., and cooperative agreement CR81470 with the Environmental Research Center "of the University
of Nevada at Las Vegas.
-------
NOTICE
The information in this document has been funded wholly or in part by the U.S. Environmental Protection Agency
It has been subjected to the Agency’s review, and it has been approved for publication as an EPA document.
Mention of trade names or commercial products does not constitute endorsement or recommendation for use
This report should be cited as follows.
Riitters, K., Papp, M., Cassell, D., and J. hazard, eds. 1991. Forest health monitoring plot design and
logistics study. EPA/ÔC}OI_-_I_. U.S. Environment.al Protection Agency, Office of Research and
Development, Research Triangle Park, NC
Contributing authors in alphabetical order:
David Alerich’
Sam Alexander
Robert Anderson
William Bechtold 3
David Cassell 4
Steven Clime 4
Aiisa Gallant 3
John I-Iazard 6
Robert Kucera’
Beverly Laws
Timothy Lewis 9
Mohammed Miah 9
Margaret Miller -Weeks’ 0
Michael Papp 9
John Pembertoti’
Kurt Rntters’
Rick Van Remortel 9
‘U.S. Forest Service, Radnor, PA.
2 Virginia Polytechnic Institute and State Uruversity, Blacksburg, VA.
3 U.S. Forest Service, Asheville, NC.
4 ManTech Environmental Technology, Inc., Corvallis, OR.
‘Colorado State University, Ft. Collins, CO.
‘Statistical Consulting Service, Bend, OR
ManTech Environmental Technology, Inc., Research Tnangle Park, NC.
‘Oregon State University, Corvallis, OR.
‘Lockheed Engineermg & Sciences Company, Las Vegas, NV.
U.S. Forest Service, Durham, NH.
-------
CONTENTS
FIGURES. 1
TABLES . 2
EXECUTIVE SUMMARY . 4
SPECIES NOMENCLATURE 8
CONVERTING FACTORS FOR COMMON UNITS OF MEASUREMENT 9
ACKNOWLEDGEMENTS 10
1 PLOT DESIGN AND LOGISTICS 11
LI CONTENT AND ORGANIZATION. 11
1.2 RATIONALE AND OBJECTIVES .. 11
1.3 APPROACH TO THE STUDY .... 12
2 SITE AND STAND DESCRIPTIONS 16
2.1 SITE DESCRIPTIONS 16
2.2 STAND DENSITY 17
2,3 SPECiES COMPOSITION 17
2.4 STAND STRUCTURE 19
2.5 SOIL AND LANDFORM 21
3 COMPONENTS OF VARIANCE AND SAMPLING EFFICIENCY 24
3.1 INTRODUCTION 24
3.2 SAMPLING EFFICIENCY 24
3.2. 1 Sample Variances 24
3.2.2 Cost Coniponents 25
3.2.3 Sample Size Calculations 26
3.2.4 Important Considerations 27
3.3 RECOMMENDED ALLOCATIONS 27
3.3.1 Visual Symptoms 27
3.3.2 Mensuration 28
3 3.3 Soils 28
3.3.4 Foliage 28
3.3.5 Vegetation Structure 28
33.6 PAR 28
3.4 SPATIAL ANALYSIS 28
3.4.1 Soils 29
3.4.2 PAR and Vegetation Structure 29
3 4 3 Spatial Statistical Methodology 29
3.4.4 Results of the Spatial Analyses 29
4 SYNTHESIS OF LOGISTICS ANALYSES .. 38
4.1 INTRODUCTION 38
11
-------
4.2 STAFFING AND PERSONNEL .
38
4.2.1 Description 38
4.2.2 Discussion and Recommendations 38
4.3 PROCUREMENT AND INVENTORY CONTROL 39
4.3.1 Description 39
4.3.2 Discussion and Recommendations 40
4.4 TRAINING 41
4.4.1 Description 41
4.4.2 Discussion and Recommendations 41
4.5 RECONNAISSANCE 41
4.5.1 Description 41
4.5.2 Discussion and Recommendations 41
4.6 SAMPLING 42
4.6.1 Description 42
4.6 2 Discussion and Recommendations 43
4.7 COMMUNICATIONS 46
4.7.1 Description 46
4 7.2 DiscuSsion and Recommendations 46
4.8 SAFETY
4.8 1 Description
4.8.2 Discussion and Recommendations 47
4.9 SUMMARY 47
6 SOIL PRODUCTIViTY
6 1 BACKGROUND AND OVERVIEW
6 2 STUDY OBJECTIVES
6.3 MATERIALS AND METHODS
6 3.1 Field Methods
6.3.2 Statistical Methods
6.4 RESULTS AND DISCUSSION
6.4.1 Optimum Allocation of Resources
6.4.2 Semivanograrns
6.4.3 Other Sources of Variation
6.5 RECOMMENDATIONS
8 VEGETATION STRUCTURE 68
8.1 INTRODUCTION 68
5 VISUAL
SYMPTOMS
5.1
INTRODUCTION
5.2
DESCRIPTION AND APPLICATION OF THE MEASUREMENTS
5.3
SUMMARY OF VISUAL SYMPTOMS
5.4
ANOVA OF VISUAL SYMPTOMS MEASUREMENTS
5.5
ANOVA OF MENSURATION VARIABLES
• 48
• . 48
• . 48
• . 49
• . 53
• 55
55
7 FOLIAR NUTRIENTS AND CHEMICAL CONTAI 4INANTS
7.1 INTRODUCTION
7.2 METHODS
7.3 STATISTICAL ANALYSIS
7.4 RESULTS
7.5 DISCUSSION
• . • 56
• . . 56
57
• . • 57
• • 57
• • • 58
• • • 58
• . . 58
• 58
• . . 58
• • . 59
• . 64
• . 64
• 64
• . 65
• . 65
66
iii
-------
68
68
68
69
69
70
70
71
71
71
71
7 ]
71
73
74
74
74
74
74
74
76
77
77
77
77
79
8 1
82
86
87
10 REFERENCES
APPENDIX A
91
96
8.2 METHODS
8.2. 1 Vegetation Structure Measurement Methods
8.2.1.1 Ocular method
8.2 1.2 Pole method
8.2.1.3 Measurement variation
8.2.2 Statistic.al Analysis
8.2.2.1 Method comparisons
8.2.2.2 Measurement variation
8.2.2.3 Spatial variation
8.2.2.4 Sample allocation
8.3 RESULTS
8.3.1 Method Comparisons
8.3.1.1 Percent foliage occupancy differences
8.3.1.2 Measurement variation
8.3.2 Pole Method
8.3.2.1 Spatial variation
8.3.2.2 San5ple allocation
8 3.2.3 Time requirements
8.4 DISCUSSION
8.4 1 Method Comparison
8.4.2 Measurement Variation
9 GROWTH EFFICIENCY
9.1 INTRODUCTiON
9.2 METHODS
9.2.1 PAR Measurements
9.2.2 Statistical Analysis
9.3 RESULTS
9.4 DISCUSSION
9.5 MEASUREMENT QUALITY OBJECTIVES.
9.6 LOGISTICS
iv
-------
FIGURES
Figure 1-1. Field study plot locations in New England and Virginia . 13
Figure 3-2. Sampling design for tree and foliar measurements 14
Figure 1-3. Sampling design for soil measurements 15
Figure 1-4. Sampling design for vegetation structure and PAR measurements 15
Figure 4-1. Organizational chart for field staffing 39
Figure 4-2. A typical day’s sampling sequence 44
Figure 4-3. Flow of information to arid from the regional project leaders 46
Figure 8-1. Relationships between stressors, response indicators, and biotic integrity endpoint for the
vegetation structure measurements 69
Figure 8-2. Frequency of differences in total foliage occupancy between the ocuhr and pole measurements by
the presence-absence method 72
Figure 8-3. Frequency of differences in total foliage occupancy between the ocular and pole measurements by
the summati9d method 72
Figure 8-4. Examples of semivanograms for three foot-levels from pole measurements by the summation
method in Virginia 75
Figure 8-5. Examples of semivanograms for three foot-levels from pole measurements by the summation
method in New England 75
Figure 9-I. Ambient PAR in relation to time of day and sky condition 78
Figure 9-2. What is meant by intercepted (Qi) arid absorbed (Qa) radiation 78
Figure 9-3. Details of PAR sampling design 79
Figure 9-4. Plot median percent transmitted PAR 83
Figure 9-5. Repeatability of PTPAR in training, testing, and field auditing 88
Figure 9-6. Repeatability of PTPAR in crew remeasurenients 89
Figure 9-7. Frequency of differences in PTPAR between trainer and crew 89
Figure 9-8. Measurement quality objectives for subplot and plot mean PTPAR 90
Figure 9-9. Field experiences with MQOs based on crew remeasurements 90
-------
TABLES
Table I-I. Summary of measurements available for each study plot location 14
Table 2-I. Descnptions of field plot locations 16
Table 2-2. Descriptions of forest types 17
Table 2-3. Mensuration of overstory and understory trees by plot 18
Table 2-4. Number of species and their proportional abundance by plot 19
Table 2-5. Frequency of over-story trees by size class and plot 20
Table 2-6. Slope, physrography, and soil parent material by plot 21
Table 2-7. Soil classification characteristics by plot 22
Table 2-8. Soil water characteristics by plot 23
Table 3-1. Sample units for different sampling stages and models 24
Table 3-2. Cost scenarios for each region 25
Table 3-3. Sample size equations for four-stage sampling 26
Table 3-4. Optimum sample sizes for scenarios of costs and variances 30
Table 3-5. Response variable definitions 37
Table 4-I. Summary of time required for crew activities 42
Table 4-2. Time allocated for data collection by measurement 43
Table 5-1. Number of measured trees and species average crown ratings in New England 50
Table 5-2. Number of measured trees and species average crown ratings in Virginia 51
Table 5-3. Number of sample trees and average crown ratings for all species by plot 52
Table 5-4. ANOVA of visual symptoms variables for t o sampling models 54
Table 5-5. ANOVA of mensuration variables 55
Table 6-i. Field plot soil variance components by region . . 60
Table 6-2. Soil parameter imprecision and bias pooled across regions 61
Table 7-1. Fohar chemistry parameters and analytical techniques 65
Table 7-2. Percent of total variance from sugar maple variance components 67
Table 7-3. Percent of total variance from loblolly pine variance components 67
Table 8-I. Pbysiognoinrc classes for vegetation structure 68
Table 8-2. Ground cover substrates 68
Table 8-3. Analysis of hypothetical pole measurements using two calculation methods 70
Table 8-4. Pole method measurement variance from training session remeasurement.s 73
Table 8-5. Time required to complete pole measurements 76
Table 9-1. Ambient PAR and sky conditions by plot 80
Table 9-2. Unadjusted means of ln(PTPAR+ 1) by plot and subplot 8]
Table 9-3. Linear models and expected values of mean squares for the ANOVAs 83
Table 9-4. Two-factor ANOVA of PAR by plot before and after time series analysis 84
Table 9-5. Three-factor ANOVA of PAR by forest type before and after time senes analysis 85
Table 9-6. Time required for PAR sampling 88
Table A-i. Descriptive data by subplot at each location 97
Table A-2. Over-story numbers of trees, basal area, and average diameter by species 101
Table A-3. Means and standard deviations of elemental concentrations in sugar maple foliar tissue 105
Table A-4. Means and standard deviations for elemental analyses of loblolly pine foliar samples 110
2
-------
THIS PAGE WAS INTENTIONALLY LEFT BLANK
3
-------
EXECUTIVE SUMMARY
BACKGROUND
The Environmental Protection Agency’s Environmental Monitoring and Assessment Program (EMAP-Forests)
has joined the U.S. Department of Agriculture Forest Service and other government agencies in an effort to monitor
and assess the condition of the nation’s forested ecosystems in relation to natural and manmade stresses A long-
term and multi-tiered strategy for morutoring and assessment includes elements for detecting, evaluating, and
explaining changes in forest condition. To improve the efficiency of monitoring, part of the strategy is to test and
optimize field measurement procedures.
A preliminary set of mea.surernents was chosen for testing in the detection phase of monitoring based on
workshops, a review of the literature, expert opinion, and reports from studies done elsewhere. The interagency
Forest Health Monitoring (FHM) program is now conducting research to expand that set and to verify the
capabilities of measurements to accurately represent and respond to changes in forest condition over time. Research
is needed to optimize the deployment of selected measurements, because any per-unit cost reductions will be
multiplied many times in a nationwide program.
Thus, the FHM program conducted a field test of plot design and logistics for previously-selected measurements
in 1990. Not all of the possible measurements were tested, and not all of the questions that have been asked about
the selected measurements were asked in this study. The objectives of the field test were:
• to evaluate plot design and subsampling procedures,
• to quantif r time and resource requirements,
• to assess the relative efficiency of competing methods in some cases, and
• to supply mformat on to improve the national FHM program.
The purpose of this report is to summarize the results that were obtained Recommendations are made to guide
irther planning of the momtoring program.
Section 1 provides an overview of the rationale, objectives, and approach to the field study. Section 2 describes
the site and stand characteristics of the forests that were sampled. Sections 3 and 4 are the main focus of the report,
they summarize the results obtained for plot design and logistics. Recommendations are based on the detailed results
obtained for different groups of measurements as reported in Sections 5 through 9. These measurements include
the following.
• Visual Symptoms - selected mensurational variables such as tree size and the percentage of live crown, and
tree crown dieback, transparency, discoloration, defoliation, and density.
• Soil Productivity - soil chemical and physical properties.
• Foliar Nutrients/Chemical Contaminants - chemical analyses of foliage from sample trees
• Vegetation Structure - vertical vegetation structure.
• Growth Efficiency - canopy transmittance of photosynthetically active radiation.
The starting point for measurements and plot design was based on several interagency committee reports, the
FHM measurement strategy, and the current plans for monitoring. From this starting point, a suite of measures
of forest site and stand condition, broadly defined, were selected for testing. The potential efficacy of these
measures has been established through peer review. The question of their efficiency is important because many are
likely to be included in the collection of measurements deployed in the field The type of information needed about
these measurements includes, for example, how many sites should be measured, how frequently should
measurements be made, bow should measurements be physically arranged in the field, how much do different
methods cost, and what infrastructures are required to make the measurements.
4
-------
The field test was conducted in the New England and the Southeast regions of the United States In New
England, 20 field plots were located in northern hardwoods forest types. In Virginia, 20 field plots were located
in loblolly pine-hardwoods forest types. The plot locations were selected systematically (Virginia) or randomly
(New England) from candidate locations defined by the EMAP sampling grid and by the U.S. Forest Service forest
inventory system. At each selected location, a plot consisting of four subplots was established. Within subplots,
further aubsampling rules were devised according to the particular objectives for each set of measurements. This
sampling design established a multistage sample framework to address the objective of sources of variance It also
established a realistic setting for the test of logistics.
in New England, the field plots were located on a wide vanety of site conditions, but all were contained in the
maple-beech-birch forest-type group. In Virginia, the coastal plain sites were less variable, but both the loblolly-
sbortleaf pine and the oak-pine forest-type groups were sampled. The field plots are representative of uniform, fully
stocked, mature stands on typical soil types in these forest types and regions. Stand density (basal area) ranged
from 19 to 48 mY/ba in New England and from 13 to 47 mY/ha in Virginia. Stem density was between 500 and 1300
trees/ba in New England and between 400 and 1550 tree.s/ha in Virginia. Species composition was different for
the two regions, but the range and average number of overstory species were similar between regions The New
England stands had structures approaching uneven-aged, and the Virginia stand structures were suggestive of
multi stoned stands.
PLOT DESIGN
Standard statistical procedures were used to estimate the optimum number of sample units for different
measurements and stages of sampling under two sets of cost assumptions These results should be considered
guidelines rather than rules for sampling. In most cases, the sample designs developed for the 1990 and 1991 field
tests are adequate for a regional monitoring program, but generalizations to untested species and regions may be
tenuous. The optimum solutions were not particularly sensitive to cost reductions in the final stage of sampling
In comparison to locating field plots and establishing subplots, the costs of the final stages are less important than
the information gained, suggesting that the measurement effort not be unduly constrained by logistical considerations
once personnel are on the plot. Spatial correlation was not an important factor to consider when calculating
optimum sample sizes for measurements that were made on systematic grids within plots.
The recommended sample allocations suggest that the current design of four subplots per field plot location is
more than adequate for most of the measurements that were tested. The estimates of sample allocation suggest that
two trees per each of three or four subplots should be sufficient for visual symptoms measurements for an single
species. If separate statistics are desired for each species, then the total number of trees measured at each location
depends on the number of species present. Analyses of typical mensuration variables suggested that two suhplots
were si.ifuicient for characterizing average tree sizes and total stand basal area, but the subplots may be too small
because not enough trees were present on them to adequately portray stand structure.
The soils data suggested that two to three soil pits will be sufficient These pits should be systematically
arranged so as to represent the entire field plot location. The foliage chemistry data were quite variable and only
two species were tested. The suggested allocation is for five to six branches from each of one to two trees taken
from each of two to three subp!ots, or a total of between 15 and 30 branch samples per species per plot location
if interest centers on the subset of macronutrients, then only one-third as many branches are required. The larger
sample sizes would be required mainly to characterize the heavy metals in foliage.
A conservative estimate of the number of vegetation structure measurements required in the forest types tested
S 5iX subploLs and four measurement stations per subplot. The results for photos)nthetically active radiation (PAR)
suggest between two and six subptots, more on cloudy days and fewer on clear days, and two measurement stations
per subplot. Both the vegetation structure and PAR measurements need to be tested under a wider variety of forest
canopy types before firm recommendations can be given, because the results obtained were highly conditioned upon
the subplots being rotated into similar canopy conditions
5
-------
LOGISTICS
The elements of logistics that were examined included: staffing and personnel, procurement and inventory
control, training, reconnaissance, sampling, communications, and safety. It was feasible for a five-person crew to
make the measurements on a plot in a ten-hour day. Six specific recommendations are made to improve the
efficiency of field activities:
• Specify the criteria for eliminating sampling sites prior to reconnaissance.
• Provide reconnaissance prior to field sampling to reduce the time it takes a field crew to locate a sampling
site.
• Review all equipment and consumable items with measurement coordinators to determine exactly what is
required.
• Determine staff requirements early and ensure that contracts are established before field sampling activities
commence.
• Provide an assistant to the field crews to stock equipment and consumables and to receive, maintain,
transporl, and track samples.
• Provide a better cbmrnunications network. The regional project leader or an identified assistant should be
responsible for all communication with the field crew leader. The logistics personnel should have early
and close communication with other groups (design, indicators, QA, information management) to develop
efficient field implementation.
VISUAL SYMPTOMS
Measures of visual symptoms and mensuration are important elements of forest monitoring in most countries,
but care must be taken to obtain comparable measurements for different species and locations. This study sampled
trees according to the protocols that were used for implementation of FHM in New England in 1990, and despite
the comparatively small sample sizes in this study, the results were generally similar to those reported by the F1-JM
program based on New England monitoring effort. There was no comparable sample for Virginia, however. The
variance analysis of crown density yielded results similar to those obtained in a field test in Great Britain. Most
of the variability of crown density can be attributed to tree-to-tree variability, and most of the remainder to stand-to-
stand variability. The quantitative analysis of these sources of vanance allow F}-IM to derive optimum numbers of
trees to sample within each stand Additional analyses are needed to determine which of the competing
measurement methods are most efficient or accurate. Additional analyses of root and tree increment core samples
are also needed.
SOILS
Measurements of soil physical and chemical properties are fundamental to forest monitoring The study focused
on logistical concerns and on quantifying the measurement variability that may be expected when soil measurements
become a routine part of monitoring. Two different statistical techniques were applied to a set of laboratory
chemical parameters that were measured on soil samples collected from eight intensively-sampled field plots in each
region. The results suggest that the present systematic sampling design of three soil holes per field plot is sufficient.
The variance among pits within clusters was the same as the variance among pits among clusters, indicating that
a sampling design with individual pits as the subsampling unit provides a better allocation of resources
6
-------
FOLTAR CHEMISTRY
Sugar maple (in New England) and loblolly pine (ui Virginia) foliage samples were obtained from the upper
crowns of dominant and co-dominant trees on 10 plots in each region. A suite of chemical analyses were performed
which included macro- and micronutrients, total C, N, and S. and selected trace elements. Between-plot, between’-
subplot, between-tree, and between-branch variances were calculated. Between-branch variances could not be
estimated for sugar maple because the samples were collected incorrectly. The sample optimization suggested five
to six branches from each of one to two trees on two to three subplots for most elements, but fewer branches were
required to characterize just the macronutrients. Most of the relatively large branch-to-branch variability observed
for the trace elements may be attributed to concentrations at or below the analytical detection limit
VFGETATION STRUCTURE
Physical alteration of habitats isa threat to biotic diversity. For this reason, the structural features of land use,
land cover types, and animal habitats are candidate indicators of biotic integrity. The primary objective of this study
was to compare an ocular method to a pole method for measuring the amount, arrangement, and composition of
forest veget.ation. These eihods differ in their conceptual and procedural approaches to estimate foliage occupancy
and distribution. The results indicated that ocular estimates of total foliage occupancy are significantly (3 to 13%,
p
-------
SPECIES NOMENCLATURE
Common name Latin name Common name Latin name
American basswood iTha americana L. post oak QuercusstellaiaWangerth
American beech Fagus grandifolia Ehrh. quaking aspen Populus rremukudes Miclix.
American holly Jle.r opaca Mt. red maple Acer rubrum L.
American mountain-ash Sorbus americana Marsh red pine P:nus resinosa Mt
balsam fir Abies baLsamea (L.) Mill. red spruce Picea rubens Sarg.
bigtooth aspen Populus grandidenrata rock elm Lilmus :homosii Sarg
Michx. sassafras Sassafras aibidum (Nuit)
black cherry Prunus serorina Ehrh. Nees
black oak Quercus velurina Lam. scarlet oak Quercus coccuiea
black walnut Juglan.s nigra L. Muenchh.
blackjack oak Ql4ercus marilandica Scots pine P:nu.s rylvestris L.
Muenchh sessile oak Quercus peraea
bluebeech Cazpinus caroliniana Walt. (Mattuschka) Lieblein
cherry Prunus spp shortleaf pine Pinus echinaza Mill
common choke cherry Prunus virginiana L. Sitka spruce .Picea suchensis (Bong)
eastern white pine Puius sirobus L. Carr.
eastern hemlock Tsuga canadensis (L.) sourwood Oryd.endrum arborewn (L)
Carr. DC.
eastern hopbornbeam Ostrya virgzniana southern red oak Quercusfalcaza Michx.
(Mill.) K. Koch striped maple Acer penns-yl ’anzcum L.
eastern redcedar .Juniperus vzrgziuana L. sugar maple Acer saccharum Marsh
elm Ulmus spp. sweet birch Betula lenta L.
English oak Quercus robur L. sweetbay Magnolia virgiiz:ana L
European beech Fagus rylvat:ca L sweetgum Liqu:dambarsr)racijlua L
flowering dogwood Corn u .s florida L. upland blackgum t yssa sylvatica Marsh
green ash Fraxinus penni -ylvansca Virginia pine P nus virgintana Mill
Marsh. water oak Quercus nigra L.
hickory Carya spp. white ash Frax nus americana L.
loblolly pine Finus :aeda L. white oak Quercus ciba L.
lowland blackgum Nyssa syli’azica var. white pine Pin us srrobu.s L
bjflora (Walt.) Sarg. white spruce Picea glauca (Moench)
northern red oak Quercus rubra L. Voss
northern white-cedar Thuja occidentalis L. willow oak Quercus phellos L
Norway spruce Picea abies (L.) Karst. yellow birch Berula atleghaniensis
paper birch Bezula papyrjfera Marsh Bntton
persimmon Drospyros wrgtniana L. yellow-poplar Leriodendron :ulip?fera L.
pond pine Pinu.s serozina M icl ix.
8
-------
CONVERTING FACTORS FOR COMMON UNITS OF MEASUREM ENT
Metric unit English equhalent
L ib
1 centimeter (cm) 0 39370 inches (in)
1 meter (m) 3.28083 feet (It)
I kilometer (km) 0.62137 miles (mi)
Ares
I square cenlimeter (cm 2 ) 0.15500 in 2
0001076 ft 2
I square meter (m2) 10.76387 ft 2
I hectare (ha) 2 471044 acres (ac)
0.0038610 mi 2
I square kilometer (km2) 247.104 ac
0.38610 mi 2
I m 2 fha 4.3560 ft 2 /ac
9
-------
A CKNOWLEDGEMENTS
This report is one produci of a project that was the combuied effort of more than 50 individuals from thxee Federal
Agencies, three universities, and five private contractors. Program managers were: Craig Palmer (Technical
Director of the U.S. EPA Environmental Monitoring and Assessment Program-Forests); Joe Barnard (National
Program Manager of the U.S. Forest Service Forest Health Monitoring program); Ralph Baumgardner, Barry
Martin, Ann Pitchford, and Parker Wigington (US. EPA Project Officers), and; Robert Brooks and Noel Cost
(U.S. Forest Service Project Leaders). Many scientists are acknowledged by authorship of the chapters of this
document. Additional individuals are recognized here for their special contributions:
Steve Agnew - Virginia Polytechnic Institute & State Univ., Blacksburg, VA
Moss Baldwin - Virginia Polytechnic thstitut.e & State Univ., Blacksburg, VA.
Roger Blair - U.S. Environmental Protection Agency, Corvallis, OR.
Bill Burknian- ManTech Environmental Technology, Inc., R.adnor, PA.
Mike Cook - Asplundh Co., VA.
Robert Cooke - U.S. Forest Service, Durham, NH.
Susan Cox - U.S. Forest Service, Durham, NH.
Dave Drake - U.S. Forest Service, Radnor, PA.
Kate Dwire - ManTech Environmental Technology, Inc., Corvallis, OR.
Chris Eagar - U.S. Forest Service, Radnor, PA
Dabne Eastham - U.S Soil Conservation Service, VA.
Ron Faison - Asplundh Co , VA.
Bob Gerlach - Lockheed Engineering and Science Co., Las Vegas, NV.
Andre Grady - Asplundh Co., VA.
Karl Hermann - ManTech Environmental Technology, Inc., Research Triangle Park, NC.
Brenda Huritley - ManTech Environmental Technology, Inc., Corvallis, OR.
Shirlea Johnson - U.S. Forest Service. Asheville, NC.
Ben Koonta - U.S. Forest Service, Asheville, NC.
Chuck Liff - University of Nevada at Las Vegas, NV.
Don Mason - Asplundh Co., VA.
Jamie McKinney - Davey Tree Co., NH.
Robert Mickler - Manlech Environmental Technology, Inc., Research Triangle Park, NC.
Imants Millers - U.S. Forest Service, Durham, NH.
Jim Morrison - ManTech Environmental Technology, Inc., Corvallis, OR.
Bruce Nash - The Pennsylvania State University, University Park, PA.
David Nichols - Davey Tree Co., NI-I.
Lewis Ohrnann - U.S Forest Service, Grand Rapids, MN.
Janice Braswe lt Parker- ManTech Environmental Technology, Inc., Research Triangle Park, NC
Charles Scott - U.S Forest Service, Radnor, PA.
Diane Shields - U.S. Soil Conservation Service, VA.
Renee Stang - ManTecb Environmental Technology, Inc., Corvallis, OR
William Taylor - U.S. Soil Conservation Service, MA.
John Teberg - Lockheed Engineering and Science Co., Las Vegas, NV.
Rob Tidwell - Lockheed Engineering and Science Co., Las Vegas, NV.
Jim Twaroski - U.S. Forest Service, Asheville, NC.
James Wiant - ManTech Environmental Technology, Inc., Corvallis, OR.
The editors thank Susan Frarison and two anonymous reviewers for valuable comments.
10
-------
I PLOT DESIGN AND LOGISTICS
“These reprehensible beings (statisticians] are
commonly accused of (I) demanding too many trees
per treatment and (2) being ready to base results on
samples that are not large enough.”
— S.C. Pearce (1953)
The U.S. Environmental Protection Agency’s
(EPA’s) Environmental Monitoring and Assessment
Program (EMAP-Forests) has joined the U.S.
Department of Agriculture (USDA) Forest Service
and other government agencies in an effort to monitor
and assess the condition of the nation’s forested
ecosystems in relation :to natural and manmade
stresses. A long-term and multi-tiered strategy for
monitoring and assessment (Palmer et a!., 1991)
includes elements for detecting, evaluating, and
explaining changes in forest condition.
To implement the many aspects of monitoring,
the interagency Forest Health Monitoring (FHM)
program draws upon organizations with skills in
forest inventory, pest and health management,
ecological research, and environmental regulation
Within EPA, the FHM effort is linked to the
monitoring of other resource classes (e.g.,
agricultural and and lands) and draws expertise from
groups that coordinate statistical design, measurement
selection, information management, and other
technical areas (Palmer et a! , 1991).
Because the FHM program is large and complex,
it cannot be fully implemented everywhere at once,
and therefore the strategy for implementation is
conservative. The detection phase of forest
monitoring began in 1990 with a survey of 206 field
plots in New England (Brooks et al., 1991a) and has
been expanded to the Mid-Atlantic and Southeast
regions of the United States in 1991. Because
budgets are limited, the strategy includes pilot tests,
regional demonstrations, and critical evaluations of
procedures prior to implementation. Selected
procedures for the detection phase of monitoring
were evaluated in a regional test in 1990, and the
testing was continued in the Southeast and Western
United States in 1991.
The purpose of this report is to summarize the
1990 field test. The test was designed to answer
specific questions about field plot design and logistics
for selected measurements (Palmer et al., 1990) Not
all of the possible measurements were tested, and not
all of the possible questions were asked about the
measurements that were tested. The purpose of this
document is to summarize what was learned about
plot design and logistics from that field test. The
goal is to provide a timely analysis to help guide the
expansion of the monitoring program in 1991
The purpose of this introductory section is to
provide an overview of the scope and purpose of the
field study. The contents of each section will be
reviewed, the rationale for the plot design and
logistics objectives will be given, and the approach to
the study will be described A quality assurance
report (Burkman and Mickler, 1990) and a field
methods manual (Dwire et al , 1990) provide
additional information about the study.
1.1 CONTENT AND ORGANIZATION
This report is organized into nine sections.
Section 2 describes the site and stand characteristics
at the field plot locations. Section 3 summarizes the
statistical issues and lessons learned about field plot
design and required sample sizes Section 4 revie s
logistical issues and makes recommendations for
implementing these measurements. The lessons and
recommendations are based on the test results for
different groups of measurements that are reported in
Sections 5 through 9.
1.2 RATIONALE AND OBJECTIVES
Measurement selection and plot design was based
on interagency committee reports (Forest Damage
Survey Workshop, Chapel Hill, NC, November
1987; Forest Health Task Force meeting, Harper’s
Ferry WV, July 1988; and Joint EPAIUSDA Forest
Service FHM National Coordination Meeting,
Research Triangle Park, NC, January 1990), the
EMAP measurement strategy (}lunsaker and
Carpenter, 1990), and the current morutonng plans
(Brooks et a!., 1991b)
11
-------
One important question for any measurement is
its biological accuracy, or its capability to represent
forest condition. But there is a practical difficulty
of testing this, because accuracy assessment requires
that good sites be distinguished, a priori, from
ba3 sites. That implies there is at least one metric
that is guaranteed to accurately classify condition (or
else the sites would be indistinguishable). Any
correlated metric will aiso appear accurate; any
uncorrelated metric will appear maccurate. The
difficulty is that, of a suite of measurements chosen
to characterize several dimensions of ‘condition,
some will incorrectly appear to be inaccurate only
because they measure good’ vs. bad’ on a different
scale.
For example, choosing good arid bad sites in
terms of past growth virtually guarantees that growth
will appear to be an accurate nietnc of condition. It
also makes it more likely that a correlated measure
such as crown density will appear accurate. But any
imcorrelated measure of, for example, wildlife
abundance would appear to be inaccurate, without
even receiving a/air test! This difficulty can be
overcome, but it was decided that a fair test of
accuracy of all the proposed measurements was
beyond the scope of this study.
Instead, another set of questions was asked about
selected measurements. It is reasonable to assume
that some measurements have a high probability of
being included in forest assessments, based on expert
opinions, the experiences of monitoring programs in
ether countries, and the need to simply characterize
the inorutonng sites. A small increase in the
efficiency of these measurements would translate to
substantial cost reductions in a fully-implemented
national effort.
Preference was therefore given to selected
measures of forest site and stand condition, broadly
defined, that can potentially be applied i.n any type of
forest. The type of information needed about these
measurements is rather utilitarian, for example, how
many sites should be measured, how frequently
should measurements be made, how should
measurements be physically arranged in the field,
bow much do different methods cost, and what
infrastructures are required to make the
measurements.
A study of the components of variance of forest
condition measurements was conducted in a way that
logistical questions could also be addressed in a
realistic setting. The objectives were:
to evaluate plot design and subsarnplzng
procedure sensitivity to spatial heterogeneity at
the scale of monitoring,
• to quantify the Lime and fiscal requirements of
making these measurements,
• to assess the relative efficiency of competing
measurement methods in some cases, and
• to apply the results in planning the furiher
implementation of the national PHM program
A secondary objective was to start to set up an
infrastructure for monitonng and assessments This
infrastructure is needed, for example, for information
management, quality assurance, logistics, and data
analysis Another secondary objective was to
generate a set of data that could be used in future
simulation studies to explore alternate designs for
monitoring and reporting.
1.3 APPROACH TO THE STUDY
In October 1989, EMAP-Forests was directed to
conduct a field study to augment the current Forest
Service monitoring plans in New England. The
FHM program determined in March 1990 that
EMAP-Forests would conduct a pilot test of selected
measurements in two forest types: one in the
Northeast arid the other in the Southeast regions of
the United States The test would be conducted as a
separate but related effort of the ongoing
implementation in New England
An interagency work group was set up to plan
and conduct the pilot test (Palmer et al , 1990).
Individuals were assigned lead roles for specific sets
of measurements, for quality assurance, and for
logistics. Both agencies participated in the review of
study plans, sampling designs, and field methods.
Staff from both agencies arranged the resources to
physically conduct the study. The study was set up
in the months of April and May 1990
Methods were refined and field crews were
trained and tested in typical field settings at the
Pocahontas State Forest, VA, during two weeks in
early summer. A debriefing workshop v.as
conducted after the field work was completed
12
-------
Figure 1-1. FIELD STUDY PLOT LOCATIONS P 1 NEW
ENGLAND AND VIRGINIA
The study was conducted at 40 field plot
locations: 20 in the northern hardwoods forest types
in New England and 20 in the loblolly pine.
hardwoods forest types in eastern Virginia (Fig 1-])
In Virginia, the plot locations were selected
systematically from the phase I (aerial photograph)
sample used for multiphase forest inventory by the
U.S. Forest Service (USFS) Forest Inventory and
Analysis (FIA) projects (U.S. Forest Service, 1985a)
In New England, the plot locations were selected
randomly from the stage 2 (ground plot) FIA forest
inventory sample. To maximize geographic
representation within the selected forest types and to
provide realistic tests of logistics, at most one
location was selected within one of the EMAP 40-
sampling units (Overton et al., 1990).
At each selected location, a plot consisting of
four subplots was established. The four-subplot
design was convenient because it established a
multistage sample framework at each location (needed
to address the objective of sources of vanance), and
because it was similar to the design that was planned
for monitoring in New England. Within subplots,
further subsampling rules were devised according to
the objectives for each set of measurements
These initial studies used the USFS Northeast
station’s protocols for rotating subplots. With
rotation, all mibplot.s are certain to be located in the
same forest type at each location. Rotation provided
the opportunity to obtain compatible data within plots
and simplified the analysis of resource allocation.
Table 1-I summarizes the data that were available
for this report. Note that not all measurements were
planned for every location, so the table s sparse for
some measurements For example, uitensive soil
sampling and foliar chemical measurements were
planned for S to 10 of the 20 total locations per
region. Foliage data for some locations were not
available because the laboratory analyses had not
been completed in time.
The following brief descriptions of the designs
for the measurements set the stage for the more
detailed descnptions in later sections.
• VisuaI symptoms includes forest inventory
measurements and several measures of tree
crown condition that were made on trees in the
four 0 0169-ba circular subplots (Fig 1-2)
• Sample trees includes measurements of roots,
branches, tree increment cores, insects, and
diseases that were made on 4 to 6 dominant or
co-dominant trees in the 0.0843-ha annular
subplots (Fig. 1-2). This sample was limited to
sugar maple and red maple in New England and
loblolly pine in Virginia These data are not
reported in this document.
• Foliage includes chemical analyses of samples
obtained from the sugar maple and loblolly pine
sample trees descnbed above.
• Soil includes physical and chemical
measurements made at nine pits at each plot
(Fig 1-3). At the less-intensively sampled plots,
fewer measurements were made arid they are not
reported here.
• Vegetation structure refers to the physiognoinic
class and height of vegetation at each of 16
sample stations overlaid upon two of the four
subplots (Fig. 1-4).
• PAR refers to measurements of light irradiance
at 16 sample stations overlaid upon each of the
four subplots (Fig 1-4).
13
-------
Table 1-I SUMMARY OF MEASUREMENTS AVAILABLE FOR EACH STUDY PLOT LOCATION
Ploc Visual Sample Vcgeta1 on
no symptoma tree? SoiJ Foliage azructure PAR
Plot Visual Simple Vegetation
no .ympoma tree? Soil Foliage rucrure PAR
NewFnE!end
V,r irna
16
X
X
X
X
1839
X
X
X
X
117
X
X
X
1841
X
X
X
X
122
X
X
X
X
X
X
1843
X
X
X
X
X
X
15$
X
X
X
X
1954
X
X
X
X
X
159
X
X
X
X
X
1955
X
X
X
X
X
X
160
X
X
X
X
1957
X
X
X
x
166
X
X
X
1958
X
X
X
X
X
173a
X
X
X
X
1959
X
X
X
X
X
X
174
x
x
x
x
x
x
2072
X
X
X
X
X
X
181
X
X
X
X
2074
X
X
X
X
267
X
X
X
X
X
X
2075
X
X
X
x
270
X
X
X
X
X
2185
X
X
X
X
289
X
X
X
X
2186
X
X
X
X
321
X
x
x
X
X
X
2187
X
X
X
X
X
347
X
X
X
X
2188
X
X
X
X
420
X
X
X
X
2189
X
X
X
X
X
X
508
X
X
X
X
X
2190
X
X
X
X
X
692
X
X
X
X
2191
X
X
X
X
902
X
X
X
2192
X
X
X
X
3270
X
X
X
X
2303
X
X
X
X
L.boratorydiagnosesofaample tree leaf and twig s>mptoms are reported by Nash c i al (1990) for New England species Laboratory diagnoses
of loblolly pine symptoms have not been reported
b Data (corn the 0 0014-ha aubplots arc missing for plot number 173
Figure 1-2. SAMPLING DESIGN FOR TREE AND FOL1AR MEASUREMENTS TREES < 762 cm dbh WERE MEASURED ON
REGENERATION PLOTS, AND TREES 7 62 cm dbh WERE MEASURED ON TI-It MENSURATION PLOTS SAMPLES OF
ROOTS, TREE INCREMENT CORES. AND FOLIAGE WERE OBTAINED FROM TREES ON THE SAMPLE TREE PLOTS
AZu V’ -0.”
0° —
clot c.u.-
1 ,.e a, a Stb-.o
S CDIOi Osacr D L Oil
Pici PoCus A,55
ii .
io n 2 0 0 00 ’
21 0 0159
Sa pI. irs . 1 9 3 0 0513
0.1 5 I
S i .0O 0 1
2
n’$ .,,
PI
00°
--‘
2400
5 .Cc lot Qe’ it cr1
csntsr plot c.nte’
.4
,
120°
14
-------
P gur 1-3 SAMPLING DESIGN FOR SOIL MEASUREMENTS
Figure 1-4 SAMPLING DESIGN FOR VEGETATION STRUCTURE AND PAR MEASUREMENTS VEGETATION STRUCTURE
WAS MEASURED ONLY ON SUBPLOTS I AND 2
Deta I
A
etc
A-p .o . - izor,
p rof I.
I e
crr I e
Tb 0
P lo I e
lot
2
Plot
C0 lIt l
at
Cente .-
A
3
A
0
etet e .
3
J5
-------
2 SiTE AND STAND DESCRIPTIONS
Kuil Rtitters, Rick Van Remortel, William
Bechtold, and David Ajerich
about 28% and 13%, respectively, of the —20
million ha of forestland in the south Atlantic region
(VA, NC, and SC; U.S. Forest Service, 3982).
New England
Virginia
VA Northumberland
topographic maps
Table 2-1
DESCRIPTIONS OF
(see also Table A-I)
PLOT
LOCATIONS
Plot
I SO State
Elev •
Me ss
(irs)
date
The purpose of this section is to descnbe briefly
the sites and stands that were studied Some site
attributes were recorded at each plot location, and
others at each of the four 0.0169-ha subplots (Fig.
1-2) within each location. The stand descriptions
come from the tree measurements made on trees with
16
MA
Franklin
305
8-27
diameter at breast height (dbh) 7.62 cm and larger
122
VT
ME
Cbiti.cndcn
Wsldo
244
8-15
within the four 0.0169-ha circular subplot.s
155
VT
Addison
91
518
7-31
8-13
(overstory) and on trees with dbh less than 7.62 cm
within the eight 0 0014-ha circular subplots
( understory) (Fig. 1-2).
159
160
166
173
174
NH
VT
NH
CT
‘VT
Graf’Lon
Wjndham
Ca 1l
Lnch(ieId
366
335
488
427
8-07
8-24
8-22
8-29
Table 2-I lists site data, including plot number,
181
VT
Orange
Essex
427
610
8-09
8-17
state, county, elevation, and measurement date
More detailed information about each of the four
subplots is contained in Table A-i (Appendix A). In
267
270
289
ME
NH
ME
Franklin
Memrnack
Washington
213
183
30
7-26
8.03
8-01
terms of elevation, slope, aspect, terrain position,
347
ME
NH
Ar stc k
c
396
7-20
micro-relief, and landformn, the New England sites
were generally more vanable than the Virginia sites.
All stands were of natural origin, but past land uses
are unknown. A few sites had been disturbed
420
308
692
902
3270
ME
ME
ME
ME
Pcnobscot
Somerset
Piscataquss
A iiock
183
366
213
183
8-21
7-18
7-25
9-12
7-17
recently (Table A-I). The average stand age was 43
ME
Oxford
610
9-li
years (range 30 to 70) in Virginia Stand age is not
known for the New England sites
1839
The study encountered four Society of American
1841
1843
VA
VA
Gloucester
le of
30
30
8-17
8-15
Foresters (SAF) forest types (Eyre, 1980)
1954
VA
Wight
Richmond
30
0
8-14
representing three of the USFS FJA-type groups
(Tables A-I and 2-2). The red maple-northern
hardwood and sugar maple-beech.yellow birch forest
types (SAF) that were sampled in New England are
1955
1957
1958
1959
2072
VA
VA
VA
VA
VA
King & Queen
Surty
Sussex
Southampton
charles
30
30
61
152
8-16
8-14
8
8-07
part of the FIA maple-beech-birch type group (Eyre,
2074
VA
City
Suuex
30
8-15
1980). The maple-beech-birch type group occurs
throughout the region and occupies about 24% of the
—13 million ha of forestland in New England (U.S.
Forest Service, 1982).
2075
2185
2186
2187
2188
2189
VA
VA
VA
VA
VA
VA
Greenzville
Wcslmora lsnd
Caroline
King William
}Iennco
Cbesterfieid
30
30
91
61
30
8-06
8-13
8-22
8-21
8-17
8-10
The loblolly pine and loblolly pine-hardwood
forest types (SAF) that were sampled in Virginia are
part of the FIA loblolly-shortleaf pine and oak-pine
‘type groups, respectively (Eyre, 1980). The loblolly-
shortleaf pine and oak-pine types occur primarily on
2190
2191
2192
2303
VA
VA
VA
VA
Dinwicjd 1 c
Bnjrsawick
Brunswick
Car 1 1
61
61
91
91
61
8-10
8-07
8-13
8-08
8 22
•
evation
from
U S
the Atlantic coastal plain and Piedmont and occupy
rounded up
Geological Survey
to the nearest 100 ft
16
-------
Table 2-2 DESCPJT’TlONS OF FOREST TYPES
Forest
sype
FW
type-group
Forest type dcscnpuon (Pwzrc et at , 1990)
Sugar maple-
beech-yellow
birch
Maple-bcech-
birch
Auociatza include baaiwood, red maple, hemlock, northern red oak, white ash, white pine, black
cherry, Sweet birch, American elm, rock elm, and eastern hophornbcam Sites are fertile, moist,
and well drained
Red maple-
northern
hardwood
Maple-beech
birch
The type ii dorninatedby red maple and some of the wide variety of northern hardwood associates
including sugsr maple, beech, birch, aspen, white pine, red pine, and hcmiock The t)pe is often
manmade and may be the result of repeated cuttings Sites are generally uplands
Loblotly-
hardwood
Oak-pine
Associates on moist and wet site include biackgum, .weetgum, yellow-poplar, red maple, white ash,
green ash, and American elm On drier sites the associate. include southern red oak, northern red
oak, white oak, post oak, scarlet oak, persimmon, arid hickory. Sites are usually moist to very moisi
though not wet all year, also found on drier sites
Loblolly
pine
Loblolly-
shortheaf
Associates include sweelgum, southern red oak, post oak, blackjack oak, bhackgum,
pine yellow-poplar, and pond pine Sites in MD and DE .rc both upland with abundani moisture
but good drainage arid on poorly drained deprcuioni
FIA type groups are from the USFS FIA and are the major type groups identified b) FHM for national reporting
2.2 STAND DENSITY
Table 2-3 reports stem density of overstoi-y and
nnderstory trees, and the stand density in terms of
basal area per unit area of overstory trees, Typical
sizes of mdividu.al trees are indicated by the quadratic
average dbh (the dbh of a tree with average basal
area) of overstory trees, and by the average height of
dominant and co-dominant overstory trees. The
range of overstory stem density was 334 to 1305
trees/ha in New England, and 415 to 1557 trees/ha in
Virginia. Understory stern density varied by at least
an order of magnitude in both regions. Basal area
(cross-sectional area of live trees at breast height) per
unit area was between 19 and 48 m 2 [ ha in New
England, and between 13 and 47 m 2 /ba in Virginia.
The main canopy was between 14 and 23 m tall
in New England, and between 15 and 27 m in
Virginia. The quadratic average tree dbh was about
20 cm in both regions. The largest (dbh) trees
measured in each region were an 8L8-cm northern
red oak on plot 174 in New England and a 75.2-cm
willow oak on plot 2188 in Virginia.
Overall, the stem density, stand density, average
tree size, and main canopy heights were similar for
the two regions. Almost all of these stands would be
characterized as mature, fully stocked, pole- to saw
timber-sized stands, with ample regenerat on The
stand density statistics were similar to those compiled
from the most recent USFS FIA forest inventory for
Virginia. Although no comparisons with forest
inventory were made for the New England stands, the
locations were drawn from existing FIA plots and are
therefore presumably representative of the FIA
sample.
2.3 SPECIES COMPOSITION
Table A-2 (Appendix A) sumrnanzes species
composition of the overstory trees of the stands in
terms of stein density, basal area, and quadratic
average dbh by species. Twenty-four species were
found in New England and 31 in Virginia
Compansons of quadratic average dbh suggest that
some of these species appeared mainly in the lower-
canopy positions in these forest types (New England
eastern hophornbeam, black cherry, balsam fir,
striped maple, and common choke cherry; Virginia:
American holly, sweetbay, persimmon, flowering
dogwood, sourwood, bluebeech, red maple, and
eastern redcedar). The composition of the upper-
canopy trees was not consistent between stands. In
four New England stands, the species selected for
intensive sampling (sugar maple) was not on the plot
but was available for sampling in the adjacent stand
17
-------
Tabtc 2.3. MENSURATION OF OVERSTORY AND
UNDERSTORY TREES BY PLOT
lJnder-
Plot
Over .iory
gory
No.
No.
Ba a1
Avg
Avg
Do.
reel
(ha l)
area
(mr/ha)
dbh
(cm)
height’
(m)
trees
(ha I)
New Enitand
16 652 2829 235 200 11305
117 534 28 20 25 9 21 4 18348
122 1305 38 38 19 4 21 0 93
155 593 1955 195 150 10008
159 993 30 38 19 7 20 6 9822
160 890 41 20 24 3 22 6 2780
166 771 27 72 21 4 20 1 8062
173 1023 32 34 : 20 I 20 0 —
874 786 47 52 27 7 20 4 2039
181 712 1961 187 140 12232
267 860 23.26 186 16 1 12417
270 904 35 06 22 2 19 8 3336
289 1082 19 84 15 3 17 1 2687
321 756 29 03 22 1 15 7 13529
347 904 3083 208 192 11490
420 712 25 07 21 2 18 I 26409
508 1156 3130 186 174 7228
692 890 22.37 179 168 6857
902 — — — — —
3270 1216 28 14 172 151 6394
V rgjnia
1839 949 35 54 21 8 2.3 2 3521
1841 1201 3983 205 254 4911
1843 1142 2625 171 194 3243
1954 1038 46 73 23 9 22 5 2965
1955 563 24 82 23 7 20 2 6209
1957 77! 23 35 20 5 23 4 19367
1958 415 1314 20 1 19 0 21869
1959 1156 3612 199 204 2687
2072 1260 3008 174 202 14085
2074 1260 26 38 16 3 15 0 8988
2075 801 39 24 25 0 26.2 12973
2183 1053 38 75 21 6 21 2 8803
2186 1171 3859 205 202 12880
2)87 1394 3011 166 177 8988
2188 1067 27.19 18 0 17.6 7042
2189 1245 27.77 16 8 19 2 8340
2190 964 24 15 17 9 19 8 21035
2191 1334 2700 160 204 10564
2192 1557 4260 187 180 6116
2303 756 26.39 21 2 263 5375
There are many statistics that represent species
diversity and stand species composition (e.g.,
Magurran, 1988). Two of the simpler statistics, the
numbers of species (a measure of species richness)
and the Shannon evenness index (Pielou, 1969 - a
measure of the proportional abundances of species)
will be used here to illustrate species diversity.
The range and average of the number of
overstory species per location were similar for the
two regions (Tabie 2-4), but both the largest (14) and
smallest (2) number of species per location were
found in Virginia. The average number of
understory species per location was higher in Virginia
than in New England (9.1 vs. 6.6). but a two-tailed
t-test (assuming independent samples, unequal
variances, and 17 effective degrees of freedom)
suggests that the difference is only marginally
significant (0 OS
-------
P1 i
Overitory
Under iory
No.
No
ao
CCLCS
t 1 ipecics
TabIc 2-4
PROPORTiONAL
NUMBER
OF SPECIES
AND
THEIR
Furthermore, the evenness of the New
ABUNDANCE
Y PLOT
England
species was about equal in terms of the numbers and
the basal area of individuals, suggesting that species
occupy sites in proportion to their number. In
contrast, the same comparison for Virginia suggests
that the evenness in terms of numbers of individuals
was higher than the evenness in terms of basal area;
‘ .ra lind
this could imply that sit-es are less equitably occupied
16 6
117 8
122 9
0851
0 870
0 826
0741
0 83)
o s s
8
6
0664
0634
,. ,,
by different species in the Virginia plots. In
summary, individuals of a given species were likely
to occur as both large and small individuah in a stand
155 6
0 88)
0 856
8
0 843
in New England, whereas in Virginia, some
359 7
160 9
166 7
173 11
174 7
0 895
0 780
0805
0833
0747
0 872
0 751
0.863
0.824
. 0
13
4
6
—
4
0 797
0 756
0848
—
0 634
species
were likely to occur as either large or small
individuals (but not both). There was no large
difference between regions for the values of e and c
for understory trees (Table 2.4)
181 8
072.3
. 0.518
10
0832
267 9
0776
0711
8
0837
270 10
289 9
0 871
0 685
0 856
0 752
7
4
0 800
0 920
2.4 STAND STRUCTURE
32! 5
0686
0858
6
0662
347 10
0 884
0 786
6
0 566
The numbers of overstory trees per unit area in
420 8
0773
0 751
10
0851
different size classes is shown for each plot Table
508 7
692 6
—
3270 10
0889
0 804
0888
0839
0 948
0864
9
8
7
0752
0 702
0622
2-5. In general, there were more smaller trees than
larger trees, and the majority of trees were less than
17.5 cm dbh Ideally, uneven-aged stand structure
Average 8 0
0 814
0 804
6 6
0 707
have frequency distributions that are unimodal, and
the number of trees is a decreasing function of size
class In even-aged stands, frequency distributions
Virgina
1839 6
1841 4
0 745
0 742
are also unimodal but the number of trees is not
strictly a decreasing function of size class (e g.,
1843 8
0829
}Iusch et al., 1972). In multi-aged or rregular
1954 6
0 720
stands, many forms of frequency distnbutioris
be
1955 6
0 633
may
obtained.
1957 9
0 765
1958 2
1959 9
0811
0 768
It appears that there is some structure tending
ion 14
0 757
towards an uneven-aged condition in that the
2074 11
2075 11
2185 6
2186 5
2 )87 9
0 637
0614
0562
0579
0 829
frequency distnbutions appeared neither uniform nor
typical of an even-aged condition But the evidence
does not point to either a strictly uneven- or even-
aged situation in either region. It may be speculated
2188 9
0 701
that the sample sizes were too small to accurately
2189 10
2)90 13
2)91 8
2192 7
2303 8
0720
0850
0370
0629
0 840
portray stand structure. The actual number of trees
counted ranged from 36 to 88 in New England, and
from 28 to 105 in Virginia. Thus, some plots were
too small, for the sizes of trees and mixtures of
Average 8)
0 705
species that were sampled, to permit a good
characterization of stand structure Curtis,
0 702
o 580
0 392
o 426
o 562
o 382
0 774
0 450
o 460
o 549
o 446
04)3
o 735
0 637
o 648
0717
0 317
0 388
0,553
o 536
6
4
5
2
8
13
10
5
17
8
10
13
11
12
9
11
13
9
S
8
91
0710
o 852
o 843
0 896
0 708
0 784
o 665
0 657
0 813
o ss
0 723
0 789
0671
0 842
0 866
o 847
0 860
o 873
0 864
0 706
0 79)
See text for expIan.aI on of the Shannon c-itetistics for numbers
(‘,,) and b&ut Irea ( ) of individuali per Ipecac.
19
-------
Table 2-5. FREQUENCY OF OVERSTORY TREES BY SIZE CLASS AND PLOT
New Engla
Dbh
cliii
nd
16
117
122
155
159
Plot number
160
166
173
174
181
(cm)
7.7-124
1334
741
3262
1186
2965
1483
3262
385 S
281 7
4893
12.7-175
1334
1483
3262
1779
2372
1186
1631
2076
1186
445
178.22.6
1927
890
311.3
1334
2224
2224
741
1186
445
00
229-27.7
1186
593
2224
593
1186
1186
74 1
2076
1334
445
27 9-32 8
44 5
59 3
59 3
74 I
29 7
148 3
89.0
44 5
59 3
29 7
330-378
00
297
593
297
290
741
00
297
297
593
>378
297
741
00
00
00
59.3
445
297
1186
445
Dbl
Plot number
cli i .
267
270
289
321
347
420
508
692
902
3270
(cm)
77-12 4
355 *
252 0
489 3
341 0
311.3
237 2
3707
444 8
—
652 3
127-175
1927
1927
3113
1186
2224
1927
2669
1038
—
2076
178-226
890
1779
1779
593
1631
1186
2965
1631
—
1631
229-27.7
133 4
89 0
89 0
103 8
89 0
89 0
103 8
89 0
—
4.4 5
279-328
445
741
148
148
297
297
1038
593
—
890
330-378
297
593
00
297
297
148
00
148
—
445
>378
148
593
00
890
593
297
148
148
—
148
Virginia
Dbh
P1o number
clii.
1839
1841
1843
1954
1955
1957
1958
1959
2072
2074
(cm)
77-12 4
266 9
533 7
474 4
355 8
148 3
385 5
133 4
415 1
548 6
563 4
12 7-17.5
148 3
192 7
341 0
2076
59 3
74 1
118 6
237 2
341 0
4-44 8
178-226
2669
1186
1483
1186
118 6
44 5
74 1
2520
1779
103 8
22.9-27.7
74 1
148 3
89 0
89 0
89 0
118 6
44 5
74 I
103 8
59 3
27 9-32 8
89 0
89 0
59 3
103 8
59 3
44 5
0 0
103 8
29 7
29 7
330-378
593
741
148
445
445
741
148
148
297
445
> 37.8
44.5
44 5
14 8
118 6
44 5
29 7
29 7
59 3
29 7
14 8
Dbh
Plot number
cliii
2075
2185
2186
2187
2188
2189
2190
2191
2192
2303
(cm)
7.7-12 4
177.9
222 4
192 7
637 5
578 2
489 3
415 1
785 8
578 2
296 5
127-175
118.6
341 0
311 3
2965
148 3
4003
192 7
222 4
341 0
177 9
171-226
163.1
1186
3707
2076
1483
2224
2076
890
2224
297
229-27.7
133.4
1483
1779
1927
1483
593
59 -3
133 4
2372
74 1
279-328
297
1038
890
445
297
445
445
593
1483
593
330-37.1
103.8
89 0
14 8
14 8
0 0
14 8
29 7
44 5
29 7
74 1
>378
741
297
148
00
148
148
148
00
00
445
EnX2i.ea in the table ire numbers of Uees per hectare
20
-------
16 16 NE C I ) }IS T 34
117 20 N CU HS T 14
322 3 SE Cu RI T 14fM7
155 6 S CU }IS T 14
159 7 NE FH HS T P45
160 7 NE CF OT C 14
166 22 S CU MO T 14
173 30 W CU HS T 14
174 45 SW CU MO T HO
181 2 S 611 IfS T P47
267 5 SE Gil HS T 10
270 5 N CU RI T 141M7
321 15 Sw CU HS T HO
347 25 W CU MO T YO!HO
420 6 N CU HS T HO
508 30 W CU } 1S T 34
902 43 NE CU MO T HO
1839 1 N CP B.S P4 SO
1841 1 N CF IT P4 SO
1843 1 N CP AP P4 SO
1954 1 N CP AP P4 SO
1955 1 SE CP yr P4 SO
1957 4 SE CP MT P4 SO
1958 5 W C? MT P4 SO
1959 2 SE CP AP N SO
2072 1 N C? MT P4 SO
2074 1 N C? IfS N SO
2075 2 Sw CF MT P4 50
2185 1 NE C? MT N SO
2186 5 E CP HS A SO
2187 6 W CP HS P4 SO
2188 1 W CP MT N SO
2189 2 N C? Ri M SO
2190 6 N CP HS N 10
2191 4 W P1 IlS X 14
2192 5 N P1 F A YS
2303 3 E C? HS N SO
Table 2-6 SLOPE. PHYSIOGRAPHY, AND
MATERIAL BY PLOT
SOIL PARENT
Slope Phyaognphy
Parent matenal
Plot
no Percent Aspect Reg onat Locaf
Mode Ongin 4
2.5 SOIL AND LANDFORM
Soils in the New England region have developed
mainly from glacial till derived from igneous and
metamorphic rocks on moderately to strongly sloping
and glaciated upland hilislopes. Soils in the study
region of Virginia have developed mainly from
marine deposits on level to gently sloping alluvial
plains, and in local alluviuni/residuuni derived from
crystalline rocks on gently to moderately sloping
piedmonts (Table 2-6).
Soil taxonormc classifications in the New England
region are dominated by coarse-textured Spodosols
and Inceptisols with mixed mineralogy, a frigid or
mesic soil temperature regime, and a udic or aquic
soil moisture regime. Soil taxonomic classifications
in the study region of Virginia are dominated by
loamy- to fine-textured Ultisols with mixed or
siliceous mineralogy, a therm.ic soil temperature
regime, and a udic or aquic soil moisture regime
(Table 2-7).
Soil water charactenstics in the New England
region are dominated by shallow or moderate depths
to water table, moderate to rapid saturated hydraulic
conductivity, moderate to poor drainage, and
infrequent flooding or ponding. Soil water
characteristics in the study region of Virginia are
dominated by shallow or moderate depths to water
table, moderate saturated hydraulic conductivity,
moderate to poor drainage, and infrequent flooding or
ponding (Table 2-8).
The field plot locations provide good
representation of the types of soils that occur in the
two regions. Thus, the recommended soil sampling
designs derived from these locations will be useful be
useful for soil characterization efforts as part of
future regional monitoring.
CP coastal plain, FH foothills, CF gIiof 1u ia I
laridIorm, CU glaciated upland, P1 piedmonts
AP alluvial flat (plain), BS backawarnp, FT fluvial
ierracc, HS hillside, MO mountainside. MT marine
iemce, OT outwash terrace, RI ndga
Mode of parern material deposition A sliuvium, C = glacial
outwath, M marine, T glacial till, X residuum
HO shale (unspecified) 30 igneoua (unspccifi d), 34
igneou.-grsn ic,M5 schi i (unspccified),M7 schist-basic,S0
sedimeniary (unspecified). Y0 mixed (unspecified). Y5 =
nuxed-igncous and metamorphic
21
-------
Table 2-7 SOIL CLASSTFICAT1ON CHARACTERISTICS BY PLOT
• U S Department of Agriculture Comprehensive Soil Survey System
b Particle size clau
• Mineralogy c l i i i.
Tempersture regime
Moi ure rcgimc.
P l ot
Firmly clauification
oo. Soil series Subgroup
PSC 5 MIN
T iP MR
16
Paxton
Typic Dystrochrept
Coarse-loamy
Mixed
Mesic
Udic
117
Berkshire
Typic Haplorthod
Fngid
122
Marlow
•
•
•
155
Pew
Aquic Haplorthod
.
139
Wilmington
Typic Haplaquod
.
.
Aquic
160
Adams
Typic Haplorthod
Sandy
Udic
166
Monadnoc
.
Coarse-loamy over sandy
.
.
173
Chaiticld
Typic Dyilrochrept
Coarse-loamy
.
Mesic
174
Macember
Frigid
181
Stratton
Lithic Haptaquod
Loamy
.
267
Wes lbury
Typic Fragiaquod
Coarse-loamy
.
Aguic
270
Marlow
Typic Haplorthod
.
Udic
321
Plaisted
.
•
.
.
•
347
Marlow
420
Pliustcd
Loamy
.
.
508
Berkshire
Coarse-loamy
.
902
Thorndike Var
Lithic Haplorthod
Loamy-skeletal
.
Isofr igid
1839
Tomotley
Typic Oclirsquuli
Firte.loamy
Therinic
Aquic
1841
Roanoke
.
Claycy
Silictoua
Udic
1843
Myati
Fine-loamy
.
.
.
1934
Tomotley
.
Mixed
Aquic
1955
SIagle
Aquic Hapluduk
.
Siliceous
Udic
1957
MaU.sponi
Typic Hapludult
C layey
Mixed
.
1958
Craven
Aquic Hapludult
.
.
.
1959
Portsmouth
Typic Umbrsquuli
Coarse-loamy
.
Aquic
2072
Manapom
Typic Hapluduh
Claycy
.
Udic
2074
Empona
Fine-loamy
Siliceous
.
2075
Craven
Aquic Hapludull
Claycy
Mixed
2185
Craven
.
.
2186
Tetotum
Fine-loamy
.
.
2187
Barns
Typic Hapludult
Siliceous
.
2188
Lynchburg
Aenc Paleaquuli
.
Aguic
2189
Emporsa
Typic Hapludull
.
Udic
2190
Rumford
Coarse-loamy
.
2191
Appling
Claycy
Glaucorutic
2192
Alta Vista
Aquic Hapludult
Fine-loamy
Siliceoui
2303
Xernpsvil lc
Typic Hapluduli
.
Gibbsitic
22
-------
Table 2 -8 SOIL WATER CHARACTERISTICS BY PLOT
Plot
o.
Waler
table
depth
(cm)
Hydraulic
conductivity
Drainage
dais
Frequency of
Flooding Ponding
16 >100 Moderately .10w Well draincd None None
117 >100 Moderate
122 >100 Slow
155 42 Moderately well drained
159 16 Very slow Poorly drained Rare Rare
160 > 100 Rapid Well drained None None
166 72 Moderate
173 >100
174 >100 Moderstelyslow
181 40 Moderate Poorly drained
267 >100 Slow Somewhat poorly drained Rare
270 >100 • Wcll drained None
321 >100
347 >100
420 >100
508 > 100 Moderate
902 > 100 Very slow Somewhat exeeuively drained
1839 l5 Moderate Poorly drained Occasional Frequent
184 1 > 100 Slow None Common
1843 0
1954 30 Moderately slow Rare Rare
1955 75 Moderately well drained
1957 75 • Well drained None None
1958 10 Moderately well drained
1959 90 Moderately rapid Poorly drained Rare Rare
2072 >100 Moderately slow Well drained None None
2074 45 Moderate
2075 75 Moderately well drained Rare *
2185 >100 Moderately slow None
2186 >100
2187 >100 Moderately rapid Well drained
2188 l2 Slow Some’ hat poorly drained Rare Rare
2189 >100 Moderately rapid Well drained None None
2190 >100
2191 >100 Moderately slow
2192 >100 Moderately well drained Rare
2303 >100 Moderate Well drained None
23
-------
3 COMPONENTS OF VARIANCE AND
SAMPLING EFFICIENCY
David CSSSeI], John Hazard,
and Kurt Rjitters
3.1 INTRODUCTION
One of the primary objectives of the 1990 pilot
test was to estimate the components of vanance and
cost for each stage of sampling for the measured
variables. These results provide the ingredients for
evaluating the efficiency of the sampling design arid
sampling unit design. In addition, it is possible to
investigate the spatial “anability under several
intensities of subs.amphng for several measurements.
SpatiaT variability is important because, if the data for
a set of measurements were spatially correlated, then
the observations on those measurements could not be
assumed independent for the vanance analyses.
3.2 SAMPLING EFFICIENCY
Sampling efficiency was investigated by using
standard statistical equations to combine information
obtained about sample variances with different
assumptions of sampling cost to estimate optimum
sample sizes for the different sampling stages. A
hierarchical sampling model was used in all cases.
References to different sampling stages, and reduced
or full sampling models, all pertain to certain levels
and combinations of levels from this general
hierarchical model. The actual physical entities
associated with the sampling stages in specific models
for different measurements are identified in Table
3-I.
The models are labelled A, B, or C in Table 3-1.
Model A is always the three-stage model that
incorporates plots, subplots, arid stations. Model B
is always a variant of model A that omits one of the
stages. For the soil measurements, model B includes
only plots and soil pits. For the mensuration
measurements, model A is not defined and model B
includes only plots and subplots, because the response
variables are defined on either a per-unit area or an
average per-tree basis. For foliar chemistry
measurements, model B omits subplots but adds
another sampling stage of branches within trees.
Model C is only used for examining the foliar
chemistry measurements in Virguiia, where it was
possible to use a four-stage model with plots,
subplots within plots, trees within subplots, and
branches within trees.
3.2.1 Sample Variances
Sample variances for each stage of sampling were
estimated for all response variables as reported in
Sections 5 through 9. Estimates were made
separately for the New England and Virginia plot
locations. For some measurements, the estimates
were also made separately for subsets of the data that
had apparently different variance structures.
STAGES
AND MODELS
Sample unit
Sampling model
A
B
C
Visual Symptoms
Plot
I
I
—
Subplot
Tree
2
3
—
2
—
-
Mensuration
Plot
—
I
—
Subplot
—
2
—
Soils
Plot
1
1
—
Subplot
Soilpit
2
3
—
2
—
—
Foliage
Plot
I
I
Subplot
Tree
2
3
—
2
2
3
Branch
—
3
4
Vegetation Structure
Plot
I
Subplot
Vegetation station
2
3
—
—
—
—
—
PAR
Plot
I
—
—
Subplot
PAR station
2
3
—
—
—
—
Numbers in the table refer to sample stage
24
-------
For ciample, the photosynthetically active
radiation (PAR) analyses (see Section 9) suggested
estimating sampling variances under both clear and
cloudy sky conditions, with and without adjustments
for serial correlation, because the variances were
quite different in these situations. For some other
measurements, the estimates were made for different
sampling models by deleting an intermediate sampling
stage in the analysis (e.g., reducing a three-stage
model to a two-stage model). This was useful where
an instability of vanance estimates (probably caused
by very small variance components or small sample
sizes) precluded a meamngful analysis using the fuller
model.
3.2.2 Cost Components.
The cost components (Table 3-2) are based on the
average times required to set up and make
measurements on a new plot location, a new subplot
at a given location, and further subsampling units.
The subsampling costs for each set of measurements
are as follows: digging soil pits and collecting
samples, locating and measuring PAR stations,
locating and climbing trees, and obtaining and
processing branches and foliage, mapping tree
locations and measuring dbh and height, scoring
individual trees for a variety of visual symptoms
measures, and locating and measuring vegetation
structure Stations.
Two cost scenarios were esplored for each
measurement in each region. In the first scenario,
the cost associated with a new plot was the same for
all measurements but was higher in New England
than in Virginia because it took longer to locate new
plots there ( 4). The cost associated with a new
subplot was the same for a given measurement in
both regions, but the subplot cost was higher for the
mensuration and vertical vegetation measurements
than for the other measurements. Third-stage and,
where applicable, fourth-stage sampling costs were
the same in the two regions for a given measurement,
but the costs varied for the different measurements.
The second scenario was different in that the cost
of the last stage of sampling was set equal to one-half
of the cost actually required in the field Lest. These
reduced-cost scenarios were investigated to test the
sensitivity of optimum solutions to the relatively high-
cost activities associated with each measurement
(e.g., digging soil pits, climbing trees) or by the
practice of making many concurrent (but possibly
unnecessary) measurements on each tree, Station, or
laboratory sample. Some reduction of effort would
be needed to realize the lower costs in these
scenarios
Table 3-2 COST S
CEP4ARIOS FOR E
ACH REGION’
Southeast cost
No,i easi cost
Cost 11cm
model I model 2
model I model 2
Visual Symptoms
Plot
333 333
450 450
Subplot
021 021
021 021
Tree
019 010
019 010
Mensuration
Plot
3 33 3 33
4 50 4 50
Subplot
200 100
200 100
Soils
Plot
3 33 3 33
4 50 4 50
Subplot
021 021
021 02!
Soi p i1
200 100
200 100
Foliage
Plot
333 333
450 450
Subplot
021 02!
021 021
Tree
090 045
090 045
Branch
005 005
n /a n/a
Vegetation Structure
Plot
3 33 3 33
4 50 4 50
Subplot
0 42 0 21
0 42 0 2!
Vegetation station
0 05 0 05
0 05 0 05
PAR
Plot
3 33 3 33
4 50 4 50
Subplot
021 021
021 021
PAR station
0 02 0 01
0 02 001
Hours per un it sampled
25
-------
3.2.3 Sample Size Calculations
Variances and costs were obtained for the
measurements as discussed in Sections 5 through 9
and were substituted into the optimum sample size
equations provided by Sukhatme (1954) and Cochran
(1977) (Table 3-3). Solutions to these equations are
sensitive to the ratios of costs and variances between
different stages of sampling and to the absolute values
of those variables. The equations estimate the
optimum number of second-, third-, and fourth-stage
samples for a given scenario of costs, variances, and
sampling model. The first-stage sample here refers
to the number of field plots; optimum solutions for
this stage of sampling were beyond the scope of this
study ( 32.4).
Table 3-3. SAMPLE SIZE EQUATIONS FOR FOUR-STAGE
SAMPLING
Second stage
(C 1 IC2)°’x((MS 2 -MS k)/(MS ,-MS 2 /m)r
Third stage
A., = (C 2 IC,)°’ x (MS 3 - MS 4 /l) / (MS 2 - MS 3 /k) r 3
Fourth stage
L,. (C 3 1C 4 )03 x (MS 4 I (MS - MS 4 /l) r’
L2 these eqUatiOns
ri , A..,. and i,. are the estimated optimum sample sizes,
m, k, and I are the sample sizes used to estimate vanances in
the field study.
C 1 , C 2 , C 3 , and C 4 arc the per-unit sampling costa for stages
I through 4. respectively (Table 3-2), arid,
MS 1 . MS , MS 3 , and MS 4 are the mean squares for stages 1
through 4 obtained in the field study
By convention, the second-, third-, and fourth-
stage sample sizes are denoted by in, k, and 1,
respectively. Note that for some sampling models,
only in, or m and k, are estimated. For example, a
fourth-stage recommendation would not exist in a
two-stage sampling model. Table 3-1 lists the
physical entities associated with in, k, and I for each
sampling model.
The optimization equations were applied for each
defined scenario (cost, variance, sampling model) for
each response variable. This procedure generated a
large number of optimum solutions for in, k, and 1,
and it was necessary to reduce the number of
solutions to a manageable size to amve at
recommendations. For example, grouping the soil
response variables and making one overall
recommendation for ‘soil was preferable to
recommending different numbers of soil pits for
different soil measurements. Refinements of the
analyses may be warranted by further study, and the
full set of solutions is provided (Table 3-4) for this
purpose.
Thus, the measurements and associated optimum
sample sizes were grouped into several measurement
categories, and recommendations were made for
different sampling models for each category. The
sets and the number of response variables contained
within them were.
• visual symptoms variables for all species and for
selected subsets of species (6)
• mensurational variables (3)
• soils (21)
• foliage (28)
• vegetation structure (30)
• PAR(1)
Within categories that had a large number of
measured variables, frequency distnbutions of the
optimum solutions were charted arid a suggested
allocation was made by choosing an upper limit
above which 10% of the optimum solutions were
excluded. The choice of 10% was arbitrary, but v e
believe that the suggested allocations satisfy at least
90% of the solutions which fall in each set of
measurements There will be cases for which the
suggestions may not be optimum In other cases,
allocations were suggested by judgement There
seems to be little alternative to experienced judgement
when it comes down to making a recommendation
• The equations are condensed for three- and iwo-stage sampling
models
26
-------
3.2.4 Important Considerations
The optimum first-stage sample size is not
provided by the equations. The first-stage sample
size corresponds to the number of plot locations.
This determination is possible given the desired
precision for a parameter over the population of plots
under study, or given a total fixed cost for surveying
the population. But the solution should utilize
alternate optimization techniques that are appropriate
for the FHM first-stage sampling design. The
resource allocation formulas used here are
appropriate for simple random sampling; the FHM
design uses systematic sampling with post-
stratification which reqi ires different formulas.
These techniques have no been specified in detail at
this time. Although the Sukhatme-Cochran
estimators could be used, but the solutions would
ignore much of the design philosophy of FHM.
This emphasizes that the objective of this study —
determining an optimum plot design once a plot
location has been selected — gives little insight as to
how many plots are needed to characterize regional
forest health at some specified level of precision. On
the other hand, an optimum single-plot design is valid
no matter bow many plots are ultimately selected for
measurement.
Another important consideration is that the
optimality of plot design is evaluated for the objective
of estimating a population parameter over many plot
locations, and not for the objective of estimating that
parameter for any particular plot contained in that
population. Optirnality of plot design for a regional
survey does not imply that every site-specific estimate
will meet other precision requirements. This trade-
off between obtaining precise answers for each site
versus for all sites in a population is a fact that at
once makes large-scale surveys practical and limits
the inferences that can be made about any one
location.
A final consideration is that statistical optimality
is not the only criterion for plot design. For
example, it is sometimes necessary to ensure that
some measurements cover the area so that they
may be related to other measurements made in the
area. For example, suppose that the optimum
number of soil pits was two and the optimum number
of mensuration subplots was six. Judgment is needed
to decide whether two soil pits are sufficient to
represent the soil on all six mensuration subplots. In
many instances, a case can be built for over-
sampling to obtain the required coverage of the plot
The optimization analysis does not preclude this
effort, but it does make it clear why extra effort is
being made. These situations should be rationalized
individually as the roles and inter-relationships of the
measurements are clarified.
3.3 RECOMMENDED ALLOCATIONS
Table 3-4 lists, for each measurement that could
be analyzed as a continuous variable, the values of
the mean squares for each stage of sampling and the
calculated optimum values of m (m,,), k (k ,). and 1
(li,,) for the two cost models. Results are given for
different regions and sampling models To assist
interpretations of Table 3-4, subheadings within the
table identify the physical entities associated with the
sampling stages for the particular sampling models,
and Table 3-5 provides abbreviated definitions of
response variables.
Table 3-4 gives values of m , k ,, and 1 ,,, to the
second decimal place. This is so that the relative
sizes of the computed values may be seen In
practice, fractional sample sizes are rounded up to the
next largest integer. If the computed numbers are
between 0 and 1, and it is desired to retain that stage
in the sampling design, then the number is rounded
up to 2 so that a variance can be estimated for that
stage of sampling.
3.3.1 Visual Symptoms
Under the three-stage sampling model for all
species, the visual symptoms allocations indicate one
to two trees per each of three to four subplots, or a
total of three to eight trees per field plot location.
The two-stage sampling model yields comparable
values of two to five trees per field plot location, a
reflection of the fact that although subplot-to-subplot
variation is sometimes significant, it is not practically
important ( 5). However, a two-stage sample design
cannot be recommended automatically based on these
results because the two-stage analysis ignores any
clustering that was built into the three-stage sampling
design that was used. In general, these results
compare favorably with the Innes and Boswell (1990)
recommendations ( 5).
27
-------
The results obtained for selected species were
similar to the results for all species combined. Thus,
if interest centers on estimation by species, then the
recommendation should be applied for each species,
and this will provide more than sufficient numbers
for the estimates based on all species combined. On
the other hand, a plan for sampling just one tree per
each of six species would probably not give reliable
estimates for any one species alone.
3.3.2 Mensuration
The two-stage sampling model suggested an
allocation of one to two subplots. This result is
highly dependent on the “rot.ation of siibplots into
uniform forest types, ai d more subplots would
probably be required if ubplots were not rotated.”
In 1991, a decision was i nade to nor rotate plots, and
so the suggested allocations would apply to each
identified forest type at each plot location
3.3.3 Soils
Analyses of the soil response variables suggest
that two to three soil pits will be sufficient, regardless
of the region or cost model. The model of the nine
soil pits as three clusters of three pits was compared
with the model treating all nine pits as the second
stage of sampling. In this comparison, no evidence
was found to indicate that clusters of pits were more
efficient than single pits. It is therefore
recommended that three soil pits be systematically
arranged so as to represent the field plot location.
3.3.4 Foliage
The four-stage sampling model including
branches was estimable only for the Virginia data,
and in that case the suggested allocation is for 5 to 6
branches from each of 1 to 2 trees taken from each
of 2 to 3 subplots, or a total of 15 to 30 branch
samples. The three-stage model that did not include
subplots suggested about the same number of
branches per tree but fewer trees overall, apparently
subplot-to-subplot variation was relatively
unimportant. The three-stage sampling model for
New England suggested one to two trees per each of
two to three subplot.s, or a total of about five trees.
The branch-to-branch variation was not estimable for
New England ( 7). Generalizations of these results
to untested species may be tenuous.
If interest centers on a subset of elements, for
example the macronutrients N, P, K, Ca, and Mg,
then the four-stage model for Virginia suggests that
only two branches from each of two to three trees per
plot are sufficient (except for potassium, which
exhibited relatively large branch-to-branch variation).
If interest centers on heavy metals, then the same
model suggests relatively large numbers of branches
per tree (at least four), but not necessarily more trees
or subplots It has been suggested that a distinction
between routine monitoring of foliar nutrients and
special monitoring of foliar contaminants will enable
a practical number of branches to be sampled for
routine monitoring.
3.3.5 Vegetation Structure
Table 3-4 indicates that four stations per subplot
and six subplots will satisfy 90% of the 30 vertical
vegetation variables under cost model 1, and that
three stations per subplot and nine subplots will
satisfy cost model 2 in both regions. If variables
FF1, FF11, and FF21 (Table 3-5) are used as
examples, then three stations and five subplots will
satisfy the optimization requirements.
3.3.6 PAR
The PAR measurements require between two and
six subplots, with a minimum of two sampling
stations per subplot. Under ideal sampling
conditions (i.e , clear sky and homogeneous canopy),
two subplot.s are sufficient. More subplots are
required under cloudy sky conditions, and six
subplots would be appropriate for the most variable
canopies that were encountered in the study. The use
of 4 subplots and 19 stations per subplot in the 1991
pilot test is more than adequate for regional survey
purposes. The extra effort is justifiable if PAR will
ultimately be combined with growth data obtained
over a relatively large area at each field plot location
and the extra PAR measurements are needed to
represent this larger area.
3.4 SPATIAL ANALYSIS
There were three types of measurements that
were made on regular gnds that were fixed across all
plots: soils, vegetation structure, and PAR. The
consistent grid across plots permitted combining
information across plots to explore spatial structure
28
-------
3.4.1 Soils
Some soil measurements have been shown to
have spatial structure that can be effectively modeled
by spatial statistical methods (e.g , Robertson, 1987).
Therefore, the possibility of spatial con-elation was
explored in this study. The data for these analyses
came from the eight field plot locations in each
region at which nine additional soil pits were sampled
(Fig. 1-3). The nine pits were arranged in isosceles
triangles about each of the outlying subplots, which
permitted a spatial analysis.
3.4.2 PAR and Vegetation Structure
There was concern .tliat the 3.66-rn separation
among sample stations used for these measurements
might not be enough to avoid spatial con-elation
among stations. The natural structure of the forest
stands and its associated gap-phase dynamics might
cause measurements taken close together to be
con-elated.
The PAR and vertical vegetation structure
measurements were both sampled on 4 x 4 square
grids overlaid on the subplots (Fig. 1-4) Data for
the spatial analyses of PAR came from all four
subplots at each plot location, whereas data from
subplots I and 2 only were available for vegetation
structure analyses.
3.4.3 Spatial Statistical Methodology
The methodology used to analyze these problems
is the standard method of computing semivanograms
(e.g., Ripley, 1981). A semivanogram allows one to
visualize the variability between observations as a
function of the distance between the observations In
essence, for a pair of data points a given distance
apart, the variance is calculated as half the square of
their difference. Then the variance for all pairs of
data points a given distance apart is calculated as the
average of these pairwise calculations.
It is reasonable to calculate the variance for a
pair of numbers as half the square of their difference.
Consider two numbers, x and y, with a mean of z.
Then z must be halfway between x and y, and the
variance of the pair {x,y} is:
(x - z)’ + (y - z) 2
= ((x - y)12) 2 + ((x - y)/2)
= (1/4) X (x - y) 7 -4- (1/4) X (x - y) 2
= (1/2) X (x-y) 2
(2)
(3)
The semivariograms permit one to visualize at
what distance the spatial con-elation levels off, that
is, how far away the measurements can be when the
con-elation is roughly the same as for two points on
opposite sides of the plot
3.4.4 Results of the Spatial Analyses
The semivanogram analysis indicated that the
measurements examined in this study were taken
sufficiently far apart to be regarded as independent
Thus, spatial con-elation of the measurements does
not affect the recommended sample allocation
calculations. Also, the independence of
measurements implies that geost.atistical estimation
will not be necessary for the aggregation of mdix idual
measurements to the subplot level. This result may
or may not be obtained in other forest or soil types
that were not sampled in this study.
The soil measurements analyzed in this report
consistently showed no spatial structure. This is due
largely to the types of variables analyzed The soil
measurements that have been traditionally analyzed
with geostatistical methods are measurements for
which substantial theory exists to indicate spatial
structure, for example hydraulic conductivity. In
addition, the shortest distance between soil pits in this
study was approximately 42.2 m, and thus the spatial
analysis would not detect con-elation at a finer scale
All of the soil measurements appeared to be free of
spatial structure that could have complicated the
development of the plot sampling design. Thus, the
soil sampling design for the 1991 field study appears
to be reasonable.
29
-------
The vegetation s(ructure measurements were
analyzed separately for each of the 30 vertical foot
levels. The semivariograms consistently indicated
that the 3.66-rn distance between sample stations was
sufficient to avoid spatial correlation problems. This
makes sense if one envisions the gap-phase dynamics
of the forest stands as a homogeneous stochastic
process over an area. In this case, distances between
gaps are random and will therefore not appear
consistently at specified distances. Thus, spatial
correlation is probably not a problem for this set of
measurements.
Ve eeariori Sirucrure
The PAR measurements were analyzed in terms
of percentage of transmitted PAR. They showed a
strong spatial structure under cloudy sky conditions
but not under clear sky conditions. The spatial
structure detected under cloudy sky conditions was an
artifact; what was detected as spatial structure was
actually the result of serial (temporal) correlation on
cloudy days ( 9). It may be speculated that
stochastic gap-phase dynamics operate in the forest
stands sampled (see above). Another possibility is
simply that there were no gaps in the canopy that
were large enough to be detected as such by the
sampling protocols. In any event, the absence of a
spatial structure in these stands at distances greater
than 3.66 in indicates that spatial correlation will not
affect the recommended sample sizes.
Menn,r i ion
Table
3-4 OPTIMUM
SAMPLE SIZES FOR SCENARIOS
OF
COSTS AND VARIANCES ’
Respan e
variable
Sampling
model
Foreat
region
Mean aquare for atage
Coal model I
Coat model 2
I , ,, k,, rn,,,
i , , , k,, rn,,
1
2 3
4
subp!o
subplor
BA/ha
B
VA
259 829
77 265
—
—
—
0 73
—
1 03
BA/ha
B
NE 220 872
127 036
—
—
—
—
1.2.3
—
—
1 74
AVG_WF_D&CD
B
VA
30766
8013
—
—
—
—
068
—
096
AVG_HT_D&CD
B
NE
26 178
5 910
—
—
—
—
0 73
—
—
1 04
AVG_ liT_ALL
B
VA
12 893
3 095
—
—
—
—
0 65
—
—
0 92
AVO_HT_ALL.
B
NE
10 489
3 548
—
—
—
—
0 91
—
—
1 29
station
sub-
£LQ
station
sub-
212!
Ln(PTPAR-+- I)
Senal correlation
A
VA
0 56552
0 05870
0 00361
—
—
0 81
1 30
114
130
Serial core. adj
A
SE
0 54102
003182
000324
—
—
1 04
097
—
I 47
097
Seri*l c
A
NE
I 03478
0 17032
000753
—
—
0 68
1 92
—
0 96
1 92
Serial core. idj.
A
NE
1 04607
0 09865
0 00542
—
0 76
I 44
1 08
1 44
Max ubplol var.
A
VA
0 56552
0 54000
0 03300
—
—
0 80
4 45
—
114
4 45
Max bpIot var.
A
NE
I 04607
0 95566
0 05929
—
—
0 81
5 03
—
114
5 03
Fri
A
NE
1 830
0 225
0 128
—
—
2 22
116
—
1 57
1 65
FT2
A
NE
038972
026406
0 15052
—
—
2 23
3 2.5
—
1 58
4 60
FT3
A
NE
0 563816
0209375
0 124792
—
—
2.28
2 17
—
I 61
3 07
}74
A
NE
0 2891
0 1000
0 1017
—
—
3 02
2 05
—
2 14
2 90
FF5
A
NE
01184
01313
00825
—
—
234
506
166
715
FF6
A
NE
0 179194
0 101563
0 066354
—
—
2 39
2 85
—
1 69
4 03
FT7
A
NE
0244326
0270313
0083229
—
—
1 62
510
115
721
FF8
A
NE
0 17072
0 16875
007688
—
—
1 98
451
—
1 40
638
F 9
A
NE
0 158799
0 070313
0073646
—
—
3 07
2 39
—
2 17
3 37
FF10
A
NE
0 15551
0 06406
008760
—
—
3 54
2 2.5
—
2 51
3 19
ET II
A
NE
0102878
0 107813
0079688
—
—
255
474
- —
180
671
30
-------
Table 3-4 (CONTINUED)
1aIw,t £1 !
Response
srLsblc
Sampling
model
Forest
region
Mean square
for stage
Coal
4
model
I
m,,
Coal
1.,.
model
k,,
2
m
1 2
3
Vegetation Stricture
L ±
Iiano’
RL !
FF12
A
NE 0 07730
0 06875
0 07208
—
—
3 07
4 00
—
2 17
5 66
Fy13
A
NE 0211431
0089063
0090313
—
302
231
—
2 13
327
FT I4
A
NE 008)799
0023438
0060521
—
509
172
—
360
242
FT IS
A
NE 0102878
0 123438
0092604
—
—
257
553
—
182
783
FF16
A
NE 0 292352
0067188
0069896
—
—
3 06
2 06
—
2 16
2 91
FTI7
A
NE 0 190461
0 231230
0 103750
—
—
2 64
3.27
—
1 87
4 63
FT I S
A
NE 0070395
0190625
0071042
—
—
••
••
—
••
••
FF19
A
NE 0 197368
0 106250
0080833
—
—
2 59
2 74
—
I 83
3 88
P720
A
NE 0122615
0145313
0077813
—
—
216
549
—
153
776
FF22
A
. E 0 179194
0 085938
0069688
—
2 68
2 53
—
1 89
3 58
FT22
A
NE 0 089145
0 081250
0 065623
—
—
2 67
4 13
—
1 89
5 84
Fr2.3
A
NE 0 057237
0 103125
0 068750
—
—
2 42
13 66
—
I 71
19 32
FF24
A
NE 0114803
0140623
0079583
—
222
572
—
157
808
Ff25
A
NE 0 261431
0 123438
0 092813
—
—
2 57
2 51
—
2 82
3 55
FT26
A
NE 0 053947
0 031250
0047917
—
—
3 77
2 31
—
267
3 97
Fy27
A
NE 0 106826
0 057813
0070104
—
—
3 32
2 71
—
2 35
3 83
FF28
A
NE 0 129934
0 056230
0070833
—
—
3 39
2 34
—
2 40
3 30
FF29
A
NE 0079852
0107813
0074896
—
247
653
—
175
92.3
Ff30
A
NE 0 035855
0 015625
0 027500
—
—
4 08
2 31
—
2 88
3 26
FTL
A
VA 1 338816
0 465625
0 199375
—
—
1 92
1 80
—
1 36
255
FF2
A
VA 0 630263
0 237500
0 150625
—
—
2 36
1 88
—
I 67
2 66
PT)
A
VA 0356168
0 123438
0 109896
—
—
2 81
I 77
—
1 99
2 51
}T4
A
VA 0230263
0 043750
0 107917
—
—
495
119
—
3 50
1 68
FF5
A
VA 0349013
0065625
0110833
—
—
398
121
—
282
172
FF6
A
VA 0223355
0 131250
0 102708
—
— .
263
250
1 86
354
PT7
A
VA 0 173026
0 084375
0 076875
—
—
2 85
2 20
—
2 01
311
FF8
A
VA 0 162500
0 090625
0098333
—
—
3 13
2 39
—
2 21
338
FF9
A
VA 0 136431
0 120313
0094271
—
—
263
3 45
—
I 86
488
FF10
A
VA 0 087171
0 228225
0 096875
—
—
2 58
6 47
—
1 83
9 25
P711
A
VA 0 101563
0 082813
0077813
—
—
2 90
3 21
—
205
4 53
FF12
A
VA 0087089
0148438
0081563
—
—
219
940
—
155
1329
FF13
A
VA 0101563
0120313
0076563
—
—
236
470
167
665
P 114
A
VA 0129194
0164063
0085104
—
—
2 22
5 27
—
150
731
FF15
A
VA 0074013
0078123
0071875
—
—
286
409
—
202
578
FF16
A
VA 0 128290
0 103125
0083125
—
—
267
3 18
—
1 89
450
FF17
A
VA 0123273
0120313
0084063
—
—
248
380
—
175
538
FTI8
A
VA 0 236284
0 109375
0 093333
—
—
2 75
3 17
—
1 95
4 49
P729
A
VA 0155510
0204688
0091563
—
—
279
276
—
197
390
FF20
A
VA 0 072697
0 096875
0 076875
—
—
2 65
5 49
—
1 87
7 76
FF22
A
VA 0129194
0114063
0078646
—
—
246
346
—
274
490
FF22
A
VA 0178947
0087500
0092708
—
—
309
2 19
—
218
310
FF23
A
VA 0 144408
0 068750
0086458
—
—
3 39
2 24
—
2 39
3 02
FF24
A
VA 0061184
0078125
0063750
—
—
269
515
—
190
729
FT2.5
A
VA 0075658
0 222875
0084167
—
—
246
793
—
I 74
11 21
FT26
A
VA 0 126563
0 085938
0082396
—
—
2 93
2 77
—
2 07
3 91
FF27
A
VA 0.319984
0073438
0 100313
—
—
3 54
1 37
—
2 50
1 94
FF28
A
VA 0086431
0120313
0051354
—
—
192
594
—
136
841
FF29
A
VA 0085855
0065625
0066230
—
—
301
303
—
213
429
FT3 O
A
VA 0152878
0 242188
0082396
—
—
223
364
—
159
515
31
-------
Table 3-4 (CONTIN(JED)
Mean aqua
re (or agc
Co
ia model
I
C
oil mode
2 2
Responic
variable
Sampling
model
Forest
region
—
1,.
k ,
1 2
3
4 1,.
k.,.
m
Visual Sym Ioma
sub-
PLC A
VA 144224
53806
15997
—
—
058 253
—
080 253
A
NE 544 614
265 484
78 898
—
—
0 58 3 41
—
0 go 3 41
PDB A
VA 8 292
16 697
3 397
—
—
—
A
NE 86 385
30 704
27312
—
—
1 05 2 87
—
1 45 287
PTRN A
VA 186 102
105 758
59 569
—
—
0 84 3.34
—
115 3 34
A
NE 49373
72254
24726
—
063 1058
—
087 1058
PD IS A
VA 00
00
00
—
—
—
•‘
A
NE 16 122
9258
6.131
—
—
089 3 98
—
I 2.3 3 98
PDEN A
,VA 1460263
112 223
115 429
—
—
1.20 100
—
165 100
A
‘NE 683 954
162 730
102 994
—
—
0 87 2 31
—
1 20 2 31
PDEFOL A
‘ VA 329 458
153 493
125 160
—
—
1 04 2 84
—
1 43 2 84
A
NE 329 458
153 493
123 160
—
—
1 00 3 42
—
138 3 42
PLC- lobIol Iy A
VA 8377 049
182 96
68 572
—
—
0 66 1 45
—
0 91 1 45
PDB- loblol ly A
VA 0 879
1 987
0609
—
—
.
PrkN- lobroI ly A
VA 176 534
43 063
50 109
—
—
1.35 1 77
—
1 86 1 77
PDLS-loblo l ly A
VA 0 0
0 0
00
—
—
—
“
PDEN-loblolly A
VA 2543 082
52 362
81 321
—
—
1 68 0 58
—
2 31 0 58
PDEFOL-loblol ly A
VA 278 854
19 647
58 739
—
—
3 62 0 54
—
4 99 0 54
PLC-map+bch+brch A
NE 374 486
118 243
50 132
—
—
0 70 2 65
—
0 97 2 65
PDB-map+bch+brch A
NE 91 816
32 038
33 303
—
—
1 25 2 59
—
1 72 2 59
PTRN-map+bch+brch A
NE 18538
16307
12917
—
—
104 5.19
—
144 519
PDIS-map4bch+breh A
NE 16 524
7 236
7687
—
—
I 26 297
—
1 74 2 97
PDEN-map+bch+brch A
NE 483 698
78 125
73 604
—
—
817 1 70
—
1 61 1 70
PDEFOL-map+bch*breM
NE 280 485
91 305
108 446
—
—
1 37 2 42
—
8 88 2 42
PLC-VA oaks A
VA 305 313
192 574
89 824
—
—
0 74 3.36
—
1 02 3 36
PLC-other VA hdwd, A
VA 446 716
312 414
140524
—
—
072 357
—
100 357
PLC-NE rucc+fir A
NE 495 338
157 560
37440
—
—
0 55 2 55
—
075 2 55
PLC-other NEbdwdi A
NE 204 376
46354
38 181
—
1 24 1 74
—
1 71 1 74
rree
iree
PLC B
VA 1442 2.39
176 138
—
—
—
— 1 46
—
— 2 02
B
NE 544 624
88 479
—
—
—
— 1 96
—
— 2 7J
PDB B
VA 8 292
4 925
—
—
—
— 3 33
—
— 4 59
B
NE 86 385
27 548
—
—
—
— 2 78
—
— 3 83
rrR.N B
VA 186 102
64 878
—
—
—
— 2 52
—
— 3 47
B
NE 49 373
28 046
—
—
—
— 3 73
—
— 515
PDLS B
VA 00
00
—
—
—
—
—
—
B
NE 16 122
6 349
—
—
—
— 3 09
—
— 426
PDEN B
VA 1460 263
115061
—
—
—
— 118
—
— 163
B
NE 683 954
107 167
—
—
—
— 1 94
—
— 2 67
PDEFOL B
VA 198 023
71128
—
—
—
— 2 56
—
— 3 52
B
NE 329 458
127 139
—
—
—
— 3 06
—
— 4 22
il
L L
£
2 RiQ!
CEC_OAC A
VA 127 878
24 914
13 148
—
—
0 26 1 65
—
037 1 65
A
NE 475 463
44 075
82 683
—
—
0 73 0 88
—
1 03 0 82
CA_OAC A
VA 10 905
3 096
3 823
—
—
0 47 1 71 .
—
0 66 1 71
A
NE 84079
12 824
9478
—
—
032 1 61
—
045 1 61
.MG_OAC A
VA 0238
0026
0042
—
—
061 091
—
086 091
A
NE 0586
0 118
0 068
—
—
0 27 1 93
—
039 193
32
-------
Teble 3-4 (CONTINUED)
Responae
variable
SampIrn
model
Foreat
region
Mean aqua
re for wage
Co
mode
Ii
Coat mode
1 . ,. k,,,
2
m ,,
4 I.,.
k ,
at . , .
I 2
3
KOAC
A
VA 0 0)67
0 0022
0 0051
—
—
1 03
0 70
—
1 46
0 70
A
NE 0 0702
0 0073
00053
—
—
0 32
1.32
—
0 45
1 32
NAOAC
A
VA 000144
000034
000028
—
—
035
172
049
172
A
NE 000)78
000025
0000)7
—
—
030
156
043
156
AC_BACL
A
VA 243 228
54 888
12 203
—
—
0 16
1 89
—
0 22
1 89
A
NE 510 444
54 898
65473
—
—
046
1 20
—
064
1 20
AC_KCL.
A
VA 14981
400)
1494
—
—
021
202
—
030
202
A
NE 21 344
6 340
3 300
—
—
0 26
2 42
0 36
2 42
AL_KCL
A
A
YA 17.640
NE 24507
3 265
5 686
1 498
2208
—
—
—
—
0 24
022
1 63
2 17
—
—
0 34
031
I 63
2 17
N_MIN
A
VA 207.576
3 258
2 742
—
—
0 26
0 58
—
0 36
0 58
A
NE 9452 287
647 900
538 829
—
—
0 35
I 04
—
0 49
1 04
P_B1
A
VA 96 612
44 747
56 369
—
—
048
2 24
—
0 68
2 24
A
NE 399 674
169 447
299 584
—
—
0 67
2 08
—
0 95
2 08
S04_H20
A
VA 80220
17)02
143l7
—
—
035
162
—
049
162
A
NE 105331
28858
16688
—
027
228
—
039
228
S04_P04
A
VA 134464
7471)
37945
—
—
025
300
—
036
300
A
NE 19866 780
1802580
2070202
—
—
044
111
—
063
111
C_TOT
A
VA 8 479
2 359
1 566
—
—
0 30
1 95
—
0 42
1 95
A
NE 85 808
5 738
5 930
—
—
0 41
0 98
—
0 58
0 98
N_TOT
A
VA 00153
0 003)
00033
—
—
0 42
1 49
—
0 59
I 49
A
NE 03055
00178
00189
—
—
042
091
—
059
091
S_TOT
A
VA 0 00027
0 00005
0 00006
—
—
0 46
I 37
—
0 65
1 37
A
NE 000291
000019
000021
—
—
043
095
—
061
095
SAND
A
A
VA 4116619
Nt 442 004
198 276
60 462
88 244
59 041
—
—
—
—
0 2.3
0 39
081
1 44
—
—
033
0 55
081
1 44
SILT
A
A
VA 2858 736
NE 4.04 156
165 406
42 252
54 300
46 091
—
—
—
0 20
0 42
0 91
1 22
—
—
028
060
0 91
1 22
CLAY
A
A
VA 132583
NE 17878
11416
6941
15044
6512
—
—
—
—
050
038
089
2 56
—
—
070
054
089
2 56
ECH2O
A
VA 00440
00017
00013
—
033
068
—
046
068
A
NE 00343
00009
00014
—
—
058
052
—
082
052
PH_H20
A
A
VA 1 4.400
NE 2 3301
0 3130
0 2421
0 1092
0 1486
—
—
—
—
0 20
0 28
1 81
I 35
—
—
029
040
I 81
1 35
P}iO IM
A
VA 1 1413
02690
0 1096
—
—
022
1 87
—
031
187
A
NE 2.5002
0 2406
0 1707
—
—
0 31
I 28
—
0 44
1 28
U
CEC_OAC
B
VA 127818
16339
—
—
—
046
—
—
066
B
NE 475 .463
71 451
—
—
—
—
0 59
—
—
0 83
CA_OAC
B
B
VA 10905
NE 84 079
3 626
10 451
—
—
—
—
—
—
—
—
0 76
0 53
—
—
—
—
1 07
0 75
MG_OAC
B
VA 0238
0038
—
—
—
—
052
—
—
074
K_OAC
B
B
B
NE 0586
VA 0 0)67
NE 00702
0083
00043
00059
—
—
—
—
—
—
—
—
—
—
—
—
057
066
044
—
—
—
—
—
—
080
094
062
)4A_OAC
B
B
VA 000144
NE 000178
000029
000019
—
—
—
—
—
—
—
—
059
049
—
—
083
070
AC_EACL
AC_XCL
B
B
B
VA 243 228
NE 5)04.44
VA 14981
23 779
62396
2186
—
—
—
—
—
—
—
—
—
—
—
—
0 41
053
050
—
—
—
—
—
—
057
073
070
B
NE 21 344
4 185
—
—
—
—
0.67
—
—
095
33
-------
Table 3-4 (CONTINUED)
Responac
variable
Sampling
model
Forest
region
Mean aqua
re for cage
Co
4 1.,.
. mode
k...
I I
m .,
Co
1, .
i mode
k
1 2
m,
2
3
ALKCL
B
B
VA 17640
NE 24 507
1977
3 220
—
—
—
—
—
—
0 55
—
—
0 77
0 23
N_MIN
B
B
VA 207 576
NE 9452 287
3 424
570 559
—
—
—
—
—
—
—
—
0 37
—
—
—
—
0 52
P_B1
B
B
VA 96 612
NE 399 674
53 217
261 727
—
—
—
—
—
—
—
—
0 99
1 26
—
—
—
—
1 78
0 80
504_H20
B
B
VA 80 220
NE 105331
IS 072
20294
—
—
—
—
—
—
—
—
0 57
067
—
—
—
—
094
SO4 4
C_TOT
B
B
B
B
VA 134464
NE 19866 780
. VA 8 479
NE 85 808
47915
1990 907
1 781
5 874
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
079
0 48
0 60
0 39
—
—
—
—
—
—
—
—
0 68
0 85
0 56
084
N_TOT
B
B
VA 00153
NE 03055
00032
00186
—
—
—
—
—
—
—
—
060
037
—
—
—
—
053
0 87
S_TOT
B
B
VA 0 00027
NE 000291
0 00006
000020
—
—
—
—
—
—
—
—
62
039
—
—
—
—
056
031
SAND
B
B
VA 4116619
NE 442004
117586
59447
—
—
—
—
—
—
—
—
022
055
—
—
—
—
078
031
SILT
B
B
VA 2858736
NE 404 156
83928
44 994
—
—
—
—
—
—
—
022
050
—
—
—
—
0 71
060
CLAY
B
B
VA 132583
NE 17 878
14077
6 635
—
—
—
—
—
—
—
—
042
093
—
—
—
—
1 32
033
EC_} 120
B
B
VA 0044
NE 00343
00014
00013
—
—
—
—
—
—
—
—
029
—
—
—
—
041
0 62
PH_H2O
B
B
VA I 4400
NE 2 3301
0 1645
0 1758
—
—
—
—
—
—
—
—
044
041
—
—
—
—
0 59
067
PH_OIM
B
B
VA 11413
NE 2 5002
01528
0 1911
—
—
—
—
—
—
—
—
048
042
—
—
—
—
0 59
Foliage
sub .
Iree
sub-
£Lc !
C
A
NE 2006804
0234368
0246129
—
—
072
111
—
102
025
111
242
N
A
NE 0405718
0107428
0013414
—
—
018
242
—
57
1 87
S
A
NE 0 006545
0001348
0 000699
—
—
0 40
1 87
—
099
SI
A
NE 8373542
524503
301267
—
—
043
099
—
061
082
268
HG
A
NE 294788
142000
118611
—
—
058
268
—
0 71
5 33
ZN
A
NE 42 71078
51 86667
36 34722
—
—
0 50
5 33
—
1 23
P
A
NE 34532.3 4
63904 5
80735 6
—
—
0 89
1 25
—
1
1 28
FE
A
NE 657 9451
102 1333
108 4722
—
—
0 73
1 28
—
1
3 04
CU
A
NE 6 266667
4 966667
5 944444
—
—
0 83
3 04
—
266
MN
A
NE 6145787
2665808
1853607
—
—
050
266
—
42
MG
A
NE 358595 5
120480 4
67028 7
—
—
0 42
2 42
—
0 60
2 15
NA
A
NE 32.341 31
10385 77
832867
—
—
056
2 15
079
220
CO
A
NE 4962
2200
2486
—
—
078
220
—
110
052
AL
A
NE 1313 930
65 633
98 903
—
—
119
052
—
1
04
1 74
NI
A
NE 9 507
2 633
2 819
—
—
0 73
1 74
—
075
CA
A
NE 85852128
3703944
2950420
—
—
056
075
—
1 48
0 80
K
A
NE 12520476
1203495
1686706
—
—
1 05
0 80
—
058
083
TI
A
NE 5884363
254333
13.4306
—
—
041
083
—
0 74
0 87
CR
A
NE 135 3069
7 5333
5 5972
—
—
0 53
0 87
-
1 78
074
PB
A
NE 48051
51 70
7990
—
—
1 26
074
300
200
CD
A
NE 496177
1.70000
1.76389
—
071
200
—
067
091
SR
A
NE 2010593
113 167
72847
—
—
047
091
—
34
-------
Table 3.4 (CONTINUED)
sub-
branch itsi 212!
sub-
branch W 2121
Reaponae
var iable
Sampling
model
Forts !
region
Mean square
for Mage
Coal
model
1
Coa l model
1,. k,.
2
in
4 1 ,,
k,,
m,.
1 2
3
— — 072
— — 131
— — 065
— — 070
— — 041
— — 038
273
0 92
177
204
272
1 94
— 102
— 186
— 092
— 099
— 059
— 053
2 73
0 92
177
204
2 72
1 94
branch itLt
branch f.
Fo li a je
AS
MO
U
BA
V
B
C
N
S
S I
HG
Z N
p
FE
CU
MN
M C
NA
CO
AL
NI
K
11
CR
PB
CD
SR
A S
MO
LI
BA
B
CA
V
C
N
$
SI
HG
ZN
p
F E
CU
MN
MG
NA
A NE 3984
A NE 45 47157
A NE 6014464
A NE 1121359
A NE 5790
A NE 7784637
B . VA 6 539817
B . VA 0130519
B VA 0 085073
B VA 1106330
B VA 88548
B VA 499 9286
B VA 2056935
B VA 2216468
B VA 2254492
B VA 962181 3
B VA 4000427
B VA 38771 44
B VA 631746
B VA 2564009
B VA 16 57937
B VA 16736017
B VA 259 873
B VA 221 873
B VA 4942341
B VA 1680397
B VA 304 0833
B VA 207 9563
B VA 28 67857
B VA 856 1786
B VA 8938 329
B VA 1163016
B VA 3917473
B VA 5151567
C VA 6539817
C VA 0130519
C VA 0 085073
C VA11063300
C VA 88 54762
C VA499 92860
C VA20569350
C VA 2216468
C VA 2254 492
C VA 962181 3
C VA 4000427
C VA 3877144
236 1
7 93333
1537 267
3951 37
2.3 70
166 1667
0 536545
0 035704
0 051016
301815
219 169
149 8476
65038
414 171
120 352
487185
23364 41
10301 2.3
1 09524
29565 3
2 78810
1163007
22 893
8 581
34 1476
198167
8 7190
196 8595
8 29762
832119
6985 776
38 2286
382916
37 78268
o 612767
0 046871
0050651
3174805
253 22620
263 01190
53448 8
484 036
267 345
60898 5
1513624
1110426
249 5
1248611
1464 181
4056 135
12 76
77 6806
0 217495
0 003601
0 050076
72292
192 393
31 9286
4709
31 810
79 762
3591.4
8486 63
7591 25
0 76191
1448 9
0 77381
1333591
2.3 773
12 857
38 2.381
32 3452
12 7619
108 8690
8 10714
50 7024
6435 548
10 5952
57114
26 81707
0 485730
0 028260
0051260
291371 3
196 46430
74 40476
72764 8
367 595
22 357
40598 5
28849 96
9765 87
—
—
303
050
—
214
070
—
—
138
tOO
—
098
142
—
—
589
112
—
417
158
—
—
221
096
—
156
136
—
—
531
296
—
375
418
—
—
207
102
—
147
144
—
—
116
034
—
082
048
—
—
120
083
—
085
117
—
—
422
037
—
299
052
—
—
117
043
—
083
060
—
—
283
140
—
200
198
—
—
4.58
081
—
324
114
—
—
438
066
—
3)0
093
—
—
095
065
—
067
092
—
—
241
074
—
170
105
—
—
696
033
—
492
047
—
—
624
040
—
441
056
—
—
1037
019
—
733
027
—
—
677
034
—
479
048
—
—
1264
029
—
894
040
—
—
991
017
—
701
024
—
—
371
173
—
262
245
—
—
586
076
—
415
107
—
—
397
050
—
281
071
—
—
554
1.34
—
392
189
—
—
241
105
—
170
149
—
—
170
058
—
120
082
—
—
445
141
—
315
200
0 217495
0 003601
0 050076
72292 I
192 39290
31 92857
4709 6
31810
79 762
3591 4
8486 63
7591 25
3 22
157
5 86
2 26
5 88
3 14
110
I 28
1 29
2 49
4 78
0 49
0 43
0 49
o 59
0 39
0 25
0 98
0 52
a.
0 47
2 84
0 47
0 96
2 13
2 41
1 65
24 37
2 95
0 36
1 52
a.
0 83
0 57
1 68
2 28
111
4 15
1 60
4 16
2 22
0 78
0 90
a.
91
1 76
3.38
0 69
061
0 70
0 83
0 55
0 35
139
0 74
0 •
0 67
4 02
0 67
0 96
213
241
2 65
24 37
2 95
0 36
1 52
..
0 83
057
1 68
35
-------
Table 3-4 (CONTINUED)
Response
variable
Samphng
model
Forest
region
Mean sq
usre for agc
C
I
os* model
k
1
m
Ii,.
Cost modc
k,
I 2
m
1
2
3
4
L
£th
[ oria2e
branch
LE
branch
LQJ
CO
C
VA
631746
1 48810
0 83333
076191
551
031
1.71
389
044
1 71
AL
C
VA
256400 9
27653 7
30839 6
1448 9
0 93
0 76
0 89
0 66
1 07
0 89
Ni
C
VA
16 57937
4 27381
1 79762
0 77381
3.14
0 31
1.88
2 22
0 44
1 88
I C
C
VA
16736017
1207083
113362.3
1333591
7 17
041
0 79
5 07
0 58
0 79
TI
C
VA
2.59.87
20 11
24 75
2.3 77
5 77
0 62
0 70
4 08
0 88
0 70
CR
C
VA
221 873
10 738
7 143
12 857
18 00
0 15
0 72
12 73
0 22
0 72
PB
C
VA
494 2341
35 4048
33 3095
38 2381
6 96
0 42
0 79
4 92
0 59
0 79
CD
C
VA
1680397
194167
200833
323452
1220
031
096
863
044
096
SR
C
VA
3040833
54405
109048
127619
••
••
••
•s
•
e•
AS
C
VA
207 956
240 345
167 869
108 869
4 16
0 41
4 40
2 94
0 58
4 40
MO
C
. VA
2867857
1008333
7.10714
810714
691
033
202
489
047
202
LI
C
. VA
856 1786
70 1905
91 8929
507024
370
080
068
262
113
068
BA
C
VA
8938 329
8061 905
6268 357
6435 548
6 16
0 38
3 54
4 36
054
3 54
B
C
VA
116 30 16
276071
453095
105952
2 1$
1 37
086
1 54
1 94
086
CA
C
VA
3917473
2.53486
469202
57114
1 53
233
028
1 08
3 30
028
V
C
VA
5) 51567
36 65909
38 53175
26 81707
4 38
0 58
2 65
3 10
0 82
2 65
Ast ns1s n i eample size column indicate that the sample size equations yielded indel irute solutions Sec Table 3.1 for deuirutiorta of sampling
models and stages See Table 3.2 for dcuinjuons of cost scenarios
36
-------
Table 3-5 RESPONSE VARIABLE DEFINITIONS ’
Reapcnae
Responac
vunabic Description
v.nable Deacription
V eual Symptoms Folta2e
PLC Percentage of live crown C Carbon
PDB Percentage of crown dieback N Nitrogen
PTR.N Percentage of foli.ted crown S Sulfur
transparency SI Silicon
PDIS Percentage of foliar discoloration HG Mercury
PDEN Percentage of crown density ZN Zinc
PDEFOL Percentage of crown defoliation P Phosphorus
FE Iron
CU Copper
MN Manganese
MG Magnesium
CEC_OAC Catibri exchange capacity by NA Sodium
amrnoruum acetate CO Cobalt
CA_OAC Exchangeable calcium in ammonium AL Aluminum
acetate NI Nickel
MG_OAC Exchangeable magnesium in ammonium K Potassium
acetate TI Titanium
K_OAC Exchangeable posauium in ammonium CR Chromium
acetate PB Lead
NA_OAC Exchangeable sodium in emmonium CD Cadmium
acetate SR Strontium
ACBACL Exchangeable acidity in barium AS Arsenic
chlonde inethanolarnine MO Molybdenum
ACKCL Exchangeable acidity in potassium LI Lithium
chionde BA Barium
AL_KCL Exchangeable aluminum in potassium B Boron
chlonde CA Calcium
NMIN Mineralszable nitrogen by incubation V Vanadium
(ammonium nitrogen)
P_El Extractable phosphorus in Bray & Mensuration
Kurtz No I cxtractznt
504 fl20 Extractable sulfate in deionized water BA/ha Basal area per hectare
504_P04 Extractable sulfate irs sodium phosphate AVG_HT_D&CD Average height of dominant and co-
C_TOT Total carbon dominant trees
N_TOT Total nitrogen AVG_HT_ALL Average height of all treeS
STOT Total sulfur
SAND Sample fraction with particle diameter ItFiictlJre
between 0 05 mm and 2 00 mm
SILT Sample fraction with particle diameter FT.r,x 1,2, 3 30 Presence or absence of vegetation (any
between 0 002 mm and 0 050 mm class) at the indicated height (ft)
CLAY Sample fraction with particle diameter
Jeu than 0 002mm Growth efficiency
EC_R20 Eiectncal conductivity in deionized
water Ln(FTPAR+ I) Natural loganthm 01(1 + the estimated
PH_R20 pH in deionized water fraction of PAR tran3miiled through the
PHOIM pH in 0 OIM calcium chloride canopy)
Scc Scctiona 5 through 9 for details
37
-------
4 SYNTHESIS OF LOGISTICS
ANALYSES
Mike Papp and John Pemberton
4.1 INTRODUCTION
One of the objectives of the pilot test was to
quantify the time and fiscal constraints of making the
measurements (Palmer et al., 1990). It is also
important to report on the development of the
infrastructure for field implementation and to
recommend improvements. This section will discuss
the operations that occurre at the following stages of
logistics:
• staffing and personnel
• procurement and inventory control
• training
• reconnaissance
• sampling
• communications
• safety
Much of the discussion and many
recommendations are based on a debriefing of field
crews that was conducted in Richmond, VA, on
September 5 through 7, 1990, near the end of the
field season.
4.2 STAFFING AND PERSONNEL
Figure 4-1 depicts the organizational structure of
the field staff.
4.2.1 Description
During the field measurement phase of the study,
the project manager ensured that monitoring was
accomplished in the region and that information vital
to the field monitoring effort was disseminated
properly.
Regional project leaders supervised the daily field
monitoring activities. These individuals disseminated
information (progress and problems) to the project
manager and to the field crew. They were available
for daily communications with the field crews. The
Virginia project leader also performed reconnaissance
and assisted the crews by shipping samples.
A member of the field crew was designated a
crew leader. The crew leader supervised all field
operations and resolved any discrepancies or issues at
the measurement site. Other duties of the crew
leader were:
• maintaining and revising sampling schedules and
travel itineranes,
• assigning duties to crew members according to
sampling priori ties,
• ensuring that all sampling protocols were
followed,
• ensuring proper use and maintenance of field
equipment,
• maintaining the integrity of the site and samples
collected,
• reporting any problems or difficulties to the
proper management staff, and
• returning all field equipment and supplies.
Sampling required a five-person field crew
comprised of the following individuals
• Two foresters for measuring core variables and
visual symptoms These individuals were FIA
foresters with work-related experience in
mensurational type measurements.
• One soil scientist for sampling soils and
describing pedons. These individuals were Soil
Conservation Service (SCS) soil scientists that
were familiar with National Soil Survey
characterization and sampling methods.
• Two foliage samplers for obtaining foliage
samples and for measuring vegetation structure
and PAR The individuals were professional tree
climbers.
One field crew was used in New England and t\ o
field crews were used in Virginia.
4.2.2 Discussion and Recommendations
The organizational structure of the field staff was
appropriate. The field crews were adequately staffed
to meet the measurement objectives of the project.
Two professional tree climbers worked in each crew.
They were obtained from a contractor and made
several measurements. Two (rather than just one)
were hired for safety precautions but this may not be
mandatory. The climbers had no experience
identifying species for the vegetation structure
measurements, but this knowledge would have been
beneficial. The tree climbers were the most
38
-------
Figure 4-I ORGANIZATIONAL CHART FOR FIELD STAFFING.
expensive personnel, and the measurements they
made could have been made less expensively by
forest technicians or undergraduate forestry students.
Since tree climbers were required only for foliage
sampling, it is recommended that methods other than
tree climbing be evaluated for collecting foliage. If
tree climbing is the only viable method, then it may
be beneficial to train personnel within the Forest
Service to climb trees. If another foliage sampling
method can be developed, then forest technicians or
undergraduate foresters are a better choice as field
crew members.
The SCS soil scientists worked through an
Interagency Agreement with the USFS. Funding had
uot been transferred to the SCS by the time sampling
began. It is recommended that requirements for the
project be evaluated with enough lead time to ensure
the transfer of funds before the field work begins.
The field crews were responsible for prepanng
and shipping samples of roots, branches, tree cores,
foliage, and soil. This task was accomplished after
field sampling each day and it took one person
approximately 1 b to complete. Shipping these
samples to the appropriate laboratories was time-
consuming and it occurred at the convenience of the
field crew. Sample integrity can be improved if
samples are shipped at more frequent intervals. It is
recommended that the crews be assisted by another
person in the preparation and shipment of samples
Perhaps one person could assist a number of crews
working in a specified area.
4.3 PROCUREMENT AND INVENTORY
CONTROL
4.3.1 Description
EPA and the USFS procured supplies for the
study. About 3 months before field sampling began,
a letter was sent to each measurement coordinator
asking for a list of the specific equipment and
consumables that would be needed for the
measurements. This list was compiled and sent to the
39
-------
regional project leaders, who in turn indicated what
equipment was already available for use in the study.
The remaining equipment and consumables were then
purchased. The list of all equipment and
consumables is reproduced in Dwire et al. (1990).
Two weeks before field sampling activities began,
all equipment and consumables (except coolers and
gel-pacs) were sent to the project leaders An
inventory that was provided with each shipment was
checked upon receipt. The equipment was separated
by field measurement and then divided among the
field crews. After sampling activities were
completed, EPA equipment and consumables were
shipped back to the Environmental Monitoring
Systems Laboratory at Lis Vegas (EMSL-LV) for
inspection and inventory. The majonty of items were
available before field sampling and in appropriate
quantities. Therefore, calls for resupply were
infrequent.
4.3.2 Discussion and Recommendations
Measurement coordinators were fairly responsive
in supplying information on the equipment and
consumable needs for their measurements. Three
items — sample labels, disposable gloves for foliage
sampling, and the vegetation structure data entry
form — were in short supply at the beginning of
sampling and were supplied later. This was caused
by insufficient lead time for acquiring these
consumables. It is recommended that measurement
coordinators be more specific about the items they
require and that all their equipment and consumable
requirements be provided to the logistics coordinators
at least 3 months prior to the start of field sampling.
One of the radiometers malfunctioned during field
sampling and was replaced with a spare that was on
hand. Certain items were lost during field activities
(garden trowels, knives, etc.), but these were not
critical items. It is recommended that spare
equipment be available for all items; a rule of thumb
for expensive and/or critical equipment (portable data
recorders, radiometers, laptop computers, etc.) may
be one extra unit for each four field crews.
Replacements for less expensive equipment should be
available in higher quantities. Replacement
equipment should be located strategically to be
available to field crews within I day. The proposed
assistant (discussed in the staffing section) could
facilitate the resupply of lost or damaged equipment.
Root sampling equipment can be cleaned by
either alcohol or bleach. Due to the fire hazard of
alcohol, a 1:5 dilution of bleach was the more
acceptable cleaning agent. In Virginia, the spray
bottle that held the bleach leaked and stained some
clothing. Alcohol should be re-evaluated as an agent
for cleaning root sampling equipment.
The coolers and gel-pacs needed for storing and
shipping samples were resupplied to the crews from
the receiving laboratory. It was difficult to maintain
the gel-pacs in a frozen state. The refrigerators that
were supplied to each field crew were not identical.
The New England field crew received a refrigerator
in which the entire compartment could be frozen,
thereby maintaining the gel-pacs in a frozen state.
The Virginia refngerators had smaller freezing
compartments and could not always maintain frozen
gel-pacs All refrigerators were cumbersome, yet
they had to be moved into and out of hotel rooms.
Gel-pacs had to be changed daily to preserve the
samples.
The proposed assistant should be made
responsible for receiving samples from field crews
and maintaining them. The crew should be supplied
wth a refrigerator/freezer that can be used to
maintain gel -pacs or samples for 1 or 2 days until the
assistant receives the samples. The assistant can also
resupply the field crews with frozen gel-pacs
After the completion of field sampling, the field
sampling equipment and consumables were sent
back to the agency that purchased the equipment or
back to the regional Forest Service Experiment
Station. This equipment will be used within each
region in the future. Once a region is ready for full
implementation, equipment for field sampling should
be stored in one location in each region and be
inspected and inventoned at the completion of
sampling. Defective equipment should be repaired or
replaced as soon as possible. New items and
consumables should be ordered for sampling in
following years.
In some instances, the vehicles used to transport
personnel and equipment were not large enough or
did not contain an enclosed area where equipment and
consumables could be protected One vehicle, with
an enclosed area large enough to transport equipment,
consumables and samples should be available for each
field crew to prevent loss, theft, or destruction.
40
-------
4.4 TRAINING
4.4.1 Description
A pretraining workshop and a training workshop
were conducted prior to field activities. The
pretraining session enabled trainers (the measurement
coordinators) to test the planned field sampling
methods under reahstic conditions. Two weeks after
the pretraining session, the trainers conducted the
field crew training session This meeting included an
overview of the FIlM project, discussions of each
measurement and of quality assurance and logistics.
The field training of crew members included a full
day sampling exercise and testing and certification
(Burkman and MickJer, 1990). The last day of
training was used for debneling crews and revising
methods as needed.
4.4.2 Discussion and Recommendations
The pretraining workshop was useful for
identifying potential sampling problems and for
assisting the trainers in developing their training
procedures. As more regions become involved in
FHM, more trainers wilt be needed. It is
recommended that the pretraining session be
continued to train a set of trainers since measurement
coordinators cannot be present at all regional training
sessions. This procedure will ensure consistent
training and sampling protocols across regions.
The training workshop accomplished its objective
of training field crews in the correct sampling
methodologies. All field crews members were
evaluated by the trainers and certified for competency
in the methods for which they were responsible.
Crew members did not receive training in every
sampling method nor were they briefed in general
about other methods or the samples deriving from
those sampling methods. Field sampling in Virginia
began about I month after training. The interval
between training and implementation led to some
confusion during the first week of sampling as to the
number and types of samples needed as well as the
protocol for sample collection.
The training session should allow time for crew
members to observe all sampling techniques in order
to get a better “feel” for what the field crew must
accomplish in a day. It may also be worthwhile to
train at least two crew members for each sampling
technique in case one crewperson becomes ill during
the field season. Additional training of crew leader
responsibilities is necessary. If sampling methods or
any information affecting field sampling are revised
during training, these revisions should be discussed
up during the debnefing session, documented, and
sent to all field crew members for insertion into the
field sampling manual. It is recommended that the
training session be conducted shortly before the field
season begins. In addition, safety training should be
included in the training agenda
4 5 RECONNAISSANCE
4.5.1 Description
Reconnaissance activities include acquiring the
most recent aerial photos, topographic maps, arid
prior field tally sheets (if applicable), determining
cover types from the aerial photos, identifying and
contacting landowners to gaui access, and marking
the travel routes to the plots for the field crews to
follow.
In New England, reconnaissance included the
relocation of the FIA plot monuments In Virginia,
new plots were established and it was not necessary
to relocate plot monuments.
4 5 2 Discussion and Recommendations
The Virginia regional project leader performed
the reconnaissance activities for Virginia. in New
England, reconnaissance was not conducted for the
nine plots in Maine, but a person was identified to
help with reconnaissance for the remainder of the
plots.
Reconnaissance proved to be beneficial and cost
effective. The Virginia crews did not have to spend
any time locating the starting point” before
proceeding to the plot location. In Maine, on the
other hand, it required an average of 1.5 h to locate
the starting point. This translates to 7.5 h of lost
sampling time for a five-person crew. Some starting
points were never located by the New England crew
without reconnaissance. Although the New England
and Virginia regions bad different distances between
plots and different weather conditiois, the difference
in the weekly average number of plots completed by
a crew in New England (2.3 without reconnaissance
4]
-------
and 3.0 with reconnaissance) as compared to Virginia
(3.7 with reconnaissance for all sites) partly reflect
the difference in reconnaissance. With
reconnaissance, more of the field crew’s time is
available for sampling, and this may improve data
quality.
Reconnaissance information should include
lodging facilities closest to the plot, express mail
facilities in the vicinity, and emergency facilities.
Some of this information may be provided by a
logistical support group at regional or national
centers.
Because the plots in Maine were not
reconnoitered before sampling, the field crews spent
time locating these plots ralher than measunng them.
I n some instances, plotswere not found even after
considerable effort. One location that was never
located was on a steep slope and this made it difficult
for crew members to carry some of the heavy and
cumbersome equipment to the starting point. Because
certain crew members were under contract for either
a fixed period of time (SCS soil scientists) or for a
fixed funding limit (professional tree climbers) the
crew leader was sometimes uncertain as to how much
time to expend finding a site vs. moving to an
alternate site. The crew leader was asked to balance
the constraints of the sampling design against the
measurement objectives that required 20 sites be
visited within the time arid funding constraints.
Criteria for elinunatirig plot locations should be
developed by the measurement coordinators in
cooperation with the statistical design and logistics
coordinators. Accessibility, hazardous situations, or
the length of time or funds required to locate and
sample a site may be considered as criteria to
determine the appropriateness of a given site
4.6 SAMPLING
4.6.1 Description
Plot sampling included the following activities
(Table 4-I):
• travel to site
• site location
• plot establishment
• data collection activities
• sample maintenance and transport
• data saving and transmission
Travel to the site included outfitting (checking for
all equipment and consumables needed at the site) as
well as the travel from the place of lodging to the
place where walking was necessary. Two vehicles
were typically used for travel to the sites Travel
times were comparable in New England and in
Virginia. Travel between sites was greater in New
England because the sites were widely distributed
across New England whereas in Virginia the sites
were confined to the eastern third of the state.
Site location included the time required to find
the starting point and the plot center once out of the
vehicle. The FM foresters took responsibility for
plot establishment. This task entailed locating all
four subplot center points, establishing four
vegetation structure-PAR plots, locating three soil
sampling holes, and identifying four or six sample
trees. The sequence of plot establishment as as
follows (Dwire et al., 1990).
I. Establish subplot number 1 center location.
2. Locate the first soil sampling hole and subplot
number I sample trees.
3. Set up witness trees at subplot number I and
collect data on regeneration plot and vegetative
profile. Establish PAR grid on subplot 1.
4. Locate subplots 2, 3, and 4. Establish subplot
centers and locate soil sampling holes 2 and 3
and remaining sample trees. Tally regeneration
plots and select sample trees. Install PAR grid
Table 4-1 SUMMARY OF TIME REQUIRED FOR CREW
ACTIVITIES
Activity
Region
New England
Virginia
SLart time
7 00 a m
7 30 a m
Travel time (r 1)
3 ii
3 h
Plot location
I 5 Ii
20 mm ’
Plot eatablishment
50 nun
50 mm
Data collection
6 h
6 h
Average length
of workday
113 Ii
102 Ii
Sample maintenance
arid transport
I Ii
0 h’
Dat . saving and
ira nIffuiliOn
20 nun
20 nun
Number of site, per
crew per week
2 751
3 7
• Accomplished by. non-crev reconnaisaancc person
Average tncludei time with and without reconn.isu.snce (see text)
42
-------
Data collection activities included the
measurements and sample collections for the five sets
of measurement variables (Section 1). Table 4-2
highlights the approximate sampling times allocated
for different measurements. Sampling crews used
more than one measurement technique for some
indicators. This information is presented in Sections
S through 9 in more detailed discussions of time and
feasibility.
The two foresters were responsible for
mensuration arid visual symptoms measurements.
The soil scientist collected pedon data and soil
samples. The two tree climbers measured vegetation
structure and PAR, and collected samples of foliage,
branches, tree cores, •ahd roots. Figure 4-2
delineates the sequence of these sampling activities.
The samples of roots, tree cores, branches,
leaves, and soils were transported from the field and
stored in coolers with gel-pacs or refrigerators to
preserve the samples. The samples were boxed with
gel-pacs and shipped by overnight express to various
locations for preparation and analysis. Express
shipping was necessary to maintain sample integrity.
Maps of the Federal Express main offices were given
to the crew leaders to assist in locating drop-off
facilities. Shipping labels and packing slips were
prepnnted to prevent shipping errors.
At the end of a field day, the crews transferred
the measurements stored on the portable data recorder
(the tree tally and soil characterization data) and the
radiometer (PAR data) to a laptop computer. These
files were then transferred via modem to a computer
center. Data transfer was required to store data at
a central location. It was also useful for sending data
to the quality assurance staff so that venfication
programs could be run that could riot have been run
in the field. Data transfer can be used for sample
shipment and tracking in the future.
The general work schedule was five 10-h days
(50 h/week). Table 4-I shows that a day averaged 11
h and 20 mi ii in New England and 10 h and 10 nun
in Virginia. The difference was the time required to
locate plots. This time estimate does not include
sample maintenance and transport, or data transfer,
activities that required just one person and I b and 20
mm in New England and 20 mm in Virginia. The
Virginia estimate is lower because the regional
project leader performed this activity.
Table 4-2 TIME ALLOCATED FOR DATA COLLEC7ION
BY MEA5UR 4ENT
Mc4luremcnl let
No people
Houri/
person
Toizl
hours
Soils
1
8
8
Foliage
1
3
3
Vegetation l irucwre
2
25
3
Growih eff cien y (PAR)
1
2
2
Mensuration and
visual syrnptonu
2
5
tO
Total
28
4.6.2 Discussion and Recommendations
In New England, travel time was higher because
of a lack of extensive road systems and a lack of
knowledge of the lodging facilities that were close to
the plots. Reconnaissance should identify adequate
lodging facilities as close to the plots as possible. In
the future, sampling crews may attempt to sample a
number of sites from one lodging location. This is
considered efficient if lodging is equidistant from all
sampling sites.
In New England, site location averaged 1.5 h
(range 0.5 to 5 h). Most of this time was for
locating the starting points from the FIA tally sheets.
In some instances, the crew left their vehicles and
walked to the starting point, at other sites, crews
drove to the starting point. Crews had difficulty
finding starting points for various reasons Some
starting points had been removed by logging or
natural disturbances. Crews received insufficient
information about the starting points. For example,
factors such as lack of aerial photographs or
insufficient detail on sketch maps made locating
starting points problematic. Obscure tree scribes also
made the task difficult. Once the starting point was
located, travel to the plot center was accomplished
quite efficiently
The Virginia project leader located sites prior to
sampling. There, approximately 20 nun was required
for the crew to reach plot center.
43
-------
ACTIVITY
TIME OF DAY
7
30
8
30
9
30
10
30
113
0123
0 1
30
2
30
3
30
43
0 5
30
630
FOREST SERVICE PERSONWEL 1 & 2
Travel to sit.
PLot center location
Plot establishment
Measure regeneration (alt points)
•
Measure growth/visual • mpton aLl points
Indicator plant condition
T
— — —
j ;
I
;
;i
-
t;
—
=
=
— —
==
FOLIAGE SAMPLER #1
E
E
E T
Vertical vegetation measurements
—
FOLIAGE SAMPLER 2
I
I I
Foliage sanpling
VerticaL vegetation
I
j
Root excavation and core sai Ling
SOIL SAMPLER
—
—
i
—
—
i
i
I
o e Lxcava on
Pedon Description
( L!
Sotl/Coresw içting
Figure 4-2. A TYPICAL DAY’S SAMPLING SEQUENCE.
It is recommended that reconnaissance of the
sampling sites occur prior to sampling. If necessary,
new aerial photos and more detailed information on
starting points should be acquired. New methods for
establishing starting points that remain anonymous to
the public (e.g., global positioning systems) should be
investigated.
Crew leaders found that it was more efficient to
set up the complete plot before the FIA foresters
collected any of their data. It is recommended that
the measurement procedures proceed with plot
establishment before data collection activities.
However, an evaluation needs to be made as to
whether or not field crew members who are
responsible for vegetation structure measurements
should be allowed on subplots before the regeneration
plots are tallied (to prevent trampling).
The time requirements for data collection were
similar in New England and Virginia. When specific
methods have been approved, the time requirements
for some measurements should be reduced. The
sequence for the sampling activities was appropriate.
The most time-consuming activities were soil
sampling and tree climbing (when climbing spikes
were not used). With a reduction in data collection
for other measurements, field crew members may be
able to help complete these activities. It is
recommended that alternative methods of foliage
collection be investigated as well as the use of tree-
climbing spikes.
44
-------
Because the project had two types of plots (half
that had foliage sampling and intensive soil
sampling, and half that did not), field crew members
were sometimes uncertain about the types and
numbers of samples that were needed on a particular
plot. It is recommended that all data and samples to
be collected on a plot be listed concisely and placed
on a lamln2ted sheet for each field crew member,
and/or programmed on the portable data recorder for
the field crew leader to check before leaving the plot.
In New England, the field crew was responsible
for maintauung and transporting samples. After a
day of sampling, a crew member accounted for all
samples collected and placed them either in the
refngerator or in coolers- that contained frozen gel-
pacs. This procedure took approximately I h. Some
samples had to be maintained for as long as 4 days
under these conditions. Crew members shipped
samples only when inclement weather forced a
si utdown of field work or when they had a free time
(after flmshing a plot in half a day or during travel to
another site). Express shipping facilities were usually
not difficult to locate; more facilities were available
than were identified on the maps that were provided.
In Virguiia, the regional project leader handled
the maintenance and transport of the samples. Crew
members who had already worked a 10-h day were
relieved of this responsibility.
It is recommended that crews be assisted by an
individual who maintains equipment, stocks
consumable goods, and receives, maintains, and
transports samples. This person could visit the crews
daily or every other day.
A sample tracking and shipment form was used
to identify samples that were shipped. The plot
location identification and the number of samples
were recorded on the form. This information was
placed in one of the transported containers. Upon
arrival at its destination, the samples were checked
against the shipping form. This procedure
accomplished the goals of sample tracking and chain-
of-custody, but due to the number of samples,
different types of samples, and the number of
facilities to which they were sent, this task was time
consuming. When samples arrived at the receiving
facility, the labels had to be re-entered into a
computer system.
It is recommended that a bar coding system be
developed to label, track, and archive samples In
the field, the crews would place bar codes on samples
and the portable data recorder would associate the
written label (state, plot location, sample number,
etc.) with the bar code. This information would then
be stored in the portable data recorder and the laptop
computer. When a shipment is made, the laptop
computer would identify what types of samples, the
number of samples, and the locations that need to be
shipped. The samples would be located and the bar
codes scanned as they are packed. The information
would then be compared to the original sample list to
assure that all samples were accounted for. The
information would be pnnted as well as sent
electronically to the receiving facility. At the
receiving facility, the bar codes on the samples would
be scanned and electronically compared to the
shipping file. This would make sample tracking and
chain-of-custody more efficient and reliable. Also,
the sample identification would not have to be re-
entered at the receiving facility.
The data saving and transfer procedure was
efficient, usually requiring only 20 mm to complete.
During the first transfer in New England (plot
number 902), data was lost during the transfer from
the portable data recorder to the laptop computer
because of a programim.ng error that did not allow
commas to be entered into the free-form field note ’
field. A new program was sent to the field cre to
fix the problem At times, data could not be
transferred daily because the phones at some hotels
were incompatible.
It is recommended that all data venfication
programs be available on the portable data recorder
or the laptop computer This will allow on-site
venficat ion and correction of data before transfer and
would alleviate the need for immediate transfer in the
event phones are incompatible.
Unless field assistants are provided and the
measurements are reduced, the field sampling
activities described in this report cannot be completed
in I day. Long field days arid inclement weather can
induce field crew ‘burnout’ which, in turn, can lead
to careless sampling techniques, safety problems and
poor data quality.
45
-------
PROJECT MANAGERS
I MEDIA]
I LOGISTICS ]
LANDOWNERS
INDICATORS
Figure 4-3 FLOW OF INFORMATION TO AND FROM THE REGIONAL PROJEC7 LEADERS
It is recommended that crew leaders be aware of
field crew burnout and its effect on data quality.
A suggestion would be to task the crew with a certain
number of plots to sample and allow the crew the
flexibility of completing these within a specified time.
In any case, work schedules should be designed to
maintain high field crew morale.
4.7 COMMUNICATIONS
4.7.1 Description
The communications network in this project was
similar to the organizational chart (Fig. 4.1). Details
of the communications procedure are discussed by
Dwire et a!. (1990). Generally, communication
would lead from the crew members to the crew
leader who would then contact the regional project
leader. The regional project leaders would attend
weekly conference calls with national technical
coordinators. At this level, discussions would include
progress on all operational phases, problems, and
protocol changes. Issues would be resolved and the
results disseminated consistently to all field crew
leaders.
4.7.2 Discussion and Recommendations
The communications for the project was adequate
in most cases. Logistics personnel attempting to send
supplies sometimes had difficulty locating field
crews. As the FHM program expands, strategically
located, regional base sites may be necessary to send
equipment and consumables to field crews.
The most important communications link is
between the field crew leader and the regional project
leader. The field crew leader informs regional
project leaders about sampling progress and
communicates problems (e.g., equipment damage or
supplies needed etc.) or emergencies.
Crew leaders should submit weekly travel
itineraries to the regional project leaders and
immediately inform the regional project leader of
changes. Crew leaders should also be responsible for
the direct communication of emergencies to the
appropnate authonties unless personally injured. All
field crew members should be trained to handle
emergencies properly.
EMERGENCIES REGIONAL PROJECT LEAD DESIGN
PREP LAB
CONTRACTORS I GIS
FIELD CREW LEADERS
46
-------
Regional project leaders are important links
between the field and other groups (Fignre 4-3).
They should be available for phone communication
during the hours of sampling. Due to other
commitments, the New England regional project
leader was not always available to field crews or for
wee 1y conference calls. At the debriefing session,
it was mentioned that the regional project leaders
were unsure of where to find the appropriate
expertise when specific questions arose.
It is recommended that the regional project
leaders ideutifS’ replacements when they are not
available. Also, phone recorders should be used so
that field crew leaders can leave messages during
evening hours. It may be advantageous to develop an
electronic mail system. t is also recommended that
the leaders of each group in Figure 4-3 be readily
avaitab e during sampling to respond to questions
4.8 SAFETY
4.8.1 Description
Emphasis must be placed on safety in any field
operation. Field personnel must be aware of
potential safety hazards, follow all project safety
protocols and equipment guidelines, and be prepared
for emergency situations. A safety plan, developed
specifically for the program, was intended to address
the potential safety hazards of field sampling and to
identify required safety protocols This safety plan
was developed from EPA and USFS information.
All participants in the study were required to abide
by specific agency safety regulations.
4.8.2 Discussion and Recommendations
During the training workshops the safety plan
was mentioned but was not reviewed. Proper safety
gear (e.g., hardhats, safety glasses, gloves) was
available, but was not often used. Some of the FIA
crew members had been trained in basic first aid, but
this was not a prerequisite.
It is recommended that more emphasis be placed
on safety and that the safety plan be reviewed as part
of training. Crew members should be required to
read the plan and to document their review. Itis
recommended that safety gear be supplied to all crew
members for mandatory use. It is also recommended
that more than one individual on each crew be trained
in basic first aid including cardiopulmonary
resuscitation
Maps that identify fire and police stations in the
vicinity of the plot locations were given to the field
crews. Hospital information was not supplied but
should be in the future. It is recommended that the
regional project leaders arid crew leaders identify the
kinds of information they would like to have on
maps.
4.9 SUMMARY
In New England and in Virginia, it is feasible for
a five-person crew to make the measurements
described in this report in a 10-h day. If certain
sampling methods can be streamlined, then a four-
person crew may be feasible. The following
recommendations are made to improve the efficiency
of field activities
• Determine the critena for eliminating sampling
sites prior to reconnaissance
• Provide reconnaissance prior to field sampling to
reduce the time it takes a field crew to locate a
sampling Site.
• Review all equipment and consumable items with
measurement coordinators to determine exactly
what is required
• Determine staff requirements early and ensure
that contracts are established before field
sampling activities commence.
e Provide an assistant to the field crews. This
individual will stock equipment and consumables
and will receive, maintain, transport and track
samples.
• Provide a better communications network. Field
crews must be able to get answers to their
questions rapidly. The regional project leader or
an identified assistant should be responsible for
all conirnunication with the field crew leader in
order to maintain the proper lines of
com_rnunication.
• The logistics teams need to maintain close
communication with other groups (design,
indicators, QA, information management) in
order to develop efficient field implementation.
These groups’ plans and their logistics
arrangements need to be identified and worked
into logistics plans with as much lead tune as
possible
47
-------
5 VISUAL SYMPTOMS
Kurt Riitters, Sam Alexander, Robert Anderson,
Robert Kucera, and Margaret Miller-Weeks
5.1 INTRODUCTION
Large-scale forest monitoring systems commonly
evaluate tree health in terms of the outward
appearance of tree crowns and boles (e g , United
Nations Economic Commission for Europe, 1987;
Magasi, 1988). Despite the inherent problems of
interpretation (e g , Innes, l988a, 1988b; see below),
these measures are potentially useful because they are
easy to implement and because their meaning is
universal. In one sense, a tree that looks unhealthy
probably is unhealthy . provided that health is
gauged with reference to a baseline or normal
condition for that species, growth stage, and location.
Appearances can be misleading, because trees
cannot move about to seek their best exposure, soil,
moisture regime, nutrition, or protection. The
requirements and therefore the outward appearances
of trees differ over time and space, even when
conditions are normal everywhere. The health of any
particular tree must be assessed in relation to
normality for that tree’s circumstances, and this
makes it difficult to use visual symptoms for site-
specific assessments.
The aggregate appearance of a healthy regional
population of trees may be more stable If all
individuals of a population appear less healthy over
time, relative to their previous appearances, then
either their health has changed, or the circumstances
that define their normality have changed. The
capability to detect of either of these events would be
valuable, and would lead to follow-up studies and
assessments of what caused the changes.
The FHM program has given a high priority to
measures of appearance (e.g., Wentworth and Joyner,
1987; Riitters et al., 1990, Brooks et al , 1991b).
This reflects their potential utility, their pnor testing,
and their relative ease of application. Yet apparently
only Innes and Boswell (1990) have investigated their
reliability. Therefore, one objective of the field
study was to develop estimates of reliability with a
iiew towards deciding upon appropriate plot designs.
The purpose of this section is to summarize the
sampling optimization results obtained for the
measurements that could be analyzed as continuous
variables. Categorical variables were not analyzed
here. Average values obtained for the different
variables in each forest type will be given The
analyses of variance of these measurements will be
documented as they are the basis for plot design
recommendations made in Section 3. Finally, the
analyses of variance of selected mensurational
measurements will be documented for application in
Section 3.
5.2 DESCRIPTION AND APPLiCATION OF
THE MEASUREMENTS
As used here, the term visual symptom? refers
to a suite of measurements and observations of
appearance that were made on trees 7.62 cm dbh
within the 0 0169-ha subplots at each location (Fig.
1-2). Detailed descriptions of each measurement are
given by Dwire et al (1990). Briefly, on all live
trees on all four subplots, the percentage of live
crown (PLC) was estimated, and the occurrence of
crown and bole damage (CBD) was noted. Live
crown percent arid crown and bole damage are fairly
standard measures in forest inventory (U.S. Forest
Service, 1985a) arid in forest damage surveys
(Millers and Lachance, 1989, Alexander and Carison,
1988).
Additional measurements were made on
dominant, co-dominant, and open-grown trees on two
of the four subplots The additional measurements
included the percentage of crown dieback and
percentage of foliated crown transparency (PDB and
PTRN, both adapted from Millers and Lachance,
1989), the percentage of foliar discoloration (PD IS,
Alexander and Carlson, 1988), the percentage of
crown density (PDEN, Anderson and Belanger,
1986), and the percentage of crown defoliation
(PDEFOL, adapted from the United Nations
Economic Commission for Europe, 1987)
Defoliation, transparency, dieback, and related
measures are non-specific measures of condition
Although they may be useful for detecting change in
tree condition over time, they are not diagnostic
measurements and they reflect the particular
combination of stresses, natural and iriarimade, that
also changes over time.
48
-------
Three additional measures that will not be
reported here are the years of needle retention, the
percentage of discoloration by the ‘European
method, and the horizontal crown diameter (see
Dwire at al. 1l990 for a description of these
measures.)
In practice, these measurements are sometimes
combined as indices of crown condition. For
example, defoliation and discoloration are combined
to estimate a single damage index in the European
system (United Nations Economic Commission for
Europe, 1987), but that index has not been generally
adopted (e.g., hines and Boswell, 1987, 1990). The
measurements of crown dieback and transparency are
sometimes combined (Mi1 ers and Lachance, 1989)
but that practice also appears to be optional (e.g.,
Brooks at al., 1991b). In another example, it is a
common practice to estimate crown volume by
coinbirung measurements of the percent live crown,
total tree height, and crown diameter. Because it
appears that there is no agreement on any of these
combined indices, the measurements will be analyzed
separately in this section
It is expected that these measures of tree
appearance would be augmented by concurrent or
follow-up visits by trained forest pathologists to
evaluate the importance of observed changes and, if
possible, to identify plausible causes (e.g., insects
and diseases) of those changes. These procedures
were not tested in this study, but they are an
important part of the overall FHM program (Palmer
at a!., 1991).
5.3 SUMMARY OF VISUAL SYMPTOMS
Tables 5-I and 5-2 summarize the measurements
of the visual symptoms variables. The total number
of trees sampled and the number of live trees are
shown (columns I and 2). The number of live trees
is the sample size for the variable PLC. The number
of trees rated by other crown measures is also shown
(column 4); this is the sample size for the variables
PDB, PTRN, PDIS, PDEN, and PDEFOL. The
number of trees for which crown and bole damage
was observed is shown (column 10). Finally, the
arithmetic averages of the vanables PLC, PDB,
PTRN, PDIS, PDEN, PDEFOL, and CBD are shown
(columns 3, 5-9, and 11).
It was not the objective of this study of variance
components to use these measurements to classify
forest hea1tb, because such interpretations require
more complete assessments of all biotic pathogens,
insects, and other indicators of plant stress (Skelly ci
a!., 1987). But a brief review of the quantitative
results obtained for the subset of measurements tested
will put into perspective the analyses of variance
components (Section 54).
Considering first the New England stands, most
species had (on average) less than 30% live crowns,
and (not surprisingly) shade-intolerant species
appeared to have less than shade-tolerant species
(Tables 5-1 and 5-2). For the species with at least 10
crown-rated trees, it is notable that nearly 80% of the
American beech had at least one (but rarely more)
incident of crown or bole damage; values for other
species were considerably less. Also for these
species, average dieback was less than about 10%
and average discoloration was less than about 3%.
Average transparency was typically less than 15%,
average defoliation was consistently about 30 to 40%,
and average crown density was between 30 and 60%.
The variety among the averages obtained for the
different species suggests, at least, that the
percentages are not directly comparable measures of
health, yet they still may be highly correlated (hines
and Boswell, 1990). Without more information, it is
not possible to ascribe values of good or “poor
health for these variables.
Overall, the average percentages of live crown
appeared larger in Virginia (Tables 5-1 and 5-2).
For 6 of the 7 species measured in both regions (red
maple, American beech, bigtooth aspen, cherry,
white oak, and northern red oak, but not yellow
birch), the average percent live crown values in
Virginia were at least twice the values in New
England. Deciding whether this represents a real
difference among regions or is simply a difference
among crew measurement techniques requires further
analysis.
The Virginia species appeared to exhibit the same
pattern of percent live crown for shade-tolerant vs
shade-intolerant species. Loblolly pine was the only
Virginia specie.s for which a substantial number of
trees were crown-rated. The results for dieback,
discoloration, transparency, defoliation, and crown
density were similar to result-s mentioned earlier for
selected New England species, with the same caveats.
49
-------
Table 5-1 NUMBER OF MEASURED TREES AND SPECIES AVERAGE CROWN RATINGS IN NEW ENGLAND
Crown
Trees
Avg
Totel
Live
Live
riLed
Trans.
Discolor.
Crown
with
no of
Speciri trees
(no)
trees
(no)
crown
(PLC)
(%)
i,ces
(rio)
Diebick
(PDB)
(%)
pirency
( RN)
(%)
auoo
(PDZS)
(%)
dcruity
(PDEN)
(%)
Defoliation
(PDEFOL)
(%)
symptoms
(no)
syn-iptorns
(CBD)
(no)
(I)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
8 ,slsam 6r 92
87
22 5
4
5 0
8 8
2.5
55 0
32 5
9
1 4
it spruce 5
5
39.0
1
100
250
50
700
600
3
10
Red spruce 95
78
18 5
11
5 5
86
0.5
56 8
30 5
24
1.1
Ess&crn whitc pine 36
31
19 7
8
7.5
14 4
0 6
39 4
31 9
4
1 0
Northern wbiie-cedjr 3
3
28.3
0
—
—
—
—
—
1
I 0
Eutembemlock 126
119
279
9
44
61
00
639
194
13
10
Striped maple 29
22
16 1
3
3 3
11 7
3.3
500
33 3
13
11
Red maple 215
205
19 1
76
80
115
1.2
442
371
77
10
Su sr maple 180
17l
242
38
6 7
95
1.2
505
286
39
11
Yellow birch 96
: 89
23 5
37
8 0
10 7
1 6
46 2
31 9
33
1 2
Sweet birch 17
17
20 6
1
5 0
10.0
0 0
400
35 0
2
1 0
Paper birch 53
49
180
24
5 4
11.3
2.5
442
31 5
7
11
Anicrican beech 105
99
23 1
17
106
127
09
453
432
80
1 2
White ash 20
20
15 8
5
70
14 0
1 0
41 0
34 0
5
1 0
Easternbophornbeam I I
11
300
1
50
ISO
5.0
350
300
4
1 0
Biglooth sepen 27
27
13 5
17
5 9
15 3
0 0
32 1
35 0
6
1 0
Quabngupen 21
20
115
15
63
187
03
440
343
2
10
Prurtusspp. 4
4
125
3
200
233
00
283
383
3
10
Black cherry 22
17
179
9
83
122
22
400
339
11
13
Common choke-cherry 26
26
14 4
2
5 0
15 0
5 0
27 5
30 0
13
I 0
Whitcoak 4
3
18.3
1
50
100
00
500
2.50
2
10
Northern red oak 19
18
18 3
7
5 0
22 9
2 I
393
44 3
8
1 3
Arnencan mounizin .sah 7
7
12 9
2
7 5
15 0
0 0
37 5
35 0
7
1 0
Amcncsnbsuwood 1
1
150
0
—
—
—
—
—
0
—
The possibility of regional differences in percent
of live crown was also suggested by the data
expressed on an average per-plot basis (Table 5-3).
Other compansons between the two regions are
probably not worthwhile, since they would in effect
be comparisons of loblolly pine (in Virgmia) with all
sampled species (in New England). The vanation
among plots within the same region is a topic for the
analyses of variance in Section 5.4.
The possibility of regional differences in the
implementation of the same measurements was
suggested for the results for defoliation and crown
density in Table 5-3. In Virginia, the simple
correlation between the two variables was -0.843
(pO.l). It seems logical that
defoliation and crown density are inversely related,
and therefore one may question the New England
result. But the available evidence does not
measurements or the
for the differences
logic for finding a
correlation may vanish when data are aggregated to
plot averages of different species, as they were in this
case. It has been suggested that several similar
measures of crown condition would permit cross-
validation of each, but these procedures have not yet
been specified These and other associations among
the different measures of crown condition will be
reported in a separate publication.
The data in Tables 5-1 through 5-3 are not
stnctly comparable to any other survey results from
these regions, because sampling designs and
measurement methods may have been different. An
appropriately qualified comparison with the
monitoring results obtained by Brooks and others
(199 Ia, 1991b) in New England may be wortb hile,
however.
necessanly imply that either the
crews are somehow at fault
observed. For example, the
50
-------
Table 5-2 NUMBER OF MEASURED TREES AND SPECIES AVERAGE CROWN RATINGS IN VIRGINIA
Crown
Trees
Avg
Toeal
Live
Live
rated
Tr.nj- Discolor-
Crown
with
no of
3pec cs
trees
(no)
(1)
trees
(no)
(2)
crown
(PLC)
(%)
(3)
trees
(no)
(4)
Dieback
(PDB)
(%)
(5)
patency
(PTRN)
(%)
(6)
stion
( PD1S)
(%)
(7)
dert3lty
(PDEN)
(%)
(8)
Defoliation
(PDEFOL)
(%)
(9)
symptoms
(no)
(10)
symptoms
(CBD)
(no)
( II)
Eastern tedeedar
6
5
58 0
0
—
—
—
—
—
I
I 0
Sborticaf pine
8
8
37.5
4
0 0
16 3
0 0
48 8
25 0
0
—
Loblolly pine
641
636
23.5
157
0 1
19 9
0 0
54 8
22 6
132
1 2
Virgirus pine
96
96
20 6
8
0 0
22 5
0 0
62 3
22 5
35
1 3
Redm ap lc
I II
110
378
3
00
133
00
78.3
100
58
13
Yellow birch
7
7
20 0
0
—
—
—
—
—
2
1 0
Blucbccch
9
9
38 9
0
—
—
—
—
—
0
—
H ickoryspp
13
13
362
1
00
150
0.0
65.0
150
8
13
Flowenng dogwood
8
8
42 5
0
—
—
—
—
—
4
2 0
Cornmonperiimmon
2
2
32.5
0
—
—
—
—
—
2
1 0
An icncsnbeech
8
8
556
1
00
15.0
00
700
300
5
10
American holly
IS
18
51 4
0
—
—
—
—
—
4
13
Black alnuc
1
1
65 0
0
—
—
—
—
—
0
—
Sweetgum
185
185
36 7
5
6 0
26 0
00
55 0
2! 0
45
1 2
Yellow-poplar
78
78
37 5
5
0 0
9 0
0 0
70 0
12 0
4
1 5
Sweetbay
5
5
30 0
0
—
—
—
—
—
I
1 0
Upland blackgum
1
1
40 0
0
—
—
—
—
—
0
—
ow1andbIackg m
23
23
35 9
0
—
—
—
—
12
I 2
Sourwood
13
13
41.2
0
—
—
—
—
4
13
Eigtoothaspcn
10
10
375
I
00
100
00
600
150
0
—
black cherry
1
1
45 0
0
—
—
—
- .
—
1
1 0
White oak
59
58
43 5
2
0 0
12 5
0 0
60 0
12 5
17
11
Scarlet oak
5
5
39 0
0
—
—
—
—
—
1
2 0
Southcrnrcdoak
58
58
388
3
00
267
00
600
183
5
10
Water oak
13
13
41 2
0
—
—
—
—
—
9
11
Wji low oak
17
17
444
3
00
217
00
683
133
9
14
Northcrnre4 oak
1
1
400
0
—
—
—
—
—
0
—
Post oak
3
3
45 0
0
—
—
—
—
—
2
1 5
Black oak
15
15
38 7
1
0 0
10 0
0 0
60 0
15 0
7
1 3
Sasiafras
2
2
37 5
0
—
—
—
—
—
1
1 0
Elm
14
14
33 6
0
—
—
—
—
—
3
I 0
These comparisons are qualified by the fact that
different forest types were sampled, yet similar
sampling designs and the same observation methods
were used for measurements of percentage dieback,
percentage transparency, percentage discoloration,
and crown and bole damage. Brooks et al. (1991a,
1991b) noted an apparently high percentage of
dzeback in American beech and tentatively attributed
it to the beech-bark disease complex. They also
reported that this species had the highest percentage
of trees with crown and bole damage. These results
are consistent with the data in Table 5-1; American
beech had the highest average percentage dieback of
any species (10 6%) and a very high percentage of
trees with crown and bole damage.
In terms of transparency, Brooks et al (1991a,
1991b) report the possibility of unusual values for
yellow birch, American beech, and northern red oak.
The average transparency values reported for these
species in Table 5-1 are unremarkable in comparison,
and the high value for northern red oak must be
tempered by the rather small sample size. Brooks et
al (199]a, ]991b) report no indications of health
concerns based on discoloration, a result that is
consistent with the low average values of
discoloration obtained here. Full interpretation is
difficult without a reference by which to gauge
changes over time.
51
-------
Table 5-3 NUMBER OF SAMPLE TREES AND AVERAGE CROWN RATINGS FOR ALL SPECIES BY PLOT
Crown
Treci
Avg
Plot
Total
Live
Live
rated
Trani-
I) .color-
Crown
with
no of
no.
Veci
Veel
crown
(PLC)
ices
Dicback
(PDB)
parcncy
(PTRN)
ation
(PDLS)
dcnisty
(PDE? 4)
Defoliation
(PDEFOL)
symptoms
symptoms
(CBD)
(no)
(no)
(%)
(no)
(%)
(%)
(%)
(%)
(%)
(no)
(no)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
New Encland
16
117
122
153
159
160
166
173
174
181
267
270
289
321
347
420
308
692
3270
i n
1839
1841
1843
1954
1955
1957
1958
1959
2072
2074
2075
2185
2186
2187
2188
2189
2190
2191
2192
2303
48
44
201
IS
50
127
Ii
377
277
11
12
36
36
257
I I
50
86
27
514
318
10
Ii
93
88
196
19
79
118
16
537
324
25
I I
48
40
274
16
128
113
2.2
422
388
24
11
70
67
217
15
63
117
1.7
497
330
23
10
69
60
178
11
50
105
05
468
35.0
18
11
56
52
217
8
100
119
00
413
338
25
11
75
69
3 0
10
7 0
12 0
4.0
37 5
36 5
18
1 2
54
53
211
7
50
I I 4
07
536
329
14
1 2
50
48
262
16
94
116
1.9
397
281
24
11
64
58
190
24
63
119
23
47.1
427
28
II
67
61
2.39
10
45
190
25
420
425
10
10
73
73
399
27
54
130
00
33.5
337
I I
10
58
51
264
13
104
lOS
23
535
396
27
13
73
61
17 8
16
9 4
12 5
1 6
46 9
35 0
30
11
50
48
208
15
57
103
03
503
293
15
10
84
78
24 1
30
6 7
14 0
0 7
52 0
30 3
32
1 1
63
60
187
8
125
113
00
469
369
8
20
83
82
173
20
63
113
00
385
293
13
11
64
64
258
6
00
167
00
692
ISO
36
11
81
81
367
12
00
196
00
429
271
4
13
80
77
373
8
00
169
00
463
250
13
10
70
70
261
11
00
173
00
736
159
30
13
38
38
28 2
3
00
13 3
0 0
68 3
11 7
17
1 3
52
52
333
6
50
250
00
692
175
34
14
28
28
32 0
6
1 7
28 3
0 0
40 8
25 8
18
1 4
78
78
31 1
11
00
241
00
691
218
30
12
85
85
301
10
00
195
00
750
145
32
13
85
85
329
13
00
296
00
481
27.7
8
10
54
54
2.37
11
00
200
00
686
196
33
14
71
71
207
13
00
208
00
746
162
27
14
81
79
317
13
00
119
00
485
219
3
10
94
94
330
12
00
163
00
488
233
14
10
72
72
289
8
00
188
00
538
169
34
14
84
84
350
11
00
205
00
500
250
7
10
65
65
35.3
5
0 0
22 0
0 0
49 0
24 0
6
1 0
93
90
336
8
00
181
00
488
275
13
11
105
105
33.7
19
0.3
176
00
46 8
2.3 7
8
1 0
51
51
38.3
8
00
175
00
450
238
5
12
52
-------
5.4 ANOVA OF VISUAL SYMPTOMS
MEASUREM ENTS
The components of sampling variation were
estimated for the percentages of live crown, dieback,
bansparency, discoloration, crown density, and
defoliation. They were not estimated for crown and
bole damages because that variable is based on counts
and was not amenable to the analyses of variance
(ANOVA) models employed for the other variables.
The ANOVA models utilized were two- and a three-
stage nested, completely random designs. The two-
stage model included the effects of plots and trees
within plot. The three-stage model had the effects of
plots, su’bploLs within plot and trees within subplot.
Each region was ari lyzed separately. In one
analysis, both the two- and three-stage models were
applied to all species combined In another analysis,
subsets of well-represented species groups (loblolly
pine in Virginia, and maple plus beech plus birch
species in New England) were selected and analyzed
with the three-stage model only. Although it may be
inappropriate to combine species, the other species
were not sampled intensively enough to warrant this
separate analysis.
An exception was for the percent of live crown,
for which there was a sufficient sample size to
warrant a separate three-stage analysis of other
species groups (all oak species arid all non-oak
deciduous species in Virginia, and spruce plus fir
species and all deciduous species other than maple,
beech, or birch in New England). Sample sizes were
riot uniform for these analyses and therefore average
sample sizes need to be applied in developing
sampling recommendations The results are
discussed together with the results from other
measurements in Section 3.
The results may also be compared to those
obtained by h ines and Boswell (1990) for a
measurement (‘crown density’) that is defined
similarly to the variable PDEFOL in this study.
Table 1 in Jones and Boswel l (1990) contains
estimates of within- and between-stand variance
componeaLs for five specks based on data from the
British long-term motiitonng program on forest
conditions (Jones and Boswell, 1987). The species
are SitU spruce, Norway spruce, Scots pine, oak
species (i.e., sessile oak, English oak, and hybrids
between the two), and European beech. About half
(range 48 to 65%) of the total variance is attributable
to tree-to-tree variation within stands, and there are
important (F-statistics between 6 and 27) stand-to-
stand differences in average ‘crown density.’
In comparison, the three-stage ANOVA’s of the
variable PDEFOL (Table 5-4) yield estimates of the
percentages of variance attributable to tree-to-tree
variation within stands of about 62% for loblolly pine
only, 85 % for the three species group maple + beech
+ birch, 89% for all New England species, and 82%
for all Virginia species (these results are not shown)
The loblolly pine result is within the range reported
by Inne,s and Boswell (1990), the other results are
probably inflated simply because more than one
species is included. The F-statistics (not shown) for
the null hypothesis of no stand-to-stand variation in
average PDEFOL range from about 2 for all New
England species to about 14 for loblolly pine only
Again the loblolly pine result is consistent with the
British study; results for other species groups are
probably again affected by including more than one
species.
Jones and Boswell (1990) also reported the
percentages of variance attributable to different
‘edges’ within stands; these ‘edges’ are (roughly)
equivalent to the ‘subplots’ in the present study.
Although ‘edge’ differences were significant, the
percentages of the total variance were comparatively
small (range 4 to 14%). In comparison, the
percentage of total variance attributable to subplots in
the present study ranged from 0 (for loblolly pine
only and for maple plus beech plus birch species) to
about 4 % (for all Virginia species combined) Both
of these lines of evidence suggest that subplot-to-
subplot variance is relatively unimportant for
PDEFOL.
53
-------
Table 5-4 ANOVA OF VISUAl. SYMPTOMS VARIABLES FOR TWO SAMPLING MODELS
FT RN New B All
England A All
A Maple+bcech+birch
Virginia B All
A All
A Loblolly pine
18 — 1110
IS 57 1053
12 20 50
18 57 576
18 27 lOS
19 — 1403
19 60 1343
19 58 558
13 26 130
19 56 432
— 272
19 253
18 159
— 174
20 154
17 120
— 272
19 233
18 159
— 274
20 154
17 120
544614 — 88479
544 614 265 484 78 898
495 338 137 560 37 444
374 486 318 243 50132
204 376 46 354 38181
1442 2.39 — 176 138
1442 239 538 057 159 968
1377.049 182 960 68 572
305 313 192 574 89824
446716 322414 140524
86 384 — 27 548
86.384 30704 27312
91 816 32038 33 303
8292 — 4925
8292 16697 3397
0879 1987 0609
PDEFOL New B All
England A All
A Maple+beech+birch
Virginia B All
A MI
A Loblolly pine
18 — 272
18 19 233
18 18 159
19 — 274
19 20 254
19 17 120
— 272
19 253
18 159
— 374
20 233
17 120
16122 —
16122 9258
16524 ‘7236
0000 —
0000 0000
0000 0000
6 349
6 132
7 687
0 000
0 000
0 000
Forest
Species
Degrees of frtedom
Mean square for
Vansbte region Model group
Plot
Subplot
Tree
Plot Subplot Tree
PLC New B All
England A All
A Spruce+ fir
A Msplc+beech+birch
A Other hardwoods
Virgirue B All
A All
A Loblolly pine
A Oak.
A Other hardwood.
PDB New B ‘All
England A All
A Maple+becch+birch
Virginia B All
A All
A Loblolly pine
IS
18
18
19
19
19
18
18
18
19
19
19
49373 — 28046
49 373 72 254 24 726
18538 16307 12917
186 102 — 64 878
186 102 105 758 59569
176 534 43 063 50 109
P013 New B All
England A All
A Maple+beech+birch
Virginia B All
A All
A Lobloll) pine
PDEN New B All
England A All
A Maple + bcech+ birch
Virginia B All
A All
A Loblolly pine
— 272
29 233
18 159
— 174
20 154
17 120
18
18
18
19
19
19
18
18
18
19
19
19
683 954 — 107 167
683 954 162 730 102 994
483 698 78 125 73 604
1460263 — 115061
1460263 112223 125429
1543 082 52 362 81 321
329 458 — 127 239
329 458 153 493 123 260
280 485 91305 108 446
198023 — 71128
198 023 83 598 69 508
278 854 19 647 58 739
fle three-stage model (A) h .s effect.. of plots, l iibploLa within plot, md reel within subplot, and the Two .stage model (B) has cffcct.s of plots
and reel within plot The variablcs arc percent live crown (PLC), percent dicback (POE), percent transparency (PTRN), percent discoloration
‘DLS), percent crown density (PDEN), and percent defoliation (PDEFOL) See text for explanations of srsab1cs arid apecies groups
54
-------
5.5 ANOVA OF MENSURATION
VARIABLES
The mensurational characteristics of the stands
have been discussed in Section 2 of this document.
This section presents the results of the ANOVAs of
wand basal area (BA); average height of dominant,
co-dominant, and open-grown trees only (HTDCD);
and average height of aU trees (Hi). The basic data
were the live trees larger than 7.62 cm dbh from the
0.0169-ha subplots, and the analysis was done on a
per-subplot basis. Thus, the response variables are
subplot BA, J-ITDCD, and HT. Sample sizes were
balanced for these variables, with observations on
four subplots within each plot. A two-stage, nested,
completely random design with the effects of plots
and subploLs within p 1 oc were used in the ANOVA.
Each region was analyzed separately (Table 5-5).
The results are discussed in Section 3.
Tible
5-5 ANO
VA OF
MENSURA
TION VARiABLES ’
Fore i
vegLon
Source
df
Mean square
BA
HTDCD HT
New
P101
18
220 8724
26 1784 104891
England
Subplot
Total
57
75
127.0360
5 9098 3 5483
Vii gua
Plot
Subplot
Total
19
60
79
259 8292
77 2651
30 7659 12 8925
80131 3 0950
• Basal ires (BA. m’ s), iversge height of dominant, co-
dotniriant, and open-gro ntrees (1-ITDCD, m l, md average height
of all trees (HT, m) A completely random, nested design a
assumed Response vanabics are on a per-subplot basis
5.6 DISCUSSION
Visual symptoms measurements should be made
by trained professionals with stnct adherence to
quality assurance procedures. The suggestions of
regional differences in associations among the visual
symptoms measurements found in this study point to
a need for greater emphasis on training, testing, and
inter-regional comparalive studies. It makes little
difference that the visual measurements are relatively
inexpensive and easy to justify if, in application, the
numbers obtained are not comparable
Comparability also requires that the visual
symptoms measures be gauged relative to an
established baseline or normal condition. Although
a reference can be established over time based on the
accumulated data from monitoring, it may be
worthwhile to conduct additional tests to judge the
efficacy of these measures under different
environmental conditions.
Ultimately, the visual symptoms measures must
be augmented by observations of natural stresses,
such as insects and diseases, that are known to cause
changes in visual symptoms. Unless this is done, it
will not be possible to ascribe any plausible causes to
the observations of chancing visual conditions. These
additional observations must be made by trained
pathologists at the right time of the growing season
It will probably be worthwhile to include trained
pathologists on the regional monitoring staff, perhaps
even on every field crew, to ensure that these
measurements are made appropriately.
55
-------
6 SOIL PRODUCTIVITY
Rick D. Van Remortel and Mohammed J. Miah
This gection describes preliminary results for the
soil productivitymeasurements that were made during
the 1990 pilot study in the eastern United States.
Measurements of key soil physical and chemical
parameters have been used to optimize ’ the plot
sampling design for varLab ibty, logistical, and cost-
benefit considerations. The data are also being used
in simulations to evaluate possible interactions of soil
productivity with other measurements of forest
condition, and to estimate soil productivity baseline
condition.
6.1 BACKGROUND AND OVERVIEW
Soil productivity is generally defined as the
capacity of a given volume of soil to produce a
vegetative response under a specified system of
management (SEA-AR. 1981). Productive capacity
is affected by many physical and chemical factors.
Topographic features such as slope, aspect, and
elevation have been incorporated into models to
predict stand composition (Fralish, 1988) and have
been shown to influence growth responses of
Douglas-fir (Steinbrenner, 1963). it is expected that
these parameters could affect forest growth responses
because they contribute to the overall hydrologic
characteristics of a site.
The soil drainage classification, along with other
moisture charactenstics, has long been recognized as
vital information in estimating soil productivity
(Storie and Weislander, 1948; Mader, 1976;
Hamilton and Krause, 1985; Green et. al., 1989).
Topographic parameters are important in estimating
the hydrologic contnbutions of runoff and lateral
water flow (Hewlett, 1961), as well as such
characteristics as soil texture, rock fragments, and
bulk density. These parameters are also important
for their effects on nutrient availability (Mader,
1976), aeration (Steinbrenner, 1963, Mader, 1976),
and root distribution (Blanchar et al., 1978; Hillel,
1980), all of which directly affect vegetative
response.
Various chemical parameters of soil sample
analyses are expected to be useful for monitoring the
soil nutrient status at each plot. Many of the analyses
are routinely performed as part of standard soil
testing programs because of their importance for
plant nutrition and nutrient cycling. Parameters such
as exchangeable cations, cation exchange capacity,
pH, and exchangeable acidity have all been
incorporated into response studies for species such as
jack pine (Pawluk and Arneman, 1961; Hanulton and
Krause, 1985) and Dougla.s .fir (Green et al., 1989)
Total carbon, nitrogen, and sulfur have been used to
characterize the soil organic matter, which is a key
component of the forest ecosystem (Witde, 1964;
Mader, 1976).
Iron, manganese, magnesium, copper, zinc, and
boron are all essential elements to tree growth, and
metals such as lead, cadmium, nickel, chromium, arid
vanadium are measured to evaluate toxic stresses
Extractable sulfate can be an important constituent of
the soil solution and, together with electrical
conductivity, can be used to estimate ionic strength of
the solution (Griffen and Junriak, 1973). Ionic
strength is used to calculate the activity of ions in
solution, thereby allowing study of chemical
equilibria in soil samples and modeling of Long-term
chemical weathering of soil minerals (Lindsay, 1979)
Based on previous variability studies, it was
recognized that significant soil spatial variability can
be present on virtually all micro- arid m.acro-areal
scales (Ar-msori, 1977, Mausbach et al., 1980) It is
possible, however, for uncertainty in the soul
productivity values at a plot to be greatly reduced, by
the use of a coniposite sample design that captures
a significant amount of the within-plot variability in
soil characteristics. It is anticipated that an optimal
sampling design can be adopted whereby the samples
that are collected control the within-plot uncertainty
to a level that is negligible with respect to other
sources of uncertainty, for example, from the
measurement system or from the regional data
aggregation framework. The resulting level of data
quality would allow the data users to focus on
discerning •real ’ temporal changes in soil
productivity within a highly variable population of
forest plots
56
-------
6.2 STUDY OBJECTIVES
The overall goals of the soil productivity study
were to estimate the within-plot spatial vanahility in
soil characteristics and to test the overall feasibility of
implementing the soil productivity sampling design on
a inultiregional scale. Using the results of this study,
ills likely that an optimal sampling design can
Wtirnately be dentified that allows the control” of
within-plot data uncertainty to some level that is
acceptable to users of FI lM data Towards this end,
the data analysis will eventually include an evaluation
of uncertainty stemming from single-hole sampling as
opposed to multiple-bole sampling, and the
appropr iateness (e.g., costfbeaefit, utility, etc.) of
conipositing master horizon samples for laboratory
analysis purposes. Ed addition, optimization
algorithm.s have been applied to the data to evaluate
whether the provision of destructive sampling zones
outside the penmeters of the fixed-radius subplots
will allow the collection of soil samples that are
representative with respect to the subplot vegetative
measurements.
Specifically, the soil sampling issues to be
resolved as a result of the study include the
determination of the number of soil holes within a
plot that should be sampled, and the resources (e.g.,
time, personnel, funding, equipment, etc) required
to perform field characterization and sampling on a
given plot within a region In addition, we will later
use these and other data to determine: (I) the depth
of sampling and the types of horizons that should be
sampled, (2) whether samples will be composited and
at what stage; (3) the utility of laboratory preparation
and analytical methods that have been selected; (4)
parameter reporting units for the many different soil
parameters; (5) ancillary data that may be needed to
link the component parameters of the soil productivity
indicator; and (6) the utility of various soil sample
aggregation schemes in the stratification of data for
interpretive reporting.
6.3 MATERIALS AND METHODS
6.3.1 Field Methods
Soil classification data for the selected study plots
were obtained by on-site soil excavation and
characterization augmented by existing soil survey
information where available. Each plot was
thoroughly characterized for physical soil parameters
and landform features according to accepted National
Cooperative Soil Survey standards. The 20 plots in
each region were sampled in the manner intended for
full implementation of the monitoring design in
subsequent years. Eight of these same 20 test p 1 ois
in each region were designated as variability plots in
order to assess plot variability issues. The required
staffing was one soil scientist per day per ‘routire
plot’ and two soil scientists per day per vanability
plot”.
The soil sampling plot design was as shown in
Figure 1-3. The three soil holes were located
equidistant between the subplot centers. Excavation
of soil holes for characterization of the soil horizons
and collection of soil samples for laboratory analysis
is considered destructive sampling with respect to
long-term ecological momtonng on forest plots.
Therefore, soil sampling was done outside of the
established fixed-area subplots but was intended to
represent the soil charactenstics across the whole plot
area.
The field soil scientist was instructed to collect,
where possible, composite samples for the 0, A, E,
B, and C master horizons from each of the 20 field
plots in the 2 regions. The composite samples were
combined using equal volumes of soil material from
the three holes Each of the holes was excavated to
a I in vertical depth (or to a restnctive layer,
whichever was shallower) and a long-axis horizontal
diameter of approximately 0.5 m A detailed profile
descnption of the soil horizons on the plot was made
for the most representative (‘modal”) hole.
En order to estimate within-plot variability of soil
parameters, individual master horizon samples were
collected from the eight vanability plots in each of
the two regions. On these plots, three additional
holes encircling each of the three perimeter fixed-
radius subplots were excavated to expose the
A horizon (Fig 1-3). At each of these nine points,
individual soil samples were collected from the
A horizon, where possible, or from an E or
B horizon if the A horizon was not present or could
not be clearly delineated. The sampling strategy
assumed that the estimates of variability about the
three perimeter subplots would be comparable to the
variability about the center subplot.
57
-------
The field measurement and sampling methodology
is documented in a field methods manual (Dwire et
*1., 1990). The soil samples were prepared and
analyzed according to procedures described in a
laboratory methods manual (Byers and Van Remortel,
1991). The field and laboratory parameters that were
measured are listed in Table 6-1.
6.3.2 Statistical Methods
To meet the study objectives, two statistical
techniques were applied to the venfied and validated
soils database:
1. a sampling theory algorithm to optimize resource
allocations within plots (see Section 3.2),
2. semivariograms to study spatial structure (see
Section 3.4.3).
To assess data quality and sources of
measurement variation, statistical techniques
described by Byers et al. (1989) and Byers et al.
(1990) were used to explore the precision of
measurements. A two-tiered system was used for
characterizing measurement quality objectives
(MQOs) and results. The MQO for the first tier (at
low concentrations) is a constant value of the standard
deviation. The MQO of the second tier (at higher
concentrations) is a constant value of the relative
standard deviation and thus vanes as a linear function
of the mean. A knot value is the concentration at
which the MQOs for the two tiers coincide
Data quality was assessed by calculating the
percent of samples that had either standard deviations
or relative standard deviations, as appropriate for the
mean concentration, above the MQO. The relative
importance of analytical laboratory, preparation
laboratory, and measurement system as sources of
measurement variation was gauged by recalculating
the data quality statistics for each of those sources.
For comparison, the standard deviations and bias
were calculated for the pooled field samples
6.4 RESULTS AND DISCUSSION
6.4.1 Optimum Allocation of Resources
This technique utilized a cost-benefit algorithm to
analyze variance components. Summary results of
the analysis for 20 of the 22 primary soil physical
and chemical parameters, configured in eight
parameter groups, are contained in Table 6-1. For
all groups, it was found that the variance among pits
within clusters was the same as the variance among
pits among clusters. This indicates that the present
sampling design with individual pits as the
subs.ampling unit provides the better allocation of
resources.
6.4.2 Sem.ivanograms
This technique examined the spatial correlations
within plots to determine whether the systematic
design for soil pits provided sufficient separation
between pits so that the soil for a single master
horizon could be treated as coming from independent
units. The calculations are discussed in Section 3.4.
A distance value of zero was used as a reference
value for examination of the semivanograms. This
value was derived from QA data from the
DirectfDelayed Response Project which used similar
sampling protocols and a probability sample for
selecting soil pedons (Byers et al , 1989, Byers et al
1990). As a result, this value was uniformly lower
than the estimates for the other distances.
None of the semivariograms gave evidence that
the shorter distances had higher spatial correlation
In fact, some of the semivanogram charts appeared
to have an up and down behavior because of the
low value at zero distance and chance fluctuations
downward at higher distances The semivanograms
suggest that the current distance between soil pits is
sufficiently large to avoid undesirable correlations
between measurements. There was no data on
shorter distances, so it cannot be determined whether
soil pits could be located even closer together should
that become necessary.
6.4.3 Other Sources of Variation
An estimate of measurement uncertainty that is
achievable for each soil analytical parameter may be
derived from existing data (e.g., Van Remortel et al
1988; Byers et al., 1989, Byers et al , 1990). For
most of the analytical laboratory measurements, an
average coefficient of variation (CV) of 10% or less
is typical for replicate samples. The expected
laboratory bias is ± 5% or less of the reference
value. For the sample measurement system as a
whole (sampling, preparation, and analysis), an
average CV of 20% or less is typical.
58
-------
In most cases, QA data from this study support
these estimates (see Table 6-2). The precision and
accuracy measurement quality objectives (MQOs)
specified in the study plan were satisfied for most
parameters. However, relatively few degrees of
freedom for the measurement quality samples used in
this study may restrict the reliability of the
imprecision and bias estimates.
Although an assessment of seasonal variability
was not an objective of the study, this source of
heterogeneity may be significant for some soil
parameters. It may be possible to control (to some
degree) seasonal variation at each plot by conducting
the cyclical sampling of each plot during a specified
interval of the index period.
6.5 RECOMMENDATIONS
An issue for ongoing investigation is how to best
minimize sources of vanability so that re.al changes
in soil productivity can be detected at regional levels.
The evaluation techniques described in this Section
have provided statistical estimates of the optimum
number of samples that must be collected from a plot
to control the overall variability of the mean of the
measurements, as part of a multistage FHM sampling
design. To address different scenarios relating to the
intensity of sampling that is required to accomplish
the FHM regional monitoring objectives, detailed
assessments have been made of within-plotvanability
by statistically varying the number of soil samples
collected from each vanability plot. The data users
ultimately must choose reasonable levels of
uncertainty that can be tolerated and still allow the
overall project objectives to be satisfied. Preliminary
drafts of these data quality objectives will be
developed following the 1991 field season.
Another issue that must continue to be addressed
is the selection of soil physical and chemical
parameters that should be measured in order to
provide a sufficiently high resolution snapshot of soil
productivity status. The soil productivity parameters
defined in the laboratory methods manual (Byers and
Van Remortel, 1991) are presently being measured to
assess their utility for productivity associations with
other indicators. These parameters have been
identified through interactions with more than 60
forest soil scientists in the United States arid Canada.
Additional parameters, inappropriate parameters, or
alternate methods might be identified as a result of
information gained during these ongoing assessments
As part of the sampling design, archived soil will be
available for additional future analysis if it is deemed
necessary.
59
-------
Table 6-2 FIELD PLOT SOIL VARIANCE COMPONENTS BY REOJOIP
Plot
Psran ter
Parameter MS V
Subplot
Hole
MS V %
MS V
Particle
sjzg
SAND,rs
SAND
SILT
Sfl..TVA
CLAY
CLAYVA
442 0
43166
404 2
29587
17 88
132S8
47 8
4615
45 4
3172
1 37
1428
45
78
50
77
17
49
60 5
3983
42 3
1654
6 94
1142
0 5
390
< 0
394
0 16
< 0
I
7
0
10
2
0
39.0
882
46 1
543
6 51
1504
59 0
882
46 1
543
6 51
1504
54
15
50
33
81
51
Soil pII
Exchangeable
eat iona
PHH2ONI
PH_}I2O,
PN_01M, 5
PH01M
CA_OAOMU
CA_OAC
MO_OACKB
MO_OACVA
K_OAC,
K_OACVA
NAOAC,rs
NA_OACVA
2 330
1.440
2 500
3143
84 08
10 90
0 586
0 238
0 0702
0 0167
0 0038
0 0034
0 265
0 134
0 287
0.104
9 03
0 94
0.059
0 026
0 0080
0 0017
0 0002
0 0001
59
42
59
38
46
20
40
38
57
28
49
30
0242
0333
0241
0269
12 82
3 10
0 138
0 026
0 0073
00022
00002
0 0003
0 037
0.073
0027
0.057
1.31
< 0
0 039
< 0
0 0008
<0
0 0000
0 0000
36
23
6
21
7
0
13
0
7
0
10
5
0 149
0 109
0 1 71
0110
948
3.82
0 068
0 042
0 0053
00051
0 0002
0 0003
0 149
0 309
0 371
0110
9 48
3 82
0068
0 042
0 0053
00051
0 0002
0 0003
25
35
35
41
47
80
47
62
36
72
41
65
Exchange
capacity
sod
sctditict
CEC_OAC , ,
CECOACSJA
AC_SACI.,rs
AC_BACLJA
ACKCL,
ACKC L.
AL...KC I..,rs
ALKCLJA
475 5
127 9
510 4
243 2
21 34
14 98
24 51
1764
55 3
12 3
58 I
22 4
188
I 33
2 36
1 72
40
43
47
45
30
35
40
45
44 1
24 9
54 9
54 9
6 34
4 00
5 69
327
< 0
4.3
< 0
15 4
119
0 92
I 36
064
0
14
0
31
18
23
23
17
82 7
13 1
65 5
12 2
3 30
2 49
2 21
1.50
82 7
13 1
65 5
12 2
3 30
1 49
2 21
I 50
60
45
53
24
52
40
37
28
Mincra l tzable
nitrogen
N_M1N
NM1N 5
9452
2076
1119
242
66
87
6479
526
42 6
091
2
3
538 8
274
538 8
274
32
10
Extractable
phoaphonia
P_B1
P_El V A
399 7
96 61
30 5
6 23
9
10
169 5
44 75
< 0
< 0
0
0
299 6
56 37
299 6
56 37
91
90
Extractable
mutfate
SO4_H2O,
SO4H2OVA
SO4_PO4
SO4_PO4,,
105 33
80 22
19867
134 46
9 75
7 54
2342
7 07
31
33
53
12
28 86
17.10
1803
74 71
4 86
1 00
< 0
13 25
16
4
0
2.3
16 69
14 32
2070
37 95
16 69
14 32
2070
37 95
53
63
47
65
Total
slcmcnta
C_TOT
C_TOT,.JA
N_TOT II
N_TOTVA
S_TOT,
S_TOTVA
83 81
848
0306
00153
0 0029
0 0003
30 19
073
0037
00015
0 0003
0 00003
63
28
66
31
63
29
5.74
2.36
0018
00031
0 0002
0 0001
< 0
029
<0
<0
< 0
< 0
0
II
0
0
0
0
5 93
157
0019
00033
0 0002
0 0002
S 93
1 57
0019
00033
0 0002
0 00001
37
61
34
69
37
71
• MS — mean Iquare, V vanarice component, % percent of total variance The approximate degrees of fretdom (per region) are 7 for
plots, 16 for iubplot* within plots, and 41 for boles within subplote See Table 3 -5 for brief descriptions, and Dwirc ci ii (1990) (or complete
descriptiona of the parameters The .ubscripta NE and VA on parameters iefcr to New England arid Virgirus
60
-------
Table 6-2 SOIL PARAMETER IMPRECISION AND BIAS POOLED ACROSS RECIONS
Below
Samp lc Parameter d l SI)
the ksio
Above
the k .noi
>MQO Biaa
%
MQO
>MQO
%
df %RSD
MQO
Par,meier2roup Part, 1e IE!e ( )
ALMR MOIST
CLAY
SAND
SILT
PLHD MOIST
CLAY
SAND
SILT
MSFD MOIST
CLAY
SAND
SILT
PLOT MOIST
CLAY
SAND
SILT
7 05689
5 03821
6 11972
6 09794
7 01570
$ 23814
S 28745
5 37005
7 00622
5, 05235
6 07772
5 11450
114 1 1701
91 3 2456
91 94951
91 8 1074
ParameTer rouo Soil oH (oH unu )
ALMR PH_1120 7 0 0205
PH_ OIM 7 00154
EC_H20 7 00256
PUW PH_H2O 7 00428
PH_O IM 7 0 1505
ECH2O 7 00038
MSFD PHH2O 7 0 0357
PH_O IM 7 00212
EC_H20 7 00100
PLOT PH_H20 114 04122
PH_OIM 114 04139
EC_H20 114 00365
no e
1 334
-2 400
C) 903
-0 023
0 013
no C5
PRrI meter rnun
ALMP. CAOAC
MG_OAC
K_OAC
NA_OAC
PLHD CA_CAC
MG_OAC
K_OAC
NA_OAC
MSFD CA_OAC
MG_OAC
K_OAC
NA_OAC
PLOT CAOAC
LIGOAC
K_OAC
}IA_OAC
7 00063 002 00
7 00034 002 00
7 00048 002 00
7 00031 002 00
4 00275 003 250
6 00474 003 167
7 00093 0.03 00
7 00054 003 00
4 00047 004 00
6 00466 004 167
7 00107 004 00
7 00060 004 00
21 00849 — —
59 00852 — —
99 00615 — —
114 00174 — —
3 215
1 11
3 234
1 01
93 12.33
55 705
15 387
15 667
15 00
20 333
20 00
-0015
-0 003
-0015
0 000
03 286
2 00
3 00
3 00
045 00
3 167
45 167
45 161
06 00
4 00
6 00
6 00
01 00
01 00
I 00
015 00
015 143
15 00
02 00
02 00
2 00
61
-------
Table 6-2 (CONTINUED)
Below
$a, ,le Parameter df SD
the kr,ol
Above
thc knot
>MQO Biu
%
MQO
>MQO
%
d( %RSD
MQO
Pararneir, roue’ E’chanee capacity and ac,diliea (megI)00 )
ALP. CECOAC
AC_BACL
ACKCL
ALKCL
PLHD CECOAC
AC_B ACL
AC_KCL
ALKCL
MSFD CECOAC
AC_EACL
AC_XCL
AL_KCL
PLOT CEC_OAC
AC_BACL
AC_KCL
ALKCL
7 02470 05 143
4 01328 05 00
1 0.2348 037 00
4 04533 15 00
6 01408 075 00
5 00982 075 00
1 00983 05 00
4’ 04749 2 00
6 02779 1 00
5 01104 1 00
7 25601 — —
85 16658 — —
84 11770 — —
7 332
7 30
73
6 77.8
3 26
1 3.9
2 68
6 197
3 15
1 17
2 9.7
114 401
107 342
29 951
30 521
20 429
10 00
15 00
15 333
15 00
15 00
22 00
20 167
20 0.0
20 00
30 00
1 847
-0401
no e t
no e t
Parameter group MineralizabJe rntro2en (meg/I 002 )
ALMR N_MEN 6 00757 1 00
PLIED N_MEN 7 0 5964 1 5 14 3
MSFD N_MEN 7 02018 2 00
PLOT N_MEN 100 04517 - -
Parameter erour, Extractable phosphorus (m2 /k2 )
3 02458 075 00
3 02777 1 00
43 12384 — —
Parameter erout, Exira:table sulfate (mg S/k2 )
ALMR S04_H20 7 0 3906 15 0 0
S04_P04 7 30487 15 286
PLHD S04_}120 5 1 4455 2 2 20 0
S04_P04 2 04046 22 00
MSF’D S04_H20 5 1 4215 3 0 0
S04_P04 2 14685 3 00
PLOT S04_H20 114 42186 — —
S04_P04 80 5 7260 —
7 54
4 58
4 141
71 1901
2 52
5 56
2 48
5 53
34 906
ALMR
PLHD
MSFD
PLOT
06 15 00
14 837 —
P BI
P BI
P BI
P BI
no eat
no est
1 697
-1 354
15 00
22 00
30 00
15 00
15 00
20 00
20 00
62
-------
Tible 6-2 (CONTINUED)
Below
the knot
Above
the knot
Bias
Sample Parameter df SD
MQO
>MQO
%
df
%RSD
MQO
>MQO
%
P,rsmeter rouc’ Total elements (wt %)
ALISm C_TOT — —
—
—
7
77
10
14.3
—
N_TOT 7 0 0073
0 015
14.3
—
—
—
—
3_TOT 700014
0002
143
—
—
—
—
—
PLHD C_TOT 4 00185
007
00
3
35
15
00
—
N_TOT 6 00033
0022
00
1
07
15
0.0
—
S_TOT 6 0 0008
0 003
0 0
1
1 7
15
0 0
—
MSFD C_TOT 4 0.0288
0 1
0 0
3
6 3
20
0 0
—
NTOT 6 0 0054
0 03
0.0
I
1 7
20
0 0
—
S_TOT 6 0 0008
0 004
00
I
I 7
20
0.0
—
PLOT C_TOT —, —
—
—
114
47.2
—
—
-0461
N_TOT 49’ 00411
—
—
65
461
—
—
-0024
S_TOT 43 00051
—
—
71
442
—
—
-0005
See lexi for explanation of the knot ALM R refers to analytical laboratory, PLHD refers to preparation laboratory (confounded), MSFD
refers to the total measurement system (confoundcd), PLOT refers to among-plot vanability from pooled samples SD standard deviation,
RSD = standard deviation divided by the mean, df — degrees of freedom, MQO a the measurement quality objective for SD or RSD,
) MQO is the perccnl of samples with SD or % RSD greater than the MQO
63
-------
7 FOLIAR NUTRIENTS AND
CHEMICAL CONTAMINANTS
Timothy E. Lewis
7.1 INTRODUCTION
This set of measurements has been termed foliar
nutrients and chemical contarninants, but will
henceforth be referred to as foliar chemistry.
Foliar chemistry is an example of an exposure-habitat
indicator. This class of indicator is designed to
quantify stressors which may be associated with
changes in forest condition (e.g., visible injury,
growth, soil productivit . The foliar chemistry
indicator is also a key component in the suite of
indicators that make up the nutrient cycling
assessment endpoint.
The elements that were determined in (char
samples during the study include macro- and micro-
nutrients (e.g.. Total N, P. K, Ca, Mg, S. Fe, Mn,
Zn, Cu, B, Mo, and Cl) and potential contam.inants
(Na, Al. F, Cd, Pb, As, V, Cr, Ni, Si, and Hg).
Some essential nutrients may also enter the system in
excessive amounts from anthropogenic sources (Total
N, Fe, Mn, Zn, Cu, B, and Cl). For example,
chromium smelters emit Mn, Cr, Fe, Al, Ca, Mg,
Na, Zn, K, Pb, Ba, Ti, Hg, Cd, Be, V, and As.
These were measured in particulates emanating from
the stacks at the Chrornasco smelter in Memphis, TN
(Bowers and Melhuish, 1987).
Nutrient deficiencies or excesses and metal
toxicity can often be detected as visual symptoms on
foliage (e.g., copper deficiency: Will, 1972, boron
deficiency: Carter et al, 1983; arsenic toxicity:
Spiers et al., 1983). When nutrient deficiencies are
severe, visible symptoms such as leaf yellowing and
scorching become apparent. Other symptoms may
include stem deformities and loss of leaves. Some
symptoms may relate to a specific nutrient limit.ation,
but they may also be confused with similar symptoms
of other Stresses. Foliar (and soil) chemical analysis
is needed to diagnose or to eliminate chemical
imbalances as a plausible cause of symptoms (Skelly
et a ]., 1987). Foliar nutrient chemistry may also
correlate with visible injury caused by gaseous
pollutants such as ozone, sulfur dioxide, oxides of
nitrogen, and peroxyacetylnitrates.
The primary objectives stated in the foliar
chemistry study plan (Dwire et al., 1990) were to
determine the within tree and within plot components
of vanance for foliar nutrients and chemical
contaminants. A secondary objective was to
determine the effect of compositing on elemental
concentration variance This information could then
be extrapolated to the regional demonstrations to
deduce estimates of cost and variance associated with
each component of sample acquisition and analysis
This section reports the findings of the foliar
chemistry study as related to the stated objectives
7.2 METHODS
Ten northern hardwood stands in New England
and 10 pine-oak stands in eastern Virginia were
sampled. Two species were selected. sugar maple in
New England and loblolly pine in Virginia. Within
each stand, two dominant or co-dominant trees were
selected at each of three of the four subp!ots To
select trees, two random azimuths were found from
a subplot center, and a transect was followed out to
the radius of the subplot The first suitable tree
encountered while moving in a clockwise direction
from the transect was designated as the sample tree
With the aid of a tree climber, two branch
samples were obtained from the outer portion (i e,
•sun leaves) of the upper third of the crown. The
foliage was removed from the branch and placed in
a plastic bag for shipment to the preparation
laboratory. The intent was to keep each branch
sample separate. But in New England, the t o
branches from each tree were combined in a single
sample bag Thus, no within-tree variability could be
estimated for sugar maple. In Virginia, foliar
material from each branch was kept separate as
planned.
The foliar material was air dried, maceraled, and
homogenized before microwave digestion in a
mixture of concentrated nitric and bydrochlonc acid.
The microwave digest was analyzed by directly
coupled plasma spectrometry for macro- and
rnicronutnents and trace elements (Table 7-1).
Fluorine, chlorine, and total carbon, nitrogen, and
sulfur were determined by the methods shown in
Table 7-1.
64
-------
Table 7-I. FOUAR CHEMISTRY PARAMETERS AND
ANALYTICAL TECHNIQUES
Parameter
Type of sample
Method
Pboiphocua
Liquid cxuict
Casacue I
Calcium
of
DCP
Magnesium
microwavcd
Poeauiutn
foliar tissue
Sodium
lion
Manganese
Copper
z
orog i
Aluminum
Lead
Chromium
Nickel
Molybdenum
Liquid extract
Cassette 2
Cadmium
of
DCP
Arsenic
microwavcd
Vanadium
fohar tissue
Mercury
Nitrogen
(oust
CNS Analyzer
Sulfur
tissue
Carbon
Fluorine
foliar
potentuome tnc
Ch lonnc
tissue
7.3 STATISTICAL ANALYSIS
Summary statistics were generated by utiliz.ing the
Statistical Analysis System program (SAS, 1985) To
estimate the components of variance, linear models
for nested designs were used in the ANOVA. Three-
and four-stage models were evaluated for the effects
of plot, subplot within plot, tree within subplot, and
branch within tree.
7.4 RESULTS
Data from seven plots in Virginia and six plots in
New England were available for reporting. Results
of chemical analyses of four additional plots are still
outstanding. Certain plots were not analyzed due to
missing samples or unbalanced design. One plot in
New England (number 270) had four trees from one
subplot sampled rather than two trees on three
subplots. Notwithstanding, these data have been
presented.
The mean and standard deviations of elemental
concentrations determined in foliar material collected
from sugar maples in New England are presented in
Table A-3 in Appendix A. Similarly, summary
statistics for elemental concentrations in loblolly pine
foliar material collected in Virginia are shown in
Table A-4 (Appendix A).
Variance components were computed using the
SAS procedure VARCOMP (SAS, 1985). For the
loblolly data, variance components were determined
using three-stage and four-stage sampling models
Only a three-stage model was used for the sugar
maple data because branch samples were not kept
separate during sampling. The vanance component
estimates for the sugar maple three-stage model were
calculated using the following equations
Source Expected Mean Square
Plot Var(Error) + 2 3529 Var(Subplot)
+ 5.6471 Var(Plot)
Subplot Var(Error) + 2 Var(Subplot)
Error Var(Error)
The vanance component estimates for the loblolly
pine three-stage model were determined using the
following equations.
Source Expected Mean Square
Plot Var(Error) + 2 Var(Tree)
+ 12 Var(Plot)
Tree Var(Error) + 2 Var(Tree)
Error Var(Error)
The variance component estimates for the loblolly
four-stage model were calculated using the following
equations.
Source Expected Mean Square
Plot Var(Error) + 2 Var(Tree)
+ 4 Var(Subplot) + 12 Var(Plot)
Subplot Var(Error) + 2 Var(Tree)
+ 4 Var(Subplot)
Tree Var(Error) + 2 Var(Tree)
Error Var(Error)
65
-------
The percent vanance attributable to each stage of
sampling was calculated by dividing each individual
variance component by the total variance. The
percent variances for the sugar maples and loblolly
pines are presented in Tables 7-2 and 7-3,
respectively.
The second objective of the study was to
determine the effects of sample compositing on the
observed vanability in elemental concentrations. The
results of elemental analyses of composited loblolly
pine samples are sumn:iarized in Table A-4.
•5 DISCUSSION
The levels of macro- azd micronutrients in sugar
maple are similar to levels found by Leaf (1973)
The concentration for many of the trace elements
(e.g., As, Cd, Cr, Hg, Ni, Pb, Sr, and Ti) were at
or below the detection limit of the analytical
laboratory (see Tables A-3 and A-4). Differences in
elemental concentrations between the New England
plots and Virginia plots may not be due regional
differences, but rather may be species-specific.
For those elements that were close to their
detection limit the var(error) (i.e., branch to-branch
for loblolly) vanance component terms were the
major contributor to the overall vanance. Precision
is inversely related to concentration. Thus, the
analytical vanability may be the major contributing
factor to the overall vanance for many of the trace
elements. Another source of variance in the
var(error) term may be poor and variable recovery of
elements from the foliar sample. This variability was
noted in the pine needle standard reference material
(SRM) analysis performed with each batch of samples
for copper and potassium.
For sulfur measurements of loblolly foliar tissue,
the largest variability was noted at the branch-to-
branch level (94%) (Table 7-3). No direct
comparison is available for sugar maple, because
branch samples were composited. The var(error)
term in the sugar maple three-stage model, which
represents tree-to-tree variability, was 36% for sulfur
(Table 7-2), whereas the equivalent error term for
loblolly (var [ Tree)) was only 1 %.
In general, the trees-within-subplot variability was
less than the between-plot variability for most
elements. However, when the branch-to-branch (or
within-tree) variability was high, the between-plot
variability became smaller. To reduce the between-
plot variability, the number of trees sampled per
subplot can be increased.
The results of individual vs. composite sample
analysis revealed that elemental values are very
similar. The ability to collect composite samples will
greatly reduce the cost of analysis of foliar tissue
collected from multiple branches.
Procedures for archiving foliar tissue samples
should be investigated. These procedures were not
tested in this study but, as is the case for soils, foliar
sample archives will provide a unique opportunity for
later retrospective analyses of long-term trends.
Archiving foliar tissue samples may also reduce the
laboratory analytical effort required as part of routine
monitonng, because not all analyses would have to be
performed on every sample.
66
-------
Table 7-2 PERCENT OF TOTAL VARiANCE FROM SUGAR MAPLE VARIANCE COMPONENTS
C N S SI HG ZN P FE Cu MN MG NA Co AL
Variance Componeni
Vsr(P Ioi) 56 45 47 77 17 0 38 48 4 21 30 29 17 69
Var(Subplot) 0 43 17 6 7 18 0 0 0 14 20 8 0 0
V ar(Ezeor) 44 12 36 17 76 82 62 52 96 65 50 63 83 31
Toul 100 100 100 100 100 100 100 100 100 100 100 100 100 100
NI K TI CR PB CD SR AS MO LI BA B CA V
Var(PIo ) 30 54 84 77 49 25 78 10 35 35 24 46 81 24
Var(Subp loi) 0 0 5 3 0 0 5 0 0 2 0 19 2 23
Var(E aycjr) 70 46 II 19 51 75 17 90 65 64 76 34 17 53
Toul IOQ 100 100 300 100 100 100 100 100 100 100 100 100 100
Tablc 7-3 PERCENT OF TOTAL VARIANCE FROM LOBLOLLY PINE VARIANCE COMPONENTS
C N S SI HG ZN P FE CU MN MG NA CO AL
Variance Component
Var(Pjo ) 56 23 5 26 0 16 81 39 54 73 10 20 30 54
Var(SubpIo ) 4 17 0 3 7 39 0 8 20 5 0 3 12 0
Var(Tree) 15 45 1 43 1 18 17 45 0 IS 49 10 3 42
Var(Ertvr) 25 13 94 28 92 27 2 9 26 4 41 67 56 4
Toi.tl 100 100 100 100 100 100 100 100 100 100 300 100 100 100
Var(Plot) 57 29 5 26 0 24 83 40 64 74 8 21 32 55
VirCTrte) 18 58 1 43 7 49 15 51 7 22 43 12 12 41
Var(Enor) 25 13 94 28 93 27 2 9 29 4 49 67 56 4
Total 100 100 100 300 100 100 100 100 300 100 100 100 100 100
NI K TI CR PB CD SR AS MO LI BA B CA V
Var(Pto ) 35 49 45 56 50 28 66 0 15 48 1 21 54 4
Var(Subp lot) 21 3 0 3 1 0 0 32 7 0 6 0 0 0
Var(Tr ec) 17 0 1 0 0 0 0 19 0 15 0 49 36 17
Var(Error) 26 50 54 41 50 72 34 70 78 37 93 30 10 79
Toul 100 100 100 100 100 100 100 100 100 100 100 lOG 100 100
V*r(P loi) 39 49 45 58 50 28 66 1 17 49 2 21 57 3
vai-crnec) 34 0 0 0 0 0 0 29 1 12 4 45 32 17
Var(Error) 26 51 55 42 50 72 34 71 82 39 94 34 II 80
Toul 100 100 100 100 100 300 100 100 100 100 100 100 100 300
67
-------
8 VEGETATION STRUCTURE
Steven P. Chne, David Cassell,
and Alisa L. Gallant
8.1 INTRODUCTION
Maintenance of biotic integrity is an assessment
endpoint. Biotic diversity, a key component of
integrity, is at risk from six major types of threats:
(I) direct population reduction from bunting and
fishing, (2) physical alteration of habitats, (3)
chemical and solid waste pollution, (4) global
atmospheric change, (5) introduction of alien species,
and (6) cumulative or thultiplicative effects of
interactions among these major threats (U.S.
Environmental Protection Agency, 1990). Physical
alteration of habitats is an immediate concern and
may exacerbate the potential impacts of future
atmospheric change. For example, habitat alteration
or destruction was identified as the greatest threat to
diversity of birds, perhaps the best studied vertebrate
taxon (U. S. Environmental Protection Agency, 1990).
Consequently, the structure of land use/land cover
types and animal habitats are leading candidate
response indicators, along with area, pattern, and
geographic extent (Fig. 8-I).
The primary objective of this study was to
compare two field methods for n’ieasunng the
vegetation structure of forests: ocular and pole. We
concentrated upon the structure of understory
vegetation, which comprises most of the plant species
diversity in forests, is more sensitive to
environmental gradients, and has a faster reaction
time/turnover than overstory trees. Analyses
included methodological differences in foliage
occupancy, measurement variation, spatial variation,
and time requirements.
8.2 METHODS
8.2.1 Vegetation Structure Measurement Methods
8.2.1.1 Ocular method
The ocular estimation method has been used
routinely during forest inventory by the USFS
Southeast FIA Unit (Cost, 1979; U.S. Forest Service,
1985b). The foliage occupancy of understory
vegetation was estimated on subplots I and 2 of 20
plots in New England and 20 plots in Virginia using
the ocular method. The ocular measurement
procedure involved four main steps: (1) define the
main vertical strata in the understory and estimate
their heights, (2) estimate the total percent of space
occupied by foliage in each vertical stratum, (3)
determine the relattve contribution (percent) of each
physiognomic vegetation class (Table 8-I) to the total
foliage occupancy in each stratum, and (4) list the
four most abundant species of each physiognomic
class in each stratum. Foliage occupancy by shrubs,
vines, grasses, forbs, and trees < 2.54 cm dbh was
estimated in the field. Normally, foliage occupancy
by trees > 2.54 cm dbh is estimated from species-
specific regression equations and added to the ocular
estimates to develop the full stand profile; however,
this data was not available for analysis for this report.
In addition, the ocular data from the New England
plots were not available for analysis.
Table 8-I PHYSIOGNOMIC CLASSES FOR VEGETATION
STR UCTLTRE
Code
Physiognom c Cia i
I
Yellow Pines
2
Other SofLwoods
3
Hard ood a
4
Tropicals
5
Shrube
6
Vines
7
Greesca and grasslike
8
Forba
9
Moasea and lichen.
Table 8-2 GROUND COVER SUBSTRATES FOR POLE
METHOD
Code
Substrate
I
Mineral ioil
2
Rock
3
Standing Wa ler
4
Stream
5
Dead wood >
10 cm diameter
6
Litter/duff
7
Lwc root or bole
68
-------
Figure -1. RELATIONSHIPS BETWEEN STRESSORS, RESPONSE INDICATORS, AND BIOTIC INTEGRITY ENDPOINT FOR
THE VEGETATION STRUCTIJRE MEASUREMENTS
8.2.1.2 Pole method
The pole or vertical line intercept method was
adapted from Short and Williamson (1986), who used
a similar method to ground-truth the presence and
position of wildlife habitat layers estimated from
aerial photos. The presence and vertical distribution
of foliage of understory vegetation was estimated on
subptots I and 2 of 20 plots in New England and 20
plots in Virginia. Pole measurements were made at
points spaced 12 ft (3.67 in) apart on a 16-point
square grid centered on each subplot (Fig. 1-4). At
each point a 30-ft (9.15 rn), telescoping pole was
raised and direct intersections (hits ’) of living leaves
and branches were recorded by physiognomic
vegetation class (Table 8-1) and by height to the
nearest foot (0.305 m). All sizes of vegetation were
recorded, but hits originating from trees > 2.54 cm
dbh were uniquely coded. The substrate upon which
the pole rested was also recorded (Table 8-2). Hits
of berbaceous vegetation, shrubs, or trees at the
ground level were assigned to the first 1-ft interval.
8.2.1.3 Measurement variation
Reduction of measurement variation was
emphasized, because it was one of the components of
total variation subject to some control. Consequent’y,
quality assurance (QA) procedures for the fieldwork
called for within- and between-crew remeasurements
to control and document the measurement variation
associated with each of the vegetation structure
methods. Even though a few between-crew
remeasurements were made (four subplots [ two each
on two plots]), a field audit revealed that inost of the
within-crew remeasurements were not being made or
were made incorrectly. For example, within-crew
ocular semeasurements were not made in either
Virginia or New England, and within-crew pole
remeasurements in New England were compromised
by a lack of independence between remeasurements.
Consequently, for the pole method reliable
remeasurement data collected during the training
session were used to estimate measurement variation.
Stressor(s)
PflysJc i Alterot Ion
Cf i rbltstS
Effects
Respor se Indicators
I 0 reese foqest erel
ncresee do’ nnsnce of
c eø petcN ty
Enapo, nt
es end e*lent of
f 01 55
cr.sse e*an petcn area
Ir creb5e oetcn ISOI tI j
Decresse eree—senslilve
Fr er te lon 5O5C ’G
Change O iperc.el of
pIbfltS bflIffSIS.
pet nogens
Cont os t I one I
Pete ogene , ty
SIepl I? ‘Cat ,orr—
LOrldGCbDe pattern Ce
area. ehaoe Or’d
errengaa nt of
Pb t C flea)
Land use Datle P1 end
UCt I-,- b I
road density—
Vegetet Ion 61ra1 on
C D ’ P JO t)f
Species rIChflesb
end pbtcfllnes
Funct ona
Pop ,,Iet on rer ge bPI
processes
I Ity rate
abunda nCQ
f -4orro9enI b1.Ion of forest
6truCture Ce
contro l)
Loss of ecaI rIaDitat
feStures (e snags)
Decreaae aDeces d,vers ,I
Ce g tree .ronoc ,. ,Ite-1..
vogetstJorl Control)
Ifl
-------
The data consisted of 126 pairs of onginal and
remeasureEnents. Three types of remeasurements
were made: within’.crew (a crew remeasures itself,
n=25), between-crew (field crews remeasure each
other, n= 25), and trainer-crew (trainer remeasures
field crews, n=76). Training remeasurement data
could not be used for the ocular method, because
only a few remeasurements were recorded.
8.2.2 Statistical Analysis
8.2.2.1 Method comparisons
The data generated by each measurement method
were paired by subplot and the differences in total
foliage occupancy (%) were determined. The data
were paired in two lifferent ways for method
comparisons. First, the oIe and ocular data for each
subplot were summarized and paired by the same
vertical strata as defined by the ocular method. For
the second way, the data were paired according to the
I-ft sirata defined by the pole method. Student’s
t-test (normal distribution) and the Wilcoxon signed
ranks test (distribution free) were used to test the null
hypothesis that the mean difference in total foliage
occupancy (%) between methods was equal to 0.
Total foliage occupancy was the sum of all
physiognoniic vegetation classes. To make the data
sets from the two methods compatible, hits from
vegetation >2.54 cm dbh were dropped from the
pole data and data from above 30 ft were dropped
from the ocular estimates. Two different methods
were used to calculate total foliage occupancy from
the pole data for comparison with the ocular data
(Table 8-3). In the first method (‘presence-absence
method) the percent cover of vegetation in each
stratum was calculated by determining the presence of
a hit within any foot level of that stratum and
dividing by the number of points per subplot (16). In
the second method (summation method) the percent
cover of vegetation in each stratum was calculated by
summing the hits within each foot level of that
stratum and dividing by the product of the number of
foot levels per stratum and the number of points per
subplot (16). For the data paired by foot levels, the
two methods of calculating total foliage occupancy
produced equivalent estimates.
Table 8.3. ANAL
YSIS OF HYPOTHETICAL POLE MEASUREMENTS USING TWO CALCULATION MET
PRESENCEJABSENCE AND METHOD 2 SUMMATION
HODS METHOD 1
Foot
level Stratum
Point n imbers
Vegetation
cover
I 2 3 4 5 6 7 8 9 30 31 12 13 14 15 16
1 1
1 0 0 1 1 0 1 I 1 I 0 0 0 0 1 0
2 1
0 0 0 0 0 0 1 0 1 1 1 0 1 0 I 0
MethodI
1 0 0 1 I 0 1 I 1 I 1 0 1 0 1 0
10/16=63%
Method2
1 0 0 1 1 0 2 1 2 2 1 0 1 0 2 0
14/32=44%
3 2
0 0 0 0 0 1 0 0 1 0 0 1 0 0 1 1
4 2
0 1 1 0 0 0 0 1 0 1 0 1 1 0 0 0
5 2
1 0 0 0 0 0 1 1 0 0 0 1 1 0 0 0
6 2
1 0 1 0 1 1 0 1 0 0 0 I I 0 1 0
Method l
1 1 1 0 3 1 I 1 1 1 0 1 1 0 1 1
1311681%
Method2
2 1 2 0 1 2 1 3 1 1 0 4 3 0 2 1
22/64=34%
7 3
1 1 0 0 0 0 0 1 1 1 1 1 1 0 0 1
8 3
1 1 0 0 0 0 0 0 1 1 1 1 0 0 1 1
9 3
0 0 3 1 1 0 1 0 1 0 0 0 0 0 I 1
10 3
1 0 1 1 1 0 1 0 0 I 0 0 1 0 0 0
II 3
0 1 0 0 0 0 0 1 0 1 1 0 1 0 1 0
12 3
1 0 0 0 0 0 0 1 0 1 1 0 1 0 0 I
13 3
1 0 1 0 1 0 1 0 0 0 0 1 1 1 0 0
14 3
1 1 0 I 0 0 0 0 1 1 1 1 1 0 1 0
15 3
0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 1
36 3
1 1 0 1 1 0 1 0 1 0 1 0 0 1 0 0
Methodi
I I 1 1 1 0 I I I I I I I I I I
151I694%
Mcthod2
7 5 3 4 4 0 4 4 5 6 6 4 6 2 5 5
70/16044%
70
-------
8.2.2.2 Measurement variation
Ocular method
Remeasurement tolerances set forth by the
Southeastern FIA unit at the training session were
± 2 ft for heights of strata and ± 15% for foliage
occupancy within strata. Between-crew
remeasurements were analyzed by tallying the
frequency with which they met these standards.
Pole method
For each pair of pole remeasurements at each foot
level, the variance was calculated using the standard
variance formula. When the rerneasurement agreed
with the original at a• particular foot level, the
variance estimate was zero, since the data at each
foot level consisted of a pair of yes or no
observations. When they differed, the variance
estimate was (l 0.5)2 + (0 O.5)2 = 0.5. Each one of
these estimates has one degree of freedom. For each
foot level, these one-degree-of-freedom variance
estimates were pooled using the standard formula for
pooling variances. This was equivalent to taking a
weighted average of the variances, using the degrees
of freedom as the weights. This analysis captured the
two main sources of measurement error at each point,
namely recording of different numbers of hits and
incorrectly assigning the recorded hits to foot levels.
A variance was calculated separately for
within-crew, between-crew, and versus trainer, as
was an overall pooled variance estimate using all 326
pairs of measurements. The within-crew and
between-crew vanance estimates were compared
across foot levels using the Wilcoxon signed ranks
test, due to the non-normality of the data.
8.2..2.3 Spatial variation
Spatial correlations among pole measurement -s
were analyzed by the common procedure of
computing and interpreting semivariograrns (Ripley,
1981). The semivariograms showed if, and at what
distance, spatial correlation of data levels off. For
each pair of data of a given distance apart, the
variance was calculated as half the square of their
difference and then averaged for all pairs of that
distance. The variance at zero distance represented
measurement error, and was estimated by the pooled
vanance discussed in Section 8.2.2.2. A separate
semivariogram was produced for each of the 30 foot
levels of the pole measurements.
8.2.2.4 Sample allocation
The variance and time requirements of the pole
measurements were analyzed to estimate the optimal
sample sizes for subplots and measurements per
subplot. The methodology described in Section 3
was used.
8.3 RESULTS
8.3.1 Method Comparisons
8 3.1.1 Percent foliage occupancy differences
Paired strata
The mean total foliage occupancy estimates
(percentage) from the ocular method were
significantly greater than estimates from the pole
method. There was a mean difference of 3%
between the ocular estimates and the pole estimates
calculated by the ‘presence-absence method’, a
significant difference for both parametric
(p-value=0.016, n= 165) and nonparametric tests
(,p-value=0.020, n= 165). Method differences
appeared to be normally distributed with about
two-thirds (67%) of the differences within ± 10%
and nearly all (97%) within ± 30% (Fig 8-2)
In contrast, there was a much larger and more
significant mean difference of 13.5% between the
ocular estimates and the pole estimates calculated by
the summation method (p-value <0 001, n= 165) A
nonparametric test was used because method
differences appeared to lack normality due to more
frequent positive differences (Fig. 8-3). Although the
ocular method usually estimated greater foliage
occupancy than the pole method, the methods often
described similar distribution patterns of foliage
occupancy. About equally often, however, the
methods differed in both the amount and distribution
of foliage occupancy.
7]
-------
50
Figure 8-2 FREQUENCY OF DIFFERENCES IN TOTAL FOLIAGE OCCUPANCY BETWEEN THE OCULAR AND POLE
MEASUREMENTS BY THE PRESENCE-ABSENCE METHOD
Figure 8-3. FREQUENCY OF DIFFERENCES IN TOTAL FOLIAGE OCCUPANCY BETWEEN THE OCULAR AND POLE
MEASUREMENTS BY THE SUMMATION METHOD
45
40
35
30
25
20
15
10
5
0
Du cEflE’CE (%)
60
50
40
p.
30
20
10
0
0I ERENCE (%)
72
-------
Paired foot levels
When the data were paired by individual foot
levels, the ocular estimates of total foliage occupancy
significantly exceeded those of the pole estimates by
an average of 12.5% (p-value<0.OOI, n=7 7 1). A
nonpiwetflC test was used because method
differences ippeared non-normal due to more
frequent positive differences. This difference was
very similar to the difference found between ocular
and pole estimates when paired by strata and
calculated by the summation method.
8.3.1.2 Measurement variation
Ocular method
In two of the four subplots remeasured in
Virginia, the number of strata defined differed (plot
1955, subplot 2: four vs. two; and plot 2186, subplot
2: three vs. four). This was considered a high rate
of error for this critical vanable, since all of the
remaining measurements (i.e., percent foliage
occupancy estimates, lists of most abundant species
by physiognomic class) were nested within the
defined strata. These differences also presented
problems for pairing strata for comparisons. When
necessary, we used the convention of pairing in order
the lowest two or three strata of each remeasurement.
For the 12 pair of strata compared, 3 (25%) exceeded
the ± 2 ft tolerance for strata height and 6 (50%)
exceeded or equalled the ± 15 % tolerance for foliage
occupancy.
Pole method
The within-crew and between-crew variance
estimates were compared across the 30 distinct foot
levels (Table 8-4). These two variances were not
significantly different (p-value>0.05). The within-
and between-crew variances were then pooled into an
overall crew variance and compared to the
versus-trainer variance. The versus-trainer variances
were significantly larger than the overall crew
variances (p-value 30 FF .07937 06000 .02000
‘See Section 8 2 2 2 for the method of calculation
73
-------
8.3.2 Pole Method
8.3.2.3 Tune requirements
8.3.2.1 Spatial variation
The semivariograrns of the individual foot levels
(e.g., 1, 11, znd 21) consistently showed that spatial
correlation leveled off by the 12-foot spacing between
pole measurements, thus the measurements may be
regarded as independent for plots in Virginia and
New England (Figs. 8-4 and 8-5). This may have
resulted because 12 ft was far enough to move
measurements between most individual crowns of
trees and tall shn.ibs and between clumps of
herbaceous vegetation. In addition, if one considers
the gap-phase dynamics pf the forest stand as a
relatively homogeneous stochastic process over an
area, the distances between gaps would be random
and would not appear consistently at specified
distances.
The asymptotic variance shown on the
5emivariograms estimated the total variation
associated with measurements at that foot level. This
variance tended to decrease with height in the
examples shown, and in general, averaging 0.422 at
I It, 0.159 at 11 It, and 0.154 at 21 ft in Virginia
and 0.249 at I ft, 0.151 at 11 It, and 0.141 at 21 ft
in New England. These numbers reflected structural
features of forests, in that the vegetation below 5 or
6 ft varied widely from dense to sparse, while
vegetation above 6 or 7 ft was less vanable, being
mostly Sparse.
8.3.2.2 Sample allocation
Conservative estimates of sample sizes for
subplot.s (m) and measurements per subplot (k) were
6 and 4, respectively (90th percentile, using cost
model I; see Section 3). A smaller median value of
3 subplots and 3 measurements per subplot, however,
might suffice for regional estimates. Similarly, the
total number of measurements per plot (m X k) bad
a smaller median value of 9 compared to 15 and 20
for the 90th and 95th percentiles, respectively. These
estimated optimal sample sizes may be
tnderestimaced since the plots were selected to be
bomogeneous and all subplots were rotated into the
same forest condition. However, even if estimates
were revised upward, sample size requirements for
the pole measurements would likely be met using the
1991 sampling design of 28 measurements per plot.
Each subplot included 18 pole measurements
(16-pointsquare sampling gndplus2 remeasuremects
per subplot). On average 34 and 33 mm per subplot
were required to complete the pole measurements in
Virginia and New England, respectively (Table 8.5)
Consequently, a single point required on average just
under 2 mm to measure Minimum and maximum
subplot times were 15 and 60 mm, respectively,
considenng both regions.
8.4 DISCUSSION
8.4.1 Method Companson
The ocular and pole methods both estimate the
amount, arrangement, and composition of forest
vegetation; however, the methods employ different
conceptual, and, ultimately, procedural approaches to
arnve at these estimates. For example, the ocular
procedure uses qualitative guidelines to define strata
In addition, foliage occupancy is estimated by
visualizing square or rectangular boxes about the
individual crowns of trees, shrubs, and tall herbs,
including open space within the crown. Then
individual crowns are summed to estimate the total
foliage occupancy of each stratum. As indicated by
the limited remeasurement results, estimates using
these procedures can potentially vary widely among
observers due to their expenence and perceptions.
Observer differences can be reduced by training,
however, measurement tolerances remain relativel)
large.
In contrast, the pole method is more
straightforward in that only hits of live vegetation by
the pole need to be recorded. Consequently, the pole
method requires little interpretation or calculation,
except in cases of “close calls”, which is simplified
by following strict protocols for placing, leveling,
and raising the pole. In addition, with the pole
method the space within a crown is not included as
occupied because the pole can, and often does, go
through a crown with few or no hits being recorded
Thus, the conceptual and resulting procedural
differences are the main cause of the differences in
results between the measurements Recognizing this,
we developed the presence/absence method of
calculating foliage occupancy from pole data to an
attempt to make the method results more compatible
74
-------
° I ? LEVEL I
: ‘ ‘ I.EVLI. II
0 10
o oe
E 1 LEVEL 2 ’ —
0 — —
01 .
12 24 36 0. 06 100 120 132 14 156
0 S1 CE ( )
Figaro 8-4. EXAMPLES OF SEMIVARIOGRAMS FOR THREE FOOT-LEVELS FROM POLE MEASUREMENTS BY THE
SUMMATION METHOD IN VERGINIA DATA FROM STEMS > 254 cm DIAMETER
0 t LEVEL. 1
03.
03.
0’.
0 ‘U LEVEL 11
4
C
001.
E T LEVEL 2’ —
0 ‘1 —
0 13
o ci
0 12 24 0 gE 1 S 120 1 2 1*4 156
0 31*CC0 (FE)
Figure 8-5 EXAMPLES OF SEMJVAR]OGRAMS FOR THREE FOOT-LEVELS FROM POLE MEASUREMENTS EY THE
SUMMATION METHOD IN NEW ENGLAND DATA FROM STEMS> 2 54 cm DIAMETER
75
-------
Table 8-3. TIME REQUIRED TO COMPLETE POLE
8.4.2 Measurement Variation
M EASUREMENTS ’
V.rgiai.
New England
Subplot
Plot
Subplo4
Plot
Plot Subplot time
time
P104
Subplot lime
lime
3839
1
—
0016
1
20
3839
2
25
—
0016
2
30
50
1841
1
45
0137
1
40
1841
2
2.5
70
0117
2
35
75
3843
1
35
0122
1
45
1843
2
35
75
0122
2
35
80
1954
1
30
0155
1
35
1954
2
30
60
0155
2
50
85
1955
1
35
0159
1
30
1955
2
—
—
0l59
2
35
65
1957
1
30
‘0160
1
30
1957
2
30
60
‘ 0160
2
25
55
1958
1
—
0166
1
15
1958
2
—
—
0166
2
25
40
1959
1
—
0173
1
30
1959
2
—
—
0173
2
35
65
2072
1
35
0174
1
40
2072
2
30
65
0174
2
45
85
2074
1
55
0181
I
33
2074
2
35
90
0181
2
35
68
2075
1
23
0267
1
35
2075
2
30
55
0267
2
—
2185
1
—
2270
1
40
2185
2
—
—
2270
2
30
70
2186
1
30
0289
1
35
2186
2
30
60
0289
2
25
60
2187
1
30
0321
1
33
2187
2
35
65
0321
2
25
58
2188
1
53
0347
1
35
2188
2
24
77
0347
2
40
75
2189
1
35
0420
I
—
2189
2
35
70
0420
2
39
2190
1
30
0508
1
35
2190
2
38
68
0508
2
25
60
2191
1
30
0692
1
25
2191
2
20
50
0692
2
40
65
2192
1
40
0902
1
25
2192
2
40
80
0902
2
20
45
2303
1
40
3270
1
30
2303
2
32
72
3270
2
60
90
Average
Sasndarddcv.
34
8
68
10
33
9
66
14
Range
20-55
50-90
15-60
40-90
The total variation in measurements will greatly
influence their usefulness as indicators of changes in
forest condition. Regional, temporal, within-plot,
within-subplot, and measurement variation are
important components of total variation
Measurement vanation was emphasized in this study
because although it can be relatively large for
subjective measurements (Gumpertz et al., 1982), it
can be controlled or reduced through QA procedures
(Cline et al., 1989). What then is the relative
contribution of measurement variation to the total
variation of ocular and pole measurements of
vegetation structure, and how can this variation be
controlled and reduced?
This question cannot be addressed adequately for
the ocular method due to a lack of remeasurement
data, but a preliminary estimate can be calculated for
the pole method. Measurement variation (Table 8-4)
was less than 20% of total variation (Fig 8-6 and 8-
7, average of asymptotic variances) in the first 5 or
6 ft levels, and usually 20 to 30% above 6 or 7 ft up
to 30 ft. These ratios must be interpreted with
caution, because we do not yet have a realistic
estimate of total variation. Total variation will
undoubtedly increase because no temporal variation
is included here. Moreover, both regional and
within-plot variation are underestimated here since
the plots in each region are from a relatively small
geographical area and the subplots were rotated into
the same homogenous stand condition. In contrast,
the measurement variances are probably robust,
because the crews were unqualified (a worst-case
scenano), the estimates were based upon a large
sample size, and procedural improvements have been
made. Consequently, there is reason to believe that
the relative contribution of measurement error to total
variation will decrease for the pole method, and
easily meet Taylor’s (1987) 10% guideline for an
acceptable level of measurement error.
A major finding of this study was the need to hire
qualified personnel to make the measurements Some
of the problems that were encountered during the
field study - little to no QA data and no plant species
identification - would have been avoided if qualified
personnel had camed out the measurements An
experienced botanist familiar with -the flora of the
Southeast United States will make the vegetation
structure measurements for the 1991 field study.
• Each subplot included 18 pole measurements (16-point square
sampling gnd plus 2 remeasurementz per subplot) Plot lime was
the m of subplot times All times art in minutes
76
-------
9 GROWTH EFFICIENCY
Kurt Riitters and Beverly Law
9.1 INTRODUCTION
Forest productivity is an important aspect of
forest health. The absorption of PAR (400 to 700
am) by canopies is an important factor influencing
tree growth and forest productivity (Kramer and
Kozlowski, 1979; Russell et al., 1989). Growth
efficiency, or growth in relation to radiation
transmittance or canopy density, is an integrative and
sensitive measure of forest productivity (Waring and
Schlesinger, 1985). Estimates of canopy density per
unit area obtained via measurements of PAR
absorption may also be related to species-specific
measurements of visual symptoms (Section 5) to help
interpret possible changes in overall canopy
condition.
Recent technological advances have made it
possible to rapidly measure radiation transmittance
and leaf area index in the field (Pierce and Running,
1988; Gower and Norman, 1991), and permit serious
consideration of these measurements for monitoring.
Yet some problems of measurement variability must
be solved to implement these measurements in a
monitoring framework.
Cloudiness during the sampling period is one of
the most important sources of variation in PAR
measurements. Clouds absorb PAR (Fig. 9-I), and
this may be confused with canopy absorption in some
sampling designs. Thus, most of the reported
measurements of PAR have been made on cloud-free
days, and PAR transmittance is typically normalized
by the ratio of under-canopy to ambient PAR (Fig.
9-2). But methods are needed for sampling
irrespective of the sky condition, because not all
applications can await cloud-free conditions. For
example, large-scale regional surveys are typically
conducted on tight schedules that allow a limited
amount of time for sampling. In another example,
measurements must sometimes be made at a certain
time for comparisons with remotely sensed data
Measurements of PAR made during this test of
regional survey procedures identify some of the
consequences of sampling different stand types under
a variety of sky conditions.
The main objective of this study was to estimate
components of sampling vanation of transmitted PAR
in two different forest types under different sky
conditions. This information could then be used with
information about the cost and feasibility of PAR
measurements to develop better sampling designs for
regional surveys ( 3) and to define measurement
quality objectives.
Also of concern was the potential that variable
cloudiness would result in inaccurate, highly variable
estimates of transmitted PAR. This section focuses
on that potential, in particular on the serial
correlation that was obtained among stations under
cloudy sky conditions. This is important because,
with senal correlation, the least-squares estimators of
means will be unbiased and consistent, but variance
estimates are likely to be biased (K.menta, 1971), and
this leads to improper hypothesis tests if the
correlation is ignored.
9.2 METHODS
9.2.1 PAR Measurements
Seventeen northern hardwood stands in New
England and 19 pine-oak stands in eastern Virginia
were sampled. Within each stand, a plot consisting
of four subplots was established (Fig. 9-3). At each
subplot, under-canopy PAR measurements were made
on a rectangular grid of 16 sampling stations
Ambient PAR was measured at a sampling station in
a nearby canopy opening or nonforested area
PAR was measured with a Sunfleck Ceptometer’
(Model SF-80, Decagon Devices, Inc., 1989). The
SF-80 device is a hand-held integrating radiometer
that contains 80 quantum sensors linked to a data
processing and storage unit. In a continuous
sampling mode, the SF-80 radiometer accumulates
averages of 80 sensor values (ixucromoles per meter
squared per second) until the operator requests one
overall average for storage. For this study, each
sample is the average of at least 30 intermediate
averages accumulated during a —30-s sampling
interval.
77
-------
Figure 9-!. AMBIENT PAR IN RELATION TO TIME OF DAY AND SKY CONDITION EACH PLOTTED POINT IS THE
AVERAGE OF 800 TO 2400 PAR SENSOR VALUES TAKEN OVER A S TO 10 mm PERIOD AT THE DUKE FOREST, DURHAM,
NC, ON TWO CLEAR AND TWO CLOUDY DAYS IN APRIL, 1990
Figure 9-2. WHAT IS MEANT BY INTERCEPTED (Q ) AND ABSORBED (Qa) RADIATION Q IS THE INCIDENT RADIATION
AND Pq THE RADIATION REFLECTED UPWARDS FROM THE VEGETATION Qt IS THE RADIATION TRANSMITTED
THROUGH THE OVERSTORY AND psQi IS THAT PART OF THE TRANSMITTED RADIATION THAT IS REFLECTED
UPWARDS BY THE SOIL AND GROUND FLORA THE INTERCEPTED FRACTION (1) IS THE COMPLEMENT OF THE
TRANSMITTED FRACTION ( I) ADAPTED FROM FIGURE 2 2 IN RUSSELL ci al (1990)
1500 -
0 --- CLEAR 5 IE5
1000 —
U
a,
a,
P
a,
Q
cf
0
500 -
- CLOUDY S(IES
‘7
L
0-
9am
rf
I I I I I
Noort
hUE OF DAY
3pm
pG
78
-------
9.2.2 Statistical Analysis
Measurements were initiated at — 1000 h zone
time at each plot and were continued for about 1 h
until all sample stations were visited. The procedure
started with 10 separate samples at the ambient
station, where the SF-80 device was held level facing
the sun. The 32 under-canopy stations on subplots 1
and 2 were then measured. At the under-canopy
stations, the SF-80 device was swept in a level circle
around the station at arm’s length and waist height,
effectively sampling about 10 &. The SF-SO
measuring device was temporarily halted when the
sweep was obstructed by tree stems or brush and
restarted without losing the accumulated data after
moving around the obstruction. Ten more samples
were then collected at the ambient station, followed
by 32 samples at under-canopy stations on subplots 3
and 4, and another 10 samj,les at the ambient station.
In this way, a total of 3O ambient and 64 under-
canopy measurements were made at each plot.
The mean and the CV of the 30 ambient
measurements were used to classify the sky
conditions encountered at each plot (Table 9-1) If
the CV was 20%, the sky condition was “clear if,
in addition, the mean was 900 j mo1fm 2 /s, and ii
was ‘overcast” if the mean was <900 mol/m 2 /s.
The sky condition was always “variably cloudy” if
the CV was >20%, The cutoff value for the mean
is about 60% of the maximum plot ambient mean
encountered (Table 9-1), and the clear sky condition
includes a few days with light, high haze. The cutoff
value for the CV was determined after a visual
inspection of the ambient data.
Figurc 9-3 DETAILS OF PAR SAMPLING DESIGN THE LARGE CIRCLES ARE THE 0017-ha SUBPLOTS UPON WHICH THE
SQUARE GRIDS (3 66 m) OF SAMPLING STATIONS ARE SUPERD4POSED SUBPLOTS ARE LOCATED 36 6 m FROM
SUBPLOT I CENTER AT AZIMUTHS OF 0, 120, AND 240 DEGREES.
Subo O1
center
3
PAR s rrVlo
St t I on
Detail 1 2
0
c cer of
58 rQl F Q
on S ,. -t 0iot
3 66 in
bet eefl stations
Off-plot
Siftifent PAR
as
/
‘10 - 100 in
PAR ss,-pies S
7) — 42
0
PAR n le5 ‘
11 - 26
PAR AerrOles I
I - 10
43 - 52
55 - 9-4
1200
PAP inp l O S I
69 - 64
PAP S 5 rOleS a
5) - 66
79
-------
Tsbk 9-1. AMBIENT PAR AND SKY CONDITIONS BY
Forrd
t g o
Plol
DO.
Ambient PAR
Sky
CV Time ’ codeb
(%)
Mean
(j.uno l/m’/.)
New
16
1299
52
0950
CL
E gtand
122
1597
4 9
1020
CL
155
416
24 6
3020
VC
160
23
191
1000
DV
173
1268
267
1050
VC
174
1181
127
1005
CL
181
88
73 I
3053
VC
267
994
7.7
0925
CL
270
1417
3 2
3025
CL
289
1113
302
1020
VC
321
347
420
71
1372
1376
340
5 3
8.2
1Q25
1020
1030
VC
CL
CL
so S
394
18 5
1040
OV
692
19
28
1130
OV
902
891
64 2
1040
VC
3270
133
468
1050
VC
Virguua
1839
1841
1843
1954
1955
1957
1958
1959
2072
2074
2185
2186
2187
2188
2189
2190
2191
2192
2303
1395
501
895
239
590
1108
349
973
3309
361
94
347
1385
269
375
1531
1202
1488
239
2 9
290
234
22 5
45.7
167
19.3
24 5
11 3
17 0
35
7.3
1.7
68 5
63 2
35
7.8
II 9
304
1025
1105
1110
1030
1020
1030
1105
1030
1035
1055
1025
1230
1055
0950
1225
1055
1030
1105
1140
CL
VC
VC
VC
VC
CL
OV
VC
CL
CV
CV
CV
CL
VC
VC
CL
CL
CL
VC
The percentage of transmitted PAR (PTPAR) was
estimated for each under-canopy sarnphng station as
follows. For stations on subplots 1 and 2, the
measured PAR values were divided by the average of
the 20 ambient PAR values measured during the first
and second visits to the ambient station. For Stations
on subplots 3 and 4, the divisor was the average of
the 20 ambient PAR values measured during the
second and third visits to the ambient station. A
10 g a r it h mi c trans for ma ti on of FTP AR,
ln(PTPAR-4. 1), was applied to normalize the
frequency distributions of PTPAR Mean values of
ln(PTPAR+ l)by subplot (n= 16) and by plot (n=64)
are shown in Table 9-2.
To estimate components of variance, linear
models for nested designs (e.g., John, 1973) were
used in the ANOVA (Table 9-3). A three-factor
model with effects of p’ot, subplot within plot, and
station within subplot was estimated by forest type.
To explore the extremes of variances, a reduced,
two-factor model was also estimated by plot Both
linear models provide estimates of the Station and
subplot variance components. Two contrasts of mean
ln(PTPAR + 1) among plots of different sky
conditions (clear vs. cloudy, and ‘overcast vs
variably cloudy) were estimated for each forest
type by using the three-factor ANOVA and sets of
orthogonal contrast coefficients. Estimates of plot
median PTPAR, and confidence intervals based on
variance estimates from the two-factor ANOVA,
were obtained by back-transformations without a
variance correction
To explore the possibility of serial correlation,
Box-Jenkins models (Box and Jenkins, 1970) were fit
to the senes of 64 values of In(PTPAR+ I) for each
plot. These analyses assumed stationary senes and
equal time intervals between series values. Each
series was first tested for the presence of serial
correlation by the Q-statistic (Ljung and Box, 1978)
with a significance level of p=O. 10. For series with
significant seiial correlation, the autocorrelation
functions were inspected to suggest appropriate
models. A parsimonious model was developed using
the Q-st.attstic to assess model adequacy. The fitted
Box-Jenkins models were then used to estimate an
adjusted series of ln(PTPAR+1) for each plot that
exhibited serial correlation The ANOVAs were then
repeated with the adjusted series in place of the
original senes for these plots.
• Zone iimc (nc 5rtrl 5 mm) it the midpoint of the sarnplin8 penod
1 Sky codes arc based on the mean (j mol/m 2 /s) and CV (%) of
ambient PAR: CL •clejr• OV overcast, and VC
vuLab 1 3 ! ckudy.
go
-------
Fore
region
Plot
no
Plot
mean
Mean for
subplot
1 2
3 4
Although the canopy was very uniform in each
The absence of serial correlation does not imply
sampled stand, it is possible that the values of
an absence of spatial correlation, but serial
ln(PTPAR+ 1) were spatially correlated. Spatial
correlation would (in this design) appear as east-west
correlation would violate one assumption needed to
axusotropic spatial correlation. Therefore, the plots
apply the usual sample size equations in Section 3.
without serial correlation provided the best tests of
Spatial correlation could also be incorrectly identified
spatial correlation For these plots, the semivanance
as serial correlation when the samplmg design
of ln(l’TPAR+ 1) was plotted over distance between
confounds the space and time dimensions. In this
stations for all station pairs (e g., Robertson, 1987).
study, each time-series of 64 values contains partially
These semivanograms showed virtually no evidence
confounded sets of spatial series at lags of one, two,
of spatial correlation at scales from —3.7 to —60 ci.
and three Stations in an east-west direction, and so
Furthermore, the variances estimated by the weighted
this possibility was explored,
means of the semivariance statistic were
approximately equal to the between-station variances
estimated by the two-factor ANOVAs. If these plots
Table 9-2. UNADJUSTED MEANS OF ln(PTPAR+ 1) BY
PLOT AND SUBPLOT
are representative, it may be concluded that spatial
correlation was not incorrectly identified as serial
correlation tn the time series analyses.
9.3 RESULTS
New 16 00630 00924 00162 00508 00928
England 322 00154 00103 00116 00222 0017$
155 02344 00462 00674 05515 02729
160 03723 04875 02174 04222 03625
173 0 0721 00358 0.1363 0 0268 0.0897
Theaverage ambient PAR and sky conditionsare
shown for each plot in Table 9-1. Average ambient
PAR ranged from only 19 to almost 1600 zmol/m 2 /s
174 0 0074 0 0046 0 0130 0 0048 0 0074
The lower values in the hardwood stands are
181 00754 0 1594 00612 00247 00562
267 0 0277 00183 0 0354 0 0286 0 0286
270 00874 00325 00872 01391 00911
289 0 1549 00324 00790 0 112.5 03958
321 0 1944 0 1364 0 2981 00879 0 2554
questionable because the field crews noted that it was
difficult to find suitable canopy openings and to
orient the radiometer towards the sun on very cloudy
days. The higher values are typical of late-morning
347 0 0262 0 0340 0 0208 0 0223 0 0276
values obtainable on cloud-free days in the summer.
420 0 0597 0 0749 0 0920 0 024.6 0 0472
508 00431 0 0377 00251 00512 00584
692 0 4463 0 5462 0.7343 0 1783 0 3263
902 00296 00254 00319 00272 00340
3270 0 1585 0 2523 0 0899 0 1050 0 1869
The time series analysis revealed that serial
correlation was never significant under clear sky
conditions (Table 9-4). Under cloudy sky conditions,
however, serial correlation was significant in 19 of
Virgims 1839 0 0477 00270 00502 00514 00623
1841 002.39 00437 00221 00069 00230
1843 0 0648 0 0402 0 0830 0 0963 0 0395
1954 0 0250 0 0213 00286 0.0288 0 0210
1955 03273 01501 04205 01970 05414
22 cases. An autoregressive, degree one (AR I)
model was appropriate for 18 of these 19 cases. A
.
parsimonious model could not be identified for one
plot (number 2189), but the AR1 model did reduce
1957 0 2527 02669 0 1968 03374 02296
serial correlation. The sharp contrast between the
1958 02462 0 1874 03346 03842 00784
1959 0 0153 0 0I57 00128 0 0091 00236
2072 0 0405 00389 00298 00837 00294
2074 0 1131 0 1434 0 1719 00625 00748
results obtained under clear sky conditions and results
obtained under cloudy conditions suggests that clouds
produced the serial correlation. This is likely
2185 0 1662 0 1814 0 1057 0 1962 0 1813
because clouds intercept PAR, and therefore any
2186 0 1472 0 1889 0 3047 0 0960 0 1992
variation in the depth or density of clouds that is
2187 0 0379 0.0488 0 0347 0 0253 0 0428
2188 00260 00424 00296 00208 00113
2189 00158 00361 00214 00024 00032
2190 00676 00680 00666 00645 0.0714
associated with their passage overhead will produce
a correlated senes of actual ambient PAR, and, thus,
of PTPAR under a homogeneous canopy.
2191 01380 01748 01912 01130 00731
2192 00213 00194 00161 00267 00228
2303 00278 0 0258 0 0245 0 0300 0 0308
81
-------
Serial correlation affected the estimates of
variance components. Comparisons of ANOVAs
before and after autoregression adjustments (Table
9-4) suggest order-of-magnitude increases in the
between-subplot and between-station variance
components as a result of serial correlation. Because
the adjustments reduced the between-subplot variance
proportionally more than the between-station
variance, the percentage of the total plot variance
attributable to stations usually increased after
adjustment. Thus, in this nested design, the
adjustments should tend to increase the precision of
plot-level and subplot-level means of In(PTPAR).
The adjustments for serial correlation also made
the between-station variance components for cloudy
sky conditions comparable to those obtained for clear
sky conditions (Table 9-4). But the percentage of
variance attributable to stations was still lower on
cloudy days because the between-subplot variances
remained much higher under cloudy sky conditions.
If it is assumed that the stands were otherwise
similar, it seems that cloud variability not accounted
for by the ARI models increased the variability of
estimated PTPAR at the subplot level, which resulted
in higher between-subplot variances relative to clear
sky conditions.
The three-factor ANOVAs before and after time-
series adjustments (Table 9-5) show the effects of
senal correlation on the precision of comparisons
among plots. The increased precision of contrasts
among plots after adjustment is due to the reduction
in between-subplot variance. The importance of this
reduction can be gauged, for example, by the
approximate doubling of F - ratios for the contrasts
of sky conditions after adjustment.
Figure 9-4 shows the estimated median PTPAR
for each plot and illustrates the time-series
adjustments on a per-plot basis. There was little
change in the plot medians, but the confidence
intervals were narrower. Because the estimates were
obtained by back-transformation of plot means and
standard errors of ln(PTPAR +1), the intervals were
based on relative error and were wider at higher
values of median PTPAR. The intervals for different
plots were also different because they were based on
plot-specific variances estimated from the two-factor
ANOVAs rather than on a common vanance
estimated from the three-factor ANOVAs.
Not much should be made of the apparently
higher transmittance under cloudy sky conditions in
the hardwood stands. This is so despite the
significant contrast of clear and cloudy sky conditions
(Table 9-5) and the visual impression of higher
transmittance under cloudy sky conditions (Fig. 9-4).
The reason for caution is that the evidence comes
mostly from stands for which the ambient PAR
measurements were unrealistically low (Table 9-1).
Ambient PAR is isotropic under cloudy sky
conditions (Smith and Clark, 1990), and small canopy
openings will yield maccurate and low averages, this
would tend to increase the estimated mean PTPAR.
Very large canopy openings were always available
near the pine stands, and in those cases the contrast
of clear and cloudy sky conditions was neither
significant (Table 9-5) nor suggested visually (Fig.
9-4).
9.4 DISCUSSION
Under clear sky conditions, there was little
variation in the percentage of transmitted PAR
between stations or between subplots in these uniform
stands. The sampling design yielded very precise
estimates of plot mean ln(PTPAR+1) and plot
median PTPAR was estimated with a relative
standard error of 0.1 to 2.7%. There was no
evidence of serial correlation under clear sky
conditions, and the usual variance estimates were
therefore appropriate. Spatial correlation, although
not detected here, could require stratification or other
analyses to derive appropriate estimates of median
PTPAR and its variance under less uniform canopies.
The serial correlation of PAR measurements
under cloudy sky conditions reduced the precision of
estimated plot median percentages of transmitted
PAR. The relative standard errors obtained were 0.2
to 12.2% before adjustment; these values were up to
four times larger than those obtained under clear sky
conditions. The relative standard errors were still
large after adjustment (0.1 to 10.4%) in comparison
to clear sky conditions. The differences were not
important for hypothesis tests in this study because
the overall precision was rather high. But the
differences could be much more important when
fewer samples are Ia1. en, or in stands where
transmitted PAR is more variable.
82
-------
Table 9-3 LINEAR MODELS AND EXPEC1ED VALUES OF MEAN SQUARES FOR THE ANOVAS
source of
Three-factor model
Two-factor model
Expected vslue of Correspondir g
Expected value of Corresponding
variance
the mean square variance component
the mean square variance component
Plot
4 + 164 + 644 4
Not in model Not ea*3nsatcd
Subplot
4 * 164
4 + 164
Sistion
4 4
4
Linear model-
ln(PTPAR+I) i + P + S +T
P — plot
S, = subplot within plot
T 9 — station within subplot
bo(P’TPAR+l) 1 _ i + S +T.
S subplot
T_ = station within subplot
PTPAR is the estimated percent of ambient PAR Uansmined through the canopy
Figure 9-4 PLOT MEDIAN PERCENT TRANSMTITED PAR MEDIANS ARE SHOWN BEFORE (0) AND AFTER (0) THE
ADJUSTMENT FOR SERIAL CORRELATION SOLID LINES INDICATE CONFIDENCE INTERVALS OF ± 2 PERCENTAGE
STANDARD ERRORS AFTER ADJUSTMENT. COMPARE TO WIDTHS BEFORE ADJUSTMENT (+) FOR THOSE PLOTS V 1TH
SERIAL AIJTOCORRELATION. PLOTS MEASURED ON CLOUDY DAYS ARE INDICATED BY ASTERISKS ON THE ABSCISSA
100
AN
PT PAP
a
00
60
40 -
20 -
bc o
0
1) 267 420 101 5270 160
PLOT 122 902 15 270 321 680
i&A EP 347 509 173 289 ¶95
NSA EiGI.* iC
1959 18 1 2303 1839 2074 2103 1955
2199 1954 9197 1942 2191 1958
2192 2108 201? 2190 9196 1937
Vi IN IA
83
-------
Table 9-4 TWO-FACTOR ANOVA OF PAR BY PLOT BEFORE AND AFTER TIME SERIES ANALYSIS
Overcasi New 160 7
England 508 7
692 7
Average
-7 021265 001259 001129
-7 000348 0000 18 000057
-7 095566 005929 000706
039060 002402 000631
47 3
76 0
106
446
0 00147 0 00261
0 00007 0 00023
0 04255 0 00926
0 01470 0 00403
64 1
75 8
179
52 6
Clear New 16 -10
England 122 -10
174 -10
267 -10
270 -10
347 -10
420 -10
Average
o 002182 000097 000634
0 0 00049 0 00002 0 00019
o 000024 000001 000006
0 000079 000000 000137
0 003039 000139 000808
0 0 00058 0 00000 0 00056
0 001421 000038 000808
0 00979 0 00040 0 00353
0 00097 0 00634
000002 000019
000001000006
000000 000137
o 00139 000808
0 00000 0 00056
0 00038 0 00808
0 00040 000353
86 8
91 2
825
100 0
853
99 8
95 5
91 6
Before
time
serica inalysii
Time
series analysis
After
time
senca analysis
Mean
Mean
Contrast
Mean
square
Mean
square
Forest
cocrncicnt
square
station
Statuor
square
station
Station 5
Sky region
P104
C 1 C 2
ibp1o4
( )
term
Model
SE( )
subplo4
ô
( )
term
ARI 043 0)2 002606
AR) 063 010 000138
AR! 089 006 069001
0 23915
Virginia 1958 7
2074 7
2185 7
2186 7
Average
Vanably New 155 7
cloudy England 173 7
181 7
289 7
321 7
902 7
3270 7
Avenge
Virginia 1841 7
1843 7
1954 7
195$ 7
1959 7
2188 7
2189 7
2303 7
Average
-8 031179 001912 000592
-8 0 04478 0 00249 0 00488
-8 002678 000154 000209
-s: 004730 000289 000101
0 10766 000651 000348
3 088178 005376 002167
3 004165 000154 001700
3 0 05435 0 00320 0 00322
3 042985 002596 001449
3 0 15564 000853 0 01924
3 0 00025 0 00000 0 00080
3 009154 000522 000806
023644 001403 001207
4 000365 000022 000012
4 001369 000078 000122
4 0 00030 0 00002 0 00002
4 054893 003324 001714
4 0 00061 0 00003 0 00010
4 000281 000017 000008
4 0 00417 0 00025 0 00014
4 0 00015 0 00000 0 00013
007179 000434 0002.37
o 68
0 55
0 41
0 81
0 85
0 66
0 70
0 69
0 59
0 69
041
0 53
0 92
o 36
0 82
0 65
009
011
0 12
0 08
0 08
009
009
009
0 10
009
o 12
011
0 07
0 12
0 08
0 10
0 10970
001375
0 00039
002611
0 03749
053518
004165
002188
0 19986
006048
0 00025
003179
0 12730
000160
0 00154
0 00007
0.33434
0 00006
0 00178
000163
0 00015
0 04265
0 00659 0 00425
000077000142
0 00020 0 00039
000157000092
000228000175
39 2
649
66 4
368
518
003247 001573
0 00154 0 01700
000127 000152
001201 000760
000316000998
000000-000080
000181 000282
0 00747 0 00792
32 6
91 7
545
387
760
1000
609
64 9
000010000006
0 00008 0 00024
000000000001
001976 001820
000000000001
000011 000005
000010000006
000000000013
000252000235
402
74 8
635
479
835
337
398
988
603
236
66 2
57-5
239
43 3
28 7
91 7
50 2
35 8
69.3
1000
60 7
62 3
35 8
61 0
55 2
34 0
75 0
32 7
35 6
98 8
53 5
86 8
91 2
82 5
1000
85 3
99 8
95 5
91 6
91 3
92 3
80 9
96 4
1000
75 6
96 1
90 4
AR)
AR)
AR!
AR!
ARI
WN
ARI
AR!
ARI
WN
ARI
AR)
ARI
AR!
AR)
AR)
AR)
ARI
WN
WN
WN
WN
WN
WN
WN
WN
WN
WN
WN
WN
W?
WN
WN
Virginia 1839 -12
1957 .12
2072 -12
2187 -12
2190 -12
2191 -12
2192 -12
Average
0 000352 000013 000139
0 004290 000153 001838
0 001370 000068 000287
0 000166 000004 000104
0 000013 000000 000403
0 004812 000252 000782
0 000033 0 00001 0.00020
001577 000070 000510
— — 002182
— — 000049
— — 000024
— — 000079
— — 003039
— — 000058
— — 0.01421
0 00979
— — 000352
— — 004290
— — 001370
— — 000166
— — 000013
— — 004812
— — 000033
0 01577
000013000139
000153 001838
0 00068 0 00287
000004 0 00104
0 00000 0 00403
0 00252 0 00782
000001 000020
000070 000510
913
923
80 9
96 4
1000
75 6
96 1
90 4
‘The time series models sre WN (white noise), implying that no t mne-senes adjuszmente were made, or AR) (first-degree auloregreeslon) ith
sutoregreulve parameter . The contrast cocfficicnt.s are used in the ANOVA repoited en Table 9.5
100 • (& I ô +è )
84
-------
Table 9.5 THREE-FACTOR ANOVA OF PAR BY FOREST TYPE BEFORE AND AFrER TIME SERIES ANALYSIS
Before Li
me series
analysis
A
fter tim
e series analysis
V an ncc
Variance
component 5
cOtnpone n
Forest
Sum of
Mean
Sum of
Mean
legion
Source
df
squares
square
F
p(>P)
Estimate %
squares
square
P
p(>F)
Estimate %
New
Plot’
16
1655652
103478
608
<001
001351 433
1673718
104607
1060
<001
001480 568
England
C3
(1)
4 95613
4 95613
29 10
< 001
— —
4 90450
4 90450
49 72
< 001
— —
C3
(I)
3.26700
3 26700
19 18
< 0.01
— —
3 48548
3 48548
35.33
< 001
—
Subplot’
Si
868612
017032
2261
<001
001017 326
503118
009865
1819
<001
000583 224
Station’
1020
7 68242
000753
—
—
0 00753 24 1
5 53261
0 00542
—
—
0 00542 20 8
Total
1087
32 92306
100.0
27.30097
100 0
Virginia
Plot
18
10 17930
0 36552
9 63
< 0 01
0 00792 52 9
9 73833
0 54102
17 00
< 0 01
0 00796 61 3
C,
(1)
0 05038
005038
0.86
0 36
— —
0.04289
0 04289
1 35
0.25
— —
C 3
Subplot
Station
(1)
57
1140
1 79104
3 4596
4 11497
1 79104
005870
000361
30 51
16.26
—
< 001
< 001
—
— —
000344 23 0
000361 241
1 87095
1.81388
3.68893
1 87095
003182
000324
58 79
9.83
—
< 001
< 001
—
— —
000179 138
000324 249
Total
1215
1764023
1000
1524114
1000
• C, is a contrast of clear and cloudy sky conditions, and C 2 is a contrast of cvcrcslt arid vanably cloudy sky conditions C, and C 2 arc
orthogonal contrasts of plots in each foreis type See Table 9-4 for contrast coefficients
Variance components for plot, nibplot, and s ti t io n estimate o , o , and t4 ,. respectively
Plolpercenl 100& l(& + ,
Subplot percent 100 iT / ( + & , + & ,)
• Station percent 100 & / (ô + â + i4 1 )
The estimates of variance components obtained in
this study can be used to select numbers of samples,
stations, and subplots in future studies of the same
general design ( 3). The higher between-subplot
variance will require more subplots on cloudy days to
achieve the same precision of plot comparisons. The
measurement of more subplots also implies additional
measurements at the ambient station to maintain the
same sampling design.
The serial correlation obtained under cloudy sky
conditions was caused by the separation in time of the
ambient PAR measurements from the series of under-
canopy PAR measurements. With this design, any
serial correlation in the actual ambient PAR owing to
cloud passage during the under-canopy sampling
period will cause serial correlation in the under-
canopy measurements. Although serial correlation
does not change the validity of the usual mean-value
estimators, time series adjustments will always be
needed to satisfy the assumptions of independence
and to estimate appropriate standard errors for
hypothesis testing. There is no guarantee that serial
correlation can be identified after the fact,
particularly in heterogeneous canopies, because
spatial and temporal variances may be confounded
For this reason, it may be worthwhile to make two
consecutive measurements at every sampling station
so that spatial and temporal variance can be cleanly
separated
The reduced efficiency of this design on cloudy
days was not fufly corrected by an AR1 model of
serial correlation. It may be that shifts in actual
ambient PAR between subplots inflated the between-
subplot variance. This variation would not be
affected much by an ARI model, but it should be
detectable as nonst.ationarity of the time series and
therefore amenable to analysis by other Box-Jer kins
models. But here, the result was that more subplots
and ambient measurements were needed to achieve
the same precision of plot mean transmittance on
cloudy days.
Cloudy sky conditions do not invalidate the
sample design, but they do maJ e it less efficient than
it could be. These analyses emphasize the
importance of obtaining comparable measurements of
ambient PAR with which to estimate transmittance on
cloudy days. Under clear sky conditions, the use of
85
-------
ambient measurements made before and after the
under-canopy measurements is acceptable because
changes in ambient PAR are relatively small, slow,
and predictable. But cloudy sky conditions (including
overcast) can cause unpredictable, large, and fast
changes in ambient PAR. The utility of the AR!
model suggests that cloud variability is important at
a scale of minutes, but it is probably also important
at other scales. Cloudy sky conditions also require
larger canopy openings to obtain accurate estimates of
ambient PAR.
Better estimates of station-by-station transmittance
will probably not be obtained by simply increasing
the frequency of visits to the ambient station, so long
as there is a time lag bet een ambient and under-
canopy measurements. Even continuous
measurement of ambient PAR cannot guarantee
comparability of measurements if the measurement
devices are separated in space, because clouds are
nonum form in space as well as in time. Solutions to
this sampling and data aggregation problem must
consider the spatial and temporal scales at which
inferences are to be made, relative to the scales at
which cloud and canopy conditions are likely to
change. One possibility is sequential sampling to
achieve a target precision in both ambient and under-
canopy situations. Another possibility is continuous
measurement of ambient PAR and the use of a time-
synchronized, running-average estimator of the ratio
of under-canopy PAR to ambient PAR. The latter
option was selected for testing in 1991.
9.5 MEASUREMENT QUALITY OBJECTIVES
This study was accompanied by a QA program to
develop standard operating procedures and a methods
manual (Dwire et al., 1990), to train, test, and field
audit data collectors, and to perform QC
remeasurements (Burkinan and Mickler, 1990).
Another aspect of the QA program that is addressed
in this section is the definition of measurement
quality objectives (MQOs) to guide future PAR
studies.
Measurement quality is descnbed by precision,
accuracy, representativeness, repeatability, and
completeness of a data set that is collected for a
specific purpose. For deciding among alternate field
data collection procedures, the most important data
attributes are accuracy, repeatability, and precision.
Precision is a function of the sampling design relative
to the degree and structure of population variability,
and is treated in detail in the discussion of vanance
components and sample design in Section 3.
Objectives for accuracy and repeatability of PAR
measurements require duplicate measurements in a
variety of situations. If one measurement gives a
true value, then the accuracy of a procedure can be
assessed. Without a known (or assumed) true value,
only repeatability can be assessed.
During the course of this study, duplicate
measurements were made by trainers and crews
during training, testing. QC visits, and audits In all
cases, the same sampling design and measurement
procedures were utilized. The duplicate
measurements from the training plot, three test plots,
and five field audit plots were used to suggest MQOs.
(The training and test plots are within the Pocahontas
State Forest, VA, in mixed loblolly pine-oak stands.
The audit plots are numbers 267, 1958, 1959, 2190,
and 2192.) Three of these 9 plots had 2 duplicate
measurements, giving a total of 12 instances where
both a trainer and a crewperson made measurements.
A second set of remneasurement data were collected
by the crewpersons on four plots (173, 267, 1955,
and 2186) during QC visits. On one of these plots,
two reme.asurements were made by the crew, giving
a total of six combinations of measurements.
Repeatability was assessed using bivanate plots of
the duplicate estimates of plot average PTPAR and
subplot average PTPAR (these estimates were not
made using the logarithmic transformation) The
correspondence of the two estimates is evidence of
the repeatability of the procedure. In Figure 9-5, the
crew estimates are plotted in relation to the trainer
estimates from the first set of remeasurements, and
there appears to be reasonably good correspondence.
In Figure 9-6, the two crew estimates from the
second set of remeasurements are plotted in relation
to each other, and there appears to be much less
correspondence.
It is reasonable to conclude that the repeatability
of plot and subplot PTPAR was less in the
operational situation than in the training, testing, and
auditing situations. Some of the discrepancy is
probably due to day-to-day variation in cloud
conditions with the second set of remeasurement data
It may also be speculated that the crews did not
obtain a high level of repeatability because they
86
-------
ru.shed the relatively simple PAR measurements.
Although 2 h were allotted for PAR measurements,
on average less than 1 h was utilized.
Assuming that the trainers obtained true values
of PTPAR, it is possible to examine the accuracy of
crew measurements by looking at the absolute value
of the differences in estimates on a plot and subplot
basis. Figure 9-7 shows the frequency distribution of
these differences obtained for the first set of
remeasurement data, and in most cases both the
crror in subplot and plot averages was 3%
(absolute) or less.
These values were used to define M QOs as shown
in Figure 9.8 and described by these equations.
Si ubj,tot MOO
Y 2% +0.2*(True% 1O%)
Plot MOO
Y,I% +01(Tn.ie%- 5%)
where Y and Y , refer to the allowable errors (i.e.,
absolute differences between true arid obtau:ied
estimates of PTPAR) for subplots and plots,
respectively. Most of the values in Figure 9-8 that
he above these lines were obtained during early
training and testing, and the later audit sample values
fell mostly below the MQO lines, suggesting that
accuracy improved during the course of the study.
Strictly, there was no independent test of whether
the crews met these suggested MQOs, because there
is no independent set of trainer and crew
remeasurements. But if the data from the second set
of (crew-only) remeasurements are used, and if the
first measurement is arbitrarily defined to be the
true value, then it appears that the MQOs were not
generally met in operational settings (Fig. 9-9). The
same reasons given for lack of repeatability would
explain the apparent lack of accuracy.
(4)
9.6 LOGISTICS
The SF-SO device has an internal clock, and every
observation is time-stamped. The analysis of these
recorded times is summarized by region in Table 9-6.
Crews usually started sampling ithm one-half
hour of the appointed time (11 n.m.). The New
England crew started as early as 10 a.m. and as late
as noon. The Virginia crews never started before
about II am. but started as late as 1 pm
The average time required for subplot sampling
(64 observations plus travel between stations within
subp lots) was 0.52 b in New England and 0.39 h in
Virginia. The maximum time required was 0.75 Ii
and the minimum was 0.23 h. Ambient samples (30
observations total) required about 0.08 h in both
regions.
Travel time to and from the ambient station (four
trips total) was relatively variable because the
ambient station was not a fixed distance away from
(5) the plot and because crews sometimes had to search
for an appropriate location. On average, about 0.25
h was required, but times as low as 0.09 h and as
high as 0.80 h were recorded. Travel time between
subplots 1 and 2, and between 3 and 4, averaged
about 0.08 h with a maximum of 0.34 h and a
minimum of 0.02 ii.
This analysis does not account the time for
break, thinking, or helpi .ng someone else
during the sampling procedures. Some of the higher
values (especially in the travel category) are
probably due to these activities, and the overall
averages are probably somewhat inflated. Yet
overall, the crews required less than I h on average
to complete the measurement procedures, and never
exceeded the 2 h allotment of time. The actual on-
station sampling time is less affected by extraneous
activities; on average, about one-half hour was
required to make and store 94 observations of PAR.
87
-------
TThIE REQUIRED FOR PAR SAMPLING
Table 9-6
Mean
(h)
Maximum
(h)
Minimum
(h)
C V
(%)
New England
Start time
10.99
11 87
9.98
4
Sampling time
52
0 75
0 36
23
-Subplota
0 08
0 09
0 06
12
-Ambien i
Travel time
o 80
0 09
73
-To/from ambient
0.23
0 04
41
-Between .ubplota
008
90
0
1.60
060
26
Total time
Virginia
Start lime
11 50
13 17
1057
6
Sampling lime
0 39
0 69
0 23
30
-Subploti
008
009
006
35
-Ambient
Travel time
0 79
0 11
60
-To/from ambient
0 28
0 34
0 02
88
-Between subplota
0 08
0 82
1 77
0 47
36
Total time
Figw 9-5. REPEATABILITY OF PTPAR IN TRAINING. TESTING, AND FIELD AUDITING EACH MEAN IS BASED ON 16
(SUBPLOT) OR 64 (PLOT) STATION MEASUREMENTS THE 11 RATIO LINE IS DRAWN FOR COMPARISON
60
50 () SLB LO1 AVEPA 5 (_) --
P LOI AVEPACES — —-
--
40
30 -
I-
‘.1 —
‘A
w
a 20 --
4 _,
a —
4 --
4 C, SI
U - --
10 ,,
t,
0 20 40
4VEQADE %TPAS ESTIMATED 9Y IQANEP kEASU hENTS
88
-------
80 -
0
0
Figure 9-6 REPEATA ILITY OF PTPAR Thi CREW REMEASUREMENTS EACH MEAN IS BASED ON 16 (SUBPLOT) OR 64
(PLOT) STATION MEASUREMENTS THE I I RATIO LINE IS DRAWN FOR COMPARISON (COMPARE TO FIGURE 9.5)
Figure 9-7. FREQUENCY OF DIFFERENCES IN FTPAR B WEEN TRAINER AND CREW THE TOTAL NUMBER OF PLOTS IS
12 AND THE TOTAL NUMBER OF SUBPLOTS IS 48 SEE ALSO FZGZJRE 9-5
() 5UBPLOI AVERAGES
60 - PLOT AVERAGES
0 0
- 0
40- ----------- 0
20- O© - 0
0
0
20 3 rD
TPAR AT FJ ST MEA5 P5 €NT
20
oc pLoTs
OR
SLOPLOTS
10
SLJDPLOTS
PLOTS
0
Ho- - _
0—11—22—33—4 4 5 5-86-77—88—88-jo io
ABSOLUTE DIFTERENCE II AVERAGE WTPAR
89
-------
Figure 9-8 MEASUREMENT QUALITY ORJECTFVES FOR SUBPLOT AND PLOT MEAN PTPAR THE SUBPLOT MQO IS A 2%
ABSOLUTE ERROR FOR PTPAR LESS THAN 10%, AND AN ADDITIONAL 20% RELATIVE ERROR FOR PTPAR GREATER
THAN 10%. THE PLOT MQO IS A 1% ABSOLUTE ERROR FOR PTPAR LESS THAN 5%, AND AN ADDITIONAL 10%
RELATIVE ERROR FOR PTPAR GREATER THAN 5% ALSO SHOWN ARE ERRORS IN SUBPLOT AND PLOT MEAN PTPAR
ESTIMATES OBTAINED DURING TRAINING, TESTING, AND FIELD AUDITING
Figure 9-9. FIELD EXPERJENCES WITH MQOS BASED ON CREW REMEASUREMENTS ERROR CALCULATIONS ASSUME
THE FIRST CREW MEASUREMENT IS ‘TRUE’ THE MQOS ARE AS IN FIGURE 9-8
40
30
0
25
“a
z
U-
U
a
15
10
5
20
o SU LOT
) PLOT AV8RAGES
0
0
0
0
0
a
0 20 40 60
TRUL PE CE 4T G TRAN ITEO PAR
80
90
-------
10 REFERENCES
Aiexander, S.A., and Carlson, J.A. 1988. National
Vegetation Survey project manual for the visual
damage survey pilot test. Department of Plant
Pathology, Virginia Polytechnic Institute and State
University, Blacksburg, VA.
Anderson, R.L., and Belanger, R.P. 1986. A crown
rating method for assessing tree vigor of loblolly and
shortleaf pines. Proc. Fourth Biennial Silvicultural
Research Conference, Atlanta, GA. pp. 538-542.
Armsori, K.A. 1977. Forest Soils: Properties and
Processes. Univ. of Toropto Press, Toronto, ON.
Blanchar, R.W., Edmonds, C.R., and Bradford,
J.M. 1978. Root growth in cores formed from
fragipan and B2 horizons of Hobson sod Soil Sci.
Soc. Am. 3. 42.437-440.
Bowers, Li., and Melhuish, JR. Jr. 1987.
Elemental analysis of red oak and loblolly pine
growing near an inactive chromium smelter.
Presented at the Sixth Central Hardwood
Forest Conference, Knoxville, TN, February 24-26,
1987.
Box, G.E.P., and Jenkins, G.M. 1970. Time series
analysis: forecasting and control. Holden-Day, San
Francisco.
Brooks, R.T., Miller-Weeks, M., and Burkman, W.
1991a. Summary report: Forest Health Monitoring,
New England, fiscal year 1990. USDA Forest
Service, Northeastern Forest Experiment Station,
Information Report NE-]NF-94-91, Radnor PA.
Brooks,R.T., Dickson, D.R., Burkman, W., Miller,
I., Miller-Weeks, M., Cooter, E , and Smith, L.
1991b. New England Forest Health Monitoring: 1990
annual report. USDA Forest Service, Northeastern
Forest Experiment Station, Resource Bulletin NE-
XX, Radnor PA. (in press).
Burkman, B., and Mickler, B. (eds). 1990. Quality
assurance auditing and activities report for the 1990
Forest Health MonitoringlEMAP forest research
project. Internal report, US Environmental Protection
Agency, and US Forest Service, Radnor PA.
Byers, G.E., Van Remortel. R D., Teberg, i.E.,
Miah, M.J., Palmer, C.3., Papp, M.L., Cole, W.H.,
Tansey, A.D., Cassell, D.L., and Shaffer, P.W.
1989. DirectfDelayed Response Project. Quality
Assurance Report for Physical and Chemical
Analyses of Soils from the Northeastern United
States. EPA/600/4-89/037. U.S. Environmental
Protection Agency, Las Vegas, NV. 216 pp.
Byers, G.E., Van Remortel, R.D., Miah, MJ.,
Teberg, I.E., Papp, M.L., Schumacher, B.A
Conkling, B.L, Cassell, DL., and Shaffer, P.W.
1990. DirectlDelayed Response Project: Quality
Assurance Report for Physical and Chemical
Analyses of Soils from the Mid-Appalachian Region
of the United States. EPA/600/4-90/00l. U S.
Environmental Protection Agency, Las Vegas, NV.
337 pp.
Byers, G.E., and Van Remortel, R.D. (eds). 1991.
Environmental Monitoring and Assessment Program -
Forests: Laboratory Methods Manual for the Forest
Health Monitoring Demonstration and Pilot Projects.
Internal Report U.S. Environmental Protection
Agency, Las Vegas, NV.
Carter, R.E., Otchere-Boateng, I., and Klwka, K.
1983. Dieback of a 30-year-old Douglas-fir
plantation in the Brittain River Valley, British
Columbia symptoms and diagnosis Forest Ecol
Manage 7:249-263.
Cline, S.P., Burkman, W.G., and C.D. Gerori
1989. Use of quality control procedures for assessing
variation in measurements of forest canopy condition
In Effects of air pollution on western forests. Edited
by R.K. Olson and A.S. Lefohn. Air and Waste
Management Association, Anaheim, CA. pp 379-394.
Cochran, W.G. 1977. Sampling techniques Third
edition. John Wiley, New York, NY.
Cost, N.D. 1979. Ecological structure of forest
vegetation. In Forest resources inventory workshop
proceedings, Volume 1. Edited by W.E Frayer,
Colorado State University, Ft. Collins, CO pp 29-
37.
91
-------
Curtis, R.O. 1983. Procedures for establishing and
maintaining permanent plots for silvicultural and yield
research. U.S. Forest Service, Pacific Northwest
Forest and Range Experiment Station, General
Technical Report PNW-155.
Decagon Devices, Inc. 1989. Sunfleck ceptometer
user’s manual. Decagon Devices, Inc., Pullman,
WA.
Dwire, K., Huntley, B., Miller-Weeks, M., and
Brooks, R. (eds.). 1990. Forest Health Monitoring
plot design and logistics study — methods manual for
field measurements and sample collection. Internal
report, US Environmental Protection Agency, Las
Vegas NV and US Forest Service, Radnor PA.
Eyre, F.H. (ed). 1980. Forest cover types of the
United States and Canada Society of American
Foresters, Washington DC.
Fralish, J.S. 1988. Predicting potential stand
composition from site characteristics in the Shawnee
Hills Forest of illinois. Amer. Mid. Nat. 120.79-101.
Gower, S.T., and Norman, 3M. 1991. Rapid
estimation of leaf area index in conifer and broad-leaf
plantations using the Li-Cor LAI-2000. Ecology (in
press).
Green, R.N., Marshall, P.L., and Khnka, K. 1989.
Estimating site index of Douglas-fir from ecological
variables in southwestern British Columbia For. Sci.
35:50-63.
Griffen, R.A. and Juririak, J.J. 1973. Estimation of
activity coefficients from the electrical conductivity of
natural aquatic systems and soil extracts. Soil Sci.
116:26-30.
Gumpertz, M.L., D. Tmgey, and W. Hogsett 1982.
Precision and accuracy of visual foliar injury
assessments. I. Environ. Qual. 11:549-553.
Hamilton, W.N. and Krause, N.H. 1985.
Relationship between jack pine growth and site
variables in New Brunswick plantations. Can J. For.
Res. 15:922-926.
Hewlett, J.D. 1961. Soil moisture as a source of base
flow from steep mountain watersheds. U.S. Dept.
Agr. For. Serv. Res. Pap. SE-132.
Hillel, D. 1980. Fundamentals of soil physics.
Academic Press, New York.
Hunsaker, C.T., and Carpenter, D .E. (eds.). 1990.
Ecological indicators for the Environmental
Monitoring and Assessment Program. EPA 600/3-
90/060. U.S. Environmental Protection Agency,
Office of Research and Development, Research
Triangle Park, NC.
Husch, B., Miller, C.I., and Beers, T.W 1972
Forest mensuration. Second edition. The Ronald
Press, New York, NY.
limes, J.L. 1988a. Forest health surveys — a critique.
Environ. Pollut. 54:1-15.
Innes, 3.L. 1988b. Forest health surveys problems in
assessing observer objectivity. Can 3. For Res
18:560-565.
Innes, IL., and Boswell, R.C. 1987. Forest health
surveys 1987. Part 1: Results. Forestry Commission
Bulletin 74, Forest Research Station, AJice Holt
Lodge, Wrecclesham, Faruham, Surrey, Great
Britain.
Innes, J.L., and Boswell, R.C. 1990. Reliability,
presentation, and relationships among data from
inventones of forest condition Can. J. For Res
20:790-799.
John, P.W.M 1971. Statistical design and analysis of
expenments Macrmllan, New York
Kmenta, 3. 1971. Elements of econometrics.
Macmillan, New York.
Kramer, P.3., and Kozlowski, T.T. 1979. Physiology
of woody plants Academic Press, New York.
Leaf, A.L. 1973. Plant analysis as an aid in
fertilizing forests. in Soil Testing and Plant
Analysis. Edited by L.M. Walsh arid J.D.
Beaton. Soil Sci Soc. Am., Madison, WI, pp.
427-454.
Lindsay, W.L. 3979. Chemical equilibna in soils. 3.
Wiley & Sons, New York.
92
-------
Ljung. G.M., and Box, G.E.P. 3978. On a measure
of lack of fit in time senes models. Biometnka 65:
297-303.
Mader, D.L. 1976. Soil-site productivity for natural
stands of white pine in Massachusetts. Soil Sci. Soc.
Am. 1. 40:112-115.
Magasi, L.P. 3988. Acid rain national early warning
system — manual on plot establishment and
monitoring. Canadian Forestry Service, Inf. Rep.
DPC-X-25, Ottawa, Canada.
Magurran, A.E. 1988. Ecological diversity and its
measurement. Princeton University Press, Princeton,
NJ.
Mausbach, M.J., Brasher, BR., Yeck, R.D., and
Nettleton, W.D. 1980 Variability of measured
properties in morphologically matched pedons. Soil
Sd. Soc. Am. J. 44:358.363.
Millers, I., and Lachance, D. 1989. North American
Sugar Maple Decline Project -- Cooperative field
manual. U.S. Department of Agriculture, Forest
Service, Northeastern Area State and Private
Forestry, Durham, NH.
Nash, B.L., Davis, D.D., and Skelly, J.M 1990.
Laboratory diagnosis for symptoms and causal agents
on sugar maple and red maple leaves and twigs
collected at EMAP field plots, 1990. Department of
Plant Pathology, The Pennsylvania State University,
University Park, PA.
Overton, W.S., White, D., and Stevens, D.L. 1990
Design report for EMAP, Environmental Monitoring
and Assessment Program, Part I. Draft. US
Environmental Protection Agency, Corvallis OR.
Palmer, C., Barnard, 1., Brooks, R., and Cost, N.
1990. Forest health monitoring plot design and
logistics: a joint EPA/USFS study plan. Internal
Report, US Environmental Protection Agency,
Corvallis, OR, and USDA Forest Service, Research
Triangle Park, NC.
Palmer, C.J., Rjitters, K.H.. Strickland, T., Cassell,
D.L., Byers, G.E., Papp, M.L., and Liff, C.l. 3991.
Monitoring and research strategy for forests —
Environmental Monitoring and Assessment Program
(EMAP). EPA/60014-91/_. US Environmental
Protection Agency, Washington, DC.
Pawluk, S., and Arneman, HF. 1961. Some forest
soil characteristics arid their relationship to jack pine
growth. For. Sci. 7: 160-173.
Pearce, S.C. 1953. Field experimentation with fruit
trees and other perennial plants Tech. Commun. No
23, Commonwealth Bureau of Horticulture and
Plantation Crops, Commonwealth Agncultural
Bureaux, East MaIling, Maidstone, Kent, Great
Britain.
Pielou, E.C. 1969. An introduction to mathematical
ecology. Wiley, New York, NY.
Pierce, L.L., and Ruruung, S.W. 1988 Rapid
estimation of coniferous forest leaf area using a
portable integrating radiometer. Ecology 69: 1762-
1767.
Raitters, K., Law, B., Kucera, R., DeVelice, R
Gallant, A., and Palmer, C. 1990. Indicator strategy
for forests. In Ecological indicators for the
Environmental Monitoring and Assessment Program
Edited by C.T. Hunsaker and D E. Carpenter. EPA
600/3-90/060 U S. Environmental Protection
Agency, Office of Research and Development,
Research Triangle Park, NC Chapter 6.
Ripley,B.D. 1981. Spatial Statistics John Wiieyand
Sons, New York, NY.
Robertson, G.P. 1987. Geostatistics in ecology
interpolating with known variance. Ecology 68. 744-
748.
Russell, G., Jarvis, P.G., and Monteith, J.L. 1989.
Absorption of radiation by canopies and stand
growth. In Plant canopies their growth, form and
function. Edited by (3. Russell, B. Marshall, and
P.G. Jarvis. Cambndge University Press,
Cambridge.
SAS. 1985. SAS procedures guide for personal
computers. Version 6. SAS Institute, Cary, NC.
93
-------
SEA-AR. 1981. Soil erosion effects on soil
productivity: a research perspective. Science &
Education Administration - Agricultural Research. 3.
Soil & Water Cons. 36:82-90.
Short, H.L., and Williamson, S C. 1986. Evaluating
the structure of habitat for wildlife, in Modeling
habitat relationships of terrestrial vertebrates. Edited
by 3. Verner, M.L. Morrison, and C.i. Ralph.
University of Wisconsin Press, Madison, WI.
Skelly, 3M., Davis, D.D., Memli, W., Cameron,
E.A., Brown, H.D., Drum.inond, D.B., and L.S.
Dochinger (eds). 1987. Diagnosing injury to eastern
forest trees. Department of Plant Pathology, College
of Agriculture. The Pennsylvania State University,
University Park, PA.
Smith, N.J., and Clark, D.R. 1990 Estimating salal
Leaf area index and leaf biomass from diffuse light
attenuation. Can. J. For. Res. 20: 1265-1270.
Spiers, G.A., Packee, E.C., Louiser, J.D., Dudas,
M.J.. and Gagnon, D. 1983. Trace element levels in
young Douglas-fir plantations exhibiting “dieback”
symptoms, West Coast, Vancouver Island. (abstract)
Proc. Soc. Am. Foresters Ann. Meeting, 1983. Soc.
Am. Foresters, Washington, DC.
Steinbrenner, E.C. 1963. The influence of individual
soil and physiographic factors on the site index of
Douglas-fir in western Washington. in Forest-Soil
Relationships in North America Edti’ed by C.T.
Youngberg. Oregon State University Press, Corvallis,
OR. pp. 261-27g.
Stone, R. E., and Weislander, A,E. 1948. Rating
soils for timber sites. Soil Sci. Soc. Am Proc.
13:499-509.
Sukhatme, P.V. 1954. Sampling theory of surveys,
with applications. Iowa State College Press, Ames,
IA.
Taylor, 3.K. 1987. Quality assurance of chemical
measurements. Lewis Publishers, Chelsea, Ml.
United Nations Economic Commission for Europe
1987. Manual on methodologies and criteria for
harmonized sampling, assessment, monitoring, and
analysis of the effects of air pollution on forests.
Convention on Long-Range Transbouadary Air
Pollution, International Cooperative Programme on
Assessment and Monitoring of Air Pollution Effects
on Forests, United Nations Environment Programme
and the Economic Commission for Europe, Brussells,
Belgium.
U.S. Environmental Protection Agency. 1990,
Threats to biological diversity in the United States
Office of Policy, Planning, and Evaluation, U S
Environmental Protection Agency, Washington, DC
U.S. Forest Service. 1982. An analysis of the timber
situation in the United States 1952-2030. U.S.
Department of Agriculture, Forest Service, Forest
Resource Report No. 23, Washington, DC.
U.S. Forest Service. 1985a, Forest Service resource
inventory: an overview. U.S. Department of
Agriculture, Forest Service, Forest Resources
Economics Research Staff, Washington, DC.
U.S. Forest Service. 1985b. Field instructions for the
southeast U.S. Department of Agriculture, Forest
Service, Southeastern Forest Experiment Station,
Asheville, NC.
Van Remortel, R.D., Byers, G.E., Teberg, J E,
Miah, M 3., Palmer, C.3., Papp, M.L , Bartling,
M H., Tansey, A.D , Cassell, D.L., and Shaffer,
P.W. 1988. Direct/Delayed Response Project
Quality Assurance Report for Physical and Chemical
Analyses of Soils from the Southern Blue Ridge
Province of the United tates. EPA/600/8-S8/100.
U.S. Environmental Protection Agency, Las Vegas,
NV. 251 pp.
Waring, RH., and Schlesinger, W.H. 1985. Forest
ecosystems: concepts and rnanagemen
Academic Press, Orlando.
Wentworth, T.R., and Joyner, K.C. (eds). 1987.
Report on the forest damage survey workshop. U S.
Department of Agriculture, Forest Service,
November 2-4, 1987, Chapel Hill, NC Available
from the FHM Program Manager, U.S. Forest
Service, Research Tnangle Park, NC.
94
-------
Wilde, S.A. 1964. Changes in soil productivity
induced by pine plantations. Soil Sci. 97:276-278.
Will, G.M. 1972. Copper deficiency in radiata pine
planted on sands at Mangawhai Forest. New Zealand
3. For. Sd. 2:217-221.
Zedaker, SM., and Nicholas, N.S. 1988. Quality
assurance methods manual for site classification and
field measurements. Department of Forestry, Virginia
Polytechnic Institute and State University,
Blacksburg, VA.
95
-------
APPENDIX A
SUPPLEMENTAL TABLES
96
-------
Tsbk A-I DESCRWTIVE DATA BY SUBPLOT AT EACH LOCATION
16 1 18 112 mid-slope pI.nar hesdilope
2 12 *42
3 20 *07
4 22 121 s dc s lope
122 1 6 278 upper slope ndgetop
2 25 348 • convex sidealope
3 5 80 planar ndgclop
4 7 250 sides lope
155 1 6 191 lower slope
2 8 192 mid slope
3 0 0 flatland • none
4 0 0 -
159 1 8 43 mid-slope sideslopc
2 5 20 lower slope
3 6 58 mid.slope
4 10 50 upper slope
160 1 10 205 mid-slope concave headsiope
2 6 226 upper slope convex sideslope
3 6 243 complex headalope
4 7 238 mid-slope concave
166 1 15 *77 lower slope planar headalope
2 18 In mid-slope sideslope
3 10 175 lower slope concave headslope
4 *6 130 mid-slope complex
173 I 10 222 lower slope planar
2 5 285
3 5 190
4 35 300 mid-slope
side slope
sugar maple-beech-yello birch
red maple-northern hardwood
174 1 52 290
2 35 261
3 86 305 upper slope
4 50 295 mid-slope
*81 1 15 197
2 17 193
3 6 190
4 5 152
• ndgewp
• sideslope
sugar rnaple-becch-yellow birch
Silt prep
267 1 14 149
2 iS 65
3 5 159 lower slope
4 20 260 mid-slope
headslope red maple-northern hardwood
side slope
headilope
sid tlope
P104
Terrain
Micro-
LO Subplot Slope
Aspecib
positiorf
relief’ Landform Forest type’ ist tirbance’
percent
degrees
1*7 3 13 63
2 12 50
3 23 47
4 IS 72
red maple-northern hardwood
sugar maple-bcech-yellow birch
insects
97
-------
270 1 7 284 lower slope concave
2 15 303 planar
3 10 285 mid-slope
4 22 310 lower slope headalope
289 1 13 102 mid-slope planar iideslope
2 7 76
3 9 88
4 15 87 upper slope
321 1 24 174 mid-slope -
2 20 160 upper slope
3 20 163 mid-slope
4 20 168
347 I 16 291 concave headsiope
2 38 295 upper slope planar
3 15 290 mid-slope
4 18 285
420 1 5 352 lower slope
2 0 0 bouomiand
3 5 22 lower slope
4 0 0 flaiJand
508 1 5 250 lower slope s ides lope
2 8 230 upper slope
3 9 230
4 0 0 flatland
692 I 5 22 lower slope
2 0 0 flatland
3 5 64 lower slope -
4 17 15 mid-elope
3270 1 12 55 mid-elope convex aideslope
2 15 84 planar
3 7 30
4 24 87
1 0 0 fiatlsnd
2 0 0
3 0 0
4 0 0
1 0 0
2 0 0
3 0 0
4 0 0
I 0 0
2 0 0
3 0 0
4 0 0
Table A-I (CONTINtJED)
P loe
Terrsrn
Micro-
no. Subptoe Slope
Aspectb
poiition
relief’ Laridform Forest lype’ Distuthance
percent
degree.
weather
weather
sidealope
none
s idealope
none
red maple-nonhern hardwood
sugar rnaple-beech-ycllow birch
red maple-northern hardwood
sugar maple-beech-yellow birch
red maple-northern hardwood
sugar maple-beech-yellow birch
none
side slope
1839
1841
1843
none loblolly
98
-------
Table Al (CONTINUED)
1 0 0
2 6 210
3 0 0
4 9 65
1 5 115
2 5 120
3 5 115
4 5 120
1 0 0
2 0 0
3 0 0
4 0 0
1 0 0
2 0 0
3 0 0
4 0 0
1 0 0
2 0 0
3 0 0
4 0 0
botiomiand concave drsw
upper slope convex head lope
mid-slope convex sideilope
nose elope
Ioblol ly-h ardwood
loblolly
Iob lo lly-hardwood
loblol ly-hsrdwood
loblolly
loblolly-hardwood
2185
1 6 325
2 0 0
3 0 0
4 5 330
upper ilope convex stdeslope
bench complex hesdilope
upper slope convex iideslope
loblo lly
2186 1 0 0 flatland planar none
2 0 0
3 0 0
4 8 164 imd-slope concave sideslope
I 5 322 planar
2 8 3 10
3 6 25
4 7 320
Plod
Terrain
Micro-
no. Zubplo Slopc
Aspecib
pos LiOn
relief 1 Landform Foresi type Distuthancei
perceni
degrees
1954 1 0 0
2 0 0
3 0 0
4 0 0
loblolly
1 0 0
2 0 0
3 0 0
4 0 0
1955
1957
1958
1959
2072
2074
2075
fladand
weather
fire
weai.he r
planar none
loblolly
1 0 0
2 0 0
3 0 0
4 0 0
2187
99
-------
T
able A-I
(CONTINUED)
Ploc
Terrain
Micro-
.
Subplot
Slope
percent
Aspects position’
degrees
relief’
Lindform
Forest type’
D,sturbsnce
2188
1
0
0 flseland
•
none
2
0
0
Ioblolly-hardwood
3
0
0
4
0
0
loblolly
2189
1
0
0
2
10
310 mid-slope
complex
uideelope
3
5
170 flatland
planar
none
4
5
2.55
.
.
fire
2190
1
8
329 top ofalope
headsiope
2
9
320 lower slope
concave
draw
3
9
245 mid-slope
aidealope
4
10
222
2191
1
5
306 mid-slope
planar
aideslope
loblolly
2
5
313
3
0
0 flatland
none
4
5
300 mid-slope
sideslope
2192
1
5
21 lower slope
concave
2
5
357 mid-slope
3
10
299
planar
.
4
10
19
.
2303
1
5
45
complex
.
2
5
4 .4
.
3
5
305 flailand
4
5
203 mid-slope
Percentage slope at subplot center (Zeda1 er and Nicholas, 1988)
Direction of exposure for land surcsces that have s 1 opc >3 (Zcdaker and Nicholas, 1988)
• The location of the subplot along the slope profile (Zedaker and Nicholas, 1988)
The shape of a slope or slope position on v.hich the subplot is located (Zedaker and Nicholas, 1988)
‘The position of the subplot with respect to surrounding topography (Zedaker and Nicholas, 1988)
‘Subjective judgemeni based on plurnlit of spec ics in dominant and co-dominant crov n classes Sec Table 3 for descnptiont (From D ire
at .1 , 1990).
‘Evidence of named type of disrurbance on the subplot
100
-------
Table A-2 OVERSTORY NUMBERS OF TREES, BASAL AREA, AND AVERAGE DIAMETER BY SPECIES
Plot
bO. Species
No of Basal Avg
tr eei area dbh’
Plot
no Species
No of Easel Avg
Iz ’eel area dbh
be’
cm
ha’ ma/ha cm
16 eastern white pine
red maple
paper birch
white ash
black cherry
Dorthcrn red oak
122 red spruce
eastern hemlock
red maple
sugar maple
yellow birch
paper birch
American beech
white ash
bigtooth aspen
15$ balsam fir
red spruce
red maple
sugar maple
yellow birch
black cherry
159 balsam fir
eastern hemlock
red maple
sugar maple
yellow birch
American beech
white ash
160 balsemfir 30
red spruce 30
eastern hemlock 252
red maple 326
sugar maple 104
yellow birch $9
American beech 15
eastern hophornbeam 30
black cherry 44
267 1395 258
163 7.32 23 9
59 2.77 24 4
30 069 17,2
89 308 210
44 048 147
206 243
342 313
967 263
480 227
256 210
5.18 472
025 103
025 447
44 133 195
415 1345 203
119 414 211
222 4.72 164
30 046 141
44 179 226
282 608 166
15 048 203
133 592 238
89 112 126
15 012 102
119 632 260
222 421 155
59 3.82 28 6
89 397 238
148 176 123
44 383 33.1
222 6.98 200
148 240 144
267 ‘7.27 186
15 026 150
148 7.88 26.0
036 125
208 299
855 208
1745 261
350 207
415 299
045 lI.2
077 181
419 346
166 balsam fir
red spruce
red maple
sugar maple
yellow birch
Amencan beech
white ash
1’73 eastern white pine
eastern hemlock
red maple
sugar maple
yellow birch
sweet birch
paper birch
Arnencan beech
white ash
black cherry
northern red oak
174 eastern hemlock
sugar maple
yellow birch
paper birch
Amencan beech
c*atcrn hophornbcam
northern red oak
267 balsam fir
white spruce
red spruce
striped maple
red maple
sugar maple
yellow birch
paper birch
white ash
30 129 235
133 246 153
15 119 320
119 550 243
89 941 367
341 632 154
44 154 210
146 458
1081 182
360 227
065 137
172 222
071 142
109 153
170 135
122 187
219 306
739 398
311 961 198
252 1005 225
15 137 343
30 336 380
119 670 268
15 015 112
44 1629 683
011 97
044 112
012 102
907 193
855 303
008 84
009 89
115 91
59 122 162
15 103 297
59 148 159
178 128 96
356 1184 206
30 027 108
119 367 198
30 268 339
15 010 91
117 eastern bemiock 44
red maple 44
sugar maple 178
yellow birch 119
paper birch 74
American beech 30
eastern hophornbearn 30
American basswood 45
74
415
89
44
44
44
59
119
44
30
59
181 balsam fir 15
red spruce 44
red maple IS
sugar maple 311
yellow birch 419
Sweet birch 15
Amencan beech 15
commonchoke cherry 178
101
-------
Table A.2 (CONTINUED)
P b s
no. Specie.
No. of Basal Avg
trees area dbh
ha’ &/ba em
Plot
no. Specie.
No of Basal Avg
bees irca dbh
ha’ m’/ha
cm
270 sagern white pine
stem hemlock
red maple
aigai maplc
yellow birch
sweet birch
paper birch
American beech
i thiw oak
uonhcrn red oak
289 balsam fir
red spruce
striped maple
nd maple
yellow birch
paper birch
American beech
biglooth aspen
quaking .spen
321 red spruce
,cd maple
sugar maple
American beech
eastern hophornbeam
30 260 334
267 838 200
104 254 176
59 061 115
30 0.51 14 8
178 718 227
30 054 152
44 285 286
44 265 275
119 7.21 278
104 1.86 151
30 1.15 222
15 018 124
534 557 115
IS 011 99
59 185 199
59 106 151
232 7.58 196
15 048 203
44 256 271
44 698 447
415 795 156
237 1072 240
15 083 267
5.20 211
465 153
064 165
162 216
387 2.35
1156 315
009 86
094 164
154 182
074 112
126 190
6.52 39 3
271 483
0.72 11.1
1135 231
106 302
097 204
068 340
508 balsam fir 89
white spruce 59
red maple 313
sugar maple 237
American beech 348
eastern hophornbeam 30
quaking aspen 282
692 balsam fir
rcd spruce
eastern white pine
northern white-cedar
red maple
big ooth aspen
3270 balsam fir 133
red spruce 104
striped maple 30
red maple 104
sugar maple 104
yellow birch 193
paper birch 297
American beech 15
common choke cherry 208
Amencan mountain-ash 30
1839 loblolly pine
red maple
.weetgum
lowland blackgum
white oak
willow oak
1841 loblolly pine
red maple
swcetgu m
elm
400
193
15
44
74
74
267
74
1.07 124
4.26 30 3
3234 21 3
744 200
390 183
029 Ill
320 320
121 108
260 178
035 80
400 222
393 220
705 216
507 148
010 94
397 110
204 296
1630 228
285 137
014 109
054 124
047 90
042 85
460 148
094 127
400 480 324
178 619 210
89 307 230
44 151 208
163 442 186
15 238 452
347 balsam fir 148
red spruce 252
striped maple 30
red maple 44
gar maple 89
yellow birch 148
sweet birch 15
paper birch 44
cherry 59
American mountain-ash 74
420 balsam fir 44
red spruce 222
sastemn hemlock 15
striped maple 74
red maple 267
yellow birch 25
American beech 30
eastern hophombeam 44
489 2184 238
237 283 123
44 047 116
59 074 126
30 054 352
89 913 361
341 2587 311
30 305 362
682 859 127
148 233 141
1843 loblolly pine
red maple
American holly
.wcetgurn
iwecibay
lowland blackgumn
white oak
water oak
302
-------
Table A-2 (CONTINUED)
hi’ iTi /ha cm
311 1217 223
104 098 110
ha’ mi/ha cm
Plot
No
of
Basal
Avg
Plot
No
of
Basal
Avg
no. Species
trees
area
dbh
no. Species
trees
area
dbh
1954
ioblo!ly pine
474
37 77
31 8
2074
loblolly pine
549
17 98
20 4
red maple
282
3 96
13 4
red maple
15
0 08
8 1
persimmon
15
0 Il
9 9
American holly
44
0 45
11 3
uwcetgum
208
3 59
14 8
swcetglim
59
072
12 4
IowlaridbLsckgum
15
0 16
11 7
yellow-poplar
15
0 19
12 7
aim
44
313
18 0
sourwood
whilcoak
89
15
152
035
14 8
173
1955
Ioblolly pine
311
19 93
28 5
scarlet oak
15
0 10
9 1
V irgiruap ,ne
15
061
229
southcrnrcd oak
415
459
119
red maple
178
2 20
12 5
water oak
15
0 23
14 0
black walnut
15
099
292
willow oak
30
0 19
90
Pwec1gi. m
30
0 97
20 5
lowland blackgum
15
0 12
10 2
2075
loblolly pine
redmaple
489
15
22 67
014
24 3
109
1957
loblolly pine
193
15 36
3! 9
bluebeech
15
0 09
8 6
red maple
74
2 40
203
hickory
59
1 66
18 9
flowering dogwood
74
0 64
10 5
American holly
15
0 09
8 9
persimmon
15
0 09
8 6
swee tgum
30
3 61
39 4
American holly
44
0 32
9.5
southern red oak
IS
0 17
II 9
iwec tgum
311
488
141
water oak
15
023
140
lowland bIsckgiim
15
011
99
willow oak
59
641
37 1
ourwood
IS
029
15 7
post oak
15
0 10
9 I
white oak
30
1 26
23 3
black oak
74
4 08
26 5
1958
loblolly pine
2185
eastern redcedar
44
0 24
8 3
red maple
loblolly pIne
Virginia pine
297
652
21 75
36 17
30 6
37 8
1959
loblofly pine
534
16 24
19 7
swec tgum
15
0 23
34 0
red maple
178
6 26
21 2
lowland blackgum
30
0 30
11 3
American beech
104
2 77
18 4
black cherry
iS
0 07
7 9
American holly
89
1 34
13 8
Iweetgum
104
2 37
17 0
2186
loblolly pine
801
31 00
22 2
sourwood
44
047
II 6
Virginia pine
252
5 42
16 5
white oak
15
2 10
42 4
red maple
44
0 30
9 3
water oak
74
3 84
25 7
aweelgurn
44
0 48
11 7
post oak
15
074
2.5 1
yellow-poplar
30
1 39
24 4
2072
loblolly pine
474
21 58
24 2
2187
loblolly pine
‘356
II 23
20 0
Virginia pine
35
0 13
10.7
Virginia pine
400
8 35
16 3
red maple
119
1 08
10 8
yellow birch
104
098
11 0
hickory
74
0 99
13 1
American holly
15
0 07
7.6
flowering dogwood
15
0 08
8 1
Iweelgum
74
1 83
17 7
American beech
15
0 21
13 5
yellow-poplar
208
4 84
17 2
American holly
35
008
8 4
upland blackgum
15
0 35
13 2
swec tgum
333
1 50
12 0
southern red oak
163
1 47
10 7
yellow-poplar
89
0 85
11 0
black oak
59
1 20
16 0
whitcoak
208
240
12.1
rearletoak
15
029
157
aouthernrc doak
44
041
108
postosk
IS
009
86
blackoak
30
038
128
203
-------
Table A-2 (CONTINUED)
ha • m’/hs cm
ha’ n’i’/ha cm
‘Diameter of the tree of average basal area
Plot
No
of
Basal
Avg Plot
no. Species
trees
arcs
dbh no Species
No of Basal
trees area
Avg
dbh
2188
loblolly pine
534
12 94
17.6
2191
thortlcsf pine
30
0.62
16 3
red maple
104
0 61
8 7
loblolly pine
1067
22 67
16 4
avecigum
208
2 98
13 5
hickory
15
0.13
10 7
lowland blackgum
74
060
10 1
uweetgum
163
I 93
123
white oak
30
1 04
21 2
lowland blsckgum
15
0 11
9 7
soothem red oak
25
0 65
23 6
uourwood
15
007
7 6
wat er oak
15
0 07
7 9
white oak
15
1 35
34 0
willow oak
74
8 04
37 1
black o .k
15
0 12
102
bl ackosk
15
023
147
2189
eastern redcedar
15
0 09
8 6
2192
eastern rcdcedar
lobloIly pane
15
830
0 17
32 74
I I 9
22 4
ahofli cjf p ane
89
6 54
30 6
red maple
30
0 16
8 3
loblolly pine
593
13 34
16 9
swee tgum
297
3 00
11 3
red maple
15
0 24
14 5
yellow-poplar
341
6 19
15 2
hickory
30
0 15
8 1
sourwood
30
0 22
9 7
yellow-poplar
104
1 22
12 3
southern red oak
15
0 13
104
b igtoothsspcn
148
268
152
white oak
193
2 67
13 3
2303
loblolly pine
237
16 30
29 6
scarlet oak
44
0 59
12 9
Virgiri a pine
89
5 25
27 4
black oak
15
0 24
14 5
flowenng dogwood
sweetgum
15
148
0 08
2 26
8 4
13 9
2190
loblofly pine
148
8 90
27 6
yellow-poplar
163
1 81
ii 9
red maple
15
0 58
22 4
southern red oak
59
0 36
8 8
bluebeech
119
176
137
black oak
15
007
76
hickory
15
010
91
sassafras
30
027
107
tlowerang dogwood
25
0 ii
9 7
American holly
30
0 18
8 9
sweelgum
148
4 12
18 8
yc llow-pop lar
208
4 26
16 2
lowland blackgum
44
0 48
11 8
whit.eoak
59
045
98
southern red oak
133
2 62
15 8
northernr e4o ak
15
041
188
elm
15
017
122
104
-------
Table A-3. MEANS AND STANDAP.D DEVIATIONS OF ELEMENTAL CONCENTRATIONS IN SUGAR MAPLE FOLLAR TISSUE
P1cc Subplot Mean SD Plot Subplot Mean SD
AJun i num Sshccr i
122 1 51 6 122 1 1898 283
2 45 9 2 1915 373
3 56 4 3 1805 134
139 1 65 12 159 1 2828 126
3 73 1 3 2504 503
4 56 4 4 4459 955
174 1 70 14 174 I 2598 735
2 72 8 2 2588 318
3 66 9 3 1830 697
267 I 59 6 267 1 3057 1332
2 63 9 2 3025 757
3 53 24 3 2831 717
270 3 41 10 270 3 444 35
321 1 26 8 321 1 381 56
2 26 7 2 347 83
4 35 0 4 341 25
Carbon MervLLry
122 1 46 8 1 04 122 1 n d n d
2 473 058 2 nd nd
3 465 024 3 nd nd
159 1 46 0 028 159 I n d n d
3 466 015 3 r id
4 464 064 4 4 7
174 1 462 066 174 1 5 8
2 460 020 2 rid rid
3 460 043 3 6 5
267 1 43 8 0 48 267 1 n 4 ii d
2 456 025 2 rid rid
3 464 037 3 33 43
270 3 47 5 0 45 270 3 n 4 ri 4
321 1 46 8 064 321 I rid rid
2 47.5 043 2 rid rid
4 471 038 4 rid rid
Nitrogen Zinc
122 1 1.57 0 074 122 1 23 1 4
2 147 0078 2 21 14
3 169 0163 3 36 85
159 1 1 33 0 237 159 1 21 4 9
3 160 0212 3 29 35
4 145 0081 4 28 71
174 I 210 000.4 174 1 32 49
2 188 0000 2 31 49
3 129 0.035 3 33 35
267 1 1.51 0 106 267 1 30 1 4
2 179 0067 2 36 163
3 199 0159 3 2.5 21
270 3 1 42 0 114 270 3 24 2 6
321 1 209 0050 321 1 24 113
2 221 0110 2 30 28
4 217 0007 4 24 14
105
-------
Table A-3 (CONTINUED)
Ploc Subplo4 Mean SD Plot Subplot Mean SD
Pboaphorua Mangancac
122 1 1392 458 122 1 1134 150
2 1074 113 2 898 315
3 1375 84 3 1379 242
139 1 820 76 159 1 526 62
3 899 78 3 531 115
4 888 66 4 636 26
174 1 1380 245 174 1 813 198
2 1104 88 2 491 75
3 1227 91 3 1158 84
267 1 1586 330 267 I 812 66
2 1938 781 2 1468 1039
3 1322 94 3 317 41
270 3 1142 285 270 3 1401 681
321 I 1311 396 321 I 1691 687
2 1172 52 2 1526 369
4 1101 8 4 930 25
Iron Magr esium
822 1 49 1 122 995 21
2 52 8 2 905 6
3 61 4 3 1381 365
159 1 53 4 139 1 1491 194
3 70 6 3 1275 250
4 54 1 4 1064 199
174 1 85 13 174 I 1752 383
2 74 6 2 1434 231
3 77 15 3 1927 36
267 1 49 6 267 1 1331 477
2 65 7 2 1742 180
3 54 3 3 1984 26
270 3 74 2 270 3 1244 132
321 1 73 33 321 1 1428 415
2 69 9 2 1251 429
4 80 12 4 1535 251
Copper Sodium
122 1 8 2.1 122 1 70 47
2 5 00 2 282 14
3 10 4.2 3 184 25
159 8 8 1 4 159 1 164 65
3 8 07 3 328 64
4 6 00 4 167 139
874 1 8 14 174 1 200 49
2 9 07 2 228 264
3 8 21 3 204 173
267 1 8 2 1 267 1 265 113
2 9 28 2 154 6
3 12 07 3 141 68
270 3 7 24 270 3 69 63
321 I 10 2 8 321 1 7 159
2 10 64 2 66 44
4 8 00 4 48 55
106
-------
Table A-3 (CONTINUED)
Plot Subplot Mean SI) Plot Subplot Mean SD
co6ak Titanium
122 I 1 I 4 122 1 14 S 3 5355
2 3 28 2 Dd
3 4 3.5 3 nd nd
159 1 2 07 159 1 195 07071
3 4 07 3 18 42426
4 2 07 4 205 07071
174 1 4 0 0 174 1 26 5 0 7071
2 3 07 2 235 106066
3 4 0.7 3 28 28284
267 1 2 00 267 1 21 5 2 1213
2 3 07 2 265 21213
3 3 15 42426
270 3 4 05 270 3 nd nd
321 I 4 2.8 321 1 nd nd
2 4 0.7 2 nd nd
4 4 14 4 nd nd
Nickc l Chrorruum
122 1 3 07 122 1 11 42
2 5 28 2 8 14
3 5 28 3 10 07
159 1 3 0.7 159 1 11 1 4
3 3 00 3 16 07
4 3 07 4 14 14
174 1 5 0.0 374 1 16 2 8
2 4 00 2 15 00
3 7 00 3 14 00
267 1 3 0.0 267 1 14 07
2 4 1.4 2 16 35
3 2 14 3 12 07
270 3 S 1.7 270 3 2 05
321 1 7 42 321 I 6 35
2 6 07 2 8 64
4 5 14 4 3 07
P o t auium Lead
122 1 7705 757 122 1 28 3
2 8490 515 2 27 7
3 9606 872 3 22 6
159 1 8445 243 159 1 21 0
3 8887 328 3 31 14
4 9441 2.351 4 32 15
174 1 8362 2193 174 1 41 0
2 7372 112 2 41 0
3 5948 1241 3 31 14
267 1 10303 2199 267 1 21 0
2 10245 270 2 31 14
3 10703 2244 3 31 14
270 3 5751 909 270 3 14 5
321 1 9060 677 321 1
2 7983 1387 2 18 15
4 8266 1379 4 11 3
107
-------
Table A-) (CONTD UED)
Plot
Subp!o4 - — Pe.z
- Plot
Subplot Mean
SD
Cadrn um
122
270
321
Sirontiun,
122
159
174
267
270
321
1 39 141
2 32 35
3 41 64
1 40 57
3 38 85
4 30 42
1 90 99
2 82 28
3 78 2.8
I 36 127
2 44 92
3 40 57
3 62 73
1 49 177
2 39 85
4 66 07
1 103 41
2 78 26
3 122 30
I 138 54
3 167 21
4 92 26
1 178 66
2 184 30
3 142 0
1 132 64
2 169 60
3 145 16
3 88 15
1 91 52
2 60 14
4 125 47
159
174
267
1
2
07
2
ad
ad
3
ad
ad,
I
nd
ad
3
2
07
4
2
07
1
1
00
2
2
07
3
2
07
I
ad
ad
2
ad
ad
3
ad
nd
3
ad
ad
1
ad
ad
2
ad
nd
4
ad
ad
I ad
2 n.d.
3 1
1 4
3 6
4 5
1 6
2 12
3 8
1 5
2 6
3 11
3 3
1 4
2 3
4 5
ad
ad
14
14
07
07
21
64
49
28
14
10 6
26
07
14
14
Molybdenum
122
159
174
267
270
321
bthium
122
159
174
267
270
321
Banu rn
122
Arscruc
122
159
174
267
770
321
1
21
35
1
1550
7354
2
ad
ad
2
1175
4455
3
ad
ad
3
1665
2616
1
27
14
159
1
131.0
3677
3
21
21
3
1310
3536
4
36
28
4
1425
1061
I
n.d
ad
174
1
1195
1768
2
27
156
2
91.5
919
3
ad
ad
3
975
495
I
ad
ad
267
1
1700
4950
2
32
92
2
2765
14071
3
ad.
ad
3
885
1202
3
a d
a 6
270
3
162 3
52 83
1
18
7.8
321
1
2365
11950
2
a.d
ad
2
233.5
12940
4
19
8.5
4
2205
212
08
-------
Table A-S (CONTINUED)
Plot Subplot Mean SD Plot Subplot Mean SD
Boron Sulfur
122 1 49 0 122 I 023 0021
2 42 2 2 019 0032
3 49 8 3 073 0039
159 I 51 9 159 1 0.2.5 0021
3 57 I 3 022 0042
4 59 3 4 023 0011
174 1 61 I 174 1 023 0004
2 71 13 2 022 0011
3 37 6 3 018 0014
267 1 36 1 267 1 024 0014
2 40 4 2 027 0007
3 30 16 3 027 0039
270 3 41 15 270 3 016 0031
321 1 31 9 321 1 022 0011
2 26 1 2 017 0042
4 20 4 4 014 0018
Cakium Vanadium
122 I 6044 2416 122 1 nd
2 4758 2249 2 nd rid
3 6313 1436 3 rid rid
159 1 11801 2051 159 1 13 2
3 11204 742 3 i5 3
4 8808 1860 4 18 2
174 1 14683 1308 174 1 16 0
2 14623 1221 2 13 4
3 13125 727 3 12 1
267 1 10810 3400 267 1 Ii
2 13634 2439 2 16 1
3 14186 1773 3 nd. rid
270 3 5414 245 270 3 rid rid
321 1 5643 2258 321 I rid rid
2 4871 1459 2 rid rid
4 7614 373 4 13 5
109
-------
Tible A-4. MEANS AND STANDARD DEVIATIONS FOR ELEMENTAL ANALYSES OF LOBL.OLLY PINE FOLIAR SAMPLES
P1o Subplot Mean 1 SD 1 Meanc SD
SD 1
Plot Subplot Mc n 1 SD 1 Meant SD SD ,
I 218 39 244 97
2 309 18 266 49
3 240 60 273 102
1 563 91 565 114
3 437 25 457 5
4 451 74 444 49
1 563 186 556 151
2 412 170 388 155
4 322 58 320 71
1 163 17 157 6
2 273 9 266 2
3 288 2 279 12
1 618 62 627 78
2 514 247 330 298
3 565 122 563 106
1 593 243 603 222
3 609 158 612 170
4 505 23 490 33
1 467 22 474 16
2 567 30 563 4
3 718 266 717 253
1 516 016 520 OIl
2 51.3 072 513 070
3 515 028 522 022
1 506 070 509 084
3 511 049 509 001
4 509 045 510 016
1 509 036 507 048
2 509 072 509 043
4 507 060 504 035
I 493 026 493 060
2 495 079 486 014
3 492 001 494 018
1 510 012 513 046
2 494 055 495 020
3 509 045 511 012
1 51.1 026 508 002
3 507 048 507 047
4 511 049 526 250
1 515 001 516 021
2 516 016 517 016
3 512 087 51.3 119
1 0 97 0 042 0 97 0 050 0 022
2 1 02 0 039 1 00 0.042 0 034
3 0 98 0 083 0 99 0 088 0 046
1 1 31 0 186 1 30 0 145 0 084
3 1 31 0050 1 31 0021 0013
4 1.13 0072 113 0057 0045
1 117 0 050 112 0 035 0 040
2 0 82 0 152 0 81 0 180 0 060
4 110 0 101 116 0 106 0052
1 118 0 175 1 21 0053 0 046
2 1 26 0 348 1 28 0375 0 173
3 I 08 0 012 1 05 0 042 0 081
1 I 00 0 078 1 03 0 092 0 026
2 1 13 0131 111 0113 0045
3 0 86 0 067 0 86 0 021 0025
1 1 08 0037 1 07 0 156 0 085
3 0 93 0 080 0 93 0 127 0 065
4 102 0069 103 0007 0006
I 095 0021 095 0007 0022
2 1 08 0 057 1 06 0 071 0 021
3 097 0113 096 0141 0022
1 380 90 463 168 34
2 556 9 529 98 45
3 427 39 463 144 40
1 915 12 871 15 48
3 1207 138 1210 40 142
4 903 214 960 167 32
1 1747 788 1389 264 398
2 980 19 776 202 108
4 696 241 756 214 232
1 608 5 517 77 41
2 1042 764 1106 667 215
3 973 230 945 622 806
1 1888 23 2023 54 85
2 1210 765 1372 529 528
3 1293 561 1272 457 214
I 990 85 1313 67 106
3 1130 •240 1169 269 63
4 1015 786 1395 846 385
1 878 138 921 7 81
2 823 219 882 227 191
3 1122 65 1102 54 178
Aluminum
1843
3954
1955
1959
2072
2187
2189
Carbon
1843
1954
1955
1959
2072
2187
2189
16
55
34
31
22
11
39
5
22
14
15
51
17
113
12
23
25
20
38
15
55
0 33
0 69
O 69
044
0 OS
0 16
o 53
006
0 63
0 20
O IS
0 78
0 57
o 30
0 78
004
0 27
O 29
040
o 49
O 57
N itrogcn
1843
1954
1955
1959
2072
2187
2189
Silicon
1843
1954
1935
1959
2072
2187
2189
110
-------
Table A-4 (CONTh JUED)
Plol
Subpto
Mean 1
SD 1
Mc.n
SD
SD,
Ploc Subplcx
Mean 1
3D 1
Meanc
SD
S1X
Phoaphorus
Mercury
1843
I
ad
ad
nd
ad.
ad.
3843
1
483
16
360
116
4
2
cd
ad
ad.
ad
ad
2
377
124
467
33
26
3
ad
nd
ad
ad
ad.
3
591
53
587
74
37
1954
1
ad
ad
nd
ad
nd
1954
I
3772
659
1758
704
106
3
ad
ad
nd
ad
ad
3
1927
99
2018
59
90
4
22
39
nd
ad.
ad
4
3807
27
1805
24
61
1955
I
3
7
-2
2
5
1955
1
1137
145
1151
134
79
2
ad
ad
ad
ad
ad
2
711
42
723
63
26
4
6
5
ad
ad
ad
4
891
165
886
168
63
1959
1
a d
a d
a d
a d
a d
1959
1
950
50
953
47
92
2
ad
ad
13
20
ad
2
1084
391
1086
379
109
3
ad
ad
ad
ad
ad.
3
985
69
938
113
133
2072
1
rid,
ad
ad
ad
ad
2072
1
724
36
728
36
2.5
2
ad
ad
7
10
ad
2
905
161
866
144
64
3
ad
ad
ad
ad.
ad
3
742
159
737
132
63
2187
I
ad
ad
ad
ad
ad
2187
1
1343
4
1143
7
83
3
a d
a d
a d
a d
a d
3
929
89
934
75
50
4
8
15
a d
a d
20
4
901
126
872
143
73
2189
1
ad
ad
nd
ad.
ad
2189
1
672
16
681
8
6
2
ad
ad
ad
ad
rid
2
799
94
786
86
13
3
ad
ad
ad
ad
rid
3
667
57
662
46
9
Zinc
Iron
1843
1
30
42
26
07
57
1843
1
69
17
61
18
7
2
30
2 5
26
2 1
3.2
2
67
6
72
8
8
3
24
6 7
22
8 5
4 6
3
67
5
72
6
10
1954
1
27
4 9
29
‘7 8
2 9
1954
1
69
20
68
16
2
3
30
12 0
32
14 1
2 5
3
62
8
63
8
5
4
30
4 9
28
5 7
1 0
4
77
7
76
6
4
1955
1
35
14 8
35
14 1
6 7
1955
1
99
2
96
1
7
2
27
6 4
26
6 4
3 6
2
106
45
106
42
3
4
54
64
49
21
54
4
60
18
60
13
4
1959
1
32
0 7
32
3 5
51
1959
1
42
3
43
2
4
2
47
8 8
46
4 9
5.2
2
49
4
50
8
4
3
63
2 8
56
7 1
19 1
3
43
6
45
3
6
2072
1
34
21
36
1 4
25
2072
1
54
8
58
II
3
2
35
2 5
33
1 4
1 8
2
43
32
45
13
7
3
29
07
27
1 4
3 6
3
68
13
63
35
10
2187
1
37
49
36
2.1
1 4
2181
1
59
4
58
4
7
3
31
00
31
0.7
14
3
59
4
62
8
1
4
29
11
29
0.3
21
4
62
Ii
62
9
8
2189
1
27
46
27
07
63
2189
1
62
2
62
4
3
2
34
3 7
33
3 5
2 9
2
65
5
66
10
4
3
37
6 4
36
4 9
2.2
3
56
4
57
6
3
111
-------
Table A-4 (CONTrNUED)
Ploi Subplo Mean, SD, Mean SD SD , Plol Subplol Mean 1 SD 1 Meinc SD SD
Copper Magnecium
1843 32 28 9 2 1 35 1843 1 625 251 324 118 58
2 7 04 9 07 1 8 2 658 29 689 30 100
3 6 07 6 14 25 3 536 87 541 123 33
1954 1 7 1 8 7 0 7 1 5 1954 1 499 72 489 96 16
3 7 2 1 7 2 8 3 6 3 5!! 38 533 22 62
4 8 1 4 7 1 4 3 0 4 477 92 473 78 14
1955 1 7 0 0 6 0 0 2 2 1955 1 619 254 635 272 2.35
2 6 07 5 00 36 2 630 97 618 91 45
4 17 78 II 07 122 4 648 64 637 34 98
1959 1 30 4 2 32 7 8 7 3 3959 1 561 61 559 35 77
2 39 42 33 320 205 2 602 164 596 171 93
3 67 11 61 78 226 3 695 17 662 51 224
2072 1 S 18 10 00 21 2072 1 672 99 670 95 40
2 32 1 8 13 5 7 2 5 2 593 52 576 56 72
3 21 64 ii 2 1 18 8 3 765 128 779 93 68
2187 1 13 73 7 1 4 80 2187 1 606 100 610 103 13
3 8 11 10 0 7 0 5 3 707 56 703 57 45
4 7 1 8 7 04 2 5 4 648 106 648 89 76
2189 1 10 32 7 00 70 2189 1 509 142 505 127 14
2 10 1 I 8 1 4 3 8 2 643 49 641 38 28
3 9 25 6 1 4 1.1 3 669 178 655 173 86
Manganeac Sodium
1843 I 69 31 62 24 14 1843 I 137 84 68 161 47
2 68 24 67 18 9 2 Ill 76 73 71 55
3 61 22 62 26 10 3 200 49 207 23 115
1954 1 174 9 175 19 6 1954 1 256 51 283 14 66
3 249 53 264 59 40 3 263 50 214 5 41
4 313 130 306 118 52 4 244 40 157 66 60
1955 1 311 167 314 369 91 1955 1 315 86 276 57 110
2 323 264 317 257 34 2 205 8 279 211 90
4 55 27 54 23 7 4 190 51 128 13 104
1959 I 46 8 46 5 11 1959 1 131 13 64 33 85
2 76 19 67 8 9 2 175 51 577 693 98
3 151 5 137 30 32! 3 178 17 381 51 19
2072 1 976 133 940 91 59 2072 1 213 53 309 29 71
2 611 351 614 368 69 2 237 76 188 62 112
3 683 14 717 59 107 3 398 .2.33 236 37 168
2187 1 753 152 757 157 66 2187 1 103 43 190 63 44
3 467 93 465 96 19 3 175 5 244 32 93
4 416 40 430 46 43 4 173 42 363 48 329
2189 1 595 160 597 139 59 2189 1 175 10 116 14 40
2 708 203 691 220 63 2 136 24 202 112 75
3 734 253 722 264 113 3 159 18 265 137 56
112
-------
7eble A-4 (CONTINUED)
Plot Subplot M e a n
SD Meant SD SD
Plot Subploi Mean
SD Mcaac SD SD
Cobafl
Po tann2m
1143
1
2
1.1
3
00
15
1143
1
3553
1401
3288
1183
211
2
4
07
3
00
16
2
3616
977
3613
1084
689
3
3
04
3
14
11
3
4003
339
4209
637
911
1954
1
2
11
nd.
ad
05
1954
1
6010
328
6021
162
460
3
1
00
1
14
07
3
6750
1220
5936
93
1825
4
I
1 4
2
0 7
1 0
4
6152
239
5961
400
102
1953
1
2
04
3
07
05
1955
1
5705
248
5382
244
387
2
2
04
2
00
05
2
4260
3060
4469
1356
48
4
1
00
1
14
1.0
4
3368
1148
5457
427
775
1959
1
I
00
2
07
00
1959
1
6589
952
6495
926
405
2
2
04
2
07
05
2
8179
751
8371
911
834
3
2
07
I
00
00
3
6673
952
7033
260
1897
2072
1
2
0 4
3
1 4
05
2072
1
4575
845
5217
713
409
2
1
0 4
1
2 1
I 5
2
3855
172
5422
2.344
3965
3
2
04
1
00
1.1
3
4580
307
4361
100
732
2187
1
2
I I
1
07
05
2187
1
4597
650
5110
383
340
3
2
04
2
07
05
3
4752
17
4940
15
892
4
1
04
1
03
05
4
5146
515
5028
358
238
2189
1
1
11
2
07
05
2189
1
4407
34.4
4107
617
155
2
1
04
2
07
11
2
5248
472
5621
1024
199
3
1
0 0
1
0 0
0 7
3
4293
687
4629
67
665
‘hcke1
Ti amum
1843
1
3
11
4
07
05
1843
1
ad
ad
ad
ad
ad
2
3
07
3
00
07
2
ad
ad
ad
ad
ad
3
4
00
4
07
07
3
ad
nd
ad
d
ad
1954
1
5
1 4
4
2 1
07
1954
1
16
7 8
n d
a d
73
3
4
14
4
14
2,2
3
12
46
ad
nd
52
4
3
07
3
00
07
4
ad
nd
15
71
ad
1953
1
5
07
5
1 4
07
1955
1
19
1 1
20
3 5
2 5
2
3
00
3
00
07
2
19
21
18
21
25
4
3
11
3
14
11
4
15
04
16
49
11
1959
1
4
07
4
00
07
1959
1
10
21
10
1.4
25
2
4
00
4
00
10
2
15
32
17
92
46
3
4
07
5
07
IC
3
13
21
10
14
35
2072
1
6
II
7
28
11
2072
1
11
25
15
42
110
2
8
25
8
35
05
2
10
60
ad
ad
23
3
4
07
3
00
07
3
14
35
nd
ad
50
2187
1
2
07
3
07
07
2187
1
15
32
14
21
34
3
3
00
3
00
07
3
14
13
14
49
11
4
2
04
2
II
0.5
4
12
07
14
07
36
2189
1
3
11
3
07
03
2189
1
ad
ad
ad
d
nd
2
4
07
5
07
07
2
ad
ad
ad
nd
ad
3
3
04
4
07
05
3
10
21
11
92
81
113
-------
T b1c A-4 (CONTINUED)
Chro nium
8843
Lead
1843
1 6 14 3 14 53
2 3 07 5 42 10
3 5 32 6 35 40
1 12 14 8 00 28
3 10 04 4 07 18
4 20 II 12 14 22
I 33 18 35 07 05
2 13 04 33 23 21
4 12 00 12 34 07
1 13 25 8 3.5 15
2 14 28 14 07 07
3 14 38 12 00 05
1 15 21 17 28 3.5
2 17 28 14 14 67
3 15 25 18 35 76
1 12 07 14 07 51
3 13 04 34 49 11
4 7 42 14 18 65
I 18 04 13 92 25
2 18 04 11 57 I I
3 17 1.1 17 1.4 32
1 11 1 II 3 4
2 II 5 36 5 2
3 8 1 9 1 7
1 27 3 21 2 5
3 27 1 18 1 5
4 26 7 26 4 7
I 23 1 2.3 I 0
2 22 2 22 2 1
4 18 7 23 0 10
1 22 1 30 9 18
2 24 1 31 16 2
3 29 2 2.3 2 8
1 22 0 22 0 0
2 20 3 22 0 4
3 22 1 22 0 1
1 31 1 29 1 13
2 30 2 30 0 2
4 29 I 30 1 4
1 22 0 22 0 0
2 32 14 22 0 0
3 22 0 32 13 1
1 1 0.7
2 bd md.
3 3 0.7
1 2 1.1
3 nd nd
4 1 04
I nd rid
2 rid rid
4 rid nd
1 1 07
2 rid rid
3 3 00
1 6 18
2 11 64
3 8 32
I rid rid
3 rid nd
4 rid nd
1 2 3.2
2 7 85
3 12 74
1 5 21
2 6 II
3 5 11
1 rid rid
3 3 07
4 rid r d
3 4 39
2 4 00
4 rid rid
1 2 12
2 3 11
3 2 04
1 11 18
2 11 25
3 32 25
1 4 .04
3 6 28
4 5 07
1 34 46
2 14 35
3 17 57
I 07 10
rid rid rid
rid. rid 30
1 07 18
1 00 rid
2 14 05
1 00 nd
1 00 rid
I 07 11
rid rid 25
rid rid rid
rid rid 16
6 64 82
8 92 71
rid nd 65
rid rid rid
rid rid rid
rid rid rid
rid rid 55
1! 156 125
6 85 170
4 21 07
7 14 15
5 14 11
rid rid rid
3 21 00
rid nd rid
4 35 25
3 34 00
rid rid rid
3 07 II
rid rid 21
2 00 11
II 14 21
8 85 46
14 00 15
4 07 05
7 21 00
5 08 14
13 57 05
15 99 86
12 00 120
riot Subplot Me.
SD Mciri SD S
Plot Subplot Mcari S D 1 Mcan SD sr ,
1954
1955
1959
2072
2387
2189
C*dmium
1843
‘954
1955
1959
2072
2187
2389
Strontium
1843
1954
1955
1959
2072
2187
2189
1954
1955
1959
2072
2187
2189
114
-------
Tablc A-4 (CONTINUED)
Plot Subplot Mcan
SD Meanc SD SD
Plot Subplot Mcan
SD Meinc SDc SD
Arscnk
Lithium
1843
1
17
8.5
24
17 7
3 5
1843
1
23
4
19
8
3
2
ad.
ad.
ad
nd
nd
2
23
13
27
8
1
3
ad
nd.
nd
nd
ad
3
20
1
27
I
7
1954
1
3
4
ad.
ad.
ad.
nd.
ad
ad.
ad
ad
ad
ad
ad
,id
ad
ad
ad
1954
1
3
4
28
37
28
1
6
6
36
38
30
9
9
6
3
1
3
1955
1
2
4
ad
22
rid
ad
64
rid
ad
ad
nd
ad
ad
ad
rid
3.8
ad
1955
1
2
4
27
21
19
8
4
7
31
28
17
23
11
4
7
2
3
1959
1
2
3
nd.
ad
ad
rid
ad
rid
ad
nd
16
rid
ad
28
ad
ad
rid
1959
1
2
3
18
23
73
2
4
9
20
23
27
4
5
6
4
10
5
2072
3
2
3
rid.
ad.
nd
rid
rid
ad
ad
ad
ad
ad
ad
rid
rid
ad
ad
2072
1
2
3
21
23
33
4
16
4
23
23
35
5
17
13
7
5
15
2187
1
3
4
ad
16
ad.
ad
67
ad
ad
19
19
nd
99
57
ad
11 9
ad.
2187
1
3
4
46
41
39
2
9
0
44
38
26
4
16
27
7
5
13
2189
1
2
3
ad.
ad.
rid.
rid,
ad
ad
ad
ad
rid
ad
ad
ad
ad
ad
ad
2189
1
2
3
44
39
36
8
2
6
34
38
30
5
5
8
12
11
7
Molybdc num
Barium
1843
1
2
3
rid
rid.
nd.
ad
ad
rid
ad
ad
2
ad
nd
14
ad
nd
ad
1843
3
2
3
193
240
158
1379
354
389
145
275
195
778
071
636
304
791
650
1954
1
3
4
n.d.
rid
ad
ad.
ad.
ad
a d
ad
1
ad
ad
28
ad
ad
ad
1954
1
3
4
4 5
85
50
071
283
071
5 0
75
55
1 41
354
212
1 00
071
158
1955
I
2
4
4
2
4
3 8
28
49
1
ad
ad
1 4
ad
ad
3 0
35
2.2
1955
1
2
4
23 8
185
58
2510
424
177
22 5
140
55
23 33
566
212
14 53
158
150
1959
1
2
3
3
7
ad.
1.1
2 1
ad
4
2
ad
49
21
ad
3.2
7.6
ad
1959
1
2
3
80
202 5
193
071
248 19
1096
85
9 0
95
071
1 41
071
255
365 35
1701
2072
1
2
3
a d
2
4
a d.
07
35
5
6
2
7 1
64
34
a d
22
30
2072
3
2
3
67 3
255
3125
18 03
2334
26*7
69 0
270
3305
19.80
2404
2192
15 08
361
828
2187
1
3
4
3
3
5
00
2!
14
5
5
5
2 8
49
44
1 6
30
25
2187
1
3
4
19 8
315
238
.2 48
.1626
1591
21 5
330
123
7 78
1697
240
391
200
2115
2189
1
2
3
2
5
2
1 4
07
18
ad
I
ad
ad
14
ad
1 0
43
21
2189
I
2
3
58 3
340
438
3! 47
990
389
54 0
320
440
25 46
2273
14!
658
530
1383
115
-------
TibI A . .4 (CONTINUED)
P1o Subplot Mc n, SD Mc.n 0 SD SOP,,
Plot Subplot Mean 1 SD, Mcan0 - SD SD ,
1843
1
12
5
10
4
2
3843
I
ad
ad
ad. ad
ad
2
13
3
34
I
8
2
ad.
ad
ad. ad
ad
3
14
3
15
1
3
3
ad
ad.
ad cd
ad
1954
3
34
8
14
6
1
1954
1
ad.
n.d
ad. ad
ad
3
11
1
10
2
1
3
ad
ad
ad. ad
ad
4
9
2
10
2
2
4
ad
ad
ad. ad
ad
3935
1
30
2
12
3
1
1955
1
ad
ad
ad ad
ad
2
14
1
34
1
1
2
103
2 1233
ad ad
223607
4
12
3
19
2
2
4
ad
ad
rid ad
nd
3959
1
23
11
22
4
4
1959
1
ad
ad
ad, ad
ad
2
12
I
13
1
3
2
125
2 8284
ad ad
4 52769
3
13
0
12
2
1
3
ad.
ad
ad ad
ad
2072
1
32
1
12
3
2
2072
I
ad.
ad
ad rid
ad
2
15
0
14
0
2
2
ad.
rid
ad ad
ad
3
12
2
38
6
2
3
ad
rid
ad ad
ad
2387
1
12
1
10
0
2
2187
I
ad
ad
17 28284
ad
3
15
5
15
4
4
3
ad
rid
ad ad
ad
4
11
2
11
1
2
4
ad
ad
ad ad
ad
2189
3
19
31
19
10
4
2189
1
ad
ad
ad rid
ad
2
23
I
21
1
4
2
ad
rid
ad ad
ad
3
20
9
IS
8
7
3
ad
rid
ad ad
ad
Ca1ciun ’
Sulfur
1843
I
956
635
829
528
147
1843
1
0 066
0 019
0 055 0 014
0 021
2
3037
373
3076
230
102
2
ad
ad.
0073 0013
ad
3
946
240
994
337
115
3
0063
0004
0058 0011
0007
1954
1
14.66
359
3437
243
15
1954
1
0116
0016
0)28 0004
0003
3
1948
328
2058
535
162
3
0 330
0 000
0 123 0038
0 005
4
1248
268
1237
397
120
4
0125
0018
0328 0011
0018
t955
1
2058
1245
2086
1312
774
1955
1
0 165
0 004
0.163 0 00.4
0 008
2
1517
583
1474
548
93
2
0.129
0.005
0155 0007
0045
4
1129
170
1150
164
43
4
0148
0032
0143 0018
0041
3959
1
789
33
830
91
192
1959
1
0104
0002
0318 0004
0025
2
3306
48
1057
21
128
2
0.113
0031
0113 0004
0005
3
1085
128
1090
120
242
3
0103
0004
0105 0.000
0038
2072
1
2003
147
2036
153
101
2072
1
0089
0005
0095 0014
0006
2
1630
3009
3657
1049
275
2
0093
0031
0093 0004
0000
3
1765
195
1798
151
94
3
0093
0034
0090 0014
0031
2187
1
2386
303
2390
127
193
2187
1
0.103
0,004
0.103 0004
0010
3
2334
670
21)0
604
54
3
0095
0 004
0098 0004
0004
4
2354
38
2342
17
226
4
0 099
0 005
0 099 0 005
0 009
2189
1
2388
496
2353
499
106
2189
1
0080
0007
0.078 0011
0005
2
2275
297
224.0
355
347
2
0086
0002
0083 0004
0006
3
2620
612
2607
701
427
3
0.079
0003
0085 0007
0003
Mean, arid SD , are the mcan and standard dcv,ai ,on of the
individual foliage samples, Mean 0 and SD 0 are the mean and
andard deviation of the compocited foliage lamplee, tad SD , , is
the w ,thiri ’trec 0*twecn-branch) standard dcveaiion
116
------- |